diff --git "a/Luminia-8B-RP/trainer_state.json" "b/Luminia-8B-RP/trainer_state.json"
--- "a/Luminia-8B-RP/trainer_state.json"
+++ "b/Luminia-8B-RP/trainer_state.json"
@@ -10,39 +10,39 @@
   "log_history": [
     {
       "epoch": 0.0004801920768307323,
-      "grad_norm": 0.7035335898399353,
+      "grad_norm": 0.6976720094680786,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 1.3448,
+      "loss": 1.3449,
       "num_input_tokens_seen": 81920,
       "step": 10
     },
     {
       "epoch": 0.0009603841536614646,
-      "grad_norm": 0.9122879505157471,
+      "grad_norm": 0.9117342233657837,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.3391,
+      "loss": 1.339,
       "num_input_tokens_seen": 163840,
       "step": 20
     },
     {
       "epoch": 0.0014405762304921968,
-      "grad_norm": 0.8231492638587952,
+      "grad_norm": 0.8280390501022339,
       "learning_rate": 1.5e-06,
-      "loss": 1.3258,
+      "loss": 1.3256,
       "num_input_tokens_seen": 245760,
       "step": 30
     },
     {
       "epoch": 0.0019207683073229293,
-      "grad_norm": 0.928112804889679,
+      "grad_norm": 0.9282752871513367,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 1.4027,
+      "loss": 1.4026,
       "num_input_tokens_seen": 327680,
       "step": 40
     },
     {
       "epoch": 0.0024009603841536613,
-      "grad_norm": 0.6168057918548584,
+      "grad_norm": 0.6142542362213135,
       "learning_rate": 2.5e-06,
       "loss": 1.1825,
       "num_input_tokens_seen": 409600,
@@ -50,7 +50,7 @@
     },
     {
       "epoch": 0.0028811524609843936,
-      "grad_norm": 0.9915170073509216,
+      "grad_norm": 3.0495574474334717,
       "learning_rate": 3e-06,
       "loss": 1.0617,
       "num_input_tokens_seen": 491520,
@@ -58,55 +58,55 @@
     },
     {
       "epoch": 0.0033613445378151263,
-      "grad_norm": 1.3197886943817139,
+      "grad_norm": 1.313432216644287,
       "learning_rate": 3.5000000000000004e-06,
-      "loss": 1.2087,
+      "loss": 1.2085,
       "num_input_tokens_seen": 573440,
       "step": 70
     },
     {
       "epoch": 0.0038415366146458585,
-      "grad_norm": 0.9965929985046387,
+      "grad_norm": 0.9920127391815186,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 1.2009,
+      "loss": 1.2014,
       "num_input_tokens_seen": 655360,
       "step": 80
     },
     {
       "epoch": 0.004321728691476591,
-      "grad_norm": 1.3831547498703003,
+      "grad_norm": 1.3916656970977783,
       "learning_rate": 4.5e-06,
-      "loss": 1.2692,
+      "loss": 1.2691,
       "num_input_tokens_seen": 737280,
       "step": 90
     },
     {
       "epoch": 0.004801920768307323,
-      "grad_norm": 1.1937536001205444,
+      "grad_norm": 1.2180395126342773,
       "learning_rate": 5e-06,
-      "loss": 1.187,
+      "loss": 1.1872,
       "num_input_tokens_seen": 819200,
       "step": 100
     },
     {
       "epoch": 0.005282112845138055,
-      "grad_norm": 1.3311412334442139,
+      "grad_norm": 1.3460369110107422,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 1.21,
+      "loss": 1.2103,
       "num_input_tokens_seen": 901120,
       "step": 110
     },
     {
       "epoch": 0.005762304921968787,
-      "grad_norm": 1.3286609649658203,
+      "grad_norm": 1.3270734548568726,
       "learning_rate": 6e-06,
-      "loss": 1.4533,
+      "loss": 1.4538,
       "num_input_tokens_seen": 983040,
       "step": 120
     },
     {
       "epoch": 0.00624249699879952,
-      "grad_norm": 1.8026959896087646,
+      "grad_norm": 1.7999413013458252,
       "learning_rate": 6.5000000000000004e-06,
       "loss": 1.3365,
       "num_input_tokens_seen": 1064960,
@@ -114,287 +114,287 @@
     },
     {
       "epoch": 0.0067226890756302525,
-      "grad_norm": 1.4594871997833252,
+      "grad_norm": 1.4551628828048706,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 1.0839,
+      "loss": 1.0837,
       "num_input_tokens_seen": 1146880,
       "step": 140
     },
     {
       "epoch": 0.007202881152460984,
-      "grad_norm": 1.2625036239624023,
+      "grad_norm": 1.265320062637329,
       "learning_rate": 7.5e-06,
-      "loss": 1.0646,
+      "loss": 1.0649,
       "num_input_tokens_seen": 1228800,
       "step": 150
     },
     {
       "epoch": 0.007683073229291717,
-      "grad_norm": 1.7347521781921387,
+      "grad_norm": 1.7354531288146973,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 1.366,
+      "loss": 1.3666,
       "num_input_tokens_seen": 1310720,
       "step": 160
     },
     {
       "epoch": 0.00816326530612245,
-      "grad_norm": 1.4462337493896484,
+      "grad_norm": 1.464363932609558,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 1.2509,
+      "loss": 1.2506,
       "num_input_tokens_seen": 1392640,
       "step": 170
     },
     {
       "epoch": 0.008643457382953182,
-      "grad_norm": 1.3763086795806885,
+      "grad_norm": 1.3760008811950684,
       "learning_rate": 9e-06,
-      "loss": 1.0235,
+      "loss": 1.0234,
       "num_input_tokens_seen": 1474560,
       "step": 180
     },
     {
       "epoch": 0.009123649459783913,
-      "grad_norm": 1.7035270929336548,
+      "grad_norm": 1.6841614246368408,
       "learning_rate": 9.5e-06,
-      "loss": 0.9992,
+      "loss": 0.9989,
       "num_input_tokens_seen": 1556480,
       "step": 190
     },
     {
       "epoch": 0.009603841536614645,
-      "grad_norm": 1.4041584730148315,
+      "grad_norm": 1.3923676013946533,
       "learning_rate": 1e-05,
-      "loss": 0.8424,
+      "loss": 0.8422,
       "num_input_tokens_seen": 1638400,
       "step": 200
     },
     {
       "epoch": 0.010084033613445379,
-      "grad_norm": 1.4125291109085083,
+      "grad_norm": 1.4108831882476807,
       "learning_rate": 1.05e-05,
-      "loss": 1.0181,
+      "loss": 1.0182,
       "num_input_tokens_seen": 1720320,
       "step": 210
     },
     {
       "epoch": 0.01056422569027611,
-      "grad_norm": 1.3839924335479736,
+      "grad_norm": 1.378661036491394,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.8846,
+      "loss": 0.8848,
       "num_input_tokens_seen": 1802240,
       "step": 220
     },
     {
       "epoch": 0.011044417767106842,
-      "grad_norm": 1.4403977394104004,
+      "grad_norm": 1.4397894144058228,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 1.1465,
+      "loss": 1.1464,
       "num_input_tokens_seen": 1884160,
       "step": 230
     },
     {
       "epoch": 0.011524609843937574,
-      "grad_norm": 1.1681145429611206,
+      "grad_norm": 1.1656988859176636,
       "learning_rate": 1.2e-05,
-      "loss": 1.0085,
+      "loss": 1.0084,
       "num_input_tokens_seen": 1966080,
       "step": 240
     },
     {
       "epoch": 0.012004801920768308,
-      "grad_norm": 2.7823386192321777,
+      "grad_norm": 2.9114668369293213,
       "learning_rate": 1.25e-05,
-      "loss": 1.0097,
+      "loss": 1.0096,
       "num_input_tokens_seen": 2048000,
       "step": 250
     },
     {
       "epoch": 0.01248499399759904,
-      "grad_norm": 1.2464739084243774,
+      "grad_norm": 1.2588117122650146,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.968,
+      "loss": 0.9685,
       "num_input_tokens_seen": 2129920,
       "step": 260
     },
     {
       "epoch": 0.012965186074429771,
-      "grad_norm": 2.1030478477478027,
+      "grad_norm": 2.1231954097747803,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 1.1094,
+      "loss": 1.109,
       "num_input_tokens_seen": 2211840,
       "step": 270
     },
     {
       "epoch": 0.013445378151260505,
-      "grad_norm": 1.006523609161377,
+      "grad_norm": 1.0035645961761475,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 1.0791,
+      "loss": 1.079,
       "num_input_tokens_seen": 2293760,
       "step": 280
     },
     {
       "epoch": 0.013925570228091237,
-      "grad_norm": 2.395143508911133,
+      "grad_norm": 2.3785393238067627,
       "learning_rate": 1.45e-05,
-      "loss": 1.0844,
+      "loss": 1.0842,
       "num_input_tokens_seen": 2375680,
       "step": 290
     },
     {
       "epoch": 0.014405762304921969,
-      "grad_norm": 1.1959385871887207,
+      "grad_norm": 1.2208366394042969,
       "learning_rate": 1.5e-05,
-      "loss": 1.0286,
+      "loss": 1.0287,
       "num_input_tokens_seen": 2457600,
       "step": 300
     },
     {
       "epoch": 0.0148859543817527,
-      "grad_norm": 2.8415608406066895,
+      "grad_norm": 2.929558038711548,
       "learning_rate": 1.55e-05,
-      "loss": 1.1747,
+      "loss": 1.1743,
       "num_input_tokens_seen": 2539520,
       "step": 310
     },
     {
       "epoch": 0.015366146458583434,
-      "grad_norm": 1.2111806869506836,
+      "grad_norm": 1.2139575481414795,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 1.1556,
+      "loss": 1.1558,
       "num_input_tokens_seen": 2621440,
       "step": 320
     },
     {
       "epoch": 0.015846338535414166,
-      "grad_norm": 1.1664150953292847,
+      "grad_norm": 1.1486806869506836,
       "learning_rate": 1.65e-05,
-      "loss": 1.0936,
+      "loss": 1.0934,
       "num_input_tokens_seen": 2703360,
       "step": 330
     },
     {
       "epoch": 0.0163265306122449,
-      "grad_norm": 1.121214747428894,
+      "grad_norm": 1.1211186647415161,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 1.0638,
+      "loss": 1.0637,
       "num_input_tokens_seen": 2785280,
       "step": 340
     },
     {
       "epoch": 0.01680672268907563,
-      "grad_norm": 1.1795384883880615,
+      "grad_norm": 1.207706093788147,
       "learning_rate": 1.75e-05,
-      "loss": 1.171,
+      "loss": 1.1708,
       "num_input_tokens_seen": 2867200,
       "step": 350
     },
     {
       "epoch": 0.017286914765906363,
-      "grad_norm": 1.1597260236740112,
+      "grad_norm": 1.1747910976409912,
       "learning_rate": 1.8e-05,
-      "loss": 0.9427,
+      "loss": 0.9425,
       "num_input_tokens_seen": 2949120,
       "step": 360
     },
     {
       "epoch": 0.017767106842737093,
-      "grad_norm": 0.9535462856292725,
+      "grad_norm": 0.9403015971183777,
       "learning_rate": 1.85e-05,
-      "loss": 0.9715,
+      "loss": 0.9716,
       "num_input_tokens_seen": 3031040,
       "step": 370
     },
     {
       "epoch": 0.018247298919567827,
-      "grad_norm": 0.946639895439148,
+      "grad_norm": 0.9536274075508118,
       "learning_rate": 1.9e-05,
-      "loss": 1.0749,
+      "loss": 1.0748,
       "num_input_tokens_seen": 3112960,
       "step": 380
     },
     {
       "epoch": 0.01872749099639856,
-      "grad_norm": 0.9363052845001221,
+      "grad_norm": 0.9436410069465637,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.9876,
+      "loss": 0.9879,
       "num_input_tokens_seen": 3194880,
       "step": 390
     },
     {
       "epoch": 0.01920768307322929,
-      "grad_norm": 1.0535475015640259,
+      "grad_norm": 1.080519199371338,
       "learning_rate": 2e-05,
-      "loss": 1.105,
+      "loss": 1.1051,
       "num_input_tokens_seen": 3276800,
       "step": 400
     },
     {
       "epoch": 0.019687875150060024,
-      "grad_norm": 2.634197950363159,
+      "grad_norm": 2.6441071033477783,
       "learning_rate": 2.05e-05,
-      "loss": 1.1199,
+      "loss": 1.1197,
       "num_input_tokens_seen": 3358720,
       "step": 410
     },
     {
       "epoch": 0.020168067226890758,
-      "grad_norm": 1.0603679418563843,
+      "grad_norm": 1.0869808197021484,
       "learning_rate": 2.1e-05,
-      "loss": 1.0458,
+      "loss": 1.046,
       "num_input_tokens_seen": 3440640,
       "step": 420
     },
     {
       "epoch": 0.020648259303721488,
-      "grad_norm": 1.0156742334365845,
+      "grad_norm": 1.0044586658477783,
       "learning_rate": 2.15e-05,
-      "loss": 1.023,
+      "loss": 1.0229,
       "num_input_tokens_seen": 3522560,
       "step": 430
     },
     {
       "epoch": 0.02112845138055222,
-      "grad_norm": 0.9424085021018982,
+      "grad_norm": 0.9314221143722534,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 1.1381,
+      "loss": 1.138,
       "num_input_tokens_seen": 3604480,
       "step": 440
     },
     {
       "epoch": 0.021608643457382955,
-      "grad_norm": 0.843125581741333,
+      "grad_norm": 0.846156895160675,
       "learning_rate": 2.25e-05,
-      "loss": 0.9443,
+      "loss": 0.9448,
       "num_input_tokens_seen": 3686400,
       "step": 450
     },
     {
       "epoch": 0.022088835534213685,
-      "grad_norm": 0.7700474262237549,
+      "grad_norm": 0.7679842114448547,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 0.9857,
+      "loss": 0.9858,
       "num_input_tokens_seen": 3768320,
       "step": 460
     },
     {
       "epoch": 0.02256902761104442,
-      "grad_norm": 0.8095147013664246,
+      "grad_norm": 0.8113387227058411,
       "learning_rate": 2.35e-05,
-      "loss": 1.2423,
+      "loss": 1.242,
       "num_input_tokens_seen": 3850240,
       "step": 470
     },
     {
       "epoch": 0.02304921968787515,
-      "grad_norm": 0.7245414853096008,
+      "grad_norm": 0.7278649210929871,
       "learning_rate": 2.4e-05,
-      "loss": 1.0458,
+      "loss": 1.046,
       "num_input_tokens_seen": 3932160,
       "step": 480
     },
     {
       "epoch": 0.023529411764705882,
-      "grad_norm": 0.7768824100494385,
+      "grad_norm": 0.765138566493988,
       "learning_rate": 2.45e-05,
       "loss": 1.0759,
       "num_input_tokens_seen": 4014080,
@@ -402,87 +402,87 @@
     },
     {
       "epoch": 0.024009603841536616,
-      "grad_norm": 0.6364564895629883,
+      "grad_norm": 0.6442419290542603,
       "learning_rate": 2.5e-05,
-      "loss": 1.0799,
+      "loss": 1.0796,
       "num_input_tokens_seen": 4096000,
       "step": 500
     },
     {
       "epoch": 0.024489795918367346,
-      "grad_norm": 0.7478921413421631,
+      "grad_norm": 0.7583214640617371,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 1.0056,
+      "loss": 1.0057,
       "num_input_tokens_seen": 4177920,
       "step": 510
     },
     {
       "epoch": 0.02496998799519808,
-      "grad_norm": 0.5253103375434875,
+      "grad_norm": 0.5175316333770752,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.9158,
+      "loss": 0.9161,
       "num_input_tokens_seen": 4259840,
       "step": 520
     },
     {
       "epoch": 0.025450180072028813,
-      "grad_norm": 0.7591320872306824,
+      "grad_norm": 0.7704029679298401,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 1.0108,
+      "loss": 1.0109,
       "num_input_tokens_seen": 4341760,
       "step": 530
     },
     {
       "epoch": 0.025930372148859543,
-      "grad_norm": 0.7796912789344788,
+      "grad_norm": 0.7691550850868225,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.9604,
+      "loss": 0.9605,
       "num_input_tokens_seen": 4423680,
       "step": 540
     },
     {
       "epoch": 0.026410564225690276,
-      "grad_norm": 0.7369773983955383,
+      "grad_norm": 0.7306321263313293,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 1.0828,
+      "loss": 1.0826,
       "num_input_tokens_seen": 4505600,
       "step": 550
     },
     {
       "epoch": 0.02689075630252101,
-      "grad_norm": 0.7800397276878357,
+      "grad_norm": 0.7934399843215942,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 1.2004,
+      "loss": 1.2013,
       "num_input_tokens_seen": 4587520,
       "step": 560
     },
     {
       "epoch": 0.02737094837935174,
-      "grad_norm": 1.0041433572769165,
+      "grad_norm": 0.9972386360168457,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 1.0362,
+      "loss": 1.0361,
       "num_input_tokens_seen": 4669440,
       "step": 570
     },
     {
       "epoch": 0.027851140456182474,
-      "grad_norm": 0.7719876170158386,
+      "grad_norm": 0.7794674038887024,
       "learning_rate": 2.9e-05,
-      "loss": 1.1366,
+      "loss": 1.1374,
       "num_input_tokens_seen": 4751360,
       "step": 580
     },
     {
       "epoch": 0.028331332533013204,
-      "grad_norm": 0.7241286039352417,
+      "grad_norm": 0.7303441762924194,
       "learning_rate": 2.95e-05,
-      "loss": 1.1366,
+      "loss": 1.1363,
       "num_input_tokens_seen": 4833280,
       "step": 590
     },
     {
       "epoch": 0.028811524609843937,
-      "grad_norm": 1.1306092739105225,
+      "grad_norm": 1.1157978773117065,
       "learning_rate": 3e-05,
       "loss": 0.8576,
       "num_input_tokens_seen": 4915200,
@@ -490,15 +490,15 @@
     },
     {
       "epoch": 0.02929171668667467,
-      "grad_norm": 0.6735727190971375,
+      "grad_norm": 0.6717653274536133,
       "learning_rate": 3.05e-05,
-      "loss": 1.2392,
+      "loss": 1.2395,
       "num_input_tokens_seen": 4997120,
       "step": 610
     },
     {
       "epoch": 0.0297719087635054,
-      "grad_norm": 1.3449808359146118,
+      "grad_norm": 1.3313277959823608,
       "learning_rate": 3.1e-05,
       "loss": 0.9339,
       "num_input_tokens_seen": 5079040,
@@ -506,151 +506,151 @@
     },
     {
       "epoch": 0.030252100840336135,
-      "grad_norm": 0.7179974317550659,
+      "grad_norm": 0.7310436367988586,
       "learning_rate": 3.15e-05,
-      "loss": 0.9704,
+      "loss": 0.9699,
       "num_input_tokens_seen": 5160960,
       "step": 630
     },
     {
       "epoch": 0.030732292917166868,
-      "grad_norm": 0.7083514928817749,
+      "grad_norm": 0.708554208278656,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.9718,
+      "loss": 0.9722,
       "num_input_tokens_seen": 5242880,
       "step": 640
     },
     {
       "epoch": 0.031212484993997598,
-      "grad_norm": 1.1827001571655273,
+      "grad_norm": 1.094826102256775,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.9997,
+      "loss": 0.9991,
       "num_input_tokens_seen": 5324800,
       "step": 650
     },
     {
       "epoch": 0.03169267707082833,
-      "grad_norm": 0.553023099899292,
+      "grad_norm": 0.5501854419708252,
       "learning_rate": 3.3e-05,
-      "loss": 0.8748,
+      "loss": 0.8745,
       "num_input_tokens_seen": 5406720,
       "step": 660
     },
     {
       "epoch": 0.032172869147659065,
-      "grad_norm": 0.7951409816741943,
+      "grad_norm": 0.7503916025161743,
       "learning_rate": 3.35e-05,
-      "loss": 0.8658,
+      "loss": 0.8664,
       "num_input_tokens_seen": 5488640,
       "step": 670
     },
     {
       "epoch": 0.0326530612244898,
-      "grad_norm": 0.7278563976287842,
+      "grad_norm": 0.6931608319282532,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.8799,
+      "loss": 0.88,
       "num_input_tokens_seen": 5570560,
       "step": 680
     },
     {
       "epoch": 0.033133253301320525,
-      "grad_norm": 0.4969690442085266,
+      "grad_norm": 0.48539698123931885,
       "learning_rate": 3.45e-05,
-      "loss": 1.0723,
+      "loss": 1.0718,
       "num_input_tokens_seen": 5652480,
       "step": 690
     },
     {
       "epoch": 0.03361344537815126,
-      "grad_norm": 0.621137797832489,
+      "grad_norm": 0.6209288239479065,
       "learning_rate": 3.5e-05,
-      "loss": 1.0756,
+      "loss": 1.0753,
       "num_input_tokens_seen": 5734400,
       "step": 700
     },
     {
       "epoch": 0.03409363745498199,
-      "grad_norm": 0.6362459063529968,
+      "grad_norm": 0.6421590447425842,
       "learning_rate": 3.55e-05,
-      "loss": 0.967,
+      "loss": 0.9675,
       "num_input_tokens_seen": 5816320,
       "step": 710
     },
     {
       "epoch": 0.034573829531812726,
-      "grad_norm": 0.7109155654907227,
+      "grad_norm": 0.7048439979553223,
       "learning_rate": 3.6e-05,
-      "loss": 0.9767,
+      "loss": 0.9765,
       "num_input_tokens_seen": 5898240,
       "step": 720
     },
     {
       "epoch": 0.03505402160864346,
-      "grad_norm": 0.6260324120521545,
+      "grad_norm": 0.6135404706001282,
       "learning_rate": 3.65e-05,
-      "loss": 0.9352,
+      "loss": 0.9347,
       "num_input_tokens_seen": 5980160,
       "step": 730
     },
     {
       "epoch": 0.035534213685474186,
-      "grad_norm": 0.6420788168907166,
+      "grad_norm": 0.6425515413284302,
       "learning_rate": 3.7e-05,
-      "loss": 0.9214,
+      "loss": 0.9209,
       "num_input_tokens_seen": 6062080,
       "step": 740
     },
     {
       "epoch": 0.03601440576230492,
-      "grad_norm": 0.6354939937591553,
+      "grad_norm": 0.6536908149719238,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 1.0334,
+      "loss": 1.0337,
       "num_input_tokens_seen": 6144000,
       "step": 750
     },
     {
       "epoch": 0.036494597839135653,
-      "grad_norm": 0.5470936298370361,
+      "grad_norm": 0.5051663517951965,
       "learning_rate": 3.8e-05,
-      "loss": 1.0507,
+      "loss": 1.051,
       "num_input_tokens_seen": 6225920,
       "step": 760
     },
     {
       "epoch": 0.03697478991596639,
-      "grad_norm": 0.8862899541854858,
+      "grad_norm": 0.8759781122207642,
       "learning_rate": 3.85e-05,
-      "loss": 1.0,
+      "loss": 0.9995,
       "num_input_tokens_seen": 6307840,
       "step": 770
     },
     {
       "epoch": 0.03745498199279712,
-      "grad_norm": 0.6771734952926636,
+      "grad_norm": 0.7036904096603394,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 1.0441,
+      "loss": 1.0438,
       "num_input_tokens_seen": 6389760,
       "step": 780
     },
     {
       "epoch": 0.037935174069627854,
-      "grad_norm": 0.6661419868469238,
+      "grad_norm": 0.6532811522483826,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 1.0698,
+      "loss": 1.0693,
       "num_input_tokens_seen": 6471680,
       "step": 790
     },
     {
       "epoch": 0.03841536614645858,
-      "grad_norm": 0.6281525492668152,
+      "grad_norm": 0.6256266832351685,
       "learning_rate": 4e-05,
-      "loss": 0.9719,
+      "loss": 0.9723,
       "num_input_tokens_seen": 6553600,
       "step": 800
     },
     {
       "epoch": 0.038895558223289314,
-      "grad_norm": 0.8006066083908081,
+      "grad_norm": 0.8010514974594116,
       "learning_rate": 4.05e-05,
       "loss": 1.1387,
       "num_input_tokens_seen": 6635520,
@@ -658,223 +658,223 @@
     },
     {
       "epoch": 0.03937575030012005,
-      "grad_norm": 0.6581395268440247,
+      "grad_norm": 0.6819067001342773,
       "learning_rate": 4.1e-05,
-      "loss": 1.0304,
+      "loss": 1.0296,
       "num_input_tokens_seen": 6717440,
       "step": 820
     },
     {
       "epoch": 0.03985594237695078,
-      "grad_norm": 0.5860041379928589,
+      "grad_norm": 0.6031268239021301,
       "learning_rate": 4.15e-05,
-      "loss": 0.9735,
+      "loss": 0.9731,
       "num_input_tokens_seen": 6799360,
       "step": 830
     },
     {
       "epoch": 0.040336134453781515,
-      "grad_norm": 0.6659871339797974,
+      "grad_norm": 0.6935213804244995,
       "learning_rate": 4.2e-05,
-      "loss": 0.9434,
+      "loss": 0.9446,
       "num_input_tokens_seen": 6881280,
       "step": 840
     },
     {
       "epoch": 0.04081632653061224,
-      "grad_norm": 0.6158783435821533,
+      "grad_norm": 0.6307564377784729,
       "learning_rate": 4.25e-05,
-      "loss": 0.9919,
+      "loss": 0.9918,
       "num_input_tokens_seen": 6963200,
       "step": 850
     },
     {
       "epoch": 0.041296518607442975,
-      "grad_norm": 1.7883119583129883,
+      "grad_norm": 1.976820468902588,
       "learning_rate": 4.3e-05,
-      "loss": 0.9973,
+      "loss": 0.9964,
       "num_input_tokens_seen": 7045120,
       "step": 860
     },
     {
       "epoch": 0.04177671068427371,
-      "grad_norm": 0.6660032272338867,
+      "grad_norm": 0.6267603635787964,
       "learning_rate": 4.35e-05,
-      "loss": 1.0275,
+      "loss": 1.0334,
       "num_input_tokens_seen": 7127040,
       "step": 870
     },
     {
       "epoch": 0.04225690276110444,
-      "grad_norm": 0.7275106906890869,
+      "grad_norm": 0.717856228351593,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.8301,
+      "loss": 0.8306,
       "num_input_tokens_seen": 7208960,
       "step": 880
     },
     {
       "epoch": 0.042737094837935176,
-      "grad_norm": 0.7131916880607605,
+      "grad_norm": 0.677495002746582,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 1.3488,
+      "loss": 1.3483,
       "num_input_tokens_seen": 7290880,
       "step": 890
     },
     {
       "epoch": 0.04321728691476591,
-      "grad_norm": 0.6134268045425415,
+      "grad_norm": 0.6296849250793457,
       "learning_rate": 4.5e-05,
-      "loss": 1.1191,
+      "loss": 1.1189,
       "num_input_tokens_seen": 7372800,
       "step": 900
     },
     {
       "epoch": 0.043697478991596636,
-      "grad_norm": 0.6196214556694031,
+      "grad_norm": 0.5991678833961487,
       "learning_rate": 4.55e-05,
-      "loss": 1.179,
+      "loss": 1.1789,
       "num_input_tokens_seen": 7454720,
       "step": 910
     },
     {
       "epoch": 0.04417767106842737,
-      "grad_norm": 0.626262903213501,
+      "grad_norm": 0.6427794098854065,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 1.0239,
+      "loss": 1.0259,
       "num_input_tokens_seen": 7536640,
       "step": 920
     },
     {
       "epoch": 0.0446578631452581,
-      "grad_norm": 0.7423630952835083,
+      "grad_norm": 0.7176418900489807,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 1.0599,
+      "loss": 1.0611,
       "num_input_tokens_seen": 7618560,
       "step": 930
     },
     {
       "epoch": 0.04513805522208884,
-      "grad_norm": 0.6767206788063049,
+      "grad_norm": 0.6509612798690796,
       "learning_rate": 4.7e-05,
-      "loss": 0.9898,
+      "loss": 0.9896,
       "num_input_tokens_seen": 7700480,
       "step": 940
     },
     {
       "epoch": 0.04561824729891957,
-      "grad_norm": 0.8346232175827026,
+      "grad_norm": 0.7317907214164734,
       "learning_rate": 4.75e-05,
-      "loss": 0.9133,
+      "loss": 0.9137,
       "num_input_tokens_seen": 7782400,
       "step": 950
     },
     {
       "epoch": 0.0460984393757503,
-      "grad_norm": 0.5462753176689148,
+      "grad_norm": 0.567057728767395,
       "learning_rate": 4.8e-05,
-      "loss": 0.8747,
+      "loss": 0.875,
       "num_input_tokens_seen": 7864320,
       "step": 960
     },
     {
       "epoch": 0.04657863145258103,
-      "grad_norm": 0.6095471978187561,
+      "grad_norm": 0.6136640310287476,
       "learning_rate": 4.85e-05,
-      "loss": 1.0154,
+      "loss": 1.0156,
       "num_input_tokens_seen": 7946240,
       "step": 970
     },
     {
       "epoch": 0.047058823529411764,
-      "grad_norm": 0.555296778678894,
+      "grad_norm": 0.5537669658660889,
       "learning_rate": 4.9e-05,
-      "loss": 0.8588,
+      "loss": 0.8589,
       "num_input_tokens_seen": 8028160,
       "step": 980
     },
     {
       "epoch": 0.0475390156062425,
-      "grad_norm": 0.6341934204101562,
+      "grad_norm": 0.6746324300765991,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.8407,
+      "loss": 0.8402,
       "num_input_tokens_seen": 8110080,
       "step": 990
     },
     {
       "epoch": 0.04801920768307323,
-      "grad_norm": 0.6310316324234009,
+      "grad_norm": 0.5586228370666504,
       "learning_rate": 5e-05,
-      "loss": 0.8889,
+      "loss": 0.8892,
       "num_input_tokens_seen": 8192000,
       "step": 1000
     },
     {
       "epoch": 0.048499399759903965,
-      "grad_norm": 0.6559818983078003,
+      "grad_norm": 0.7242605686187744,
       "learning_rate": 4.9999968610581127e-05,
-      "loss": 1.0226,
+      "loss": 1.0213,
       "num_input_tokens_seen": 8273920,
       "step": 1010
     },
     {
       "epoch": 0.04897959183673469,
-      "grad_norm": 0.6584986448287964,
+      "grad_norm": 0.6588850021362305,
       "learning_rate": 4.9999874442403314e-05,
-      "loss": 1.0976,
+      "loss": 1.0984,
       "num_input_tokens_seen": 8355840,
       "step": 1020
     },
     {
       "epoch": 0.049459783913565425,
-      "grad_norm": 0.9989052414894104,
+      "grad_norm": 1.087486982345581,
       "learning_rate": 4.999971749570305e-05,
-      "loss": 1.1658,
+      "loss": 1.167,
       "num_input_tokens_seen": 8437760,
       "step": 1030
     },
     {
       "epoch": 0.04993997599039616,
-      "grad_norm": 0.6315760612487793,
+      "grad_norm": 0.6291378736495972,
       "learning_rate": 4.999949777087444e-05,
-      "loss": 0.9763,
+      "loss": 0.9767,
       "num_input_tokens_seen": 8519680,
       "step": 1040
     },
     {
       "epoch": 0.05042016806722689,
-      "grad_norm": 0.6868481636047363,
+      "grad_norm": 0.711715579032898,
       "learning_rate": 4.999921526846925e-05,
-      "loss": 0.8407,
+      "loss": 0.8419,
       "num_input_tokens_seen": 8601600,
       "step": 1050
     },
     {
       "epoch": 0.050900360144057626,
-      "grad_norm": 0.6122430562973022,
+      "grad_norm": 0.6777693629264832,
       "learning_rate": 4.9998869989196885e-05,
-      "loss": 0.9688,
+      "loss": 0.9681,
       "num_input_tokens_seen": 8683520,
       "step": 1060
     },
     {
       "epoch": 0.05138055222088835,
-      "grad_norm": 0.6200319528579712,
+      "grad_norm": 0.6924950480461121,
       "learning_rate": 4.99984619339244e-05,
-      "loss": 0.9459,
+      "loss": 0.9462,
       "num_input_tokens_seen": 8765440,
       "step": 1070
     },
     {
       "epoch": 0.051860744297719086,
-      "grad_norm": 0.6406499147415161,
+      "grad_norm": 0.6903403997421265,
       "learning_rate": 4.999799110367648e-05,
-      "loss": 1.0055,
+      "loss": 1.0044,
       "num_input_tokens_seen": 8847360,
       "step": 1080
     },
     {
       "epoch": 0.05234093637454982,
-      "grad_norm": 0.5609921813011169,
+      "grad_norm": 0.5596834421157837,
       "learning_rate": 4.999745749963545e-05,
       "loss": 1.0218,
       "num_input_tokens_seen": 8929280,
@@ -882,79 +882,79 @@
     },
     {
       "epoch": 0.05282112845138055,
-      "grad_norm": 0.6337623596191406,
+      "grad_norm": 0.7687702775001526,
       "learning_rate": 4.999686112314127e-05,
-      "loss": 0.8797,
+      "loss": 0.8808,
       "num_input_tokens_seen": 9011200,
       "step": 1100
     },
     {
       "epoch": 0.053301320528211286,
-      "grad_norm": 0.6383864879608154,
+      "grad_norm": 0.633491575717926,
       "learning_rate": 4.999620197569155e-05,
-      "loss": 1.1579,
+      "loss": 1.1594,
       "num_input_tokens_seen": 9093120,
       "step": 1110
     },
     {
       "epoch": 0.05378151260504202,
-      "grad_norm": 0.575879693031311,
+      "grad_norm": 0.6057960391044617,
       "learning_rate": 4.9995480058941483e-05,
-      "loss": 0.8893,
+      "loss": 0.8905,
       "num_input_tokens_seen": 9175040,
       "step": 1120
     },
     {
       "epoch": 0.05426170468187275,
-      "grad_norm": 0.6337655186653137,
+      "grad_norm": 0.6258108019828796,
       "learning_rate": 4.9994695374703934e-05,
-      "loss": 0.987,
+      "loss": 0.9875,
       "num_input_tokens_seen": 9256960,
       "step": 1130
     },
     {
       "epoch": 0.05474189675870348,
-      "grad_norm": 0.6570917367935181,
+      "grad_norm": 0.6195672154426575,
       "learning_rate": 4.9993847924949364e-05,
-      "loss": 0.8846,
+      "loss": 0.8852,
       "num_input_tokens_seen": 9338880,
       "step": 1140
     },
     {
       "epoch": 0.055222088835534214,
-      "grad_norm": 0.6110508441925049,
+      "grad_norm": 0.5793142914772034,
       "learning_rate": 4.999293771180584e-05,
-      "loss": 0.9093,
+      "loss": 0.91,
       "num_input_tokens_seen": 9420800,
       "step": 1150
     },
     {
       "epoch": 0.05570228091236495,
-      "grad_norm": 0.7141037583351135,
+      "grad_norm": 0.6474392414093018,
       "learning_rate": 4.999196473755905e-05,
-      "loss": 1.0216,
+      "loss": 1.0214,
       "num_input_tokens_seen": 9502720,
       "step": 1160
     },
     {
       "epoch": 0.05618247298919568,
-      "grad_norm": 0.5640223026275635,
+      "grad_norm": 0.5729678273200989,
       "learning_rate": 4.9990929004652287e-05,
-      "loss": 1.0765,
+      "loss": 1.0768,
       "num_input_tokens_seen": 9584640,
       "step": 1170
     },
     {
       "epoch": 0.05666266506602641,
-      "grad_norm": 0.4741804301738739,
+      "grad_norm": 0.4522230923175812,
       "learning_rate": 4.9989830515686434e-05,
-      "loss": 0.9421,
+      "loss": 0.9418,
       "num_input_tokens_seen": 9666560,
       "step": 1180
     },
     {
       "epoch": 0.05714285714285714,
-      "grad_norm": 0.4780801236629486,
+      "grad_norm": 0.4965748190879822,
       "learning_rate": 4.998866927341995e-05,
       "loss": 1.0168,
       "num_input_tokens_seen": 9748480,
@@ -962,303 +962,303 @@
     },
     {
       "epoch": 0.057623049219687875,
-      "grad_norm": 0.6580250859260559,
+      "grad_norm": 0.6100621223449707,
       "learning_rate": 4.9987445280768916e-05,
-      "loss": 0.9398,
+      "loss": 0.9405,
       "num_input_tokens_seen": 9830400,
       "step": 1200
     },
     {
       "epoch": 0.05810324129651861,
-      "grad_norm": 0.571909487247467,
+      "grad_norm": 0.5823714733123779,
       "learning_rate": 4.998615854080695e-05,
-      "loss": 1.0907,
+      "loss": 1.0905,
       "num_input_tokens_seen": 9912320,
       "step": 1210
     },
     {
       "epoch": 0.05858343337334934,
-      "grad_norm": 0.6233389973640442,
+      "grad_norm": 0.6303861141204834,
       "learning_rate": 4.998480905676527e-05,
-      "loss": 0.9021,
+      "loss": 0.9024,
       "num_input_tokens_seen": 9994240,
       "step": 1220
     },
     {
       "epoch": 0.059063625450180075,
-      "grad_norm": 1.2051422595977783,
+      "grad_norm": 1.1989893913269043,
       "learning_rate": 4.998339683203261e-05,
-      "loss": 1.2031,
+      "loss": 1.2022,
       "num_input_tokens_seen": 10076160,
       "step": 1230
     },
     {
       "epoch": 0.0595438175270108,
-      "grad_norm": 0.6003074645996094,
+      "grad_norm": 0.6013513803482056,
       "learning_rate": 4.9981921870155314e-05,
-      "loss": 1.0543,
+      "loss": 1.0539,
       "num_input_tokens_seen": 10158080,
       "step": 1240
     },
     {
       "epoch": 0.060024009603841535,
-      "grad_norm": 0.5767567753791809,
+      "grad_norm": 0.6614658236503601,
       "learning_rate": 4.998038417483721e-05,
-      "loss": 1.0474,
+      "loss": 1.0482,
       "num_input_tokens_seen": 10240000,
       "step": 1250
     },
     {
       "epoch": 0.06050420168067227,
-      "grad_norm": 0.6394206285476685,
+      "grad_norm": 0.7018217444419861,
       "learning_rate": 4.9978783749939705e-05,
-      "loss": 1.0161,
+      "loss": 1.0175,
       "num_input_tokens_seen": 10321920,
       "step": 1260
     },
     {
       "epoch": 0.060984393757503,
-      "grad_norm": 0.6625099778175354,
+      "grad_norm": 0.6370293498039246,
       "learning_rate": 4.997712059948171e-05,
-      "loss": 0.9022,
+      "loss": 0.902,
       "num_input_tokens_seen": 10403840,
       "step": 1270
     },
     {
       "epoch": 0.061464585834333736,
-      "grad_norm": 0.6047447323799133,
+      "grad_norm": 0.5794985890388489,
       "learning_rate": 4.997539472763964e-05,
-      "loss": 0.8682,
+      "loss": 0.8692,
       "num_input_tokens_seen": 10485760,
       "step": 1280
     },
     {
       "epoch": 0.06194477791116446,
-      "grad_norm": 0.6154036521911621,
+      "grad_norm": 0.5841014385223389,
       "learning_rate": 4.9973606138747434e-05,
-      "loss": 1.147,
+      "loss": 1.1472,
       "num_input_tokens_seen": 10567680,
       "step": 1290
     },
     {
       "epoch": 0.062424969987995196,
-      "grad_norm": 0.8171849250793457,
+      "grad_norm": 0.7204465866088867,
       "learning_rate": 4.9971754837296516e-05,
-      "loss": 0.8778,
+      "loss": 0.8769,
       "num_input_tokens_seen": 10649600,
       "step": 1300
     },
     {
       "epoch": 0.06290516206482594,
-      "grad_norm": 0.8324936628341675,
+      "grad_norm": 0.794895350933075,
       "learning_rate": 4.9969840827935776e-05,
-      "loss": 1.249,
+      "loss": 1.2477,
       "num_input_tokens_seen": 10731520,
       "step": 1310
     },
     {
       "epoch": 0.06338535414165666,
-      "grad_norm": 0.5857701897621155,
+      "grad_norm": 0.6169483661651611,
       "learning_rate": 4.99678641154716e-05,
-      "loss": 0.6743,
+      "loss": 0.674,
       "num_input_tokens_seen": 10813440,
       "step": 1320
     },
     {
       "epoch": 0.06386554621848739,
-      "grad_norm": 0.6449385285377502,
+      "grad_norm": 0.6169522404670715,
       "learning_rate": 4.9965824704867806e-05,
-      "loss": 1.0631,
+      "loss": 1.063,
       "num_input_tokens_seen": 10895360,
       "step": 1330
     },
     {
       "epoch": 0.06434573829531813,
-      "grad_norm": 0.5496826767921448,
+      "grad_norm": 0.5324148535728455,
       "learning_rate": 4.996372260124567e-05,
-      "loss": 0.8721,
+      "loss": 0.8696,
       "num_input_tokens_seen": 10977280,
       "step": 1340
     },
     {
       "epoch": 0.06482593037214886,
-      "grad_norm": 0.6244034767150879,
+      "grad_norm": 0.66362065076828,
       "learning_rate": 4.996155780988389e-05,
-      "loss": 0.7503,
+      "loss": 0.7504,
       "num_input_tokens_seen": 11059200,
       "step": 1350
     },
     {
       "epoch": 0.0653061224489796,
-      "grad_norm": 0.5835392475128174,
+      "grad_norm": 0.5489897727966309,
       "learning_rate": 4.9959330336218605e-05,
-      "loss": 0.8589,
+      "loss": 0.8579,
       "num_input_tokens_seen": 11141120,
       "step": 1360
     },
     {
       "epoch": 0.06578631452581032,
-      "grad_norm": 0.5886926054954529,
+      "grad_norm": 0.5780734419822693,
       "learning_rate": 4.995704018584334e-05,
-      "loss": 0.9261,
+      "loss": 0.9259,
       "num_input_tokens_seen": 11223040,
       "step": 1370
     },
     {
       "epoch": 0.06626650660264105,
-      "grad_norm": 0.6225190758705139,
+      "grad_norm": 0.596081018447876,
       "learning_rate": 4.9954687364508996e-05,
-      "loss": 0.7139,
+      "loss": 0.7141,
       "num_input_tokens_seen": 11304960,
       "step": 1380
     },
     {
       "epoch": 0.06674669867947179,
-      "grad_norm": 0.7748022079467773,
+      "grad_norm": 0.6452916264533997,
       "learning_rate": 4.995227187812389e-05,
-      "loss": 0.8474,
+      "loss": 0.8471,
       "num_input_tokens_seen": 11386880,
       "step": 1390
     },
     {
       "epoch": 0.06722689075630252,
-      "grad_norm": 0.625622570514679,
+      "grad_norm": 0.63129723072052,
       "learning_rate": 4.9949793732753656e-05,
-      "loss": 1.0527,
+      "loss": 1.0521,
       "num_input_tokens_seen": 11468800,
       "step": 1400
     },
     {
       "epoch": 0.06770708283313326,
-      "grad_norm": 0.5124455094337463,
+      "grad_norm": 0.48341453075408936,
       "learning_rate": 4.994725293462132e-05,
-      "loss": 0.9405,
+      "loss": 0.9402,
       "num_input_tokens_seen": 11550720,
       "step": 1410
     },
     {
       "epoch": 0.06818727490996399,
-      "grad_norm": 0.5305024981498718,
+      "grad_norm": 0.5027806758880615,
       "learning_rate": 4.994464949010722e-05,
-      "loss": 0.9588,
+      "loss": 0.9594,
       "num_input_tokens_seen": 11632640,
       "step": 1420
     },
     {
       "epoch": 0.06866746698679471,
-      "grad_norm": 0.5525708198547363,
+      "grad_norm": 0.5715320706367493,
       "learning_rate": 4.994198340574898e-05,
-      "loss": 0.9032,
+      "loss": 0.9039,
       "num_input_tokens_seen": 11714560,
       "step": 1430
     },
     {
       "epoch": 0.06914765906362545,
-      "grad_norm": 0.5921324491500854,
+      "grad_norm": 0.5546749234199524,
       "learning_rate": 4.993925468824156e-05,
-      "loss": 0.8397,
+      "loss": 0.8399,
       "num_input_tokens_seen": 11796480,
       "step": 1440
     },
     {
       "epoch": 0.06962785114045618,
-      "grad_norm": 0.5875818729400635,
+      "grad_norm": 0.5861151814460754,
       "learning_rate": 4.99364633444372e-05,
-      "loss": 0.9019,
+      "loss": 0.9013,
       "num_input_tokens_seen": 11878400,
       "step": 1450
     },
     {
       "epoch": 0.07010804321728692,
-      "grad_norm": 0.49930045008659363,
+      "grad_norm": 0.4911801815032959,
       "learning_rate": 4.993360938134537e-05,
-      "loss": 0.999,
+      "loss": 0.9985,
       "num_input_tokens_seen": 11960320,
       "step": 1460
     },
     {
       "epoch": 0.07058823529411765,
-      "grad_norm": 0.5186768770217896,
+      "grad_norm": 0.4994650185108185,
       "learning_rate": 4.993069280613282e-05,
-      "loss": 1.0207,
+      "loss": 1.0197,
       "num_input_tokens_seen": 12042240,
       "step": 1470
     },
     {
       "epoch": 0.07106842737094837,
-      "grad_norm": 0.6106103658676147,
+      "grad_norm": 0.5857521295547485,
       "learning_rate": 4.9927713626123524e-05,
-      "loss": 0.9636,
+      "loss": 0.964,
       "num_input_tokens_seen": 12124160,
       "step": 1480
     },
     {
       "epoch": 0.07154861944777911,
-      "grad_norm": 0.5551833510398865,
+      "grad_norm": 0.522347092628479,
       "learning_rate": 4.992467184879865e-05,
-      "loss": 0.9386,
+      "loss": 0.9385,
       "num_input_tokens_seen": 12206080,
       "step": 1490
     },
     {
       "epoch": 0.07202881152460984,
-      "grad_norm": 0.7018815279006958,
+      "grad_norm": 0.6849465370178223,
       "learning_rate": 4.9921567481796585e-05,
-      "loss": 1.0065,
+      "loss": 1.0057,
       "num_input_tokens_seen": 12288000,
       "step": 1500
     },
     {
       "epoch": 0.07250900360144058,
-      "grad_norm": 0.5908045768737793,
+      "grad_norm": 0.6455889940261841,
       "learning_rate": 4.9918400532912845e-05,
-      "loss": 0.9552,
+      "loss": 0.955,
       "num_input_tokens_seen": 12369920,
       "step": 1510
     },
     {
       "epoch": 0.07298919567827131,
-      "grad_norm": 0.9950196146965027,
+      "grad_norm": 0.8505333065986633,
       "learning_rate": 4.991517101010015e-05,
-      "loss": 0.9035,
+      "loss": 0.9024,
       "num_input_tokens_seen": 12451840,
       "step": 1520
     },
     {
       "epoch": 0.07346938775510205,
-      "grad_norm": 0.5551639199256897,
+      "grad_norm": 0.6128160953521729,
       "learning_rate": 4.9911878921468304e-05,
-      "loss": 1.0423,
+      "loss": 1.0432,
       "num_input_tokens_seen": 12533760,
       "step": 1530
     },
     {
       "epoch": 0.07394957983193277,
-      "grad_norm": 0.5617671012878418,
+      "grad_norm": 0.587476372718811,
       "learning_rate": 4.990852427528427e-05,
-      "loss": 0.9216,
+      "loss": 0.9204,
       "num_input_tokens_seen": 12615680,
       "step": 1540
     },
     {
       "epoch": 0.0744297719087635,
-      "grad_norm": 0.6270405650138855,
+      "grad_norm": 0.6319882273674011,
       "learning_rate": 4.9905107079972064e-05,
-      "loss": 0.9563,
+      "loss": 0.9548,
       "num_input_tokens_seen": 12697600,
       "step": 1550
     },
     {
       "epoch": 0.07490996398559424,
-      "grad_norm": 0.5399815440177917,
+      "grad_norm": 0.5549477934837341,
       "learning_rate": 4.990162734411279e-05,
-      "loss": 0.9665,
+      "loss": 0.9657,
       "num_input_tokens_seen": 12779520,
       "step": 1560
     },
     {
       "epoch": 0.07539015606242497,
-      "grad_norm": 0.55797278881073,
+      "grad_norm": 0.531639575958252,
       "learning_rate": 4.989808507644461e-05,
       "loss": 1.0059,
       "num_input_tokens_seen": 12861440,
@@ -1266,7 +1266,7 @@
     },
     {
       "epoch": 0.07587034813925571,
-      "grad_norm": 0.5673569440841675,
+      "grad_norm": 0.5933917164802551,
       "learning_rate": 4.989448028586269e-05,
       "loss": 0.9245,
       "num_input_tokens_seen": 12943360,
@@ -1274,55 +1274,55 @@
     },
     {
       "epoch": 0.07635054021608643,
-      "grad_norm": 0.5120587348937988,
+      "grad_norm": 0.5222373604774475,
       "learning_rate": 4.989081298141921e-05,
-      "loss": 0.8285,
+      "loss": 0.8293,
       "num_input_tokens_seen": 13025280,
       "step": 1590
     },
     {
       "epoch": 0.07683073229291716,
-      "grad_norm": 0.6352382898330688,
+      "grad_norm": 0.60689777135849,
       "learning_rate": 4.988708317232334e-05,
-      "loss": 0.8067,
+      "loss": 0.8068,
       "num_input_tokens_seen": 13107200,
       "step": 1600
     },
     {
       "epoch": 0.0773109243697479,
-      "grad_norm": 0.6657270789146423,
+      "grad_norm": 0.7311990261077881,
       "learning_rate": 4.988329086794122e-05,
-      "loss": 0.8029,
+      "loss": 0.8037,
       "num_input_tokens_seen": 13189120,
       "step": 1610
     },
     {
       "epoch": 0.07779111644657863,
-      "grad_norm": 0.5678449869155884,
+      "grad_norm": 0.5419260263442993,
       "learning_rate": 4.9879436077795884e-05,
-      "loss": 1.1687,
+      "loss": 1.1678,
       "num_input_tokens_seen": 13271040,
       "step": 1620
     },
     {
       "epoch": 0.07827130852340937,
-      "grad_norm": 0.5785603523254395,
+      "grad_norm": 0.6668097972869873,
       "learning_rate": 4.98755188115673e-05,
-      "loss": 0.9474,
+      "loss": 0.946,
       "num_input_tokens_seen": 13352960,
       "step": 1630
     },
     {
       "epoch": 0.0787515006002401,
-      "grad_norm": 0.5578638315200806,
+      "grad_norm": 0.5447852611541748,
       "learning_rate": 4.9871539079092344e-05,
-      "loss": 0.9421,
+      "loss": 0.9426,
       "num_input_tokens_seen": 13434880,
       "step": 1640
     },
     {
       "epoch": 0.07923169267707082,
-      "grad_norm": 0.9270123243331909,
+      "grad_norm": 0.9077025055885315,
       "learning_rate": 4.9867496890364726e-05,
       "loss": 1.0174,
       "num_input_tokens_seen": 13516800,
@@ -1330,127 +1330,127 @@
     },
     {
       "epoch": 0.07971188475390156,
-      "grad_norm": 0.4947386085987091,
+      "grad_norm": 0.5434586405754089,
       "learning_rate": 4.9863392255535e-05,
-      "loss": 1.1363,
+      "loss": 1.1374,
       "num_input_tokens_seen": 13598720,
       "step": 1660
     },
     {
       "epoch": 0.08019207683073229,
-      "grad_norm": 1.040479063987732,
+      "grad_norm": 1.1956346035003662,
       "learning_rate": 4.985922518491054e-05,
-      "loss": 0.9505,
+      "loss": 0.9507,
       "num_input_tokens_seen": 13680640,
       "step": 1670
     },
     {
       "epoch": 0.08067226890756303,
-      "grad_norm": 0.6103275418281555,
+      "grad_norm": 0.6139404773712158,
       "learning_rate": 4.9854995688955494e-05,
-      "loss": 1.0416,
+      "loss": 1.0419,
       "num_input_tokens_seen": 13762560,
       "step": 1680
     },
     {
       "epoch": 0.08115246098439376,
-      "grad_norm": 0.5951912999153137,
+      "grad_norm": 0.5825724601745605,
       "learning_rate": 4.9850703778290784e-05,
-      "loss": 1.164,
+      "loss": 1.1634,
       "num_input_tokens_seen": 13844480,
       "step": 1690
     },
     {
       "epoch": 0.08163265306122448,
-      "grad_norm": 0.5379506945610046,
+      "grad_norm": 0.5449261665344238,
       "learning_rate": 4.984634946369404e-05,
-      "loss": 1.0884,
+      "loss": 1.0925,
       "num_input_tokens_seen": 13926400,
       "step": 1700
     },
     {
       "epoch": 0.08211284513805522,
-      "grad_norm": 0.6716327667236328,
+      "grad_norm": 0.6387315988540649,
       "learning_rate": 4.984193275609964e-05,
-      "loss": 1.2702,
+      "loss": 1.2723,
       "num_input_tokens_seen": 14008320,
       "step": 1710
     },
     {
       "epoch": 0.08259303721488595,
-      "grad_norm": 0.555696964263916,
+      "grad_norm": 0.5872774720191956,
       "learning_rate": 4.983745366659859e-05,
-      "loss": 0.9906,
+      "loss": 0.9903,
       "num_input_tokens_seen": 14090240,
       "step": 1720
     },
     {
       "epoch": 0.08307322929171669,
-      "grad_norm": 0.5970556735992432,
+      "grad_norm": 0.6030364632606506,
       "learning_rate": 4.983291220643858e-05,
-      "loss": 0.9924,
+      "loss": 0.9931,
       "num_input_tokens_seen": 14172160,
       "step": 1730
     },
     {
       "epoch": 0.08355342136854742,
-      "grad_norm": 0.5664453506469727,
+      "grad_norm": 0.5457348227500916,
       "learning_rate": 4.982830838702392e-05,
-      "loss": 1.0358,
+      "loss": 1.0356,
       "num_input_tokens_seen": 14254080,
       "step": 1740
     },
     {
       "epoch": 0.08403361344537816,
-      "grad_norm": 1.6528098583221436,
+      "grad_norm": 1.6807262897491455,
       "learning_rate": 4.98236422199155e-05,
-      "loss": 0.9054,
+      "loss": 0.9048,
       "num_input_tokens_seen": 14336000,
       "step": 1750
     },
     {
       "epoch": 0.08451380552220888,
-      "grad_norm": 0.5737955570220947,
+      "grad_norm": 0.5328593850135803,
       "learning_rate": 4.9818913716830784e-05,
-      "loss": 0.985,
+      "loss": 0.9876,
       "num_input_tokens_seen": 14417920,
       "step": 1760
     },
     {
       "epoch": 0.08499399759903961,
-      "grad_norm": 0.5684435963630676,
+      "grad_norm": 0.5556053519248962,
       "learning_rate": 4.981412288964377e-05,
-      "loss": 1.0279,
+      "loss": 1.0288,
       "num_input_tokens_seen": 14499840,
       "step": 1770
     },
     {
       "epoch": 0.08547418967587035,
-      "grad_norm": 0.5723880529403687,
+      "grad_norm": 0.5741177797317505,
       "learning_rate": 4.9809269750384956e-05,
-      "loss": 1.01,
+      "loss": 1.0098,
       "num_input_tokens_seen": 14581760,
       "step": 1780
     },
     {
       "epoch": 0.08595438175270108,
-      "grad_norm": 0.5838521122932434,
+      "grad_norm": 0.5520077347755432,
       "learning_rate": 4.980435431124133e-05,
-      "loss": 1.1156,
+      "loss": 1.1153,
       "num_input_tokens_seen": 14663680,
       "step": 1790
     },
     {
       "epoch": 0.08643457382953182,
-      "grad_norm": 0.45137259364128113,
+      "grad_norm": 0.4301939010620117,
       "learning_rate": 4.97993765845563e-05,
-      "loss": 0.9487,
+      "loss": 0.9484,
       "num_input_tokens_seen": 14745600,
       "step": 1800
     },
     {
       "epoch": 0.08691476590636255,
-      "grad_norm": 0.5664239525794983,
+      "grad_norm": 0.5945805311203003,
       "learning_rate": 4.9794336582829714e-05,
       "loss": 0.9613,
       "num_input_tokens_seen": 14827520,
@@ -1458,55 +1458,55 @@
     },
     {
       "epoch": 0.08739495798319327,
-      "grad_norm": 0.7043583989143372,
+      "grad_norm": 0.7423526644706726,
       "learning_rate": 4.9789234318717784e-05,
-      "loss": 1.0248,
+      "loss": 1.0271,
       "num_input_tokens_seen": 14909440,
       "step": 1820
     },
     {
       "epoch": 0.08787515006002401,
-      "grad_norm": 0.5338053703308105,
+      "grad_norm": 0.5542834401130676,
       "learning_rate": 4.978406980503308e-05,
-      "loss": 1.0187,
+      "loss": 1.0182,
       "num_input_tokens_seen": 14991360,
       "step": 1830
     },
     {
       "epoch": 0.08835534213685474,
-      "grad_norm": 0.5779690146446228,
+      "grad_norm": 0.5887225270271301,
       "learning_rate": 4.9778843054744494e-05,
-      "loss": 0.9926,
+      "loss": 0.9922,
       "num_input_tokens_seen": 15073280,
       "step": 1840
     },
     {
       "epoch": 0.08883553421368548,
-      "grad_norm": 0.5544894337654114,
+      "grad_norm": 0.5537407398223877,
       "learning_rate": 4.977355408097719e-05,
-      "loss": 0.9672,
+      "loss": 0.9673,
       "num_input_tokens_seen": 15155200,
       "step": 1850
     },
     {
       "epoch": 0.0893157262905162,
-      "grad_norm": 0.7703062891960144,
+      "grad_norm": 0.709483802318573,
       "learning_rate": 4.9768202897012595e-05,
-      "loss": 0.8497,
+      "loss": 0.8494,
       "num_input_tokens_seen": 15237120,
       "step": 1860
     },
     {
       "epoch": 0.08979591836734693,
-      "grad_norm": 0.5235300660133362,
+      "grad_norm": 0.5220293402671814,
       "learning_rate": 4.9762789516288354e-05,
-      "loss": 1.0258,
+      "loss": 1.0266,
       "num_input_tokens_seen": 15319040,
       "step": 1870
     },
     {
       "epoch": 0.09027611044417767,
-      "grad_norm": 0.5461897253990173,
+      "grad_norm": 0.5254554748535156,
       "learning_rate": 4.97573139523983e-05,
       "loss": 0.9799,
       "num_input_tokens_seen": 15400960,
@@ -1514,271 +1514,271 @@
     },
     {
       "epoch": 0.0907563025210084,
-      "grad_norm": 2.8185510635375977,
+      "grad_norm": 2.654348373413086,
       "learning_rate": 4.9751776219092405e-05,
-      "loss": 1.0616,
+      "loss": 1.0606,
       "num_input_tokens_seen": 15482880,
       "step": 1890
     },
     {
       "epoch": 0.09123649459783914,
-      "grad_norm": 0.5489822030067444,
+      "grad_norm": 0.5544585585594177,
       "learning_rate": 4.9746176330276783e-05,
-      "loss": 0.894,
+      "loss": 0.8944,
       "num_input_tokens_seen": 15564800,
       "step": 1900
     },
     {
       "epoch": 0.09171668667466987,
-      "grad_norm": 0.5419192910194397,
+      "grad_norm": 0.49567174911499023,
       "learning_rate": 4.97405143000136e-05,
-      "loss": 1.069,
+      "loss": 1.0702,
       "num_input_tokens_seen": 15646720,
       "step": 1910
     },
     {
       "epoch": 0.0921968787515006,
-      "grad_norm": 0.5786951184272766,
+      "grad_norm": 0.582963228225708,
       "learning_rate": 4.9734790142521096e-05,
-      "loss": 1.0716,
+      "loss": 1.0714,
       "num_input_tokens_seen": 15728640,
       "step": 1920
     },
     {
       "epoch": 0.09267707082833133,
-      "grad_norm": 0.5193120837211609,
+      "grad_norm": 0.48466384410858154,
       "learning_rate": 4.9729003872173494e-05,
-      "loss": 1.1271,
+      "loss": 1.1282,
       "num_input_tokens_seen": 15810560,
       "step": 1930
     },
     {
       "epoch": 0.09315726290516206,
-      "grad_norm": 0.5859541296958923,
+      "grad_norm": 0.6081234216690063,
       "learning_rate": 4.972315550350102e-05,
-      "loss": 0.9171,
+      "loss": 0.9172,
       "num_input_tokens_seen": 15892480,
       "step": 1940
     },
     {
       "epoch": 0.0936374549819928,
-      "grad_norm": 1.0095491409301758,
+      "grad_norm": 1.0088104009628296,
       "learning_rate": 4.971724505118982e-05,
-      "loss": 0.9881,
+      "loss": 0.9888,
       "num_input_tokens_seen": 15974400,
       "step": 1950
     },
     {
       "epoch": 0.09411764705882353,
-      "grad_norm": 0.5563061237335205,
+      "grad_norm": 0.5905680060386658,
       "learning_rate": 4.971127253008194e-05,
-      "loss": 1.0533,
+      "loss": 1.052,
       "num_input_tokens_seen": 16056320,
       "step": 1960
     },
     {
       "epoch": 0.09459783913565427,
-      "grad_norm": 0.5516453385353088,
+      "grad_norm": 0.5327790975570679,
       "learning_rate": 4.970523795517532e-05,
-      "loss": 1.0697,
+      "loss": 1.0703,
       "num_input_tokens_seen": 16138240,
       "step": 1970
     },
     {
       "epoch": 0.095078031212485,
-      "grad_norm": 0.6930283308029175,
+      "grad_norm": 0.6335835456848145,
       "learning_rate": 4.969914134162368e-05,
-      "loss": 0.8598,
+      "loss": 0.8602,
       "num_input_tokens_seen": 16220160,
       "step": 1980
     },
     {
       "epoch": 0.09555822328931572,
-      "grad_norm": 0.5376786589622498,
+      "grad_norm": 0.5414095520973206,
       "learning_rate": 4.9692982704736566e-05,
-      "loss": 0.9143,
+      "loss": 0.914,
       "num_input_tokens_seen": 16302080,
       "step": 1990
     },
     {
       "epoch": 0.09603841536614646,
-      "grad_norm": 0.849349856376648,
+      "grad_norm": 0.864175021648407,
       "learning_rate": 4.968676205997925e-05,
-      "loss": 0.8093,
+      "loss": 0.8096,
       "num_input_tokens_seen": 16384000,
       "step": 2000
     },
     {
       "epoch": 0.09651860744297719,
-      "grad_norm": 0.560719907283783,
+      "grad_norm": 0.5518527626991272,
       "learning_rate": 4.9680479422972735e-05,
-      "loss": 1.0241,
+      "loss": 1.0233,
       "num_input_tokens_seen": 16465920,
       "step": 2010
     },
     {
       "epoch": 0.09699879951980793,
-      "grad_norm": 0.5490632653236389,
+      "grad_norm": 0.567816972732544,
       "learning_rate": 4.9674134809493686e-05,
-      "loss": 0.9488,
+      "loss": 0.9492,
       "num_input_tokens_seen": 16547840,
       "step": 2020
     },
     {
       "epoch": 0.09747899159663866,
-      "grad_norm": 0.45328429341316223,
+      "grad_norm": 0.4478549063205719,
       "learning_rate": 4.9667728235474396e-05,
-      "loss": 0.9864,
+      "loss": 0.9868,
       "num_input_tokens_seen": 16629760,
       "step": 2030
     },
     {
       "epoch": 0.09795918367346938,
-      "grad_norm": 0.5597791075706482,
+      "grad_norm": 0.5483748316764832,
       "learning_rate": 4.9661259717002764e-05,
-      "loss": 0.8736,
+      "loss": 0.8727,
       "num_input_tokens_seen": 16711680,
       "step": 2040
     },
     {
       "epoch": 0.09843937575030012,
-      "grad_norm": 0.6936686038970947,
+      "grad_norm": 0.6592475175857544,
       "learning_rate": 4.9654729270322234e-05,
-      "loss": 1.0361,
+      "loss": 1.0359,
       "num_input_tokens_seen": 16793600,
       "step": 2050
     },
     {
       "epoch": 0.09891956782713085,
-      "grad_norm": 0.5862544178962708,
+      "grad_norm": 0.6033288240432739,
       "learning_rate": 4.964813691183174e-05,
-      "loss": 0.9958,
+      "loss": 0.9957,
       "num_input_tokens_seen": 16875520,
       "step": 2060
     },
     {
       "epoch": 0.09939975990396159,
-      "grad_norm": 2.3423945903778076,
+      "grad_norm": 2.248500108718872,
       "learning_rate": 4.964148265808573e-05,
-      "loss": 0.8843,
+      "loss": 0.8838,
       "num_input_tokens_seen": 16957440,
       "step": 2070
     },
     {
       "epoch": 0.09987995198079232,
-      "grad_norm": 0.5327481031417847,
+      "grad_norm": 0.5395046472549438,
       "learning_rate": 4.963476652579404e-05,
-      "loss": 0.8698,
+      "loss": 0.8697,
       "num_input_tokens_seen": 17039360,
       "step": 2080
     },
     {
       "epoch": 0.10036014405762304,
-      "grad_norm": 0.5698966979980469,
+      "grad_norm": 0.5363647937774658,
       "learning_rate": 4.962798853182192e-05,
-      "loss": 1.1505,
+      "loss": 1.1534,
       "num_input_tokens_seen": 17121280,
       "step": 2090
     },
     {
       "epoch": 0.10084033613445378,
-      "grad_norm": 1.078123688697815,
+      "grad_norm": 0.936784565448761,
       "learning_rate": 4.9621148693189954e-05,
-      "loss": 1.131,
+      "loss": 1.1309,
       "num_input_tokens_seen": 17203200,
       "step": 2100
     },
     {
       "epoch": 0.10132052821128451,
-      "grad_norm": 0.4983321726322174,
+      "grad_norm": 0.4972730576992035,
       "learning_rate": 4.9614247027074024e-05,
-      "loss": 1.2084,
+      "loss": 1.208,
       "num_input_tokens_seen": 17285120,
       "step": 2110
     },
     {
       "epoch": 0.10180072028811525,
-      "grad_norm": 0.5719578266143799,
+      "grad_norm": 0.5752937197685242,
       "learning_rate": 4.960728355080527e-05,
-      "loss": 0.822,
+      "loss": 0.8207,
       "num_input_tokens_seen": 17367040,
       "step": 2120
     },
     {
       "epoch": 0.10228091236494598,
-      "grad_norm": 0.6987754106521606,
+      "grad_norm": 0.642749547958374,
       "learning_rate": 4.9600258281870046e-05,
-      "loss": 0.9156,
+      "loss": 0.914,
       "num_input_tokens_seen": 17448960,
       "step": 2130
     },
     {
       "epoch": 0.1027611044417767,
-      "grad_norm": 0.9254079461097717,
+      "grad_norm": 0.9153047204017639,
       "learning_rate": 4.959317123790988e-05,
-      "loss": 0.9211,
+      "loss": 0.9198,
       "num_input_tokens_seen": 17530880,
       "step": 2140
     },
     {
       "epoch": 0.10324129651860744,
-      "grad_norm": 0.5085683465003967,
+      "grad_norm": 0.52935391664505,
       "learning_rate": 4.958602243672145e-05,
-      "loss": 0.9979,
+      "loss": 0.9969,
       "num_input_tokens_seen": 17612800,
       "step": 2150
     },
     {
       "epoch": 0.10372148859543817,
-      "grad_norm": 0.6418617963790894,
+      "grad_norm": 0.5792514681816101,
       "learning_rate": 4.9578811896256475e-05,
-      "loss": 0.9711,
+      "loss": 0.9702,
       "num_input_tokens_seen": 17694720,
       "step": 2160
     },
     {
       "epoch": 0.10420168067226891,
-      "grad_norm": 0.5556450486183167,
+      "grad_norm": 0.5800937414169312,
       "learning_rate": 4.957153963462172e-05,
-      "loss": 0.9847,
+      "loss": 0.9843,
       "num_input_tokens_seen": 17776640,
       "step": 2170
     },
     {
       "epoch": 0.10468187274909964,
-      "grad_norm": 0.5734561085700989,
+      "grad_norm": 0.5817973613739014,
       "learning_rate": 4.9564205670078965e-05,
-      "loss": 0.9521,
+      "loss": 0.9514,
       "num_input_tokens_seen": 17858560,
       "step": 2180
     },
     {
       "epoch": 0.10516206482593037,
-      "grad_norm": 0.5195381045341492,
+      "grad_norm": 0.5222535729408264,
       "learning_rate": 4.955681002104492e-05,
-      "loss": 0.9161,
+      "loss": 0.9166,
       "num_input_tokens_seen": 17940480,
       "step": 2190
     },
     {
       "epoch": 0.1056422569027611,
-      "grad_norm": 0.5360829830169678,
+      "grad_norm": 0.5431318879127502,
       "learning_rate": 4.954935270609119e-05,
-      "loss": 0.944,
+      "loss": 0.9447,
       "num_input_tokens_seen": 18022400,
       "step": 2200
     },
     {
       "epoch": 0.10612244897959183,
-      "grad_norm": 0.5251927375793457,
+      "grad_norm": 0.5681139826774597,
       "learning_rate": 4.9541833743944244e-05,
-      "loss": 1.0233,
+      "loss": 1.0235,
       "num_input_tokens_seen": 18104320,
       "step": 2210
     },
     {
       "epoch": 0.10660264105642257,
-      "grad_norm": 0.5330286622047424,
+      "grad_norm": 0.5424164533615112,
       "learning_rate": 4.953425315348534e-05,
       "loss": 0.9329,
       "num_input_tokens_seen": 18186240,
@@ -1786,207 +1786,207 @@
     },
     {
       "epoch": 0.1070828331332533,
-      "grad_norm": 0.5502789616584778,
+      "grad_norm": 0.5445407032966614,
       "learning_rate": 4.952661095375051e-05,
-      "loss": 0.9987,
+      "loss": 1.0003,
       "num_input_tokens_seen": 18268160,
       "step": 2230
     },
     {
       "epoch": 0.10756302521008404,
-      "grad_norm": 0.6187983751296997,
+      "grad_norm": 0.611050009727478,
       "learning_rate": 4.95189071639305e-05,
-      "loss": 0.8519,
+      "loss": 0.852,
       "num_input_tokens_seen": 18350080,
       "step": 2240
     },
     {
       "epoch": 0.10804321728691477,
-      "grad_norm": 0.5929238796234131,
+      "grad_norm": 0.5931539535522461,
       "learning_rate": 4.951114180337069e-05,
-      "loss": 0.8418,
+      "loss": 0.8421,
       "num_input_tokens_seen": 18432000,
       "step": 2250
     },
     {
       "epoch": 0.1085234093637455,
-      "grad_norm": 0.4955889880657196,
+      "grad_norm": 0.4846673309803009,
       "learning_rate": 4.95033148915711e-05,
-      "loss": 0.7989,
+      "loss": 0.7993,
       "num_input_tokens_seen": 18513920,
       "step": 2260
     },
     {
       "epoch": 0.10900360144057623,
-      "grad_norm": 0.5626336336135864,
+      "grad_norm": 0.5406256914138794,
       "learning_rate": 4.949542644818631e-05,
-      "loss": 0.8756,
+      "loss": 0.8758,
       "num_input_tokens_seen": 18595840,
       "step": 2270
     },
     {
       "epoch": 0.10948379351740696,
-      "grad_norm": 1.048539638519287,
+      "grad_norm": 0.9688089489936829,
       "learning_rate": 4.948747649302542e-05,
-      "loss": 1.0697,
+      "loss": 1.0688,
       "num_input_tokens_seen": 18677760,
       "step": 2280
     },
     {
       "epoch": 0.1099639855942377,
-      "grad_norm": 0.49436965584754944,
+      "grad_norm": 0.4601312577724457,
       "learning_rate": 4.947946504605198e-05,
-      "loss": 0.9839,
+      "loss": 0.9846,
       "num_input_tokens_seen": 18759680,
       "step": 2290
     },
     {
       "epoch": 0.11044417767106843,
-      "grad_norm": 0.5616737008094788,
+      "grad_norm": 0.5519710779190063,
       "learning_rate": 4.947139212738395e-05,
-      "loss": 0.9213,
+      "loss": 0.9205,
       "num_input_tokens_seen": 18841600,
       "step": 2300
     },
     {
       "epoch": 0.11092436974789915,
-      "grad_norm": 0.5444278120994568,
+      "grad_norm": 0.5775902271270752,
       "learning_rate": 4.946325775729368e-05,
-      "loss": 1.0463,
+      "loss": 1.0461,
       "num_input_tokens_seen": 18923520,
       "step": 2310
     },
     {
       "epoch": 0.1114045618247299,
-      "grad_norm": 0.521336555480957,
+      "grad_norm": 0.5453388690948486,
       "learning_rate": 4.945506195620784e-05,
-      "loss": 0.9103,
+      "loss": 0.9106,
       "num_input_tokens_seen": 19005440,
       "step": 2320
     },
     {
       "epoch": 0.11188475390156062,
-      "grad_norm": 0.5483266115188599,
+      "grad_norm": 0.5399138927459717,
       "learning_rate": 4.944680474470731e-05,
-      "loss": 0.9041,
+      "loss": 0.9038,
       "num_input_tokens_seen": 19087360,
       "step": 2330
     },
     {
       "epoch": 0.11236494597839136,
-      "grad_norm": 0.5167329907417297,
+      "grad_norm": 0.5245352983474731,
       "learning_rate": 4.943848614352724e-05,
-      "loss": 1.1012,
+      "loss": 1.1003,
       "num_input_tokens_seen": 19169280,
       "step": 2340
     },
     {
       "epoch": 0.11284513805522209,
-      "grad_norm": 0.539591372013092,
+      "grad_norm": 0.5824449062347412,
       "learning_rate": 4.943010617355691e-05,
-      "loss": 0.8855,
+      "loss": 0.8852,
       "num_input_tokens_seen": 19251200,
       "step": 2350
     },
     {
       "epoch": 0.11332533013205282,
-      "grad_norm": 1.1016952991485596,
+      "grad_norm": 1.141217827796936,
       "learning_rate": 4.94216648558397e-05,
-      "loss": 0.9905,
+      "loss": 0.9907,
       "num_input_tokens_seen": 19333120,
       "step": 2360
     },
     {
       "epoch": 0.11380552220888356,
-      "grad_norm": 0.5307420492172241,
+      "grad_norm": 0.5422025322914124,
       "learning_rate": 4.9413162211573075e-05,
-      "loss": 0.8465,
+      "loss": 0.8463,
       "num_input_tokens_seen": 19415040,
       "step": 2370
     },
     {
       "epoch": 0.11428571428571428,
-      "grad_norm": 0.5067051649093628,
+      "grad_norm": 0.5168782472610474,
       "learning_rate": 4.9404598262108456e-05,
-      "loss": 1.3091,
+      "loss": 1.3157,
       "num_input_tokens_seen": 19496960,
       "step": 2380
     },
     {
       "epoch": 0.11476590636254502,
-      "grad_norm": 0.5648155212402344,
+      "grad_norm": 0.5548056364059448,
       "learning_rate": 4.939597302895125e-05,
-      "loss": 1.0505,
+      "loss": 1.0501,
       "num_input_tokens_seen": 19578880,
       "step": 2390
     },
     {
       "epoch": 0.11524609843937575,
-      "grad_norm": 0.5425604581832886,
+      "grad_norm": 0.5792363286018372,
       "learning_rate": 4.938728653376075e-05,
-      "loss": 0.9624,
+      "loss": 0.9628,
       "num_input_tokens_seen": 19660800,
       "step": 2400
     },
     {
       "epoch": 0.11572629051620648,
-      "grad_norm": 0.6649373173713684,
+      "grad_norm": 0.6419684886932373,
       "learning_rate": 4.9378538798350046e-05,
-      "loss": 0.882,
+      "loss": 0.8814,
       "num_input_tokens_seen": 19742720,
       "step": 2410
     },
     {
       "epoch": 0.11620648259303722,
-      "grad_norm": 0.4412856698036194,
+      "grad_norm": 0.39345890283584595,
       "learning_rate": 4.936972984468608e-05,
-      "loss": 0.802,
+      "loss": 0.8009,
       "num_input_tokens_seen": 19824640,
       "step": 2420
     },
     {
       "epoch": 0.11668667466986794,
-      "grad_norm": 0.5264973044395447,
+      "grad_norm": 0.5162023305892944,
       "learning_rate": 4.936085969488947e-05,
-      "loss": 0.8687,
+      "loss": 0.8686,
       "num_input_tokens_seen": 19906560,
       "step": 2430
     },
     {
       "epoch": 0.11716686674669868,
-      "grad_norm": 0.5381867289543152,
+      "grad_norm": 0.5232594013214111,
       "learning_rate": 4.9351928371234525e-05,
-      "loss": 0.7774,
+      "loss": 0.7769,
       "num_input_tokens_seen": 19988480,
       "step": 2440
     },
     {
       "epoch": 0.11764705882352941,
-      "grad_norm": 0.5331231355667114,
+      "grad_norm": 0.5508478283882141,
       "learning_rate": 4.934293589614917e-05,
-      "loss": 0.8777,
+      "loss": 0.8771,
       "num_input_tokens_seen": 20070400,
       "step": 2450
     },
     {
       "epoch": 0.11812725090036015,
-      "grad_norm": 0.5481956601142883,
+      "grad_norm": 0.5470507740974426,
       "learning_rate": 4.93338822922149e-05,
-      "loss": 1.0353,
+      "loss": 1.0355,
       "num_input_tokens_seen": 20152320,
       "step": 2460
     },
     {
       "epoch": 0.11860744297719088,
-      "grad_norm": 0.5417446494102478,
+      "grad_norm": 0.5525845885276794,
       "learning_rate": 4.932476758216669e-05,
-      "loss": 0.9866,
+      "loss": 0.9881,
       "num_input_tokens_seen": 20234240,
       "step": 2470
     },
     {
       "epoch": 0.1190876350540216,
-      "grad_norm": 0.5857323408126831,
+      "grad_norm": 0.5766599178314209,
       "learning_rate": 4.931559178889297e-05,
       "loss": 0.9443,
       "num_input_tokens_seen": 20316160,
@@ -1994,199 +1994,199 @@
     },
     {
       "epoch": 0.11956782713085234,
-      "grad_norm": 0.7107313871383667,
+      "grad_norm": 0.7670407295227051,
       "learning_rate": 4.9306354935435594e-05,
-      "loss": 1.0362,
+      "loss": 1.0363,
       "num_input_tokens_seen": 20398080,
       "step": 2490
     },
     {
       "epoch": 0.12004801920768307,
-      "grad_norm": 0.5205745100975037,
+      "grad_norm": 0.5062960386276245,
       "learning_rate": 4.929705704498969e-05,
-      "loss": 1.0075,
+      "loss": 1.0079,
       "num_input_tokens_seen": 20480000,
       "step": 2500
     },
     {
       "epoch": 0.12052821128451381,
-      "grad_norm": 0.5517631769180298,
+      "grad_norm": 0.5576232671737671,
       "learning_rate": 4.928769814090371e-05,
-      "loss": 0.9074,
+      "loss": 0.9076,
       "num_input_tokens_seen": 20561920,
       "step": 2510
     },
     {
       "epoch": 0.12100840336134454,
-      "grad_norm": 0.5010194778442383,
+      "grad_norm": 0.5333806872367859,
       "learning_rate": 4.927827824667929e-05,
-      "loss": 1.0916,
+      "loss": 1.0913,
       "num_input_tokens_seen": 20643840,
       "step": 2520
     },
     {
       "epoch": 0.12148859543817526,
-      "grad_norm": 0.5098908543586731,
+      "grad_norm": 0.525405764579773,
       "learning_rate": 4.926879738597122e-05,
-      "loss": 1.0548,
+      "loss": 1.0558,
       "num_input_tokens_seen": 20725760,
       "step": 2530
     },
     {
       "epoch": 0.121968787515006,
-      "grad_norm": 0.8620312213897705,
+      "grad_norm": 0.835304319858551,
       "learning_rate": 4.925925558258741e-05,
-      "loss": 1.1967,
+      "loss": 1.1955,
       "num_input_tokens_seen": 20807680,
       "step": 2540
     },
     {
       "epoch": 0.12244897959183673,
-      "grad_norm": 0.5866915583610535,
+      "grad_norm": 0.5637728571891785,
       "learning_rate": 4.924965286048879e-05,
-      "loss": 0.9201,
+      "loss": 0.9205,
       "num_input_tokens_seen": 20889600,
       "step": 2550
     },
     {
       "epoch": 0.12292917166866747,
-      "grad_norm": 1.4908480644226074,
+      "grad_norm": 1.5548111200332642,
       "learning_rate": 4.9239989243789275e-05,
-      "loss": 1.0407,
+      "loss": 1.0418,
       "num_input_tokens_seen": 20971520,
       "step": 2560
     },
     {
       "epoch": 0.1234093637454982,
-      "grad_norm": 0.483049213886261,
+      "grad_norm": 0.5124868750572205,
       "learning_rate": 4.9230264756755685e-05,
-      "loss": 1.0231,
+      "loss": 1.0242,
       "num_input_tokens_seen": 21053440,
       "step": 2570
     },
     {
       "epoch": 0.12388955582232893,
-      "grad_norm": 2.7187938690185547,
+      "grad_norm": 2.6235156059265137,
       "learning_rate": 4.9220479423807694e-05,
-      "loss": 0.9814,
+      "loss": 0.9804,
       "num_input_tokens_seen": 21135360,
       "step": 2580
     },
     {
       "epoch": 0.12436974789915967,
-      "grad_norm": 0.5453813672065735,
+      "grad_norm": 0.5854002237319946,
       "learning_rate": 4.9210633269517776e-05,
-      "loss": 1.0074,
+      "loss": 1.0083,
       "num_input_tokens_seen": 21217280,
       "step": 2590
     },
     {
       "epoch": 0.12484993997599039,
-      "grad_norm": 0.5155650973320007,
+      "grad_norm": 0.5365314483642578,
       "learning_rate": 4.920072631861115e-05,
-      "loss": 0.9252,
+      "loss": 0.9283,
       "num_input_tokens_seen": 21299200,
       "step": 2600
     },
     {
       "epoch": 0.12533013205282112,
-      "grad_norm": 0.510693371295929,
+      "grad_norm": 0.5210400819778442,
       "learning_rate": 4.919075859596567e-05,
-      "loss": 1.1132,
+      "loss": 1.1134,
       "num_input_tokens_seen": 21381120,
       "step": 2610
     },
     {
       "epoch": 0.12581032412965187,
-      "grad_norm": 0.5092229843139648,
+      "grad_norm": 0.5012139678001404,
       "learning_rate": 4.918073012661183e-05,
-      "loss": 1.1313,
+      "loss": 1.1302,
       "num_input_tokens_seen": 21463040,
       "step": 2620
     },
     {
       "epoch": 0.1262905162064826,
-      "grad_norm": 0.5227720737457275,
+      "grad_norm": 0.522561252117157,
       "learning_rate": 4.9170640935732654e-05,
-      "loss": 1.0631,
+      "loss": 1.0637,
       "num_input_tokens_seen": 21544960,
       "step": 2630
     },
     {
       "epoch": 0.12677070828331333,
-      "grad_norm": 0.5651267766952515,
+      "grad_norm": 0.5822430849075317,
       "learning_rate": 4.916049104866365e-05,
-      "loss": 0.8485,
+      "loss": 0.8478,
       "num_input_tokens_seen": 21626880,
       "step": 2640
     },
     {
       "epoch": 0.12725090036014405,
-      "grad_norm": 0.8087652921676636,
+      "grad_norm": 0.8148403763771057,
       "learning_rate": 4.915028049089274e-05,
-      "loss": 0.9922,
+      "loss": 0.9934,
       "num_input_tokens_seen": 21708800,
       "step": 2650
     },
     {
       "epoch": 0.12773109243697478,
-      "grad_norm": 0.5781594514846802,
+      "grad_norm": 0.5749411582946777,
       "learning_rate": 4.914000928806021e-05,
-      "loss": 0.9331,
+      "loss": 0.9332,
       "num_input_tokens_seen": 21790720,
       "step": 2660
     },
     {
       "epoch": 0.12821128451380553,
-      "grad_norm": 0.5171418786048889,
+      "grad_norm": 0.5240984559059143,
       "learning_rate": 4.912967746595861e-05,
-      "loss": 1.0793,
+      "loss": 1.0782,
       "num_input_tokens_seen": 21872640,
       "step": 2670
     },
     {
       "epoch": 0.12869147659063626,
-      "grad_norm": 0.5301877856254578,
+      "grad_norm": 0.5066481828689575,
       "learning_rate": 4.911928505053275e-05,
-      "loss": 1.1306,
+      "loss": 1.1305,
       "num_input_tokens_seen": 21954560,
       "step": 2680
     },
     {
       "epoch": 0.129171668667467,
-      "grad_norm": 0.5109881162643433,
+      "grad_norm": 0.5011911392211914,
       "learning_rate": 4.9108832067879574e-05,
-      "loss": 0.9962,
+      "loss": 0.9965,
       "num_input_tokens_seen": 22036480,
       "step": 2690
     },
     {
       "epoch": 0.12965186074429771,
-      "grad_norm": 0.5722172856330872,
+      "grad_norm": 0.5225823521614075,
       "learning_rate": 4.909831854424812e-05,
-      "loss": 0.8064,
+      "loss": 0.8051,
       "num_input_tokens_seen": 22118400,
       "step": 2700
     },
     {
       "epoch": 0.13013205282112844,
-      "grad_norm": 0.5787602663040161,
+      "grad_norm": 0.5528846979141235,
       "learning_rate": 4.908774450603946e-05,
-      "loss": 0.9561,
+      "loss": 0.9556,
       "num_input_tokens_seen": 22200320,
       "step": 2710
     },
     {
       "epoch": 0.1306122448979592,
-      "grad_norm": 0.5278475284576416,
+      "grad_norm": 0.5039660930633545,
       "learning_rate": 4.907710997980664e-05,
-      "loss": 1.0554,
+      "loss": 1.0562,
       "num_input_tokens_seen": 22282240,
       "step": 2720
     },
     {
       "epoch": 0.13109243697478992,
-      "grad_norm": 0.5114937424659729,
+      "grad_norm": 0.5411632061004639,
       "learning_rate": 4.906641499225457e-05,
       "loss": 0.9203,
       "num_input_tokens_seen": 22364160,
@@ -2194,111 +2194,111 @@
     },
     {
       "epoch": 0.13157262905162065,
-      "grad_norm": 0.5476466417312622,
+      "grad_norm": 0.5488880276679993,
       "learning_rate": 4.905565957024003e-05,
-      "loss": 0.9199,
+      "loss": 0.9196,
       "num_input_tokens_seen": 22446080,
       "step": 2740
     },
     {
       "epoch": 0.13205282112845138,
-      "grad_norm": 0.5202212929725647,
+      "grad_norm": 0.5375452637672424,
       "learning_rate": 4.9044843740771505e-05,
-      "loss": 0.9683,
+      "loss": 0.9674,
       "num_input_tokens_seen": 22528000,
       "step": 2750
     },
     {
       "epoch": 0.1325330132052821,
-      "grad_norm": 0.5196512341499329,
+      "grad_norm": 0.5489015579223633,
       "learning_rate": 4.9033967531009225e-05,
-      "loss": 1.1265,
+      "loss": 1.1255,
       "num_input_tokens_seen": 22609920,
       "step": 2760
     },
     {
       "epoch": 0.13301320528211286,
-      "grad_norm": 0.5623713731765747,
+      "grad_norm": 0.5846374034881592,
       "learning_rate": 4.902303096826502e-05,
-      "loss": 1.1676,
+      "loss": 1.1679,
       "num_input_tokens_seen": 22691840,
       "step": 2770
     },
     {
       "epoch": 0.13349339735894358,
-      "grad_norm": 0.6298018097877502,
+      "grad_norm": 0.5768160223960876,
       "learning_rate": 4.901203408000227e-05,
-      "loss": 0.8947,
+      "loss": 0.8948,
       "num_input_tokens_seen": 22773760,
       "step": 2780
     },
     {
       "epoch": 0.1339735894357743,
-      "grad_norm": 0.6308296322822571,
+      "grad_norm": 0.5849971771240234,
       "learning_rate": 4.9000976893835856e-05,
-      "loss": 1.1018,
+      "loss": 1.1007,
       "num_input_tokens_seen": 22855680,
       "step": 2790
     },
     {
       "epoch": 0.13445378151260504,
-      "grad_norm": 1.6783674955368042,
+      "grad_norm": 1.5311932563781738,
       "learning_rate": 4.898985943753207e-05,
-      "loss": 0.9954,
+      "loss": 0.9984,
       "num_input_tokens_seen": 22937600,
       "step": 2800
     },
     {
       "epoch": 0.13493397358943576,
-      "grad_norm": 0.5339183211326599,
+      "grad_norm": 0.5359712243080139,
       "learning_rate": 4.897868173900854e-05,
-      "loss": 0.7595,
+      "loss": 0.7598,
       "num_input_tokens_seen": 23019520,
       "step": 2810
     },
     {
       "epoch": 0.13541416566626652,
-      "grad_norm": 0.5435370802879333,
+      "grad_norm": 0.5451187491416931,
       "learning_rate": 4.89674438263342e-05,
-      "loss": 1.0,
+      "loss": 1.0037,
       "num_input_tokens_seen": 23101440,
       "step": 2820
     },
     {
       "epoch": 0.13589435774309724,
-      "grad_norm": 0.5289290547370911,
+      "grad_norm": 0.5637598633766174,
       "learning_rate": 4.8956145727729156e-05,
-      "loss": 0.9093,
+      "loss": 0.9097,
       "num_input_tokens_seen": 23183360,
       "step": 2830
     },
     {
       "epoch": 0.13637454981992797,
-      "grad_norm": 0.644011914730072,
+      "grad_norm": 0.7073699235916138,
       "learning_rate": 4.8944787471564686e-05,
-      "loss": 0.8391,
+      "loss": 0.8392,
       "num_input_tokens_seen": 23265280,
       "step": 2840
     },
     {
       "epoch": 0.1368547418967587,
-      "grad_norm": 0.5689689517021179,
+      "grad_norm": 0.558293879032135,
       "learning_rate": 4.89333690863631e-05,
-      "loss": 0.8903,
+      "loss": 0.8909,
       "num_input_tokens_seen": 23347200,
       "step": 2850
     },
     {
       "epoch": 0.13733493397358942,
-      "grad_norm": 0.5509852766990662,
+      "grad_norm": 0.5416772961616516,
       "learning_rate": 4.892189060079773e-05,
-      "loss": 0.9087,
+      "loss": 0.908,
       "num_input_tokens_seen": 23429120,
       "step": 2860
     },
     {
       "epoch": 0.13781512605042018,
-      "grad_norm": 0.6888849139213562,
+      "grad_norm": 0.6885920166969299,
       "learning_rate": 4.8910352043692806e-05,
       "loss": 0.8255,
       "num_input_tokens_seen": 23511040,
@@ -2306,55 +2306,55 @@
     },
     {
       "epoch": 0.1382953181272509,
-      "grad_norm": 0.5926205515861511,
+      "grad_norm": 0.5959227085113525,
       "learning_rate": 4.889875344402342e-05,
-      "loss": 0.9486,
+      "loss": 0.9484,
       "num_input_tokens_seen": 23592960,
       "step": 2880
     },
     {
       "epoch": 0.13877551020408163,
-      "grad_norm": 0.5350954532623291,
+      "grad_norm": 0.5337544083595276,
       "learning_rate": 4.8887094830915427e-05,
-      "loss": 1.0803,
+      "loss": 1.0808,
       "num_input_tokens_seen": 23674880,
       "step": 2890
     },
     {
       "epoch": 0.13925570228091236,
-      "grad_norm": 0.5506545305252075,
+      "grad_norm": 0.5348219871520996,
       "learning_rate": 4.8875376233645396e-05,
-      "loss": 0.9572,
+      "loss": 0.9569,
       "num_input_tokens_seen": 23756800,
       "step": 2900
     },
     {
       "epoch": 0.13973589435774308,
-      "grad_norm": 0.552919328212738,
+      "grad_norm": 0.5439090132713318,
       "learning_rate": 4.886359768164054e-05,
-      "loss": 0.9725,
+      "loss": 0.9723,
       "num_input_tokens_seen": 23838720,
       "step": 2910
     },
     {
       "epoch": 0.14021608643457384,
-      "grad_norm": 0.5363306403160095,
+      "grad_norm": 0.5317606329917908,
       "learning_rate": 4.88517592044786e-05,
-      "loss": 0.9,
+      "loss": 0.9003,
       "num_input_tokens_seen": 23920640,
       "step": 2920
     },
     {
       "epoch": 0.14069627851140457,
-      "grad_norm": 0.5855774879455566,
+      "grad_norm": 0.560670018196106,
       "learning_rate": 4.8839860831887805e-05,
-      "loss": 1.0477,
+      "loss": 1.0471,
       "num_input_tokens_seen": 24002560,
       "step": 2930
     },
     {
       "epoch": 0.1411764705882353,
-      "grad_norm": 0.5440022349357605,
+      "grad_norm": 0.5462415218353271,
       "learning_rate": 4.882790259374681e-05,
       "loss": 0.8685,
       "num_input_tokens_seen": 24084480,
@@ -2362,79 +2362,79 @@
     },
     {
       "epoch": 0.14165666266506602,
-      "grad_norm": 0.5618158578872681,
+      "grad_norm": 0.5607855319976807,
       "learning_rate": 4.881588452008456e-05,
-      "loss": 0.9956,
+      "loss": 0.9955,
       "num_input_tokens_seen": 24166400,
       "step": 2950
     },
     {
       "epoch": 0.14213685474189675,
-      "grad_norm": 0.5429749488830566,
+      "grad_norm": 0.536926805973053,
       "learning_rate": 4.880380664108032e-05,
-      "loss": 1.0358,
+      "loss": 1.0356,
       "num_input_tokens_seen": 24248320,
       "step": 2960
     },
     {
       "epoch": 0.1426170468187275,
-      "grad_norm": 0.5229565501213074,
+      "grad_norm": 0.5230121612548828,
       "learning_rate": 4.879166898706347e-05,
-      "loss": 0.8954,
+      "loss": 0.8951,
       "num_input_tokens_seen": 24330240,
       "step": 2970
     },
     {
       "epoch": 0.14309723889555823,
-      "grad_norm": 0.5274274349212646,
+      "grad_norm": 0.5175470113754272,
       "learning_rate": 4.877947158851352e-05,
-      "loss": 0.9882,
+      "loss": 0.9876,
       "num_input_tokens_seen": 24412160,
       "step": 2980
     },
     {
       "epoch": 0.14357743097238895,
-      "grad_norm": 0.5633429884910583,
+      "grad_norm": 0.5634691119194031,
       "learning_rate": 4.876721447606002e-05,
-      "loss": 0.8396,
+      "loss": 0.8405,
       "num_input_tokens_seen": 24494080,
       "step": 2990
     },
     {
       "epoch": 0.14405762304921968,
-      "grad_norm": 0.5823813676834106,
+      "grad_norm": 0.5656299591064453,
       "learning_rate": 4.875489768048247e-05,
-      "loss": 1.0854,
+      "loss": 1.0846,
       "num_input_tokens_seen": 24576000,
       "step": 3000
     },
     {
       "epoch": 0.14453781512605043,
-      "grad_norm": 0.5569640398025513,
+      "grad_norm": 0.5407316088676453,
       "learning_rate": 4.8742521232710234e-05,
-      "loss": 0.9147,
+      "loss": 0.9156,
       "num_input_tokens_seen": 24657920,
       "step": 3010
     },
     {
       "epoch": 0.14501800720288116,
-      "grad_norm": 0.5742693543434143,
+      "grad_norm": 0.5185397863388062,
       "learning_rate": 4.873008516382245e-05,
-      "loss": 0.956,
+      "loss": 0.9555,
       "num_input_tokens_seen": 24739840,
       "step": 3020
     },
     {
       "epoch": 0.1454981992797119,
-      "grad_norm": 0.5225424766540527,
+      "grad_norm": 0.5350617170333862,
       "learning_rate": 4.871758950504801e-05,
-      "loss": 0.9568,
+      "loss": 0.9566,
       "num_input_tokens_seen": 24821760,
       "step": 3030
     },
     {
       "epoch": 0.14597839135654261,
-      "grad_norm": 0.42165303230285645,
+      "grad_norm": 0.3936846852302551,
       "learning_rate": 4.870503428776544e-05,
       "loss": 0.95,
       "num_input_tokens_seen": 24903680,
@@ -2442,255 +2442,255 @@
     },
     {
       "epoch": 0.14645858343337334,
-      "grad_norm": 0.5314249992370605,
+      "grad_norm": 0.5534759759902954,
       "learning_rate": 4.869241954350281e-05,
-      "loss": 1.0183,
+      "loss": 1.0182,
       "num_input_tokens_seen": 24985600,
       "step": 3050
     },
     {
       "epoch": 0.1469387755102041,
-      "grad_norm": 0.5332600474357605,
+      "grad_norm": 0.5339453220367432,
       "learning_rate": 4.867974530393767e-05,
-      "loss": 0.9581,
+      "loss": 0.9629,
       "num_input_tokens_seen": 25067520,
       "step": 3060
     },
     {
       "epoch": 0.14741896758703482,
-      "grad_norm": 0.5184004306793213,
+      "grad_norm": 0.5030773282051086,
       "learning_rate": 4.8667011600896994e-05,
-      "loss": 0.8628,
+      "loss": 0.8619,
       "num_input_tokens_seen": 25149440,
       "step": 3070
     },
     {
       "epoch": 0.14789915966386555,
-      "grad_norm": 0.524926483631134,
+      "grad_norm": 0.551304042339325,
       "learning_rate": 4.8654218466357064e-05,
-      "loss": 0.8973,
+      "loss": 0.8969,
       "num_input_tokens_seen": 25231360,
       "step": 3080
     },
     {
       "epoch": 0.14837935174069627,
-      "grad_norm": 0.4728389382362366,
+      "grad_norm": 0.5021383762359619,
       "learning_rate": 4.86413659324434e-05,
-      "loss": 0.9561,
+      "loss": 0.9565,
       "num_input_tokens_seen": 25313280,
       "step": 3090
     },
     {
       "epoch": 0.148859543817527,
-      "grad_norm": 0.4952199459075928,
+      "grad_norm": 0.5079641342163086,
       "learning_rate": 4.8628454031430694e-05,
-      "loss": 0.8549,
+      "loss": 0.8552,
       "num_input_tokens_seen": 25395200,
       "step": 3100
     },
     {
       "epoch": 0.14933973589435776,
-      "grad_norm": 0.6020041704177856,
+      "grad_norm": 0.5812205076217651,
       "learning_rate": 4.8615482795742696e-05,
-      "loss": 1.1022,
+      "loss": 1.1013,
       "num_input_tokens_seen": 25477120,
       "step": 3110
     },
     {
       "epoch": 0.14981992797118848,
-      "grad_norm": 1.6901748180389404,
+      "grad_norm": 1.72549307346344,
       "learning_rate": 4.860245225795219e-05,
-      "loss": 1.108,
+      "loss": 1.1029,
       "num_input_tokens_seen": 25559040,
       "step": 3120
     },
     {
       "epoch": 0.1503001200480192,
-      "grad_norm": 0.5366313457489014,
+      "grad_norm": 0.5684882998466492,
       "learning_rate": 4.858936245078084e-05,
-      "loss": 1.1467,
+      "loss": 1.162,
       "num_input_tokens_seen": 25640960,
       "step": 3130
     },
     {
       "epoch": 0.15078031212484994,
-      "grad_norm": 0.5097447633743286,
+      "grad_norm": 0.5000339150428772,
       "learning_rate": 4.857621340709917e-05,
-      "loss": 1.0484,
+      "loss": 1.0493,
       "num_input_tokens_seen": 25722880,
       "step": 3140
     },
     {
       "epoch": 0.15126050420168066,
-      "grad_norm": 0.5351747870445251,
+      "grad_norm": 0.5555292963981628,
       "learning_rate": 4.856300515992646e-05,
-      "loss": 0.8896,
+      "loss": 0.8903,
       "num_input_tokens_seen": 25804800,
       "step": 3150
     },
     {
       "epoch": 0.15174069627851142,
-      "grad_norm": 0.5735185742378235,
+      "grad_norm": 0.5889732837677002,
       "learning_rate": 4.854973774243062e-05,
-      "loss": 0.9816,
+      "loss": 0.9819,
       "num_input_tokens_seen": 25886720,
       "step": 3160
     },
     {
       "epoch": 0.15222088835534214,
-      "grad_norm": 0.5087327361106873,
+      "grad_norm": 0.5141785144805908,
       "learning_rate": 4.8536411187928186e-05,
-      "loss": 0.974,
+      "loss": 0.9732,
       "num_input_tokens_seen": 25968640,
       "step": 3170
     },
     {
       "epoch": 0.15270108043217287,
-      "grad_norm": 0.5174838900566101,
+      "grad_norm": 0.5038670897483826,
       "learning_rate": 4.852302552988418e-05,
-      "loss": 0.9798,
+      "loss": 0.9799,
       "num_input_tokens_seen": 26050560,
       "step": 3180
     },
     {
       "epoch": 0.1531812725090036,
-      "grad_norm": 0.5763252973556519,
+      "grad_norm": 0.5643836855888367,
       "learning_rate": 4.850958080191205e-05,
-      "loss": 1.0298,
+      "loss": 1.0295,
       "num_input_tokens_seen": 26132480,
       "step": 3190
     },
     {
       "epoch": 0.15366146458583432,
-      "grad_norm": 0.565152108669281,
+      "grad_norm": 0.5722659826278687,
       "learning_rate": 4.849607703777356e-05,
-      "loss": 0.9407,
+      "loss": 0.942,
       "num_input_tokens_seen": 26214400,
       "step": 3200
     },
     {
       "epoch": 0.15414165666266508,
-      "grad_norm": 0.5854922533035278,
+      "grad_norm": 0.5935901403427124,
       "learning_rate": 4.8482514271378745e-05,
-      "loss": 0.7755,
+      "loss": 0.7757,
       "num_input_tokens_seen": 26296320,
       "step": 3210
     },
     {
       "epoch": 0.1546218487394958,
-      "grad_norm": 0.5151272416114807,
+      "grad_norm": 0.5209060311317444,
       "learning_rate": 4.846889253678578e-05,
-      "loss": 0.8748,
+      "loss": 0.8739,
       "num_input_tokens_seen": 26378240,
       "step": 3220
     },
     {
       "epoch": 0.15510204081632653,
-      "grad_norm": 0.5427525043487549,
+      "grad_norm": 0.5480584502220154,
       "learning_rate": 4.845521186820096e-05,
-      "loss": 0.9308,
+      "loss": 0.931,
       "num_input_tokens_seen": 26460160,
       "step": 3230
     },
     {
       "epoch": 0.15558223289315726,
-      "grad_norm": 0.5123739242553711,
+      "grad_norm": 0.524978518486023,
       "learning_rate": 4.8441472299978504e-05,
-      "loss": 0.9556,
+      "loss": 0.955,
       "num_input_tokens_seen": 26542080,
       "step": 3240
     },
     {
       "epoch": 0.15606242496998798,
-      "grad_norm": 0.6060318946838379,
+      "grad_norm": 0.5937045812606812,
       "learning_rate": 4.8427673866620615e-05,
-      "loss": 0.9465,
+      "loss": 0.9456,
       "num_input_tokens_seen": 26624000,
       "step": 3250
     },
     {
       "epoch": 0.15654261704681874,
-      "grad_norm": 0.5624286532402039,
+      "grad_norm": 0.5852760076522827,
       "learning_rate": 4.841381660277725e-05,
-      "loss": 0.8611,
+      "loss": 0.8613,
       "num_input_tokens_seen": 26705920,
       "step": 3260
     },
     {
       "epoch": 0.15702280912364946,
-      "grad_norm": 0.5059430003166199,
+      "grad_norm": 0.5103029012680054,
       "learning_rate": 4.839990054324614e-05,
-      "loss": 0.9294,
+      "loss": 0.9297,
       "num_input_tokens_seen": 26787840,
       "step": 3270
     },
     {
       "epoch": 0.1575030012004802,
-      "grad_norm": 0.5408222675323486,
+      "grad_norm": 0.5465720295906067,
       "learning_rate": 4.838592572297265e-05,
-      "loss": 1.0522,
+      "loss": 1.0529,
       "num_input_tokens_seen": 26869760,
       "step": 3280
     },
     {
       "epoch": 0.15798319327731092,
-      "grad_norm": 0.5456782579421997,
+      "grad_norm": 0.5594531297683716,
       "learning_rate": 4.837189217704968e-05,
-      "loss": 0.9068,
+      "loss": 0.9072,
       "num_input_tokens_seen": 26951680,
       "step": 3290
     },
     {
       "epoch": 0.15846338535414164,
-      "grad_norm": 0.5521805882453918,
+      "grad_norm": 0.5584396719932556,
       "learning_rate": 4.835779994071764e-05,
-      "loss": 0.8502,
+      "loss": 0.8515,
       "num_input_tokens_seen": 27033600,
       "step": 3300
     },
     {
       "epoch": 0.1589435774309724,
-      "grad_norm": 0.525742769241333,
+      "grad_norm": 0.5218715071678162,
       "learning_rate": 4.8343649049364284e-05,
-      "loss": 0.9537,
+      "loss": 0.9548,
       "num_input_tokens_seen": 27115520,
       "step": 3310
     },
     {
       "epoch": 0.15942376950780313,
-      "grad_norm": 0.7998719215393066,
+      "grad_norm": 0.7685995697975159,
       "learning_rate": 4.832943953852468e-05,
-      "loss": 0.9564,
+      "loss": 0.9565,
       "num_input_tokens_seen": 27197440,
       "step": 3320
     },
     {
       "epoch": 0.15990396158463385,
-      "grad_norm": 0.41986241936683655,
+      "grad_norm": 0.42694056034088135,
       "learning_rate": 4.831517144388109e-05,
-      "loss": 0.8895,
+      "loss": 0.8896,
       "num_input_tokens_seen": 27279360,
       "step": 3330
     },
     {
       "epoch": 0.16038415366146458,
-      "grad_norm": 0.5009111166000366,
+      "grad_norm": 0.4931061267852783,
       "learning_rate": 4.830084480126288e-05,
-      "loss": 0.765,
+      "loss": 0.7648,
       "num_input_tokens_seen": 27361280,
       "step": 3340
     },
     {
       "epoch": 0.1608643457382953,
-      "grad_norm": 0.9933325052261353,
+      "grad_norm": 1.0382879972457886,
       "learning_rate": 4.828645964664647e-05,
-      "loss": 0.9538,
+      "loss": 0.9537,
       "num_input_tokens_seen": 27443200,
       "step": 3350
     },
     {
       "epoch": 0.16134453781512606,
-      "grad_norm": 0.6056552529335022,
+      "grad_norm": 1.5314464569091797,
       "learning_rate": 4.8272016016155166e-05,
       "loss": 1.0036,
       "num_input_tokens_seen": 27525120,
@@ -2698,431 +2698,431 @@
     },
     {
       "epoch": 0.1618247298919568,
-      "grad_norm": 0.6024153828620911,
+      "grad_norm": 0.61847323179245,
       "learning_rate": 4.825751394605916e-05,
-      "loss": 1.1447,
+      "loss": 1.1458,
       "num_input_tokens_seen": 27607040,
       "step": 3370
     },
     {
       "epoch": 0.1623049219687875,
-      "grad_norm": 0.5308077335357666,
+      "grad_norm": 0.5093514323234558,
       "learning_rate": 4.824295347277537e-05,
-      "loss": 0.8392,
+      "loss": 0.8404,
       "num_input_tokens_seen": 27688960,
       "step": 3380
     },
     {
       "epoch": 0.16278511404561824,
-      "grad_norm": 0.5333995223045349,
+      "grad_norm": 0.5439700484275818,
       "learning_rate": 4.8228334632867375e-05,
-      "loss": 1.1125,
+      "loss": 1.1109,
       "num_input_tokens_seen": 27770880,
       "step": 3390
     },
     {
       "epoch": 0.16326530612244897,
-      "grad_norm": 0.5561527013778687,
+      "grad_norm": 0.5555248856544495,
       "learning_rate": 4.8213657463045344e-05,
-      "loss": 1.059,
+      "loss": 1.0581,
       "num_input_tokens_seen": 27852800,
       "step": 3400
     },
     {
       "epoch": 0.16374549819927972,
-      "grad_norm": 0.6268987655639648,
+      "grad_norm": 0.6565396785736084,
       "learning_rate": 4.819892200016588e-05,
-      "loss": 1.0134,
+      "loss": 1.0135,
       "num_input_tokens_seen": 27934720,
       "step": 3410
     },
     {
       "epoch": 0.16422569027611045,
-      "grad_norm": 0.5519012808799744,
+      "grad_norm": 0.5589664578437805,
       "learning_rate": 4.818412828123201e-05,
-      "loss": 0.8942,
+      "loss": 0.8944,
       "num_input_tokens_seen": 28016640,
       "step": 3420
     },
     {
       "epoch": 0.16470588235294117,
-      "grad_norm": 0.5266446471214294,
+      "grad_norm": 0.5469520688056946,
       "learning_rate": 4.816927634339302e-05,
-      "loss": 0.8664,
+      "loss": 0.8662,
       "num_input_tokens_seen": 28098560,
       "step": 3430
     },
     {
       "epoch": 0.1651860744297719,
-      "grad_norm": 0.5828010439872742,
+      "grad_norm": 0.5610776543617249,
       "learning_rate": 4.815436622394441e-05,
-      "loss": 0.7622,
+      "loss": 0.7626,
       "num_input_tokens_seen": 28180480,
       "step": 3440
     },
     {
       "epoch": 0.16566626650660263,
-      "grad_norm": 0.7593029141426086,
+      "grad_norm": 0.7478719353675842,
       "learning_rate": 4.813939796032779e-05,
-      "loss": 1.0249,
+      "loss": 1.0245,
       "num_input_tokens_seen": 28262400,
       "step": 3450
     },
     {
       "epoch": 0.16614645858343338,
-      "grad_norm": 0.5545393228530884,
+      "grad_norm": 0.5465115904808044,
       "learning_rate": 4.812437159013076e-05,
-      "loss": 0.9852,
+      "loss": 0.9836,
       "num_input_tokens_seen": 28344320,
       "step": 3460
     },
     {
       "epoch": 0.1666266506602641,
-      "grad_norm": 0.5645286440849304,
+      "grad_norm": 0.5847524404525757,
       "learning_rate": 4.810928715108683e-05,
-      "loss": 0.8945,
+      "loss": 0.8955,
       "num_input_tokens_seen": 28426240,
       "step": 3470
     },
     {
       "epoch": 0.16710684273709484,
-      "grad_norm": 0.6866888999938965,
+      "grad_norm": 0.7089729309082031,
       "learning_rate": 4.809414468107536e-05,
-      "loss": 0.8606,
+      "loss": 0.8607,
       "num_input_tokens_seen": 28508160,
       "step": 3480
     },
     {
       "epoch": 0.16758703481392556,
-      "grad_norm": 0.5912758111953735,
+      "grad_norm": 0.6048206090927124,
       "learning_rate": 4.8078944218121404e-05,
-      "loss": 0.7668,
+      "loss": 0.7663,
       "num_input_tokens_seen": 28590080,
       "step": 3490
     },
     {
       "epoch": 0.16806722689075632,
-      "grad_norm": 0.5021241903305054,
+      "grad_norm": 0.5017691850662231,
       "learning_rate": 4.806368580039566e-05,
-      "loss": 1.1072,
+      "loss": 1.1082,
       "num_input_tokens_seen": 28672000,
       "step": 3500
     },
     {
       "epoch": 0.16854741896758704,
-      "grad_norm": 0.5017940402030945,
+      "grad_norm": 0.5294102430343628,
       "learning_rate": 4.804836946621437e-05,
-      "loss": 0.7905,
+      "loss": 0.7908,
       "num_input_tokens_seen": 28753920,
       "step": 3510
     },
     {
       "epoch": 0.16902761104441777,
-      "grad_norm": 0.5461972951889038,
+      "grad_norm": 0.5386155247688293,
       "learning_rate": 4.803299525403919e-05,
-      "loss": 0.9979,
+      "loss": 0.9984,
       "num_input_tokens_seen": 28835840,
       "step": 3520
     },
     {
       "epoch": 0.1695078031212485,
-      "grad_norm": 0.5111615061759949,
+      "grad_norm": 0.5093580484390259,
       "learning_rate": 4.801756320247713e-05,
-      "loss": 1.1211,
+      "loss": 1.1216,
       "num_input_tokens_seen": 28917760,
       "step": 3530
     },
     {
       "epoch": 0.16998799519807922,
-      "grad_norm": 0.5483148097991943,
+      "grad_norm": 0.5753290057182312,
       "learning_rate": 4.800207335028044e-05,
-      "loss": 0.9346,
+      "loss": 0.9355,
       "num_input_tokens_seen": 28999680,
       "step": 3540
     },
     {
       "epoch": 0.17046818727490998,
-      "grad_norm": 0.6430408358573914,
+      "grad_norm": 0.6612030863761902,
       "learning_rate": 4.798652573634651e-05,
-      "loss": 0.8605,
+      "loss": 0.8603,
       "num_input_tokens_seen": 29081600,
       "step": 3550
     },
     {
       "epoch": 0.1709483793517407,
-      "grad_norm": 0.41809263825416565,
+      "grad_norm": 0.44161081314086914,
       "learning_rate": 4.797092039971779e-05,
-      "loss": 0.631,
+      "loss": 0.6316,
       "num_input_tokens_seen": 29163520,
       "step": 3560
     },
     {
       "epoch": 0.17142857142857143,
-      "grad_norm": 0.5095425844192505,
+      "grad_norm": 0.5154605507850647,
       "learning_rate": 4.7955257379581675e-05,
-      "loss": 0.8789,
+      "loss": 0.8794,
       "num_input_tokens_seen": 29245440,
       "step": 3570
     },
     {
       "epoch": 0.17190876350540216,
-      "grad_norm": 0.5259044766426086,
+      "grad_norm": 0.520716667175293,
       "learning_rate": 4.7939536715270415e-05,
-      "loss": 1.2783,
+      "loss": 1.2806,
       "num_input_tokens_seen": 29327360,
       "step": 3580
     },
     {
       "epoch": 0.17238895558223288,
-      "grad_norm": 0.5708450675010681,
+      "grad_norm": 0.5714126825332642,
       "learning_rate": 4.792375844626101e-05,
-      "loss": 0.7012,
+      "loss": 0.7021,
       "num_input_tokens_seen": 29409280,
       "step": 3590
     },
     {
       "epoch": 0.17286914765906364,
-      "grad_norm": 0.5291116833686829,
+      "grad_norm": 0.49484771490097046,
       "learning_rate": 4.790792261217512e-05,
-      "loss": 0.976,
+      "loss": 0.9755,
       "num_input_tokens_seen": 29491200,
       "step": 3600
     },
     {
       "epoch": 0.17334933973589436,
-      "grad_norm": 0.4987095296382904,
+      "grad_norm": 0.4960525929927826,
       "learning_rate": 4.789202925277895e-05,
-      "loss": 0.9292,
+      "loss": 0.929,
       "num_input_tokens_seen": 29573120,
       "step": 3610
     },
     {
       "epoch": 0.1738295318127251,
-      "grad_norm": 0.6108640432357788,
+      "grad_norm": 0.5834500789642334,
       "learning_rate": 4.787607840798317e-05,
-      "loss": 0.8974,
+      "loss": 0.8966,
       "num_input_tokens_seen": 29655040,
       "step": 3620
     },
     {
       "epoch": 0.17430972388955582,
-      "grad_norm": 0.42809537053108215,
+      "grad_norm": 0.4365158677101135,
       "learning_rate": 4.786007011784279e-05,
-      "loss": 0.9049,
+      "loss": 0.9054,
       "num_input_tokens_seen": 29736960,
       "step": 3630
     },
     {
       "epoch": 0.17478991596638654,
-      "grad_norm": 0.4873834252357483,
+      "grad_norm": 0.48445218801498413,
       "learning_rate": 4.78440044225571e-05,
-      "loss": 0.8813,
+      "loss": 0.881,
       "num_input_tokens_seen": 29818880,
       "step": 3640
     },
     {
       "epoch": 0.1752701080432173,
-      "grad_norm": 0.6227244138717651,
+      "grad_norm": 0.6551578640937805,
       "learning_rate": 4.7827881362469506e-05,
-      "loss": 0.7923,
+      "loss": 0.7921,
       "num_input_tokens_seen": 29900800,
       "step": 3650
     },
     {
       "epoch": 0.17575030012004803,
-      "grad_norm": 0.4629347324371338,
+      "grad_norm": 0.45337924361228943,
       "learning_rate": 4.781170097806751e-05,
-      "loss": 0.9563,
+      "loss": 0.956,
       "num_input_tokens_seen": 29982720,
       "step": 3660
     },
     {
       "epoch": 0.17623049219687875,
-      "grad_norm": 0.5306552648544312,
+      "grad_norm": 0.521186888217926,
       "learning_rate": 4.779546330998253e-05,
-      "loss": 0.9327,
+      "loss": 0.9339,
       "num_input_tokens_seen": 30064640,
       "step": 3670
     },
     {
       "epoch": 0.17671068427370948,
-      "grad_norm": 0.5401423573493958,
+      "grad_norm": 0.552125096321106,
       "learning_rate": 4.7779168398989826e-05,
-      "loss": 1.2167,
+      "loss": 1.2112,
       "num_input_tokens_seen": 30146560,
       "step": 3680
     },
     {
       "epoch": 0.1771908763505402,
-      "grad_norm": 0.5693386197090149,
+      "grad_norm": 0.6073199510574341,
       "learning_rate": 4.7762816286008454e-05,
-      "loss": 0.9795,
+      "loss": 0.9783,
       "num_input_tokens_seen": 30228480,
       "step": 3690
     },
     {
       "epoch": 0.17767106842737096,
-      "grad_norm": 0.5739344954490662,
+      "grad_norm": 0.5938025712966919,
       "learning_rate": 4.774640701210106e-05,
-      "loss": 1.108,
+      "loss": 1.1071,
       "num_input_tokens_seen": 30310400,
       "step": 3700
     },
     {
       "epoch": 0.1781512605042017,
-      "grad_norm": 0.5333011150360107,
+      "grad_norm": 0.5386204719543457,
       "learning_rate": 4.7729940618473854e-05,
-      "loss": 1.0156,
+      "loss": 1.0155,
       "num_input_tokens_seen": 30392320,
       "step": 3710
     },
     {
       "epoch": 0.1786314525810324,
-      "grad_norm": 0.6173415780067444,
+      "grad_norm": 0.6245675683021545,
       "learning_rate": 4.771341714647648e-05,
-      "loss": 0.9543,
+      "loss": 0.9545,
       "num_input_tokens_seen": 30474240,
       "step": 3720
     },
     {
       "epoch": 0.17911164465786314,
-      "grad_norm": 0.5696475505828857,
+      "grad_norm": 0.5611715912818909,
       "learning_rate": 4.76968366376019e-05,
-      "loss": 0.875,
+      "loss": 0.8737,
       "num_input_tokens_seen": 30556160,
       "step": 3730
     },
     {
       "epoch": 0.17959183673469387,
-      "grad_norm": 0.5061577558517456,
+      "grad_norm": 0.4791863262653351,
       "learning_rate": 4.768019913348634e-05,
-      "loss": 0.9798,
+      "loss": 0.9804,
       "num_input_tokens_seen": 30638080,
       "step": 3740
     },
     {
       "epoch": 0.18007202881152462,
-      "grad_norm": 0.454387903213501,
+      "grad_norm": 0.46026119589805603,
       "learning_rate": 4.766350467590911e-05,
-      "loss": 0.8985,
+      "loss": 0.8976,
       "num_input_tokens_seen": 30720000,
       "step": 3750
     },
     {
       "epoch": 0.18055222088835535,
-      "grad_norm": 0.5241227746009827,
+      "grad_norm": 0.5252475738525391,
       "learning_rate": 4.764675330679256e-05,
-      "loss": 0.9849,
+      "loss": 0.9886,
       "num_input_tokens_seen": 30801920,
       "step": 3760
     },
     {
       "epoch": 0.18103241296518607,
-      "grad_norm": 0.5448564887046814,
+      "grad_norm": 0.5292924642562866,
       "learning_rate": 4.7629945068201954e-05,
-      "loss": 0.9484,
+      "loss": 0.9472,
       "num_input_tokens_seen": 30883840,
       "step": 3770
     },
     {
       "epoch": 0.1815126050420168,
-      "grad_norm": 0.5225489139556885,
+      "grad_norm": 0.5213156938552856,
       "learning_rate": 4.7613080002345345e-05,
-      "loss": 0.8183,
+      "loss": 0.8187,
       "num_input_tokens_seen": 30965760,
       "step": 3780
     },
     {
       "epoch": 0.18199279711884753,
-      "grad_norm": 0.5782943367958069,
+      "grad_norm": 0.5824140906333923,
       "learning_rate": 4.759615815157352e-05,
-      "loss": 1.0653,
+      "loss": 1.0672,
       "num_input_tokens_seen": 31047680,
       "step": 3790
     },
     {
       "epoch": 0.18247298919567828,
-      "grad_norm": 0.5790293216705322,
+      "grad_norm": 0.6037180423736572,
       "learning_rate": 4.7579179558379836e-05,
-      "loss": 0.8532,
+      "loss": 0.854,
       "num_input_tokens_seen": 31129600,
       "step": 3800
     },
     {
       "epoch": 0.182953181272509,
-      "grad_norm": 0.49173110723495483,
+      "grad_norm": 0.47562384605407715,
       "learning_rate": 4.7562144265400146e-05,
-      "loss": 0.7594,
+      "loss": 0.759,
       "num_input_tokens_seen": 31211520,
       "step": 3810
     },
     {
       "epoch": 0.18343337334933973,
-      "grad_norm": 0.5832033157348633,
+      "grad_norm": 0.5508514642715454,
       "learning_rate": 4.754505231541268e-05,
-      "loss": 1.0458,
+      "loss": 1.0452,
       "num_input_tokens_seen": 31293440,
       "step": 3820
     },
     {
       "epoch": 0.18391356542617046,
-      "grad_norm": 0.539718508720398,
+      "grad_norm": 0.5534719824790955,
       "learning_rate": 4.752790375133797e-05,
-      "loss": 0.8403,
+      "loss": 0.8398,
       "num_input_tokens_seen": 31375360,
       "step": 3830
     },
     {
       "epoch": 0.1843937575030012,
-      "grad_norm": 0.5238791704177856,
+      "grad_norm": 0.5246814489364624,
       "learning_rate": 4.751069861623867e-05,
-      "loss": 0.9284,
+      "loss": 0.9282,
       "num_input_tokens_seen": 31457280,
       "step": 3840
     },
     {
       "epoch": 0.18487394957983194,
-      "grad_norm": 0.5033566355705261,
+      "grad_norm": 0.5262382626533508,
       "learning_rate": 4.749343695331952e-05,
-      "loss": 1.0164,
+      "loss": 1.0174,
       "num_input_tokens_seen": 31539200,
       "step": 3850
     },
     {
       "epoch": 0.18535414165666267,
-      "grad_norm": 0.6581845879554749,
+      "grad_norm": 0.6730586290359497,
       "learning_rate": 4.747611880592721e-05,
-      "loss": 0.8328,
+      "loss": 0.8329,
       "num_input_tokens_seen": 31621120,
       "step": 3860
     },
     {
       "epoch": 0.1858343337334934,
-      "grad_norm": 0.5361823439598083,
+      "grad_norm": 0.5353675484657288,
       "learning_rate": 4.745874421755027e-05,
-      "loss": 1.0115,
+      "loss": 1.0112,
       "num_input_tokens_seen": 31703040,
       "step": 3870
     },
     {
       "epoch": 0.18631452581032412,
-      "grad_norm": 0.5217710733413696,
+      "grad_norm": 0.517441987991333,
       "learning_rate": 4.744131323181895e-05,
-      "loss": 0.8417,
+      "loss": 0.8418,
       "num_input_tokens_seen": 31784960,
       "step": 3880
     },
     {
       "epoch": 0.18679471788715485,
-      "grad_norm": 0.5840950608253479,
+      "grad_norm": 0.5716066956520081,
       "learning_rate": 4.742382589250514e-05,
-      "loss": 0.9902,
+      "loss": 0.9913,
       "num_input_tokens_seen": 31866880,
       "step": 3890
     },
     {
       "epoch": 0.1872749099639856,
-      "grad_norm": 0.5277583599090576,
+      "grad_norm": 0.5103275179862976,
       "learning_rate": 4.740628224352222e-05,
       "loss": 1.0409,
       "num_input_tokens_seen": 31948800,
@@ -3130,151 +3130,151 @@
     },
     {
       "epoch": 0.18775510204081633,
-      "grad_norm": 0.5681836605072021,
+      "grad_norm": 0.5536543726921082,
       "learning_rate": 4.7388682328925e-05,
-      "loss": 0.9343,
+      "loss": 0.9342,
       "num_input_tokens_seen": 32030720,
       "step": 3910
     },
     {
       "epoch": 0.18823529411764706,
-      "grad_norm": 0.4854600131511688,
+      "grad_norm": 0.5033223032951355,
       "learning_rate": 4.737102619290956e-05,
-      "loss": 0.9198,
+      "loss": 0.9206,
       "num_input_tokens_seen": 32112640,
       "step": 3920
     },
     {
       "epoch": 0.18871548619447778,
-      "grad_norm": 0.527630627155304,
+      "grad_norm": 0.5292133688926697,
       "learning_rate": 4.7353313879813165e-05,
-      "loss": 0.9707,
+      "loss": 0.9706,
       "num_input_tokens_seen": 32194560,
       "step": 3930
     },
     {
       "epoch": 0.18919567827130854,
-      "grad_norm": 0.5316260457038879,
+      "grad_norm": 0.527191698551178,
       "learning_rate": 4.733554543411417e-05,
-      "loss": 1.0691,
+      "loss": 1.0686,
       "num_input_tokens_seen": 32276480,
       "step": 3940
     },
     {
       "epoch": 0.18967587034813926,
-      "grad_norm": 0.5303869843482971,
+      "grad_norm": 0.534002959728241,
       "learning_rate": 4.731772090043184e-05,
-      "loss": 0.9361,
+      "loss": 0.9365,
       "num_input_tokens_seen": 32358400,
       "step": 3950
     },
     {
       "epoch": 0.19015606242497,
-      "grad_norm": 0.5066818594932556,
+      "grad_norm": 0.5045952796936035,
       "learning_rate": 4.729984032352635e-05,
-      "loss": 0.8927,
+      "loss": 0.8931,
       "num_input_tokens_seen": 32440320,
       "step": 3960
     },
     {
       "epoch": 0.19063625450180072,
-      "grad_norm": 0.5547256469726562,
+      "grad_norm": 0.5647931694984436,
       "learning_rate": 4.728190374829854e-05,
-      "loss": 0.9308,
+      "loss": 0.9301,
       "num_input_tokens_seen": 32522240,
       "step": 3970
     },
     {
       "epoch": 0.19111644657863144,
-      "grad_norm": 0.46649324893951416,
+      "grad_norm": 0.48447665572166443,
       "learning_rate": 4.726391121978992e-05,
-      "loss": 0.9229,
+      "loss": 0.9234,
       "num_input_tokens_seen": 32604160,
       "step": 3980
     },
     {
       "epoch": 0.1915966386554622,
-      "grad_norm": 0.5034019351005554,
+      "grad_norm": 0.5057299137115479,
       "learning_rate": 4.7245862783182496e-05,
-      "loss": 0.9033,
+      "loss": 0.9042,
       "num_input_tokens_seen": 32686080,
       "step": 3990
     },
     {
       "epoch": 0.19207683073229292,
-      "grad_norm": 0.5081968307495117,
+      "grad_norm": 0.5080722570419312,
       "learning_rate": 4.722775848379866e-05,
-      "loss": 1.0315,
+      "loss": 1.0318,
       "num_input_tokens_seen": 32768000,
       "step": 4000
     },
     {
       "epoch": 0.19255702280912365,
-      "grad_norm": 0.4671435058116913,
+      "grad_norm": 0.4755553901195526,
       "learning_rate": 4.720959836710107e-05,
-      "loss": 0.8195,
+      "loss": 0.8199,
       "num_input_tokens_seen": 32849920,
       "step": 4010
     },
     {
       "epoch": 0.19303721488595438,
-      "grad_norm": 0.4735753536224365,
+      "grad_norm": 0.47308361530303955,
       "learning_rate": 4.7191382478692594e-05,
-      "loss": 0.9489,
+      "loss": 0.9483,
       "num_input_tokens_seen": 32931840,
       "step": 4020
     },
     {
       "epoch": 0.1935174069627851,
-      "grad_norm": 0.5032724142074585,
+      "grad_norm": 0.5154343843460083,
       "learning_rate": 4.7173110864316104e-05,
-      "loss": 0.814,
+      "loss": 0.8141,
       "num_input_tokens_seen": 33013760,
       "step": 4030
     },
     {
       "epoch": 0.19399759903961586,
-      "grad_norm": 0.645969033241272,
+      "grad_norm": 0.6429656744003296,
       "learning_rate": 4.7154783569854444e-05,
-      "loss": 0.811,
+      "loss": 0.8107,
       "num_input_tokens_seen": 33095680,
       "step": 4040
     },
     {
       "epoch": 0.19447779111644659,
-      "grad_norm": 0.5401909947395325,
+      "grad_norm": 0.5589474439620972,
       "learning_rate": 4.713640064133025e-05,
-      "loss": 0.9777,
+      "loss": 0.9762,
       "num_input_tokens_seen": 33177600,
       "step": 4050
     },
     {
       "epoch": 0.1949579831932773,
-      "grad_norm": 0.5023765563964844,
+      "grad_norm": 0.5128633379936218,
       "learning_rate": 4.7117962124905885e-05,
-      "loss": 0.8785,
+      "loss": 0.8789,
       "num_input_tokens_seen": 33259520,
       "step": 4060
     },
     {
       "epoch": 0.19543817527010804,
-      "grad_norm": 0.5897220373153687,
+      "grad_norm": 0.5557809472084045,
       "learning_rate": 4.709946806688329e-05,
-      "loss": 0.9837,
+      "loss": 0.9835,
       "num_input_tokens_seen": 33341440,
       "step": 4070
     },
     {
       "epoch": 0.19591836734693877,
-      "grad_norm": 0.6706360578536987,
+      "grad_norm": 0.6408475637435913,
       "learning_rate": 4.708091851370389e-05,
-      "loss": 0.8769,
+      "loss": 0.8732,
       "num_input_tokens_seen": 33423360,
       "step": 4080
     },
     {
       "epoch": 0.19639855942376952,
-      "grad_norm": 0.5493477582931519,
+      "grad_norm": 0.5550397634506226,
       "learning_rate": 4.706231351194845e-05,
       "loss": 0.9604,
       "num_input_tokens_seen": 33505280,
@@ -3282,39 +3282,39 @@
     },
     {
       "epoch": 0.19687875150060025,
-      "grad_norm": 0.5341002345085144,
+      "grad_norm": 0.5453352928161621,
       "learning_rate": 4.7043653108336994e-05,
-      "loss": 0.8028,
+      "loss": 0.8032,
       "num_input_tokens_seen": 33587200,
       "step": 4100
     },
     {
       "epoch": 0.19735894357743097,
-      "grad_norm": 0.5551480054855347,
+      "grad_norm": 0.5484746098518372,
       "learning_rate": 4.702493734972866e-05,
-      "loss": 0.9994,
+      "loss": 0.9989,
       "num_input_tokens_seen": 33669120,
       "step": 4110
     },
     {
       "epoch": 0.1978391356542617,
-      "grad_norm": 1.2934221029281616,
+      "grad_norm": 1.165541648864746,
       "learning_rate": 4.700616628312158e-05,
-      "loss": 1.0073,
+      "loss": 1.0064,
       "num_input_tokens_seen": 33751040,
       "step": 4120
     },
     {
       "epoch": 0.19831932773109243,
-      "grad_norm": 0.5070963501930237,
+      "grad_norm": 0.5196086168289185,
       "learning_rate": 4.69873399556528e-05,
-      "loss": 0.8428,
+      "loss": 0.8425,
       "num_input_tokens_seen": 33832960,
       "step": 4130
     },
     {
       "epoch": 0.19879951980792318,
-      "grad_norm": 0.49348321557044983,
+      "grad_norm": 0.506844162940979,
       "learning_rate": 4.696845841459811e-05,
       "loss": 1.0941,
       "num_input_tokens_seen": 33914880,
@@ -3322,551 +3322,551 @@
     },
     {
       "epoch": 0.1992797118847539,
-      "grad_norm": 0.48639488220214844,
+      "grad_norm": 0.46458375453948975,
       "learning_rate": 4.6949521707371965e-05,
-      "loss": 0.8673,
+      "loss": 0.8666,
       "num_input_tokens_seen": 33996800,
       "step": 4150
     },
     {
       "epoch": 0.19975990396158463,
-      "grad_norm": 0.5290814638137817,
+      "grad_norm": 0.5106430649757385,
       "learning_rate": 4.693052988152733e-05,
-      "loss": 0.8488,
+      "loss": 0.8477,
       "num_input_tokens_seen": 34078720,
       "step": 4160
     },
     {
       "epoch": 0.20024009603841536,
-      "grad_norm": 0.7159754037857056,
+      "grad_norm": 0.7916086316108704,
       "learning_rate": 4.691148298475561e-05,
-      "loss": 1.0891,
+      "loss": 1.0889,
       "num_input_tokens_seen": 34160640,
       "step": 4170
     },
     {
       "epoch": 0.2007202881152461,
-      "grad_norm": 0.5923986434936523,
+      "grad_norm": 0.5892463326454163,
       "learning_rate": 4.689238106488647e-05,
-      "loss": 0.904,
+      "loss": 0.9039,
       "num_input_tokens_seen": 34242560,
       "step": 4180
     },
     {
       "epoch": 0.20120048019207684,
-      "grad_norm": 0.4993061125278473,
+      "grad_norm": 0.521371603012085,
       "learning_rate": 4.687322416988779e-05,
-      "loss": 0.8948,
+      "loss": 0.8954,
       "num_input_tokens_seen": 34324480,
       "step": 4190
     },
     {
       "epoch": 0.20168067226890757,
-      "grad_norm": 0.48761218786239624,
+      "grad_norm": 0.49719128012657166,
       "learning_rate": 4.685401234786544e-05,
-      "loss": 0.9203,
+      "loss": 0.92,
       "num_input_tokens_seen": 34406400,
       "step": 4200
     },
     {
       "epoch": 0.2021608643457383,
-      "grad_norm": 0.4988939166069031,
+      "grad_norm": 0.5170416235923767,
       "learning_rate": 4.683474564706327e-05,
-      "loss": 0.9092,
+      "loss": 0.9093,
       "num_input_tokens_seen": 34488320,
       "step": 4210
     },
     {
       "epoch": 0.20264105642256902,
-      "grad_norm": 0.6328974366188049,
+      "grad_norm": 0.6528691053390503,
       "learning_rate": 4.681542411586294e-05,
-      "loss": 0.8973,
+      "loss": 0.8968,
       "num_input_tokens_seen": 34570240,
       "step": 4220
     },
     {
       "epoch": 0.20312124849939975,
-      "grad_norm": 0.5057374835014343,
+      "grad_norm": 0.5039422512054443,
       "learning_rate": 4.6796047802783755e-05,
-      "loss": 1.046,
+      "loss": 1.0459,
       "num_input_tokens_seen": 34652160,
       "step": 4230
     },
     {
       "epoch": 0.2036014405762305,
-      "grad_norm": 0.562879741191864,
+      "grad_norm": 0.547264039516449,
       "learning_rate": 4.6776616756482624e-05,
-      "loss": 0.8767,
+      "loss": 0.8744,
       "num_input_tokens_seen": 34734080,
       "step": 4240
     },
     {
       "epoch": 0.20408163265306123,
-      "grad_norm": 0.9288103580474854,
+      "grad_norm": 0.9196895360946655,
       "learning_rate": 4.6757131025753886e-05,
-      "loss": 0.849,
+      "loss": 0.8492,
       "num_input_tokens_seen": 34816000,
       "step": 4250
     },
     {
       "epoch": 0.20456182472989196,
-      "grad_norm": 0.5365880727767944,
+      "grad_norm": 0.5697296857833862,
       "learning_rate": 4.67375906595292e-05,
-      "loss": 1.1241,
+      "loss": 1.1252,
       "num_input_tokens_seen": 34897920,
       "step": 4260
     },
     {
       "epoch": 0.20504201680672268,
-      "grad_norm": 0.5342133641242981,
+      "grad_norm": 0.5439584851264954,
       "learning_rate": 4.671799570687743e-05,
-      "loss": 1.1088,
+      "loss": 1.1081,
       "num_input_tokens_seen": 34979840,
       "step": 4270
     },
     {
       "epoch": 0.2055222088835534,
-      "grad_norm": 0.5288349390029907,
+      "grad_norm": 0.5510857105255127,
       "learning_rate": 4.6698346217004494e-05,
-      "loss": 0.9175,
+      "loss": 0.917,
       "num_input_tokens_seen": 35061760,
       "step": 4280
     },
     {
       "epoch": 0.20600240096038416,
-      "grad_norm": 0.7718984484672546,
+      "grad_norm": 0.7729096412658691,
       "learning_rate": 4.66786422392533e-05,
-      "loss": 1.146,
+      "loss": 1.1476,
       "num_input_tokens_seen": 35143680,
       "step": 4290
     },
     {
       "epoch": 0.2064825930372149,
-      "grad_norm": 0.5124546885490417,
+      "grad_norm": 0.5175178050994873,
       "learning_rate": 4.665888382310356e-05,
-      "loss": 0.9771,
+      "loss": 0.9777,
       "num_input_tokens_seen": 35225600,
       "step": 4300
     },
     {
       "epoch": 0.20696278511404562,
-      "grad_norm": 0.52974534034729,
+      "grad_norm": 0.5331414341926575,
       "learning_rate": 4.663907101817167e-05,
-      "loss": 0.9299,
+      "loss": 0.9304,
       "num_input_tokens_seen": 35307520,
       "step": 4310
     },
     {
       "epoch": 0.20744297719087634,
-      "grad_norm": 0.50569087266922,
+      "grad_norm": 0.5062457919120789,
       "learning_rate": 4.661920387421064e-05,
-      "loss": 1.0124,
+      "loss": 1.0128,
       "num_input_tokens_seen": 35389440,
       "step": 4320
     },
     {
       "epoch": 0.20792316926770707,
-      "grad_norm": 0.5128062963485718,
+      "grad_norm": 0.5168769955635071,
       "learning_rate": 4.65992824411099e-05,
-      "loss": 0.9051,
+      "loss": 0.904,
       "num_input_tokens_seen": 35471360,
       "step": 4330
     },
     {
       "epoch": 0.20840336134453782,
-      "grad_norm": 0.549404501914978,
+      "grad_norm": 0.5486623644828796,
       "learning_rate": 4.657930676889526e-05,
-      "loss": 1.11,
+      "loss": 1.1092,
       "num_input_tokens_seen": 35553280,
       "step": 4340
     },
     {
       "epoch": 0.20888355342136855,
-      "grad_norm": 0.5476168394088745,
+      "grad_norm": 0.5371941924095154,
       "learning_rate": 4.655927690772868e-05,
-      "loss": 0.7549,
+      "loss": 0.7555,
       "num_input_tokens_seen": 35635200,
       "step": 4350
     },
     {
       "epoch": 0.20936374549819928,
-      "grad_norm": 0.5444672107696533,
+      "grad_norm": 0.5360463857650757,
       "learning_rate": 4.6539192907908204e-05,
-      "loss": 0.798,
+      "loss": 0.7976,
       "num_input_tokens_seen": 35717120,
       "step": 4360
     },
     {
       "epoch": 0.20984393757503,
-      "grad_norm": 0.7407582998275757,
+      "grad_norm": 0.735427975654602,
       "learning_rate": 4.6519054819867856e-05,
-      "loss": 1.0632,
+      "loss": 1.0636,
       "num_input_tokens_seen": 35799040,
       "step": 4370
     },
     {
       "epoch": 0.21032412965186073,
-      "grad_norm": 0.5673499703407288,
+      "grad_norm": 0.5705459117889404,
       "learning_rate": 4.649886269417746e-05,
-      "loss": 0.9968,
+      "loss": 0.9967,
       "num_input_tokens_seen": 35880960,
       "step": 4380
     },
     {
       "epoch": 0.21080432172869148,
-      "grad_norm": 0.5538674592971802,
+      "grad_norm": 0.5621838569641113,
       "learning_rate": 4.647861658154254e-05,
-      "loss": 0.8916,
+      "loss": 0.8918,
       "num_input_tokens_seen": 35962880,
       "step": 4390
     },
     {
       "epoch": 0.2112845138055222,
-      "grad_norm": 0.4925515949726105,
+      "grad_norm": 0.4890127182006836,
       "learning_rate": 4.6458316532804214e-05,
-      "loss": 1.0407,
+      "loss": 1.0406,
       "num_input_tokens_seen": 36044800,
       "step": 4400
     },
     {
       "epoch": 0.21176470588235294,
-      "grad_norm": 0.4931751787662506,
+      "grad_norm": 0.4952712655067444,
       "learning_rate": 4.643796259893899e-05,
-      "loss": 0.9456,
+      "loss": 0.9447,
       "num_input_tokens_seen": 36126720,
       "step": 4410
     },
     {
       "epoch": 0.21224489795918366,
-      "grad_norm": 0.5209604501724243,
+      "grad_norm": 0.5150204300880432,
       "learning_rate": 4.641755483105874e-05,
-      "loss": 0.9378,
+      "loss": 0.9381,
       "num_input_tokens_seen": 36208640,
       "step": 4420
     },
     {
       "epoch": 0.21272509003601442,
-      "grad_norm": 1.0745376348495483,
+      "grad_norm": 1.1830569505691528,
       "learning_rate": 4.63970932804105e-05,
-      "loss": 0.8205,
+      "loss": 0.8216,
       "num_input_tokens_seen": 36290560,
       "step": 4430
     },
     {
       "epoch": 0.21320528211284515,
-      "grad_norm": 0.5311976671218872,
+      "grad_norm": 0.5193588733673096,
       "learning_rate": 4.637657799837635e-05,
-      "loss": 0.8998,
+      "loss": 0.9,
       "num_input_tokens_seen": 36372480,
       "step": 4440
     },
     {
       "epoch": 0.21368547418967587,
-      "grad_norm": 0.5121796727180481,
+      "grad_norm": 0.5316673517227173,
       "learning_rate": 4.635600903647333e-05,
-      "loss": 0.8201,
+      "loss": 0.8199,
       "num_input_tokens_seen": 36454400,
       "step": 4450
     },
     {
       "epoch": 0.2141656662665066,
-      "grad_norm": 0.5421672463417053,
+      "grad_norm": 0.5320562124252319,
       "learning_rate": 4.633538644635326e-05,
-      "loss": 0.879,
+      "loss": 0.8799,
       "num_input_tokens_seen": 36536320,
       "step": 4460
     },
     {
       "epoch": 0.21464585834333733,
-      "grad_norm": 1.2252622842788696,
+      "grad_norm": 1.1617580652236938,
       "learning_rate": 4.631471027980262e-05,
-      "loss": 0.9865,
+      "loss": 0.9863,
       "num_input_tokens_seen": 36618240,
       "step": 4470
     },
     {
       "epoch": 0.21512605042016808,
-      "grad_norm": 0.5252002477645874,
+      "grad_norm": 0.5181916952133179,
       "learning_rate": 4.629398058874245e-05,
-      "loss": 0.8658,
+      "loss": 0.8663,
       "num_input_tokens_seen": 36700160,
       "step": 4480
     },
     {
       "epoch": 0.2156062424969988,
-      "grad_norm": 0.5939603447914124,
+      "grad_norm": 0.5579994916915894,
       "learning_rate": 4.6273197425228166e-05,
-      "loss": 0.9567,
+      "loss": 0.9576,
       "num_input_tokens_seen": 36782080,
       "step": 4490
     },
     {
       "epoch": 0.21608643457382953,
-      "grad_norm": 0.5345758199691772,
+      "grad_norm": 0.5332381725311279,
       "learning_rate": 4.62523608414495e-05,
-      "loss": 0.9567,
+      "loss": 0.9573,
       "num_input_tokens_seen": 36864000,
       "step": 4500
     },
     {
       "epoch": 0.21656662665066026,
-      "grad_norm": 1.0143563747406006,
+      "grad_norm": 1.0233074426651,
       "learning_rate": 4.623147088973031e-05,
-      "loss": 0.7952,
+      "loss": 0.795,
       "num_input_tokens_seen": 36945920,
       "step": 4510
     },
     {
       "epoch": 0.217046818727491,
-      "grad_norm": 0.5424322485923767,
+      "grad_norm": 0.5333688855171204,
       "learning_rate": 4.6210527622528465e-05,
-      "loss": 1.1468,
+      "loss": 1.1473,
       "num_input_tokens_seen": 37027840,
       "step": 4520
     },
     {
       "epoch": 0.21752701080432174,
-      "grad_norm": 0.5119237303733826,
+      "grad_norm": 0.5092252492904663,
       "learning_rate": 4.618953109243573e-05,
-      "loss": 0.8657,
+      "loss": 0.8676,
       "num_input_tokens_seen": 37109760,
       "step": 4530
     },
     {
       "epoch": 0.21800720288115247,
-      "grad_norm": 0.5329012274742126,
+      "grad_norm": 0.539064347743988,
       "learning_rate": 4.616848135217761e-05,
-      "loss": 0.9235,
+      "loss": 0.9233,
       "num_input_tokens_seen": 37191680,
       "step": 4540
     },
     {
       "epoch": 0.2184873949579832,
-      "grad_norm": 0.5090308785438538,
+      "grad_norm": 0.509436309337616,
       "learning_rate": 4.6147378454613246e-05,
-      "loss": 0.8308,
+      "loss": 0.831,
       "num_input_tokens_seen": 37273600,
       "step": 4550
     },
     {
       "epoch": 0.21896758703481392,
-      "grad_norm": 0.7635847926139832,
+      "grad_norm": 0.780189573764801,
       "learning_rate": 4.6126222452735233e-05,
-      "loss": 0.8878,
+      "loss": 0.8871,
       "num_input_tokens_seen": 37355520,
       "step": 4560
     },
     {
       "epoch": 0.21944777911164465,
-      "grad_norm": 0.48617926239967346,
+      "grad_norm": 0.5022001266479492,
       "learning_rate": 4.6105013399669564e-05,
-      "loss": 0.9695,
+      "loss": 0.9688,
       "num_input_tokens_seen": 37437440,
       "step": 4570
     },
     {
       "epoch": 0.2199279711884754,
-      "grad_norm": 0.6429051756858826,
+      "grad_norm": 0.5899694561958313,
       "learning_rate": 4.608375134867541e-05,
-      "loss": 0.8531,
+      "loss": 0.8533,
       "num_input_tokens_seen": 37519360,
       "step": 4580
     },
     {
       "epoch": 0.22040816326530613,
-      "grad_norm": 0.5649603605270386,
+      "grad_norm": 0.62712562084198,
       "learning_rate": 4.6062436353145044e-05,
-      "loss": 0.8998,
+      "loss": 0.8995,
       "num_input_tokens_seen": 37601280,
       "step": 4590
     },
     {
       "epoch": 0.22088835534213686,
-      "grad_norm": 0.5439668297767639,
+      "grad_norm": 0.5399054884910583,
       "learning_rate": 4.60410684666037e-05,
-      "loss": 1.2532,
+      "loss": 1.2521,
       "num_input_tokens_seen": 37683200,
       "step": 4600
     },
     {
       "epoch": 0.22136854741896758,
-      "grad_norm": 0.5839661359786987,
+      "grad_norm": 0.5870562195777893,
       "learning_rate": 4.601964774270941e-05,
-      "loss": 0.9984,
+      "loss": 0.998,
       "num_input_tokens_seen": 37765120,
       "step": 4610
     },
     {
       "epoch": 0.2218487394957983,
-      "grad_norm": 0.5015205144882202,
+      "grad_norm": 0.49380698800086975,
       "learning_rate": 4.599817423525292e-05,
-      "loss": 1.0361,
+      "loss": 1.036,
       "num_input_tokens_seen": 37847040,
       "step": 4620
     },
     {
       "epoch": 0.22232893157262906,
-      "grad_norm": 0.5025255084037781,
+      "grad_norm": 0.501021146774292,
       "learning_rate": 4.597664799815749e-05,
-      "loss": 1.0169,
+      "loss": 1.0166,
       "num_input_tokens_seen": 37928960,
       "step": 4630
     },
     {
       "epoch": 0.2228091236494598,
-      "grad_norm": 0.4889910817146301,
+      "grad_norm": 0.49086204171180725,
       "learning_rate": 4.595506908547881e-05,
-      "loss": 0.9112,
+      "loss": 0.9113,
       "num_input_tokens_seen": 38010880,
       "step": 4640
     },
     {
       "epoch": 0.22328931572629052,
-      "grad_norm": 0.5253650546073914,
+      "grad_norm": 0.5220022797584534,
       "learning_rate": 4.593343755140483e-05,
-      "loss": 0.9424,
+      "loss": 0.9429,
       "num_input_tokens_seen": 38092800,
       "step": 4650
     },
     {
       "epoch": 0.22376950780312124,
-      "grad_norm": 0.4819222688674927,
+      "grad_norm": 0.48921534419059753,
       "learning_rate": 4.5911753450255665e-05,
-      "loss": 1.0131,
+      "loss": 1.0135,
       "num_input_tokens_seen": 38174720,
       "step": 4660
     },
     {
       "epoch": 0.22424969987995197,
-      "grad_norm": 0.5012921690940857,
+      "grad_norm": 0.4981484115123749,
       "learning_rate": 4.589001683648343e-05,
-      "loss": 0.7262,
+      "loss": 0.7263,
       "num_input_tokens_seen": 38256640,
       "step": 4670
     },
     {
       "epoch": 0.22472989195678272,
-      "grad_norm": 0.5600073337554932,
+      "grad_norm": 0.5662268400192261,
       "learning_rate": 4.586822776467208e-05,
-      "loss": 0.9312,
+      "loss": 0.9288,
       "num_input_tokens_seen": 38338560,
       "step": 4680
     },
     {
       "epoch": 0.22521008403361345,
-      "grad_norm": 0.5474848747253418,
+      "grad_norm": 0.5699325203895569,
       "learning_rate": 4.584638628953733e-05,
-      "loss": 0.9062,
+      "loss": 0.9058,
       "num_input_tokens_seen": 38420480,
       "step": 4690
     },
     {
       "epoch": 0.22569027611044418,
-      "grad_norm": 0.614006519317627,
+      "grad_norm": 0.6078118681907654,
       "learning_rate": 4.582449246592647e-05,
-      "loss": 0.8918,
+      "loss": 0.8911,
       "num_input_tokens_seen": 38502400,
       "step": 4700
     },
     {
       "epoch": 0.2261704681872749,
-      "grad_norm": 0.5042563080787659,
+      "grad_norm": 0.5221937894821167,
       "learning_rate": 4.5802546348818264e-05,
-      "loss": 0.8067,
+      "loss": 0.8064,
       "num_input_tokens_seen": 38584320,
       "step": 4710
     },
     {
       "epoch": 0.22665066026410563,
-      "grad_norm": 0.5609814524650574,
+      "grad_norm": 0.5463407635688782,
       "learning_rate": 4.578054799332277e-05,
-      "loss": 0.9678,
+      "loss": 0.9667,
       "num_input_tokens_seen": 38666240,
       "step": 4720
     },
     {
       "epoch": 0.22713085234093638,
-      "grad_norm": 0.4998452663421631,
+      "grad_norm": 0.49847063422203064,
       "learning_rate": 4.575849745468124e-05,
-      "loss": 1.0305,
+      "loss": 1.0298,
       "num_input_tokens_seen": 38748160,
       "step": 4730
     },
     {
       "epoch": 0.2276110444177671,
-      "grad_norm": 0.5019862651824951,
+      "grad_norm": 0.4975888729095459,
       "learning_rate": 4.573639478826596e-05,
-      "loss": 0.8708,
+      "loss": 0.87,
       "num_input_tokens_seen": 38830080,
       "step": 4740
     },
     {
       "epoch": 0.22809123649459784,
-      "grad_norm": 0.5069782733917236,
+      "grad_norm": 0.5155524611473083,
       "learning_rate": 4.571424004958012e-05,
-      "loss": 1.0366,
+      "loss": 1.0355,
       "num_input_tokens_seen": 38912000,
       "step": 4750
     },
     {
       "epoch": 0.22857142857142856,
-      "grad_norm": 0.5219393372535706,
+      "grad_norm": 0.5229736566543579,
       "learning_rate": 4.5692033294257666e-05,
-      "loss": 1.1274,
+      "loss": 1.1281,
       "num_input_tokens_seen": 38993920,
       "step": 4760
     },
     {
       "epoch": 0.2290516206482593,
-      "grad_norm": 0.5168918967247009,
+      "grad_norm": 0.5159292817115784,
       "learning_rate": 4.5669774578063174e-05,
-      "loss": 0.906,
+      "loss": 0.9067,
       "num_input_tokens_seen": 39075840,
       "step": 4770
     },
     {
       "epoch": 0.22953181272509005,
-      "grad_norm": 0.554415762424469,
+      "grad_norm": 0.5559686422348022,
       "learning_rate": 4.56474639568917e-05,
-      "loss": 0.9868,
+      "loss": 0.9853,
       "num_input_tokens_seen": 39157760,
       "step": 4780
     },
     {
       "epoch": 0.23001200480192077,
-      "grad_norm": 0.5875802040100098,
+      "grad_norm": 0.6078135371208191,
       "learning_rate": 4.5625101486768626e-05,
-      "loss": 0.9674,
+      "loss": 0.9676,
       "num_input_tokens_seen": 39239680,
       "step": 4790
     },
     {
       "epoch": 0.2304921968787515,
-      "grad_norm": 0.524535596370697,
+      "grad_norm": 0.5246491432189941,
       "learning_rate": 4.560268722384956e-05,
-      "loss": 0.944,
+      "loss": 0.9445,
       "num_input_tokens_seen": 39321600,
       "step": 4800
     },
     {
       "epoch": 0.23097238895558223,
-      "grad_norm": 0.5096153020858765,
+      "grad_norm": 0.4907456338405609,
       "learning_rate": 4.558022122442016e-05,
-      "loss": 0.9393,
+      "loss": 0.939,
       "num_input_tokens_seen": 39403520,
       "step": 4810
     },
     {
       "epoch": 0.23145258103241295,
-      "grad_norm": 0.527031421661377,
+      "grad_norm": 0.5344095230102539,
       "learning_rate": 4.555770354489598e-05,
-      "loss": 0.9572,
+      "loss": 0.9579,
       "num_input_tokens_seen": 39485440,
       "step": 4820
     },
     {
       "epoch": 0.2319327731092437,
-      "grad_norm": 0.4810847342014313,
+      "grad_norm": 0.4841211140155792,
       "learning_rate": 4.5535134241822394e-05,
       "loss": 0.8217,
       "num_input_tokens_seen": 39567360,
@@ -3874,551 +3874,551 @@
     },
     {
       "epoch": 0.23241296518607443,
-      "grad_norm": 0.4917358458042145,
+      "grad_norm": 0.4927588105201721,
       "learning_rate": 4.551251337187436e-05,
-      "loss": 0.8829,
+      "loss": 0.8828,
       "num_input_tokens_seen": 39649280,
       "step": 4840
     },
     {
       "epoch": 0.23289315726290516,
-      "grad_norm": 0.4697302281856537,
+      "grad_norm": 0.47400715947151184,
       "learning_rate": 4.548984099185638e-05,
-      "loss": 1.0275,
+      "loss": 1.0273,
       "num_input_tokens_seen": 39731200,
       "step": 4850
     },
     {
       "epoch": 0.23337334933973589,
-      "grad_norm": 0.7084998488426208,
+      "grad_norm": 0.5200524926185608,
       "learning_rate": 4.546711715870227e-05,
-      "loss": 0.8638,
+      "loss": 0.8631,
       "num_input_tokens_seen": 39813120,
       "step": 4860
     },
     {
       "epoch": 0.23385354141656664,
-      "grad_norm": 0.6179664134979248,
+      "grad_norm": 0.6111912131309509,
       "learning_rate": 4.5444341929475064e-05,
-      "loss": 0.9251,
+      "loss": 0.9249,
       "num_input_tokens_seen": 39895040,
       "step": 4870
     },
     {
       "epoch": 0.23433373349339737,
-      "grad_norm": 0.5251472592353821,
+      "grad_norm": 0.5096678733825684,
       "learning_rate": 4.5421515361366854e-05,
-      "loss": 1.362,
+      "loss": 1.3656,
       "num_input_tokens_seen": 39976960,
       "step": 4880
     },
     {
       "epoch": 0.2348139255702281,
-      "grad_norm": 0.566771924495697,
+      "grad_norm": 0.6361328959465027,
       "learning_rate": 4.5398637511698665e-05,
-      "loss": 0.6601,
+      "loss": 0.6605,
       "num_input_tokens_seen": 40058880,
       "step": 4890
     },
     {
       "epoch": 0.23529411764705882,
-      "grad_norm": 0.5199572443962097,
+      "grad_norm": 0.505113422870636,
       "learning_rate": 4.5375708437920284e-05,
-      "loss": 0.9545,
+      "loss": 0.9551,
       "num_input_tokens_seen": 40140800,
       "step": 4900
     },
     {
       "epoch": 0.23577430972388955,
-      "grad_norm": 0.4832773506641388,
+      "grad_norm": 0.47399622201919556,
       "learning_rate": 4.535272819761014e-05,
-      "loss": 1.0709,
+      "loss": 1.0701,
       "num_input_tokens_seen": 40222720,
       "step": 4910
     },
     {
       "epoch": 0.2362545018007203,
-      "grad_norm": 0.575834333896637,
+      "grad_norm": 0.5620723366737366,
       "learning_rate": 4.532969684847514e-05,
-      "loss": 1.0263,
+      "loss": 1.026,
       "num_input_tokens_seen": 40304640,
       "step": 4920
     },
     {
       "epoch": 0.23673469387755103,
-      "grad_norm": 0.6102591753005981,
+      "grad_norm": 0.5764886736869812,
       "learning_rate": 4.530661444835054e-05,
-      "loss": 0.9399,
+      "loss": 0.9408,
       "num_input_tokens_seen": 40386560,
       "step": 4930
     },
     {
       "epoch": 0.23721488595438175,
-      "grad_norm": 0.4413188099861145,
+      "grad_norm": 0.4281923174858093,
       "learning_rate": 4.5283481055199784e-05,
-      "loss": 1.0041,
+      "loss": 1.0037,
       "num_input_tokens_seen": 40468480,
       "step": 4940
     },
     {
       "epoch": 0.23769507803121248,
-      "grad_norm": 0.5925584435462952,
+      "grad_norm": 0.5836651921272278,
       "learning_rate": 4.526029672711437e-05,
-      "loss": 1.1321,
+      "loss": 1.1308,
       "num_input_tokens_seen": 40550400,
       "step": 4950
     },
     {
       "epoch": 0.2381752701080432,
-      "grad_norm": 1.5264989137649536,
+      "grad_norm": 1.5391772985458374,
       "learning_rate": 4.523706152231373e-05,
-      "loss": 0.9726,
+      "loss": 0.9735,
       "num_input_tokens_seen": 40632320,
       "step": 4960
     },
     {
       "epoch": 0.23865546218487396,
-      "grad_norm": 0.5782877206802368,
+      "grad_norm": 0.5564135909080505,
       "learning_rate": 4.5213775499145e-05,
-      "loss": 1.0489,
+      "loss": 1.0495,
       "num_input_tokens_seen": 40714240,
       "step": 4970
     },
     {
       "epoch": 0.2391356542617047,
-      "grad_norm": 0.48922982811927795,
+      "grad_norm": 0.49914756417274475,
       "learning_rate": 4.519043871608297e-05,
-      "loss": 0.7904,
+      "loss": 0.7898,
       "num_input_tokens_seen": 40796160,
       "step": 4980
     },
     {
       "epoch": 0.23961584633853542,
-      "grad_norm": 0.5028977990150452,
+      "grad_norm": 0.4968912899494171,
       "learning_rate": 4.5167051231729894e-05,
-      "loss": 0.8851,
+      "loss": 0.8855,
       "num_input_tokens_seen": 40878080,
       "step": 4990
     },
     {
       "epoch": 0.24009603841536614,
-      "grad_norm": 0.4978225529193878,
+      "grad_norm": 0.49146732687950134,
       "learning_rate": 4.514361310481533e-05,
-      "loss": 0.9244,
+      "loss": 0.9249,
       "num_input_tokens_seen": 40960000,
       "step": 5000
     },
     {
       "epoch": 0.24057623049219687,
-      "grad_norm": 0.6003936529159546,
+      "grad_norm": 0.597633957862854,
       "learning_rate": 4.512012439419601e-05,
-      "loss": 0.9562,
+      "loss": 0.9561,
       "num_input_tokens_seen": 41041920,
       "step": 5010
     },
     {
       "epoch": 0.24105642256902762,
-      "grad_norm": 0.5342166423797607,
+      "grad_norm": 0.5319405794143677,
       "learning_rate": 4.509658515885568e-05,
-      "loss": 1.0229,
+      "loss": 1.0238,
       "num_input_tokens_seen": 41123840,
       "step": 5020
     },
     {
       "epoch": 0.24153661464585835,
-      "grad_norm": 0.5202224850654602,
+      "grad_norm": 0.5182055234909058,
       "learning_rate": 4.5072995457904995e-05,
-      "loss": 0.8586,
+      "loss": 0.8584,
       "num_input_tokens_seen": 41205760,
       "step": 5030
     },
     {
       "epoch": 0.24201680672268908,
-      "grad_norm": 0.5373672842979431,
+      "grad_norm": 0.5302344560623169,
       "learning_rate": 4.50493553505813e-05,
-      "loss": 0.9188,
+      "loss": 0.9193,
       "num_input_tokens_seen": 41287680,
       "step": 5040
     },
     {
       "epoch": 0.2424969987995198,
-      "grad_norm": 0.5604919195175171,
+      "grad_norm": 0.5817127823829651,
       "learning_rate": 4.502566489624855e-05,
-      "loss": 1.041,
+      "loss": 1.0409,
       "num_input_tokens_seen": 41369600,
       "step": 5050
     },
     {
       "epoch": 0.24297719087635053,
-      "grad_norm": 0.4886492192745209,
+      "grad_norm": 0.48699209094047546,
       "learning_rate": 4.50019241543971e-05,
-      "loss": 0.9901,
+      "loss": 0.9899,
       "num_input_tokens_seen": 41451520,
       "step": 5060
     },
     {
       "epoch": 0.24345738295318128,
-      "grad_norm": 0.5549675822257996,
+      "grad_norm": 0.5494588017463684,
       "learning_rate": 4.4978133184643586e-05,
-      "loss": 0.9472,
+      "loss": 0.9473,
       "num_input_tokens_seen": 41533440,
       "step": 5070
     },
     {
       "epoch": 0.243937575030012,
-      "grad_norm": 0.4997415244579315,
+      "grad_norm": 0.49519428610801697,
       "learning_rate": 4.495429204673081e-05,
-      "loss": 0.9372,
+      "loss": 0.937,
       "num_input_tokens_seen": 41615360,
       "step": 5080
     },
     {
       "epoch": 0.24441776710684274,
-      "grad_norm": 0.5093867182731628,
+      "grad_norm": 0.5150887966156006,
       "learning_rate": 4.493040080052752e-05,
-      "loss": 0.8477,
+      "loss": 0.8483,
       "num_input_tokens_seen": 41697280,
       "step": 5090
     },
     {
       "epoch": 0.24489795918367346,
-      "grad_norm": 0.5322418808937073,
+      "grad_norm": 0.5311129093170166,
       "learning_rate": 4.49064595060283e-05,
-      "loss": 1.0359,
+      "loss": 1.0361,
       "num_input_tokens_seen": 41779200,
       "step": 5100
     },
     {
       "epoch": 0.2453781512605042,
-      "grad_norm": 0.485977441072464,
+      "grad_norm": 0.4963180124759674,
       "learning_rate": 4.488246822335341e-05,
-      "loss": 1.027,
+      "loss": 1.0274,
       "num_input_tokens_seen": 41861120,
       "step": 5110
     },
     {
       "epoch": 0.24585834333733494,
-      "grad_norm": 0.6202290058135986,
+      "grad_norm": 0.6075377464294434,
       "learning_rate": 4.485842701274865e-05,
-      "loss": 1.4544,
+      "loss": 1.4537,
       "num_input_tokens_seen": 41943040,
       "step": 5120
     },
     {
       "epoch": 0.24633853541416567,
-      "grad_norm": 0.5418204665184021,
+      "grad_norm": 0.48977187275886536,
       "learning_rate": 4.4834335934585194e-05,
-      "loss": 0.8381,
+      "loss": 0.8386,
       "num_input_tokens_seen": 42024960,
       "step": 5130
     },
     {
       "epoch": 0.2468187274909964,
-      "grad_norm": 0.5167099237442017,
+      "grad_norm": 0.5131157040596008,
       "learning_rate": 4.4810195049359435e-05,
-      "loss": 0.8446,
+      "loss": 0.8449,
       "num_input_tokens_seen": 42106880,
       "step": 5140
     },
     {
       "epoch": 0.24729891956782712,
-      "grad_norm": 0.5339356064796448,
+      "grad_norm": 0.5210369229316711,
       "learning_rate": 4.4786004417692836e-05,
-      "loss": 0.9213,
+      "loss": 0.9216,
       "num_input_tokens_seen": 42188800,
       "step": 5150
     },
     {
       "epoch": 0.24777911164465785,
-      "grad_norm": 0.5605190396308899,
+      "grad_norm": 0.575391411781311,
       "learning_rate": 4.4761764100331795e-05,
-      "loss": 1.0777,
+      "loss": 1.0769,
       "num_input_tokens_seen": 42270720,
       "step": 5160
     },
     {
       "epoch": 0.2482593037214886,
-      "grad_norm": 0.5921156406402588,
+      "grad_norm": 0.6365089416503906,
       "learning_rate": 4.473747415814747e-05,
-      "loss": 0.8808,
+      "loss": 0.8806,
       "num_input_tokens_seen": 42352640,
       "step": 5170
     },
     {
       "epoch": 0.24873949579831933,
-      "grad_norm": 0.6122276186943054,
+      "grad_norm": 0.6410951018333435,
       "learning_rate": 4.471313465213562e-05,
-      "loss": 1.0989,
+      "loss": 1.0999,
       "num_input_tokens_seen": 42434560,
       "step": 5180
     },
     {
       "epoch": 0.24921968787515006,
-      "grad_norm": 0.618924617767334,
+      "grad_norm": 0.5982769727706909,
       "learning_rate": 4.46887456434165e-05,
-      "loss": 0.8832,
+      "loss": 0.8831,
       "num_input_tokens_seen": 42516480,
       "step": 5190
     },
     {
       "epoch": 0.24969987995198079,
-      "grad_norm": 0.5166974067687988,
+      "grad_norm": 0.5119848251342773,
       "learning_rate": 4.466430719323465e-05,
-      "loss": 0.843,
+      "loss": 0.8431,
       "num_input_tokens_seen": 42598400,
       "step": 5200
     },
     {
       "epoch": 0.25018007202881154,
-      "grad_norm": 0.6116747856140137,
+      "grad_norm": 0.6475332975387573,
       "learning_rate": 4.463981936295876e-05,
-      "loss": 1.096,
+      "loss": 1.0951,
       "num_input_tokens_seen": 42680320,
       "step": 5210
     },
     {
       "epoch": 0.25066026410564224,
-      "grad_norm": 0.4793613851070404,
+      "grad_norm": 0.4759005010128021,
       "learning_rate": 4.461528221408153e-05,
-      "loss": 0.8411,
+      "loss": 0.8407,
       "num_input_tokens_seen": 42762240,
       "step": 5220
     },
     {
       "epoch": 0.251140456182473,
-      "grad_norm": 0.4967063069343567,
+      "grad_norm": 0.5037009716033936,
       "learning_rate": 4.459069580821953e-05,
-      "loss": 0.9483,
+      "loss": 0.9482,
       "num_input_tokens_seen": 42844160,
       "step": 5230
     },
     {
       "epoch": 0.25162064825930375,
-      "grad_norm": 0.4823471009731293,
+      "grad_norm": 0.4769095778465271,
       "learning_rate": 4.4566060207112983e-05,
-      "loss": 0.7803,
+      "loss": 0.7797,
       "num_input_tokens_seen": 42926080,
       "step": 5240
     },
     {
       "epoch": 0.25210084033613445,
-      "grad_norm": 0.5162050127983093,
+      "grad_norm": 0.5239166021347046,
       "learning_rate": 4.454137547262566e-05,
-      "loss": 0.9638,
+      "loss": 0.9646,
       "num_input_tokens_seen": 43008000,
       "step": 5250
     },
     {
       "epoch": 0.2525810324129652,
-      "grad_norm": 0.6449117660522461,
+      "grad_norm": 0.6174294352531433,
       "learning_rate": 4.451664166674472e-05,
-      "loss": 0.9499,
+      "loss": 0.9503,
       "num_input_tokens_seen": 43089920,
       "step": 5260
     },
     {
       "epoch": 0.2530612244897959,
-      "grad_norm": 0.5185222625732422,
+      "grad_norm": 0.5097090005874634,
       "learning_rate": 4.449185885158056e-05,
-      "loss": 0.8689,
+      "loss": 0.8684,
       "num_input_tokens_seen": 43171840,
       "step": 5270
     },
     {
       "epoch": 0.25354141656662665,
-      "grad_norm": 0.5826950669288635,
+      "grad_norm": 0.5547761917114258,
       "learning_rate": 4.4467027089366625e-05,
-      "loss": 0.9125,
+      "loss": 0.9119,
       "num_input_tokens_seen": 43253760,
       "step": 5280
     },
     {
       "epoch": 0.2540216086434574,
-      "grad_norm": 0.6201736330986023,
+      "grad_norm": 0.6190077662467957,
       "learning_rate": 4.444214644245928e-05,
-      "loss": 0.9452,
+      "loss": 0.9446,
       "num_input_tokens_seen": 43335680,
       "step": 5290
     },
     {
       "epoch": 0.2545018007202881,
-      "grad_norm": 0.5122745633125305,
+      "grad_norm": 0.5137811303138733,
       "learning_rate": 4.441721697333765e-05,
-      "loss": 0.8625,
+      "loss": 0.8618,
       "num_input_tokens_seen": 43417600,
       "step": 5300
     },
     {
       "epoch": 0.25498199279711886,
-      "grad_norm": 0.5290250182151794,
+      "grad_norm": 0.5213572382926941,
       "learning_rate": 4.4392238744603464e-05,
-      "loss": 0.8692,
+      "loss": 0.8695,
       "num_input_tokens_seen": 43499520,
       "step": 5310
     },
     {
       "epoch": 0.25546218487394956,
-      "grad_norm": 0.719904363155365,
+      "grad_norm": 0.7135751843452454,
       "learning_rate": 4.436721181898088e-05,
-      "loss": 1.0077,
+      "loss": 1.0076,
       "num_input_tokens_seen": 43581440,
       "step": 5320
     },
     {
       "epoch": 0.2559423769507803,
-      "grad_norm": 0.5093972086906433,
+      "grad_norm": 0.4903344511985779,
       "learning_rate": 4.434213625931636e-05,
-      "loss": 0.9048,
+      "loss": 0.9059,
       "num_input_tokens_seen": 43663360,
       "step": 5330
     },
     {
       "epoch": 0.25642256902761107,
-      "grad_norm": 0.5186138153076172,
+      "grad_norm": 0.5091985464096069,
       "learning_rate": 4.431701212857847e-05,
-      "loss": 0.9174,
+      "loss": 0.9168,
       "num_input_tokens_seen": 43745280,
       "step": 5340
     },
     {
       "epoch": 0.25690276110444177,
-      "grad_norm": 0.5094157457351685,
+      "grad_norm": 0.5019556879997253,
       "learning_rate": 4.429183948985777e-05,
-      "loss": 0.7608,
+      "loss": 0.7605,
       "num_input_tokens_seen": 43827200,
       "step": 5350
     },
     {
       "epoch": 0.2573829531812725,
-      "grad_norm": 0.5557414889335632,
+      "grad_norm": 0.5584927201271057,
       "learning_rate": 4.426661840636662e-05,
-      "loss": 0.8841,
+      "loss": 0.8832,
       "num_input_tokens_seen": 43909120,
       "step": 5360
     },
     {
       "epoch": 0.2578631452581032,
-      "grad_norm": 0.49979162216186523,
+      "grad_norm": 0.5135828852653503,
       "learning_rate": 4.424134894143903e-05,
-      "loss": 1.0389,
+      "loss": 1.0383,
       "num_input_tokens_seen": 43991040,
       "step": 5370
     },
     {
       "epoch": 0.258343337334934,
-      "grad_norm": 0.6658138036727905,
+      "grad_norm": 0.6649297475814819,
       "learning_rate": 4.42160311585305e-05,
-      "loss": 1.0486,
+      "loss": 1.0498,
       "num_input_tokens_seen": 44072960,
       "step": 5380
     },
     {
       "epoch": 0.25882352941176473,
-      "grad_norm": 0.5858299732208252,
+      "grad_norm": 0.6021849513053894,
       "learning_rate": 4.419066512121788e-05,
-      "loss": 0.6945,
+      "loss": 0.6946,
       "num_input_tokens_seen": 44154880,
       "step": 5390
     },
     {
       "epoch": 0.25930372148859543,
-      "grad_norm": 0.5159766674041748,
+      "grad_norm": 0.5068604946136475,
       "learning_rate": 4.4165250893199176e-05,
-      "loss": 0.8972,
+      "loss": 0.898,
       "num_input_tokens_seen": 44236800,
       "step": 5400
     },
     {
       "epoch": 0.2597839135654262,
-      "grad_norm": 0.48362815380096436,
+      "grad_norm": 0.4720013439655304,
       "learning_rate": 4.413978853829342e-05,
-      "loss": 0.8574,
+      "loss": 0.8575,
       "num_input_tokens_seen": 44318720,
       "step": 5410
     },
     {
       "epoch": 0.2602641056422569,
-      "grad_norm": 0.5605273842811584,
+      "grad_norm": 0.5253759622573853,
       "learning_rate": 4.411427812044049e-05,
-      "loss": 0.9521,
+      "loss": 0.953,
       "num_input_tokens_seen": 44400640,
       "step": 5420
     },
     {
       "epoch": 0.26074429771908764,
-      "grad_norm": 0.5811429619789124,
+      "grad_norm": 0.5874719023704529,
       "learning_rate": 4.408871970370096e-05,
-      "loss": 1.0423,
+      "loss": 1.0408,
       "num_input_tokens_seen": 44482560,
       "step": 5430
     },
     {
       "epoch": 0.2612244897959184,
-      "grad_norm": 0.7415308952331543,
+      "grad_norm": 0.8202329874038696,
       "learning_rate": 4.406311335225595e-05,
-      "loss": 0.9938,
+      "loss": 0.9933,
       "num_input_tokens_seen": 44564480,
       "step": 5440
     },
     {
       "epoch": 0.2617046818727491,
-      "grad_norm": 0.6823397874832153,
+      "grad_norm": 0.6459318995475769,
       "learning_rate": 4.4037459130406923e-05,
-      "loss": 0.8559,
+      "loss": 0.8562,
       "num_input_tokens_seen": 44646400,
       "step": 5450
     },
     {
       "epoch": 0.26218487394957984,
-      "grad_norm": 0.5368736386299133,
+      "grad_norm": 0.5401067733764648,
       "learning_rate": 4.401175710257558e-05,
-      "loss": 0.9533,
+      "loss": 0.9538,
       "num_input_tokens_seen": 44728320,
       "step": 5460
     },
     {
       "epoch": 0.26266506602641054,
-      "grad_norm": 0.5195080637931824,
+      "grad_norm": 0.515424370765686,
       "learning_rate": 4.398600733330365e-05,
-      "loss": 0.8958,
+      "loss": 0.8955,
       "num_input_tokens_seen": 44810240,
       "step": 5470
     },
     {
       "epoch": 0.2631452581032413,
-      "grad_norm": 0.5098795890808105,
+      "grad_norm": 0.5169644951820374,
       "learning_rate": 4.3960209887252766e-05,
-      "loss": 1.1406,
+      "loss": 1.1404,
       "num_input_tokens_seen": 44892160,
       "step": 5480
     },
     {
       "epoch": 0.26362545018007205,
-      "grad_norm": 0.5225921273231506,
+      "grad_norm": 0.5176170468330383,
       "learning_rate": 4.3934364829204265e-05,
-      "loss": 1.0363,
+      "loss": 1.0361,
       "num_input_tokens_seen": 44974080,
       "step": 5490
     },
     {
       "epoch": 0.26410564225690275,
-      "grad_norm": 0.5075384378433228,
+      "grad_norm": 0.520937442779541,
       "learning_rate": 4.3908472224059064e-05,
-      "loss": 0.876,
+      "loss": 0.8763,
       "num_input_tokens_seen": 45056000,
       "step": 5500
     },
     {
       "epoch": 0.2645858343337335,
-      "grad_norm": 0.5186338424682617,
+      "grad_norm": 0.5109109878540039,
       "learning_rate": 4.388253213683747e-05,
-      "loss": 0.8846,
+      "loss": 0.8841,
       "num_input_tokens_seen": 45137920,
       "step": 5510
     },
     {
       "epoch": 0.2650660264105642,
-      "grad_norm": 0.5157744884490967,
+      "grad_norm": 0.525878369808197,
       "learning_rate": 4.385654463267901e-05,
       "loss": 0.841,
       "num_input_tokens_seen": 45219840,
@@ -4426,127 +4426,127 @@
     },
     {
       "epoch": 0.26554621848739496,
-      "grad_norm": 0.49746039509773254,
+      "grad_norm": 0.49905362725257874,
       "learning_rate": 4.383050977684231e-05,
-      "loss": 1.1015,
+      "loss": 1.1014,
       "num_input_tokens_seen": 45301760,
       "step": 5530
     },
     {
       "epoch": 0.2660264105642257,
-      "grad_norm": 0.5038164258003235,
+      "grad_norm": 0.5063333511352539,
       "learning_rate": 4.3804427634704885e-05,
-      "loss": 0.7614,
+      "loss": 0.7611,
       "num_input_tokens_seen": 45383680,
       "step": 5540
     },
     {
       "epoch": 0.2665066026410564,
-      "grad_norm": 0.4780644476413727,
+      "grad_norm": 0.48698627948760986,
       "learning_rate": 4.3778298271762995e-05,
-      "loss": 0.8541,
+      "loss": 0.854,
       "num_input_tokens_seen": 45465600,
       "step": 5550
     },
     {
       "epoch": 0.26698679471788717,
-      "grad_norm": 0.49261558055877686,
+      "grad_norm": 0.4989083409309387,
       "learning_rate": 4.375212175363149e-05,
-      "loss": 0.9464,
+      "loss": 0.9457,
       "num_input_tokens_seen": 45547520,
       "step": 5560
     },
     {
       "epoch": 0.26746698679471786,
-      "grad_norm": 0.256661593914032,
+      "grad_norm": 0.2386409342288971,
       "learning_rate": 4.372589814604362e-05,
-      "loss": 0.8231,
+      "loss": 0.8236,
       "num_input_tokens_seen": 45629440,
       "step": 5570
     },
     {
       "epoch": 0.2679471788715486,
-      "grad_norm": 0.7495452761650085,
+      "grad_norm": 0.7750229239463806,
       "learning_rate": 4.369962751485089e-05,
-      "loss": 0.8247,
+      "loss": 0.8256,
       "num_input_tokens_seen": 45711360,
       "step": 5580
     },
     {
       "epoch": 0.2684273709483794,
-      "grad_norm": 0.6870756149291992,
+      "grad_norm": 0.8927270174026489,
       "learning_rate": 4.367330992602289e-05,
-      "loss": 0.7793,
+      "loss": 0.7786,
       "num_input_tokens_seen": 45793280,
       "step": 5590
     },
     {
       "epoch": 0.2689075630252101,
-      "grad_norm": 0.5133636593818665,
+      "grad_norm": 0.5122426748275757,
       "learning_rate": 4.3646945445647114e-05,
-      "loss": 0.7808,
+      "loss": 0.7805,
       "num_input_tokens_seen": 45875200,
       "step": 5600
     },
     {
       "epoch": 0.2693877551020408,
-      "grad_norm": 0.48444893956184387,
+      "grad_norm": 0.47932419180870056,
       "learning_rate": 4.362053413992883e-05,
-      "loss": 1.023,
+      "loss": 1.0227,
       "num_input_tokens_seen": 45957120,
       "step": 5610
     },
     {
       "epoch": 0.2698679471788715,
-      "grad_norm": 0.4823382794857025,
+      "grad_norm": 0.47927728295326233,
       "learning_rate": 4.359407607519088e-05,
-      "loss": 0.7344,
+      "loss": 0.7345,
       "num_input_tokens_seen": 46039040,
       "step": 5620
     },
     {
       "epoch": 0.2703481392557023,
-      "grad_norm": 0.4958154261112213,
+      "grad_norm": 0.5311817526817322,
       "learning_rate": 4.356757131787353e-05,
-      "loss": 0.9653,
+      "loss": 0.9656,
       "num_input_tokens_seen": 46120960,
       "step": 5630
     },
     {
       "epoch": 0.27082833133253303,
-      "grad_norm": 2.4843976497650146,
+      "grad_norm": 2.499256134033203,
       "learning_rate": 4.354101993453429e-05,
-      "loss": 0.8851,
+      "loss": 0.8838,
       "num_input_tokens_seen": 46202880,
       "step": 5640
     },
     {
       "epoch": 0.27130852340936373,
-      "grad_norm": 0.5063963532447815,
+      "grad_norm": 0.5061020851135254,
       "learning_rate": 4.3514421991847746e-05,
-      "loss": 0.8417,
+      "loss": 0.8414,
       "num_input_tokens_seen": 46284800,
       "step": 5650
     },
     {
       "epoch": 0.2717887154861945,
-      "grad_norm": 0.5009209513664246,
+      "grad_norm": 0.49775904417037964,
       "learning_rate": 4.3487777556605446e-05,
-      "loss": 0.8548,
+      "loss": 0.8546,
       "num_input_tokens_seen": 46366720,
       "step": 5660
     },
     {
       "epoch": 0.2722689075630252,
-      "grad_norm": 0.7282830476760864,
+      "grad_norm": 0.7386242747306824,
       "learning_rate": 4.3461086695715625e-05,
-      "loss": 1.0355,
+      "loss": 1.0363,
       "num_input_tokens_seen": 46448640,
       "step": 5670
     },
     {
       "epoch": 0.27274909963985594,
-      "grad_norm": 0.5339224934577942,
+      "grad_norm": 0.5468425750732422,
       "learning_rate": 4.343434947620315e-05,
       "loss": 0.799,
       "num_input_tokens_seen": 46530560,
@@ -4554,15 +4554,15 @@
     },
     {
       "epoch": 0.2732292917166867,
-      "grad_norm": 0.5233862400054932,
+      "grad_norm": 0.5276323556900024,
       "learning_rate": 4.340756596520929e-05,
-      "loss": 0.9427,
+      "loss": 0.9422,
       "num_input_tokens_seen": 46612480,
       "step": 5690
     },
     {
       "epoch": 0.2737094837935174,
-      "grad_norm": 0.487689733505249,
+      "grad_norm": 0.4901086688041687,
       "learning_rate": 4.338073622999154e-05,
       "loss": 0.904,
       "num_input_tokens_seen": 46694400,
@@ -4570,31 +4570,31 @@
     },
     {
       "epoch": 0.27418967587034815,
-      "grad_norm": 0.5029451251029968,
+      "grad_norm": 0.5071278214454651,
       "learning_rate": 4.335386033792347e-05,
-      "loss": 1.2054,
+      "loss": 1.2078,
       "num_input_tokens_seen": 46776320,
       "step": 5710
     },
     {
       "epoch": 0.27466986794717885,
-      "grad_norm": 0.5146957635879517,
+      "grad_norm": 0.5222769379615784,
       "learning_rate": 4.332693835649461e-05,
-      "loss": 0.8869,
+      "loss": 0.887,
       "num_input_tokens_seen": 46858240,
       "step": 5720
     },
     {
       "epoch": 0.2751500600240096,
-      "grad_norm": 0.5206642150878906,
+      "grad_norm": 0.5225254893302917,
       "learning_rate": 4.329997035331015e-05,
-      "loss": 0.9947,
+      "loss": 0.9954,
       "num_input_tokens_seen": 46940160,
       "step": 5730
     },
     {
       "epoch": 0.27563025210084036,
-      "grad_norm": 0.5068042278289795,
+      "grad_norm": 0.5147578716278076,
       "learning_rate": 4.3272956396090906e-05,
       "loss": 0.8496,
       "num_input_tokens_seen": 47022080,
@@ -4602,31 +4602,31 @@
     },
     {
       "epoch": 0.27611044417767105,
-      "grad_norm": 0.5032111406326294,
+      "grad_norm": 0.5081062316894531,
       "learning_rate": 4.324589655267306e-05,
-      "loss": 0.9618,
+      "loss": 0.9616,
       "num_input_tokens_seen": 47104000,
       "step": 5750
     },
     {
       "epoch": 0.2765906362545018,
-      "grad_norm": 0.6475026607513428,
+      "grad_norm": 0.6242536306381226,
       "learning_rate": 4.321879089100805e-05,
-      "loss": 1.065,
+      "loss": 1.0649,
       "num_input_tokens_seen": 47185920,
       "step": 5760
     },
     {
       "epoch": 0.2770708283313325,
-      "grad_norm": 0.9474783539772034,
+      "grad_norm": 0.9209274053573608,
       "learning_rate": 4.319163947916234e-05,
-      "loss": 0.9001,
+      "loss": 0.9002,
       "num_input_tokens_seen": 47267840,
       "step": 5770
     },
     {
       "epoch": 0.27755102040816326,
-      "grad_norm": 0.48900583386421204,
+      "grad_norm": 0.49128785729408264,
       "learning_rate": 4.316444238531729e-05,
       "loss": 0.9649,
       "num_input_tokens_seen": 47349760,
@@ -4634,263 +4634,263 @@
     },
     {
       "epoch": 0.278031212484994,
-      "grad_norm": 0.5249919891357422,
+      "grad_norm": 0.5176426768302917,
       "learning_rate": 4.313719967776899e-05,
-      "loss": 0.9746,
+      "loss": 0.9747,
       "num_input_tokens_seen": 47431680,
       "step": 5790
     },
     {
       "epoch": 0.2785114045618247,
-      "grad_norm": 0.5116479992866516,
+      "grad_norm": 0.5055992603302002,
       "learning_rate": 4.310991142492805e-05,
-      "loss": 1.0193,
+      "loss": 1.0223,
       "num_input_tokens_seen": 47513600,
       "step": 5800
     },
     {
       "epoch": 0.27899159663865547,
-      "grad_norm": 0.4895400404930115,
+      "grad_norm": 0.4892292022705078,
       "learning_rate": 4.308257769531947e-05,
-      "loss": 0.9313,
+      "loss": 0.9318,
       "num_input_tokens_seen": 47595520,
       "step": 5810
     },
     {
       "epoch": 0.27947178871548617,
-      "grad_norm": 0.5116353034973145,
+      "grad_norm": 0.51258385181427,
       "learning_rate": 4.3055198557582445e-05,
-      "loss": 1.0058,
+      "loss": 1.0057,
       "num_input_tokens_seen": 47677440,
       "step": 5820
     },
     {
       "epoch": 0.2799519807923169,
-      "grad_norm": 0.6602368354797363,
+      "grad_norm": 0.6507702469825745,
       "learning_rate": 4.3027774080470174e-05,
-      "loss": 1.1261,
+      "loss": 1.126,
       "num_input_tokens_seen": 47759360,
       "step": 5830
     },
     {
       "epoch": 0.2804321728691477,
-      "grad_norm": 0.5150447487831116,
+      "grad_norm": 0.5292423367500305,
       "learning_rate": 4.300030433284974e-05,
-      "loss": 0.8111,
+      "loss": 0.8118,
       "num_input_tokens_seen": 47841280,
       "step": 5840
     },
     {
       "epoch": 0.2809123649459784,
-      "grad_norm": 0.49057701230049133,
+      "grad_norm": 0.499051958322525,
       "learning_rate": 4.29727893837019e-05,
-      "loss": 0.7262,
+      "loss": 0.7263,
       "num_input_tokens_seen": 47923200,
       "step": 5850
     },
     {
       "epoch": 0.28139255702280913,
-      "grad_norm": 0.5138354897499084,
+      "grad_norm": 0.5214549899101257,
       "learning_rate": 4.294522930212091e-05,
-      "loss": 0.809,
+      "loss": 0.8095,
       "num_input_tokens_seen": 48005120,
       "step": 5860
     },
     {
       "epoch": 0.28187274909963983,
-      "grad_norm": 0.5236837267875671,
+      "grad_norm": 0.5287784337997437,
       "learning_rate": 4.291762415731437e-05,
-      "loss": 0.9162,
+      "loss": 0.9164,
       "num_input_tokens_seen": 48087040,
       "step": 5870
     },
     {
       "epoch": 0.2823529411764706,
-      "grad_norm": 0.5018298625946045,
+      "grad_norm": 0.5037810802459717,
       "learning_rate": 4.288997401860303e-05,
-      "loss": 1.0343,
+      "loss": 1.0349,
       "num_input_tokens_seen": 48168960,
       "step": 5880
     },
     {
       "epoch": 0.28283313325330134,
-      "grad_norm": 0.4813729524612427,
+      "grad_norm": 0.48096582293510437,
       "learning_rate": 4.286227895542064e-05,
-      "loss": 1.36,
+      "loss": 1.3606,
       "num_input_tokens_seen": 48250880,
       "step": 5890
     },
     {
       "epoch": 0.28331332533013204,
-      "grad_norm": 0.44931960105895996,
+      "grad_norm": 0.4305453300476074,
       "learning_rate": 4.283453903731375e-05,
-      "loss": 0.8472,
+      "loss": 0.8476,
       "num_input_tokens_seen": 48332800,
       "step": 5900
     },
     {
       "epoch": 0.2837935174069628,
-      "grad_norm": 0.5551528930664062,
+      "grad_norm": 0.5652824640274048,
       "learning_rate": 4.2806754333941546e-05,
-      "loss": 1.1295,
+      "loss": 1.1286,
       "num_input_tokens_seen": 48414720,
       "step": 5910
     },
     {
       "epoch": 0.2842737094837935,
-      "grad_norm": 0.6064106822013855,
+      "grad_norm": 0.587988018989563,
       "learning_rate": 4.2778924915075704e-05,
-      "loss": 0.9477,
+      "loss": 0.9471,
       "num_input_tokens_seen": 48496640,
       "step": 5920
     },
     {
       "epoch": 0.28475390156062425,
-      "grad_norm": 0.48876288533210754,
+      "grad_norm": 0.48125171661376953,
       "learning_rate": 4.275105085060014e-05,
-      "loss": 0.93,
+      "loss": 0.9296,
       "num_input_tokens_seen": 48578560,
       "step": 5930
     },
     {
       "epoch": 0.285234093637455,
-      "grad_norm": 0.4730078876018524,
+      "grad_norm": 0.4742906093597412,
       "learning_rate": 4.272313221051094e-05,
-      "loss": 1.0398,
+      "loss": 1.0404,
       "num_input_tokens_seen": 48660480,
       "step": 5940
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 0.40384843945503235,
+      "grad_norm": 0.4079858064651489,
       "learning_rate": 4.269516906491607e-05,
-      "loss": 0.9514,
+      "loss": 0.9517,
       "num_input_tokens_seen": 48742400,
       "step": 5950
     },
     {
       "epoch": 0.28619447779111645,
-      "grad_norm": 0.5116029381752014,
+      "grad_norm": 0.5222288966178894,
       "learning_rate": 4.266716148403529e-05,
-      "loss": 0.7283,
+      "loss": 0.7284,
       "num_input_tokens_seen": 48824320,
       "step": 5960
     },
     {
       "epoch": 0.28667466986794715,
-      "grad_norm": 0.5653484463691711,
+      "grad_norm": 0.5701093673706055,
       "learning_rate": 4.263910953819993e-05,
-      "loss": 0.9808,
+      "loss": 0.98,
       "num_input_tokens_seen": 48906240,
       "step": 5970
     },
     {
       "epoch": 0.2871548619447779,
-      "grad_norm": 0.4929262697696686,
+      "grad_norm": 0.48784807324409485,
       "learning_rate": 4.2611013297852744e-05,
-      "loss": 0.9758,
+      "loss": 0.9762,
       "num_input_tokens_seen": 48988160,
       "step": 5980
     },
     {
       "epoch": 0.28763505402160866,
-      "grad_norm": 0.5140427947044373,
+      "grad_norm": 0.5211770534515381,
       "learning_rate": 4.2582872833547693e-05,
-      "loss": 0.9226,
+      "loss": 0.9233,
       "num_input_tokens_seen": 49070080,
       "step": 5990
     },
     {
       "epoch": 0.28811524609843936,
-      "grad_norm": 0.4957669973373413,
+      "grad_norm": 0.49390462040901184,
       "learning_rate": 4.255468821594981e-05,
-      "loss": 1.0878,
+      "loss": 1.0877,
       "num_input_tokens_seen": 49152000,
       "step": 6000
     },
     {
       "epoch": 0.2885954381752701,
-      "grad_norm": 0.5024486184120178,
+      "grad_norm": 0.5107914209365845,
       "learning_rate": 4.2526459515834996e-05,
-      "loss": 0.917,
+      "loss": 0.9164,
       "num_input_tokens_seen": 49233920,
       "step": 6010
     },
     {
       "epoch": 0.28907563025210087,
-      "grad_norm": 0.5399232506752014,
+      "grad_norm": 0.5360170006752014,
       "learning_rate": 4.249818680408984e-05,
-      "loss": 1.0098,
+      "loss": 1.01,
       "num_input_tokens_seen": 49315840,
       "step": 6020
     },
     {
       "epoch": 0.28955582232893157,
-      "grad_norm": 0.5091607570648193,
+      "grad_norm": 0.5045222043991089,
       "learning_rate": 4.246987015171148e-05,
-      "loss": 1.0573,
+      "loss": 1.057,
       "num_input_tokens_seen": 49397760,
       "step": 6030
     },
     {
       "epoch": 0.2900360144057623,
-      "grad_norm": 0.48257771134376526,
+      "grad_norm": 0.48786187171936035,
       "learning_rate": 4.244150962980735e-05,
-      "loss": 1.1285,
+      "loss": 1.1284,
       "num_input_tokens_seen": 49479680,
       "step": 6040
     },
     {
       "epoch": 0.290516206482593,
-      "grad_norm": 0.7823185324668884,
+      "grad_norm": 0.7572475075721741,
       "learning_rate": 4.2413105309595105e-05,
-      "loss": 0.9014,
+      "loss": 0.9011,
       "num_input_tokens_seen": 49561600,
       "step": 6050
     },
     {
       "epoch": 0.2909963985594238,
-      "grad_norm": 0.49261143803596497,
+      "grad_norm": 0.48458683490753174,
       "learning_rate": 4.238465726240233e-05,
-      "loss": 0.8398,
+      "loss": 0.8397,
       "num_input_tokens_seen": 49643520,
       "step": 6060
     },
     {
       "epoch": 0.29147659063625453,
-      "grad_norm": 0.5321176648139954,
+      "grad_norm": 0.5217098593711853,
       "learning_rate": 4.235616555966645e-05,
-      "loss": 0.879,
+      "loss": 0.8792,
       "num_input_tokens_seen": 49725440,
       "step": 6070
     },
     {
       "epoch": 0.29195678271308523,
-      "grad_norm": 0.5565729737281799,
+      "grad_norm": 0.5514656901359558,
       "learning_rate": 4.232763027293451e-05,
-      "loss": 0.7561,
+      "loss": 0.7543,
       "num_input_tokens_seen": 49807360,
       "step": 6080
     },
     {
       "epoch": 0.292436974789916,
-      "grad_norm": 0.6039222478866577,
+      "grad_norm": 0.6352956891059875,
       "learning_rate": 4.2299051473862976e-05,
-      "loss": 1.0367,
+      "loss": 1.0364,
       "num_input_tokens_seen": 49889280,
       "step": 6090
     },
     {
       "epoch": 0.2929171668667467,
-      "grad_norm": 0.38333648443222046,
+      "grad_norm": 0.42271411418914795,
       "learning_rate": 4.227042923421762e-05,
-      "loss": 0.979,
+      "loss": 0.9792,
       "num_input_tokens_seen": 49971200,
       "step": 6100
     },
     {
       "epoch": 0.29339735894357744,
-      "grad_norm": 0.5529670119285583,
+      "grad_norm": 0.5425634980201721,
       "learning_rate": 4.224176362587326e-05,
       "loss": 0.9951,
       "num_input_tokens_seen": 50053120,
@@ -4898,31 +4898,31 @@
     },
     {
       "epoch": 0.2938775510204082,
-      "grad_norm": 0.6076828241348267,
+      "grad_norm": 0.6409218907356262,
       "learning_rate": 4.221305472081365e-05,
-      "loss": 0.8707,
+      "loss": 0.8709,
       "num_input_tokens_seen": 50135040,
       "step": 6120
     },
     {
       "epoch": 0.2943577430972389,
-      "grad_norm": 0.49703189730644226,
+      "grad_norm": 0.4920530617237091,
       "learning_rate": 4.2184302591131264e-05,
-      "loss": 0.9571,
+      "loss": 0.9572,
       "num_input_tokens_seen": 50216960,
       "step": 6130
     },
     {
       "epoch": 0.29483793517406964,
-      "grad_norm": 0.5342078804969788,
+      "grad_norm": 0.5330979228019714,
       "learning_rate": 4.21555073090271e-05,
-      "loss": 1.0041,
+      "loss": 1.0038,
       "num_input_tokens_seen": 50298880,
       "step": 6140
     },
     {
       "epoch": 0.29531812725090034,
-      "grad_norm": 0.6617572903633118,
+      "grad_norm": 0.5953214168548584,
       "learning_rate": 4.2126668946810545e-05,
       "loss": 1.1052,
       "num_input_tokens_seen": 50380800,
@@ -4930,55 +4930,55 @@
     },
     {
       "epoch": 0.2957983193277311,
-      "grad_norm": 0.5139254331588745,
+      "grad_norm": 0.5212881565093994,
       "learning_rate": 4.2097787576899144e-05,
-      "loss": 0.7878,
+      "loss": 0.788,
       "num_input_tokens_seen": 50462720,
       "step": 6160
     },
     {
       "epoch": 0.29627851140456185,
-      "grad_norm": 0.5757784247398376,
+      "grad_norm": 0.5825909972190857,
       "learning_rate": 4.2068863271818455e-05,
-      "loss": 0.8359,
+      "loss": 0.8364,
       "num_input_tokens_seen": 50544640,
       "step": 6170
     },
     {
       "epoch": 0.29675870348139255,
-      "grad_norm": 0.6105610132217407,
+      "grad_norm": 0.5860311985015869,
       "learning_rate": 4.2039896104201844e-05,
-      "loss": 0.7906,
+      "loss": 0.7908,
       "num_input_tokens_seen": 50626560,
       "step": 6180
     },
     {
       "epoch": 0.2972388955582233,
-      "grad_norm": 0.5291628241539001,
+      "grad_norm": 0.5145933032035828,
       "learning_rate": 4.201088614679032e-05,
-      "loss": 0.8226,
+      "loss": 0.8223,
       "num_input_tokens_seen": 50708480,
       "step": 6190
     },
     {
       "epoch": 0.297719087635054,
-      "grad_norm": 0.5184271335601807,
+      "grad_norm": 0.520693302154541,
       "learning_rate": 4.198183347243233e-05,
-      "loss": 0.922,
+      "loss": 0.9225,
       "num_input_tokens_seen": 50790400,
       "step": 6200
     },
     {
       "epoch": 0.29819927971188476,
-      "grad_norm": 0.5568735599517822,
+      "grad_norm": 0.5607156157493591,
       "learning_rate": 4.1952738154083614e-05,
-      "loss": 0.8489,
+      "loss": 0.8484,
       "num_input_tokens_seen": 50872320,
       "step": 6210
     },
     {
       "epoch": 0.2986794717887155,
-      "grad_norm": 0.5584334135055542,
+      "grad_norm": 0.5793859362602234,
       "learning_rate": 4.1923600264806975e-05,
       "loss": 0.9279,
       "num_input_tokens_seen": 50954240,
@@ -4986,7 +4986,7 @@
     },
     {
       "epoch": 0.2991596638655462,
-      "grad_norm": 1.1234581470489502,
+      "grad_norm": 0.9440059065818787,
       "learning_rate": 4.189441987777212e-05,
       "loss": 0.8793,
       "num_input_tokens_seen": 51036160,
@@ -4994,111 +4994,111 @@
     },
     {
       "epoch": 0.29963985594237696,
-      "grad_norm": 0.5137265920639038,
+      "grad_norm": 0.542341947555542,
       "learning_rate": 4.186519706625549e-05,
-      "loss": 0.8057,
+      "loss": 0.809,
       "num_input_tokens_seen": 51118080,
       "step": 6240
     },
     {
       "epoch": 0.30012004801920766,
-      "grad_norm": 0.5186243057250977,
+      "grad_norm": 0.5198069214820862,
       "learning_rate": 4.1835931903640046e-05,
-      "loss": 1.0145,
+      "loss": 1.015,
       "num_input_tokens_seen": 51200000,
       "step": 6250
     },
     {
       "epoch": 0.3006002400960384,
-      "grad_norm": 0.5995609164237976,
+      "grad_norm": 0.5908818244934082,
       "learning_rate": 4.180662446341511e-05,
-      "loss": 0.9712,
+      "loss": 0.971,
       "num_input_tokens_seen": 51281920,
       "step": 6260
     },
     {
       "epoch": 0.3010804321728692,
-      "grad_norm": 0.5310963988304138,
+      "grad_norm": 0.5370185971260071,
       "learning_rate": 4.1777274819176154e-05,
-      "loss": 0.9031,
+      "loss": 0.9027,
       "num_input_tokens_seen": 51363840,
       "step": 6270
     },
     {
       "epoch": 0.30156062424969987,
-      "grad_norm": 0.536044716835022,
+      "grad_norm": 0.5164900422096252,
       "learning_rate": 4.1747883044624644e-05,
-      "loss": 0.9543,
+      "loss": 0.9546,
       "num_input_tokens_seen": 51445760,
       "step": 6280
     },
     {
       "epoch": 0.3020408163265306,
-      "grad_norm": 0.5187882781028748,
+      "grad_norm": 0.5132122039794922,
       "learning_rate": 4.171844921356784e-05,
-      "loss": 1.0411,
+      "loss": 1.041,
       "num_input_tokens_seen": 51527680,
       "step": 6290
     },
     {
       "epoch": 0.3025210084033613,
-      "grad_norm": 0.526749312877655,
+      "grad_norm": 0.5190037488937378,
       "learning_rate": 4.168897339991861e-05,
-      "loss": 1.0103,
+      "loss": 1.0105,
       "num_input_tokens_seen": 51609600,
       "step": 6300
     },
     {
       "epoch": 0.3030012004801921,
-      "grad_norm": 0.5082546472549438,
+      "grad_norm": 0.5217801332473755,
       "learning_rate": 4.1659455677695245e-05,
-      "loss": 1.0368,
+      "loss": 1.0367,
       "num_input_tokens_seen": 51691520,
       "step": 6310
     },
     {
       "epoch": 0.30348139255702283,
-      "grad_norm": 0.513116180896759,
+      "grad_norm": 0.4742312729358673,
       "learning_rate": 4.162989612102128e-05,
-      "loss": 0.6589,
+      "loss": 0.6583,
       "num_input_tokens_seen": 51773440,
       "step": 6320
     },
     {
       "epoch": 0.30396158463385353,
-      "grad_norm": 0.5046403408050537,
+      "grad_norm": 0.5095861554145813,
       "learning_rate": 4.160029480412529e-05,
-      "loss": 0.9936,
+      "loss": 0.9942,
       "num_input_tokens_seen": 51855360,
       "step": 6330
     },
     {
       "epoch": 0.3044417767106843,
-      "grad_norm": 0.6154318451881409,
+      "grad_norm": 0.6195290088653564,
       "learning_rate": 4.1570651801340735e-05,
-      "loss": 0.8154,
+      "loss": 0.8157,
       "num_input_tokens_seen": 51937280,
       "step": 6340
     },
     {
       "epoch": 0.304921968787515,
-      "grad_norm": 0.5283302664756775,
+      "grad_norm": 0.5015403032302856,
       "learning_rate": 4.1540967187105753e-05,
-      "loss": 1.0798,
+      "loss": 1.0806,
       "num_input_tokens_seen": 52019200,
       "step": 6350
     },
     {
       "epoch": 0.30540216086434574,
-      "grad_norm": 0.5067933797836304,
+      "grad_norm": 0.5029851794242859,
       "learning_rate": 4.151124103596295e-05,
-      "loss": 0.8155,
+      "loss": 0.815,
       "num_input_tokens_seen": 52101120,
       "step": 6360
     },
     {
       "epoch": 0.3058823529411765,
-      "grad_norm": 0.4994664788246155,
+      "grad_norm": 0.4925341308116913,
       "learning_rate": 4.148147342255926e-05,
       "loss": 0.9705,
       "num_input_tokens_seen": 52183040,
@@ -5106,95 +5106,95 @@
     },
     {
       "epoch": 0.3063625450180072,
-      "grad_norm": 0.4871218800544739,
+      "grad_norm": 0.4862985610961914,
       "learning_rate": 4.145166442164573e-05,
-      "loss": 0.849,
+      "loss": 0.8492,
       "num_input_tokens_seen": 52264960,
       "step": 6380
     },
     {
       "epoch": 0.30684273709483795,
-      "grad_norm": 0.5769742727279663,
+      "grad_norm": 0.5730807781219482,
       "learning_rate": 4.142181410807735e-05,
-      "loss": 0.9561,
+      "loss": 0.9553,
       "num_input_tokens_seen": 52346880,
       "step": 6390
     },
     {
       "epoch": 0.30732292917166865,
-      "grad_norm": 0.8338456153869629,
+      "grad_norm": 0.8465107083320618,
       "learning_rate": 4.1391922556812815e-05,
-      "loss": 0.9117,
+      "loss": 0.9121,
       "num_input_tokens_seen": 52428800,
       "step": 6400
     },
     {
       "epoch": 0.3078031212484994,
-      "grad_norm": 0.521043062210083,
+      "grad_norm": 0.5357550382614136,
       "learning_rate": 4.136198984291442e-05,
-      "loss": 0.8657,
+      "loss": 0.8654,
       "num_input_tokens_seen": 52510720,
       "step": 6410
     },
     {
       "epoch": 0.30828331332533015,
-      "grad_norm": 0.4783630073070526,
+      "grad_norm": 0.47538501024246216,
       "learning_rate": 4.133201604154779e-05,
-      "loss": 0.9356,
+      "loss": 0.9387,
       "num_input_tokens_seen": 52592640,
       "step": 6420
     },
     {
       "epoch": 0.30876350540216085,
-      "grad_norm": 0.6084679961204529,
+      "grad_norm": 0.6129695773124695,
       "learning_rate": 4.1302001227981765e-05,
-      "loss": 1.0838,
+      "loss": 1.084,
       "num_input_tokens_seen": 52674560,
       "step": 6430
     },
     {
       "epoch": 0.3092436974789916,
-      "grad_norm": 0.5003706812858582,
+      "grad_norm": 0.5037487149238586,
       "learning_rate": 4.1271945477588126e-05,
-      "loss": 0.7471,
+      "loss": 0.7468,
       "num_input_tokens_seen": 52756480,
       "step": 6440
     },
     {
       "epoch": 0.3097238895558223,
-      "grad_norm": 0.5811577439308167,
+      "grad_norm": 0.5857306122779846,
       "learning_rate": 4.124184886584148e-05,
-      "loss": 0.9374,
+      "loss": 0.9378,
       "num_input_tokens_seen": 52838400,
       "step": 6450
     },
     {
       "epoch": 0.31020408163265306,
-      "grad_norm": 0.5085806250572205,
+      "grad_norm": 0.5044699311256409,
       "learning_rate": 4.121171146831905e-05,
-      "loss": 1.0186,
+      "loss": 1.0187,
       "num_input_tokens_seen": 52920320,
       "step": 6460
     },
     {
       "epoch": 0.3106842737094838,
-      "grad_norm": 0.5559006333351135,
+      "grad_norm": 0.5601541996002197,
       "learning_rate": 4.118153336070045e-05,
-      "loss": 1.1422,
+      "loss": 1.1429,
       "num_input_tokens_seen": 53002240,
       "step": 6470
     },
     {
       "epoch": 0.3111644657863145,
-      "grad_norm": 0.7885730266571045,
+      "grad_norm": 0.7909272909164429,
       "learning_rate": 4.115131461876756e-05,
-      "loss": 1.0219,
+      "loss": 1.0201,
       "num_input_tokens_seen": 53084160,
       "step": 6480
     },
     {
       "epoch": 0.31164465786314527,
-      "grad_norm": 0.5197670459747314,
+      "grad_norm": 0.5243809223175049,
       "learning_rate": 4.1121055318404264e-05,
       "loss": 0.9806,
       "num_input_tokens_seen": 53166080,
@@ -5202,455 +5202,455 @@
     },
     {
       "epoch": 0.31212484993997597,
-      "grad_norm": 0.4977589249610901,
+      "grad_norm": 0.4948495328426361,
       "learning_rate": 4.109075553559633e-05,
-      "loss": 1.1074,
+      "loss": 1.1109,
       "num_input_tokens_seen": 53248000,
       "step": 6500
     },
     {
       "epoch": 0.3126050420168067,
-      "grad_norm": 0.46243059635162354,
+      "grad_norm": 0.4655405879020691,
       "learning_rate": 4.1060415346431134e-05,
-      "loss": 0.93,
+      "loss": 0.9303,
       "num_input_tokens_seen": 53329920,
       "step": 6510
     },
     {
       "epoch": 0.3130852340936375,
-      "grad_norm": 0.6155077219009399,
+      "grad_norm": 0.621785581111908,
       "learning_rate": 4.103003482709758e-05,
-      "loss": 0.9669,
+      "loss": 0.9677,
       "num_input_tokens_seen": 53411840,
       "step": 6520
     },
     {
       "epoch": 0.3135654261704682,
-      "grad_norm": 0.6652863025665283,
+      "grad_norm": 0.6690912246704102,
       "learning_rate": 4.0999614053885795e-05,
-      "loss": 0.7738,
+      "loss": 0.7746,
       "num_input_tokens_seen": 53493760,
       "step": 6530
     },
     {
       "epoch": 0.31404561824729893,
-      "grad_norm": 0.538733184337616,
+      "grad_norm": 0.5284923911094666,
       "learning_rate": 4.096915310318702e-05,
-      "loss": 0.8493,
+      "loss": 0.8491,
       "num_input_tokens_seen": 53575680,
       "step": 6540
     },
     {
       "epoch": 0.31452581032412963,
-      "grad_norm": 0.5088375210762024,
+      "grad_norm": 0.5016783475875854,
       "learning_rate": 4.093865205149337e-05,
-      "loss": 0.9464,
+      "loss": 0.9473,
       "num_input_tokens_seen": 53657600,
       "step": 6550
     },
     {
       "epoch": 0.3150060024009604,
-      "grad_norm": 0.5455616116523743,
+      "grad_norm": 0.5532281398773193,
       "learning_rate": 4.090811097539768e-05,
-      "loss": 0.7853,
+      "loss": 0.7859,
       "num_input_tokens_seen": 53739520,
       "step": 6560
     },
     {
       "epoch": 0.31548619447779114,
-      "grad_norm": 0.5154474377632141,
+      "grad_norm": 0.47948014736175537,
       "learning_rate": 4.087752995159327e-05,
-      "loss": 0.764,
+      "loss": 0.7647,
       "num_input_tokens_seen": 53821440,
       "step": 6570
     },
     {
       "epoch": 0.31596638655462184,
-      "grad_norm": 0.5074660181999207,
+      "grad_norm": 0.5162191390991211,
       "learning_rate": 4.084690905687379e-05,
-      "loss": 0.9871,
+      "loss": 0.988,
       "num_input_tokens_seen": 53903360,
       "step": 6580
     },
     {
       "epoch": 0.3164465786314526,
-      "grad_norm": 0.510628879070282,
+      "grad_norm": 0.529799222946167,
       "learning_rate": 4.0816248368133016e-05,
-      "loss": 1.1464,
+      "loss": 1.148,
       "num_input_tokens_seen": 53985280,
       "step": 6590
     },
     {
       "epoch": 0.3169267707082833,
-      "grad_norm": 0.6401832699775696,
+      "grad_norm": 0.701400876045227,
       "learning_rate": 4.078554796236462e-05,
-      "loss": 0.7681,
+      "loss": 0.7685,
       "num_input_tokens_seen": 54067200,
       "step": 6600
     },
     {
       "epoch": 0.31740696278511404,
-      "grad_norm": 0.5684659481048584,
+      "grad_norm": 0.5586780309677124,
       "learning_rate": 4.0754807916662055e-05,
-      "loss": 1.0146,
+      "loss": 1.0151,
       "num_input_tokens_seen": 54149120,
       "step": 6610
     },
     {
       "epoch": 0.3178871548619448,
-      "grad_norm": 0.5699182152748108,
+      "grad_norm": 0.5554243922233582,
       "learning_rate": 4.072402830821829e-05,
-      "loss": 0.9229,
+      "loss": 0.9232,
       "num_input_tokens_seen": 54231040,
       "step": 6620
     },
     {
       "epoch": 0.3183673469387755,
-      "grad_norm": 0.5109874606132507,
+      "grad_norm": 0.5143104195594788,
       "learning_rate": 4.069320921432564e-05,
-      "loss": 0.9067,
+      "loss": 0.907,
       "num_input_tokens_seen": 54312960,
       "step": 6630
     },
     {
       "epoch": 0.31884753901560625,
-      "grad_norm": 0.498054563999176,
+      "grad_norm": 0.49600183963775635,
       "learning_rate": 4.066235071237559e-05,
-      "loss": 0.8371,
+      "loss": 0.837,
       "num_input_tokens_seen": 54394880,
       "step": 6640
     },
     {
       "epoch": 0.31932773109243695,
-      "grad_norm": 0.4968014061450958,
+      "grad_norm": 0.5070446729660034,
       "learning_rate": 4.0631452879858565e-05,
-      "loss": 0.9809,
+      "loss": 0.9814,
       "num_input_tokens_seen": 54476800,
       "step": 6650
     },
     {
       "epoch": 0.3198079231692677,
-      "grad_norm": 0.49730050563812256,
+      "grad_norm": 0.5054400563240051,
       "learning_rate": 4.0600515794363774e-05,
-      "loss": 0.9101,
+      "loss": 0.9107,
       "num_input_tokens_seen": 54558720,
       "step": 6660
     },
     {
       "epoch": 0.32028811524609846,
-      "grad_norm": 0.5172863006591797,
+      "grad_norm": 0.5172276496887207,
       "learning_rate": 4.0569539533578985e-05,
-      "loss": 0.9802,
+      "loss": 0.9817,
       "num_input_tokens_seen": 54640640,
       "step": 6670
     },
     {
       "epoch": 0.32076830732292916,
-      "grad_norm": 0.5139868259429932,
+      "grad_norm": 0.5226202607154846,
       "learning_rate": 4.053852417529035e-05,
-      "loss": 0.7294,
+      "loss": 0.7322,
       "num_input_tokens_seen": 54722560,
       "step": 6680
     },
     {
       "epoch": 0.3212484993997599,
-      "grad_norm": 0.5264905691146851,
+      "grad_norm": 0.5340245962142944,
       "learning_rate": 4.050746979738218e-05,
-      "loss": 0.9096,
+      "loss": 0.9097,
       "num_input_tokens_seen": 54804480,
       "step": 6690
     },
     {
       "epoch": 0.3217286914765906,
-      "grad_norm": 0.5231399536132812,
+      "grad_norm": 0.5202445387840271,
       "learning_rate": 4.047637647783681e-05,
-      "loss": 1.0167,
+      "loss": 1.0164,
       "num_input_tokens_seen": 54886400,
       "step": 6700
     },
     {
       "epoch": 0.32220888355342137,
-      "grad_norm": 0.6286957859992981,
+      "grad_norm": 0.5725679993629456,
       "learning_rate": 4.044524429473431e-05,
-      "loss": 0.8882,
+      "loss": 0.888,
       "num_input_tokens_seen": 54968320,
       "step": 6710
     },
     {
       "epoch": 0.3226890756302521,
-      "grad_norm": 0.5430505275726318,
+      "grad_norm": 0.5457696914672852,
       "learning_rate": 4.041407332625238e-05,
-      "loss": 0.8349,
+      "loss": 0.8352,
       "num_input_tokens_seen": 55050240,
       "step": 6720
     },
     {
       "epoch": 0.3231692677070828,
-      "grad_norm": 0.5100399851799011,
+      "grad_norm": 0.5226444602012634,
       "learning_rate": 4.038286365066613e-05,
-      "loss": 0.8427,
+      "loss": 0.843,
       "num_input_tokens_seen": 55132160,
       "step": 6730
     },
     {
       "epoch": 0.3236494597839136,
-      "grad_norm": 0.48778000473976135,
+      "grad_norm": 0.48937085270881653,
       "learning_rate": 4.0351615346347804e-05,
-      "loss": 0.9561,
+      "loss": 0.9569,
       "num_input_tokens_seen": 55214080,
       "step": 6740
     },
     {
       "epoch": 0.3241296518607443,
-      "grad_norm": 0.5565239191055298,
+      "grad_norm": 0.5608055591583252,
       "learning_rate": 4.032032849176673e-05,
-      "loss": 0.9102,
+      "loss": 0.912,
       "num_input_tokens_seen": 55296000,
       "step": 6750
     },
     {
       "epoch": 0.324609843937575,
-      "grad_norm": 0.486905962228775,
+      "grad_norm": 0.524291455745697,
       "learning_rate": 4.0289003165488976e-05,
-      "loss": 0.8668,
+      "loss": 0.8669,
       "num_input_tokens_seen": 55377920,
       "step": 6760
     },
     {
       "epoch": 0.3250900360144058,
-      "grad_norm": 0.5659222602844238,
+      "grad_norm": 0.570642352104187,
       "learning_rate": 4.025763944617727e-05,
-      "loss": 0.9355,
+      "loss": 0.9358,
       "num_input_tokens_seen": 55459840,
       "step": 6770
     },
     {
       "epoch": 0.3255702280912365,
-      "grad_norm": 0.4450863301753998,
+      "grad_norm": 0.4495159387588501,
       "learning_rate": 4.0226237412590696e-05,
-      "loss": 0.9062,
+      "loss": 0.9055,
       "num_input_tokens_seen": 55541760,
       "step": 6780
     },
     {
       "epoch": 0.32605042016806723,
-      "grad_norm": 0.4924563467502594,
+      "grad_norm": 0.414616197347641,
       "learning_rate": 4.019479714358461e-05,
-      "loss": 0.8195,
+      "loss": 0.8201,
       "num_input_tokens_seen": 55623680,
       "step": 6790
     },
     {
       "epoch": 0.32653061224489793,
-      "grad_norm": 0.5010157227516174,
+      "grad_norm": 0.49962684512138367,
       "learning_rate": 4.016331871811033e-05,
-      "loss": 0.7515,
+      "loss": 0.7518,
       "num_input_tokens_seen": 55705600,
       "step": 6800
     },
     {
       "epoch": 0.3270108043217287,
-      "grad_norm": 0.49570614099502563,
+      "grad_norm": 0.4985042214393616,
       "learning_rate": 4.0131802215215025e-05,
-      "loss": 1.0522,
+      "loss": 1.0517,
       "num_input_tokens_seen": 55787520,
       "step": 6810
     },
     {
       "epoch": 0.32749099639855944,
-      "grad_norm": 0.7191053628921509,
+      "grad_norm": 0.7347798347473145,
       "learning_rate": 4.010024771404147e-05,
-      "loss": 0.9683,
+      "loss": 0.9678,
       "num_input_tokens_seen": 55869440,
       "step": 6820
     },
     {
       "epoch": 0.32797118847539014,
-      "grad_norm": 0.5185818672180176,
+      "grad_norm": 0.5248961448669434,
       "learning_rate": 4.006865529382787e-05,
-      "loss": 0.8884,
+      "loss": 0.8882,
       "num_input_tokens_seen": 55951360,
       "step": 6830
     },
     {
       "epoch": 0.3284513805522209,
-      "grad_norm": 0.6259596347808838,
+      "grad_norm": 0.6377902030944824,
       "learning_rate": 4.0037025033907635e-05,
-      "loss": 0.9416,
+      "loss": 0.9419,
       "num_input_tokens_seen": 56033280,
       "step": 6840
     },
     {
       "epoch": 0.3289315726290516,
-      "grad_norm": 0.5207152962684631,
+      "grad_norm": 0.5339779853820801,
       "learning_rate": 4.000535701370921e-05,
-      "loss": 0.9646,
+      "loss": 0.9656,
       "num_input_tokens_seen": 56115200,
       "step": 6850
     },
     {
       "epoch": 0.32941176470588235,
-      "grad_norm": 0.6173118948936462,
+      "grad_norm": 0.624208390712738,
       "learning_rate": 3.997365131275584e-05,
-      "loss": 0.9449,
+      "loss": 0.9448,
       "num_input_tokens_seen": 56197120,
       "step": 6860
     },
     {
       "epoch": 0.3298919567827131,
-      "grad_norm": 0.5023009777069092,
+      "grad_norm": 0.5037301778793335,
       "learning_rate": 3.994190801066542e-05,
-      "loss": 1.1115,
+      "loss": 1.1117,
       "num_input_tokens_seen": 56279040,
       "step": 6870
     },
     {
       "epoch": 0.3303721488595438,
-      "grad_norm": 0.7239778637886047,
+      "grad_norm": 0.7410263419151306,
       "learning_rate": 3.9910127187150246e-05,
-      "loss": 0.996,
+      "loss": 0.9973,
       "num_input_tokens_seen": 56360960,
       "step": 6880
     },
     {
       "epoch": 0.33085234093637456,
-      "grad_norm": 0.45797932147979736,
+      "grad_norm": 0.4599323868751526,
       "learning_rate": 3.9878308922016846e-05,
-      "loss": 0.887,
+      "loss": 0.8873,
       "num_input_tokens_seen": 56442880,
       "step": 6890
     },
     {
       "epoch": 0.33133253301320525,
-      "grad_norm": 0.594767689704895,
+      "grad_norm": 0.6016721129417419,
       "learning_rate": 3.984645329516578e-05,
-      "loss": 0.803,
+      "loss": 0.8034,
       "num_input_tokens_seen": 56524800,
       "step": 6900
     },
     {
       "epoch": 0.331812725090036,
-      "grad_norm": 0.502420961856842,
+      "grad_norm": 0.5103434324264526,
       "learning_rate": 3.98145603865914e-05,
-      "loss": 0.9339,
+      "loss": 0.9337,
       "num_input_tokens_seen": 56606720,
       "step": 6910
     },
     {
       "epoch": 0.33229291716686676,
-      "grad_norm": 0.4855603873729706,
+      "grad_norm": 0.4790404438972473,
       "learning_rate": 3.978263027638171e-05,
-      "loss": 1.0942,
+      "loss": 1.0943,
       "num_input_tokens_seen": 56688640,
       "step": 6920
     },
     {
       "epoch": 0.33277310924369746,
-      "grad_norm": 0.561661422252655,
+      "grad_norm": 0.5801796913146973,
       "learning_rate": 3.975066304471811e-05,
-      "loss": 0.9728,
+      "loss": 0.9727,
       "num_input_tokens_seen": 56770560,
       "step": 6930
     },
     {
       "epoch": 0.3332533013205282,
-      "grad_norm": 0.5083943009376526,
+      "grad_norm": 0.506883442401886,
       "learning_rate": 3.971865877187523e-05,
-      "loss": 0.8131,
+      "loss": 0.8136,
       "num_input_tokens_seen": 56852480,
       "step": 6940
     },
     {
       "epoch": 0.33373349339735897,
-      "grad_norm": 0.5498092770576477,
+      "grad_norm": 0.5405831336975098,
       "learning_rate": 3.968661753822071e-05,
-      "loss": 0.8142,
+      "loss": 0.8136,
       "num_input_tokens_seen": 56934400,
       "step": 6950
     },
     {
       "epoch": 0.33421368547418967,
-      "grad_norm": 0.5083803534507751,
+      "grad_norm": 0.5131149888038635,
       "learning_rate": 3.9654539424214996e-05,
-      "loss": 0.9086,
+      "loss": 0.9095,
       "num_input_tokens_seen": 57016320,
       "step": 6960
     },
     {
       "epoch": 0.3346938775510204,
-      "grad_norm": 0.4920550286769867,
+      "grad_norm": 0.4872545897960663,
       "learning_rate": 3.962242451041118e-05,
-      "loss": 0.8183,
+      "loss": 0.8184,
       "num_input_tokens_seen": 57098240,
       "step": 6970
     },
     {
       "epoch": 0.3351740696278511,
-      "grad_norm": 0.5124842524528503,
+      "grad_norm": 0.5294336080551147,
       "learning_rate": 3.9590272877454714e-05,
-      "loss": 0.9817,
+      "loss": 0.9821,
       "num_input_tokens_seen": 57180160,
       "step": 6980
     },
     {
       "epoch": 0.3356542617046819,
-      "grad_norm": 0.4871028959751129,
+      "grad_norm": 0.500049352645874,
       "learning_rate": 3.955808460608331e-05,
-      "loss": 1.0946,
+      "loss": 1.0956,
       "num_input_tokens_seen": 57262080,
       "step": 6990
     },
     {
       "epoch": 0.33613445378151263,
-      "grad_norm": 0.49150577187538147,
+      "grad_norm": 0.5052493810653687,
       "learning_rate": 3.952585977712664e-05,
-      "loss": 1.0473,
+      "loss": 1.0477,
       "num_input_tokens_seen": 57344000,
       "step": 7000
     },
     {
       "epoch": 0.33661464585834333,
-      "grad_norm": 0.5065931677818298,
+      "grad_norm": 0.49708953499794006,
       "learning_rate": 3.94935984715062e-05,
-      "loss": 0.9402,
+      "loss": 0.9404,
       "num_input_tokens_seen": 57425920,
       "step": 7010
     },
     {
       "epoch": 0.3370948379351741,
-      "grad_norm": 0.5077062249183655,
+      "grad_norm": 0.5013074278831482,
       "learning_rate": 3.9461300770235093e-05,
-      "loss": 0.678,
+      "loss": 0.6775,
       "num_input_tokens_seen": 57507840,
       "step": 7020
     },
     {
       "epoch": 0.3375750300120048,
-      "grad_norm": 0.4949451684951782,
+      "grad_norm": 0.4998069405555725,
       "learning_rate": 3.942896675441779e-05,
-      "loss": 0.9907,
+      "loss": 0.9909,
       "num_input_tokens_seen": 57589760,
       "step": 7030
     },
     {
       "epoch": 0.33805522208883554,
-      "grad_norm": 0.7275571823120117,
+      "grad_norm": 0.7399186491966248,
       "learning_rate": 3.939659650524997e-05,
-      "loss": 0.9791,
+      "loss": 0.9797,
       "num_input_tokens_seen": 57671680,
       "step": 7040
     },
     {
       "epoch": 0.3385354141656663,
-      "grad_norm": 0.4286993145942688,
+      "grad_norm": 0.4175257980823517,
       "learning_rate": 3.9364190104018307e-05,
-      "loss": 0.8087,
+      "loss": 0.8085,
       "num_input_tokens_seen": 57753600,
       "step": 7050
     },
     {
       "epoch": 0.339015606242497,
-      "grad_norm": 0.4871419668197632,
+      "grad_norm": 0.47881293296813965,
       "learning_rate": 3.933174763210024e-05,
       "loss": 0.9574,
       "num_input_tokens_seen": 57835520,
@@ -5658,303 +5658,303 @@
     },
     {
       "epoch": 0.33949579831932775,
-      "grad_norm": 0.5074589252471924,
+      "grad_norm": 0.49739906191825867,
       "learning_rate": 3.9299269170963795e-05,
-      "loss": 0.866,
+      "loss": 0.8655,
       "num_input_tokens_seen": 57917440,
       "step": 7070
     },
     {
       "epoch": 0.33997599039615845,
-      "grad_norm": 0.5097850561141968,
+      "grad_norm": 0.5103303790092468,
       "learning_rate": 3.926675480216738e-05,
-      "loss": 0.9768,
+      "loss": 0.9769,
       "num_input_tokens_seen": 57999360,
       "step": 7080
     },
     {
       "epoch": 0.3404561824729892,
-      "grad_norm": 0.4989878833293915,
+      "grad_norm": 0.4884282350540161,
       "learning_rate": 3.923420460735957e-05,
-      "loss": 0.9659,
+      "loss": 0.9666,
       "num_input_tokens_seen": 58081280,
       "step": 7090
     },
     {
       "epoch": 0.34093637454981995,
-      "grad_norm": 0.4922942817211151,
+      "grad_norm": 0.4890122711658478,
       "learning_rate": 3.920161866827889e-05,
-      "loss": 0.7939,
+      "loss": 0.7937,
       "num_input_tokens_seen": 58163200,
       "step": 7100
     },
     {
       "epoch": 0.34141656662665065,
-      "grad_norm": 0.5052926540374756,
+      "grad_norm": 0.5014382600784302,
       "learning_rate": 3.916899706675365e-05,
-      "loss": 1.1513,
+      "loss": 1.1521,
       "num_input_tokens_seen": 58245120,
       "step": 7110
     },
     {
       "epoch": 0.3418967587034814,
-      "grad_norm": 0.5977290272712708,
+      "grad_norm": 0.6726990342140198,
       "learning_rate": 3.913633988470169e-05,
-      "loss": 1.0614,
+      "loss": 1.0604,
       "num_input_tokens_seen": 58327040,
       "step": 7120
     },
     {
       "epoch": 0.3423769507803121,
-      "grad_norm": 0.45560044050216675,
+      "grad_norm": 0.4604361653327942,
       "learning_rate": 3.91036472041302e-05,
-      "loss": 1.0288,
+      "loss": 1.0293,
       "num_input_tokens_seen": 58408960,
       "step": 7130
     },
     {
       "epoch": 0.34285714285714286,
-      "grad_norm": 0.5187392234802246,
+      "grad_norm": 0.5115752816200256,
       "learning_rate": 3.907091910713553e-05,
-      "loss": 0.9384,
+      "loss": 0.9386,
       "num_input_tokens_seen": 58490880,
       "step": 7140
     },
     {
       "epoch": 0.3433373349339736,
-      "grad_norm": 0.49635642766952515,
+      "grad_norm": 0.5059329867362976,
       "learning_rate": 3.9038155675902956e-05,
-      "loss": 0.9873,
+      "loss": 0.9874,
       "num_input_tokens_seen": 58572800,
       "step": 7150
     },
     {
       "epoch": 0.3438175270108043,
-      "grad_norm": 0.47836464643478394,
+      "grad_norm": 0.47620999813079834,
       "learning_rate": 3.900535699270647e-05,
-      "loss": 0.9565,
+      "loss": 0.9545,
       "num_input_tokens_seen": 58654720,
       "step": 7160
     },
     {
       "epoch": 0.34429771908763507,
-      "grad_norm": 0.49842721223831177,
+      "grad_norm": 0.49813467264175415,
       "learning_rate": 3.8972523139908616e-05,
-      "loss": 0.9251,
+      "loss": 0.9244,
       "num_input_tokens_seen": 58736640,
       "step": 7170
     },
     {
       "epoch": 0.34477791116446577,
-      "grad_norm": 0.5168570876121521,
+      "grad_norm": 0.5196046829223633,
       "learning_rate": 3.8939654199960244e-05,
-      "loss": 0.9024,
+      "loss": 0.9025,
       "num_input_tokens_seen": 58818560,
       "step": 7180
     },
     {
       "epoch": 0.3452581032412965,
-      "grad_norm": 0.508305549621582,
+      "grad_norm": 0.5014814734458923,
       "learning_rate": 3.890675025540028e-05,
-      "loss": 0.8464,
+      "loss": 0.8459,
       "num_input_tokens_seen": 58900480,
       "step": 7190
     },
     {
       "epoch": 0.3457382953181273,
-      "grad_norm": 0.5206389427185059,
+      "grad_norm": 0.5205827355384827,
       "learning_rate": 3.8873811388855605e-05,
-      "loss": 0.909,
+      "loss": 0.9075,
       "num_input_tokens_seen": 58982400,
       "step": 7200
     },
     {
       "epoch": 0.346218487394958,
-      "grad_norm": 0.5759819149971008,
+      "grad_norm": 0.555486798286438,
       "learning_rate": 3.8840837683040766e-05,
-      "loss": 1.0237,
+      "loss": 1.0231,
       "num_input_tokens_seen": 59064320,
       "step": 7210
     },
     {
       "epoch": 0.34669867947178873,
-      "grad_norm": 0.5078734755516052,
+      "grad_norm": 0.5133519768714905,
       "learning_rate": 3.880782922075778e-05,
-      "loss": 0.8262,
+      "loss": 0.8264,
       "num_input_tokens_seen": 59146240,
       "step": 7220
     },
     {
       "epoch": 0.3471788715486194,
-      "grad_norm": 0.9097881317138672,
+      "grad_norm": 0.9706603288650513,
       "learning_rate": 3.8774786084896e-05,
-      "loss": 0.9193,
+      "loss": 0.9188,
       "num_input_tokens_seen": 59228160,
       "step": 7230
     },
     {
       "epoch": 0.3476590636254502,
-      "grad_norm": 0.4915110468864441,
+      "grad_norm": 0.5008965730667114,
       "learning_rate": 3.8741708358431774e-05,
-      "loss": 0.8414,
+      "loss": 0.841,
       "num_input_tokens_seen": 59310080,
       "step": 7240
     },
     {
       "epoch": 0.34813925570228094,
-      "grad_norm": 0.4847305119037628,
+      "grad_norm": 0.4682840406894684,
       "learning_rate": 3.870859612442837e-05,
-      "loss": 1.138,
+      "loss": 1.1382,
       "num_input_tokens_seen": 59392000,
       "step": 7250
     },
     {
       "epoch": 0.34861944777911164,
-      "grad_norm": 0.5575107336044312,
+      "grad_norm": 0.5504831075668335,
       "learning_rate": 3.86754494660357e-05,
-      "loss": 0.8703,
+      "loss": 0.8716,
       "num_input_tokens_seen": 59473920,
       "step": 7260
     },
     {
       "epoch": 0.3490996398559424,
-      "grad_norm": 0.5898226499557495,
+      "grad_norm": 0.5717173218727112,
       "learning_rate": 3.864226846649008e-05,
-      "loss": 0.8408,
+      "loss": 0.8413,
       "num_input_tokens_seen": 59555840,
       "step": 7270
     },
     {
       "epoch": 0.3495798319327731,
-      "grad_norm": 0.5211881995201111,
+      "grad_norm": 0.5430110692977905,
       "learning_rate": 3.860905320911413e-05,
-      "loss": 0.9104,
+      "loss": 0.9106,
       "num_input_tokens_seen": 59637760,
       "step": 7280
     },
     {
       "epoch": 0.35006002400960384,
-      "grad_norm": 0.5230681896209717,
+      "grad_norm": 0.5178146958351135,
       "learning_rate": 3.857580377731644e-05,
-      "loss": 0.7804,
+      "loss": 0.7808,
       "num_input_tokens_seen": 59719680,
       "step": 7290
     },
     {
       "epoch": 0.3505402160864346,
-      "grad_norm": 0.5010928511619568,
+      "grad_norm": 0.5012259483337402,
       "learning_rate": 3.854252025459144e-05,
-      "loss": 0.868,
+      "loss": 0.8681,
       "num_input_tokens_seen": 59801600,
       "step": 7300
     },
     {
       "epoch": 0.3510204081632653,
-      "grad_norm": 0.4810640811920166,
+      "grad_norm": 0.5083165764808655,
       "learning_rate": 3.8509202724519165e-05,
-      "loss": 0.9105,
+      "loss": 0.9087,
       "num_input_tokens_seen": 59883520,
       "step": 7310
     },
     {
       "epoch": 0.35150060024009605,
-      "grad_norm": 0.5282043814659119,
+      "grad_norm": 0.5158397555351257,
       "learning_rate": 3.8475851270765054e-05,
-      "loss": 0.8274,
+      "loss": 0.8279,
       "num_input_tokens_seen": 59965440,
       "step": 7320
     },
     {
       "epoch": 0.35198079231692675,
-      "grad_norm": 0.5089632868766785,
+      "grad_norm": 0.5039446949958801,
       "learning_rate": 3.844246597707972e-05,
-      "loss": 0.9016,
+      "loss": 0.9013,
       "num_input_tokens_seen": 60047360,
       "step": 7330
     },
     {
       "epoch": 0.3524609843937575,
-      "grad_norm": 0.4965154230594635,
+      "grad_norm": 0.5005216598510742,
       "learning_rate": 3.8409046927298755e-05,
-      "loss": 0.8791,
+      "loss": 0.879,
       "num_input_tokens_seen": 60129280,
       "step": 7340
     },
     {
       "epoch": 0.35294117647058826,
-      "grad_norm": 0.5418551564216614,
+      "grad_norm": 0.5384024977684021,
       "learning_rate": 3.8375594205342534e-05,
-      "loss": 1.1453,
+      "loss": 1.1444,
       "num_input_tokens_seen": 60211200,
       "step": 7350
     },
     {
       "epoch": 0.35342136854741896,
-      "grad_norm": 0.5690187215805054,
+      "grad_norm": 0.5492764711380005,
       "learning_rate": 3.834210789521598e-05,
-      "loss": 1.0104,
+      "loss": 1.0091,
       "num_input_tokens_seen": 60293120,
       "step": 7360
     },
     {
       "epoch": 0.3539015606242497,
-      "grad_norm": 0.5610761046409607,
+      "grad_norm": 0.5844592452049255,
       "learning_rate": 3.830858808100834e-05,
-      "loss": 1.0196,
+      "loss": 1.0194,
       "num_input_tokens_seen": 60375040,
       "step": 7370
     },
     {
       "epoch": 0.3543817527010804,
-      "grad_norm": 0.501844584941864,
+      "grad_norm": 0.49856579303741455,
       "learning_rate": 3.8275034846893046e-05,
-      "loss": 0.8126,
+      "loss": 0.8125,
       "num_input_tokens_seen": 60456960,
       "step": 7380
     },
     {
       "epoch": 0.35486194477791116,
-      "grad_norm": 0.5217064023017883,
+      "grad_norm": 0.5153807997703552,
       "learning_rate": 3.824144827712738e-05,
-      "loss": 1.3003,
+      "loss": 1.3012,
       "num_input_tokens_seen": 60538880,
       "step": 7390
     },
     {
       "epoch": 0.3553421368547419,
-      "grad_norm": 0.5244282484054565,
+      "grad_norm": 0.5248746275901794,
       "learning_rate": 3.82078284560524e-05,
-      "loss": 1.0134,
+      "loss": 1.0136,
       "num_input_tokens_seen": 60620800,
       "step": 7400
     },
     {
       "epoch": 0.3558223289315726,
-      "grad_norm": 0.5655602216720581,
+      "grad_norm": 0.5390773415565491,
       "learning_rate": 3.817417546809263e-05,
-      "loss": 0.9036,
+      "loss": 0.9035,
       "num_input_tokens_seen": 60702720,
       "step": 7410
     },
     {
       "epoch": 0.3563025210084034,
-      "grad_norm": 0.5483632683753967,
+      "grad_norm": 0.5516919493675232,
       "learning_rate": 3.8140489397755886e-05,
-      "loss": 1.0055,
+      "loss": 1.0054,
       "num_input_tokens_seen": 60784640,
       "step": 7420
     },
     {
       "epoch": 0.35678271308523407,
-      "grad_norm": 0.5369768738746643,
+      "grad_norm": 0.5377995371818542,
       "learning_rate": 3.810677032963307e-05,
-      "loss": 0.9698,
+      "loss": 0.9705,
       "num_input_tokens_seen": 60866560,
       "step": 7430
     },
     {
       "epoch": 0.3572629051620648,
-      "grad_norm": 0.4874258041381836,
+      "grad_norm": 0.5012087821960449,
       "learning_rate": 3.807301834839793e-05,
       "loss": 0.8928,
       "num_input_tokens_seen": 60948480,
@@ -5962,23 +5962,23 @@
     },
     {
       "epoch": 0.3577430972388956,
-      "grad_norm": 0.4989224076271057,
+      "grad_norm": 0.5082871913909912,
       "learning_rate": 3.803923353880687e-05,
-      "loss": 0.7671,
+      "loss": 0.7673,
       "num_input_tokens_seen": 61030400,
       "step": 7450
     },
     {
       "epoch": 0.3582232893157263,
-      "grad_norm": 0.5844639539718628,
+      "grad_norm": 0.5730777382850647,
       "learning_rate": 3.8005415985698754e-05,
-      "loss": 0.9805,
+      "loss": 0.9807,
       "num_input_tokens_seen": 61112320,
       "step": 7460
     },
     {
       "epoch": 0.35870348139255703,
-      "grad_norm": 0.5072601437568665,
+      "grad_norm": 0.4978314936161041,
       "learning_rate": 3.797156577399462e-05,
       "loss": 0.9943,
       "num_input_tokens_seen": 61194240,
@@ -5986,127 +5986,127 @@
     },
     {
       "epoch": 0.35918367346938773,
-      "grad_norm": 0.5117142200469971,
+      "grad_norm": 0.5063808560371399,
       "learning_rate": 3.7937682988697566e-05,
-      "loss": 0.8656,
+      "loss": 0.8648,
       "num_input_tokens_seen": 61276160,
       "step": 7480
     },
     {
       "epoch": 0.3596638655462185,
-      "grad_norm": 0.5001174211502075,
+      "grad_norm": 0.5042293667793274,
       "learning_rate": 3.790376771489247e-05,
-      "loss": 1.0121,
+      "loss": 1.0123,
       "num_input_tokens_seen": 61358080,
       "step": 7490
     },
     {
       "epoch": 0.36014405762304924,
-      "grad_norm": 0.5205296277999878,
+      "grad_norm": 0.5322972536087036,
       "learning_rate": 3.7869820037745776e-05,
-      "loss": 0.8441,
+      "loss": 0.8446,
       "num_input_tokens_seen": 61440000,
       "step": 7500
     },
     {
       "epoch": 0.36062424969987994,
-      "grad_norm": 0.5183282494544983,
+      "grad_norm": 0.5157181024551392,
       "learning_rate": 3.783584004250531e-05,
-      "loss": 0.7604,
+      "loss": 0.7602,
       "num_input_tokens_seen": 61521920,
       "step": 7510
     },
     {
       "epoch": 0.3611044417767107,
-      "grad_norm": 0.3707723319530487,
+      "grad_norm": 0.3682163655757904,
       "learning_rate": 3.7801827814500074e-05,
-      "loss": 0.8183,
+      "loss": 0.8187,
       "num_input_tokens_seen": 61603840,
       "step": 7520
     },
     {
       "epoch": 0.3615846338535414,
-      "grad_norm": 0.5028261542320251,
+      "grad_norm": 0.49639391899108887,
       "learning_rate": 3.7767783439139984e-05,
-      "loss": 0.8522,
+      "loss": 0.8516,
       "num_input_tokens_seen": 61685760,
       "step": 7530
     },
     {
       "epoch": 0.36206482593037215,
-      "grad_norm": 0.5016160011291504,
+      "grad_norm": 0.48996445536613464,
       "learning_rate": 3.77337070019157e-05,
-      "loss": 0.8296,
+      "loss": 0.8289,
       "num_input_tokens_seen": 61767680,
       "step": 7540
     },
     {
       "epoch": 0.3625450180072029,
-      "grad_norm": 0.6660007238388062,
+      "grad_norm": 0.6653403043746948,
       "learning_rate": 3.7699598588398364e-05,
-      "loss": 0.9405,
+      "loss": 0.9404,
       "num_input_tokens_seen": 61849600,
       "step": 7550
     },
     {
       "epoch": 0.3630252100840336,
-      "grad_norm": 0.4924418032169342,
+      "grad_norm": 0.48587802052497864,
       "learning_rate": 3.766545828423946e-05,
-      "loss": 0.9701,
+      "loss": 0.9699,
       "num_input_tokens_seen": 61931520,
       "step": 7560
     },
     {
       "epoch": 0.36350540216086435,
-      "grad_norm": 0.5335708260536194,
+      "grad_norm": 0.5049532055854797,
       "learning_rate": 3.7631286175170535e-05,
-      "loss": 0.8598,
+      "loss": 0.8596,
       "num_input_tokens_seen": 62013440,
       "step": 7570
     },
     {
       "epoch": 0.36398559423769505,
-      "grad_norm": 0.4859420955181122,
+      "grad_norm": 0.4853578507900238,
       "learning_rate": 3.7597082347003e-05,
-      "loss": 1.0999,
+      "loss": 1.0988,
       "num_input_tokens_seen": 62095360,
       "step": 7580
     },
     {
       "epoch": 0.3644657863145258,
-      "grad_norm": 0.4671091139316559,
+      "grad_norm": 0.47532176971435547,
       "learning_rate": 3.75628468856279e-05,
-      "loss": 0.8699,
+      "loss": 0.8688,
       "num_input_tokens_seen": 62177280,
       "step": 7590
     },
     {
       "epoch": 0.36494597839135656,
-      "grad_norm": 0.5766321420669556,
+      "grad_norm": 0.6255443692207336,
       "learning_rate": 3.7528579877015746e-05,
-      "loss": 0.8688,
+      "loss": 0.8713,
       "num_input_tokens_seen": 62259200,
       "step": 7600
     },
     {
       "epoch": 0.36542617046818726,
-      "grad_norm": 0.8806756138801575,
+      "grad_norm": 1.008660912513733,
       "learning_rate": 3.749428140721626e-05,
-      "loss": 1.0697,
+      "loss": 1.0696,
       "num_input_tokens_seen": 62341120,
       "step": 7610
     },
     {
       "epoch": 0.365906362545018,
-      "grad_norm": 0.5065363049507141,
+      "grad_norm": 0.5324845910072327,
       "learning_rate": 3.745995156235815e-05,
-      "loss": 0.8349,
+      "loss": 0.8351,
       "num_input_tokens_seen": 62423040,
       "step": 7620
     },
     {
       "epoch": 0.3663865546218487,
-      "grad_norm": 0.5004618763923645,
+      "grad_norm": 0.5093852281570435,
       "learning_rate": 3.742559042864895e-05,
       "loss": 0.9453,
       "num_input_tokens_seen": 62504960,
@@ -6114,175 +6114,175 @@
     },
     {
       "epoch": 0.36686674669867947,
-      "grad_norm": 0.46816208958625793,
+      "grad_norm": 0.4843917787075043,
       "learning_rate": 3.7391198092374726e-05,
-      "loss": 1.0237,
+      "loss": 1.0241,
       "num_input_tokens_seen": 62586880,
       "step": 7640
     },
     {
       "epoch": 0.3673469387755102,
-      "grad_norm": 0.5166922211647034,
+      "grad_norm": 0.5122877955436707,
       "learning_rate": 3.7356774639899914e-05,
-      "loss": 0.8499,
+      "loss": 0.8502,
       "num_input_tokens_seen": 62668800,
       "step": 7650
     },
     {
       "epoch": 0.3678271308523409,
-      "grad_norm": 0.5414100289344788,
+      "grad_norm": 0.525092363357544,
       "learning_rate": 3.7322320157667094e-05,
-      "loss": 0.8653,
+      "loss": 0.8658,
       "num_input_tokens_seen": 62750720,
       "step": 7660
     },
     {
       "epoch": 0.3683073229291717,
-      "grad_norm": 0.4967511296272278,
+      "grad_norm": 0.5073078274726868,
       "learning_rate": 3.728783473219676e-05,
-      "loss": 0.7981,
+      "loss": 0.7977,
       "num_input_tokens_seen": 62832640,
       "step": 7670
     },
     {
       "epoch": 0.3687875150060024,
-      "grad_norm": 0.5025008320808411,
+      "grad_norm": 0.49645790457725525,
       "learning_rate": 3.72533184500871e-05,
-      "loss": 0.7632,
+      "loss": 0.7638,
       "num_input_tokens_seen": 62914560,
       "step": 7680
     },
     {
       "epoch": 0.36926770708283313,
-      "grad_norm": 0.4785637855529785,
+      "grad_norm": 0.4900324046611786,
       "learning_rate": 3.7218771398013807e-05,
-      "loss": 0.9014,
+      "loss": 0.9013,
       "num_input_tokens_seen": 62996480,
       "step": 7690
     },
     {
       "epoch": 0.3697478991596639,
-      "grad_norm": 1.0697089433670044,
+      "grad_norm": 1.146216630935669,
       "learning_rate": 3.718419366272982e-05,
-      "loss": 0.828,
+      "loss": 0.8292,
       "num_input_tokens_seen": 63078400,
       "step": 7700
     },
     {
       "epoch": 0.3702280912364946,
-      "grad_norm": 0.4909783899784088,
+      "grad_norm": 0.4886285662651062,
       "learning_rate": 3.714958533106515e-05,
-      "loss": 0.7715,
+      "loss": 0.7722,
       "num_input_tokens_seen": 63160320,
       "step": 7710
     },
     {
       "epoch": 0.37070828331332534,
-      "grad_norm": 0.5369053483009338,
+      "grad_norm": 0.5365391969680786,
       "learning_rate": 3.7114946489926633e-05,
-      "loss": 0.9035,
+      "loss": 0.9032,
       "num_input_tokens_seen": 63242240,
       "step": 7720
     },
     {
       "epoch": 0.37118847539015604,
-      "grad_norm": 0.4767155647277832,
+      "grad_norm": 0.47952884435653687,
       "learning_rate": 3.708027722629772e-05,
-      "loss": 1.0931,
+      "loss": 1.0929,
       "num_input_tokens_seen": 63324160,
       "step": 7730
     },
     {
       "epoch": 0.3716686674669868,
-      "grad_norm": 0.4803867042064667,
+      "grad_norm": 0.4813827574253082,
       "learning_rate": 3.704557762723823e-05,
-      "loss": 0.8443,
+      "loss": 0.8444,
       "num_input_tokens_seen": 63406080,
       "step": 7740
     },
     {
       "epoch": 0.37214885954381755,
-      "grad_norm": 0.5255789756774902,
+      "grad_norm": 0.521245539188385,
       "learning_rate": 3.7010847779884204e-05,
-      "loss": 0.9003,
+      "loss": 0.8989,
       "num_input_tokens_seen": 63488000,
       "step": 7750
     },
     {
       "epoch": 0.37262905162064824,
-      "grad_norm": 0.8208502531051636,
+      "grad_norm": 0.8104525208473206,
       "learning_rate": 3.697608777144762e-05,
-      "loss": 0.9216,
+      "loss": 0.9214,
       "num_input_tokens_seen": 63569920,
       "step": 7760
     },
     {
       "epoch": 0.373109243697479,
-      "grad_norm": 0.5180743932723999,
+      "grad_norm": 0.5063729882240295,
       "learning_rate": 3.694129768921619e-05,
-      "loss": 0.966,
+      "loss": 0.9651,
       "num_input_tokens_seen": 63651840,
       "step": 7770
     },
     {
       "epoch": 0.3735894357743097,
-      "grad_norm": 0.5045250058174133,
+      "grad_norm": 0.4979282319545746,
       "learning_rate": 3.6906477620553156e-05,
-      "loss": 0.947,
+      "loss": 0.946,
       "num_input_tokens_seen": 63733760,
       "step": 7780
     },
     {
       "epoch": 0.37406962785114045,
-      "grad_norm": 0.4700779616832733,
+      "grad_norm": 0.4596295952796936,
       "learning_rate": 3.687162765289704e-05,
-      "loss": 0.9151,
+      "loss": 0.9149,
       "num_input_tokens_seen": 63815680,
       "step": 7790
     },
     {
       "epoch": 0.3745498199279712,
-      "grad_norm": 0.5987953543663025,
+      "grad_norm": 0.5804716348648071,
       "learning_rate": 3.683674787376148e-05,
-      "loss": 0.8509,
+      "loss": 0.8513,
       "num_input_tokens_seen": 63897600,
       "step": 7800
     },
     {
       "epoch": 0.3750300120048019,
-      "grad_norm": 0.4968700408935547,
+      "grad_norm": 0.4932875633239746,
       "learning_rate": 3.6801838370734945e-05,
-      "loss": 0.7108,
+      "loss": 0.7109,
       "num_input_tokens_seen": 63979520,
       "step": 7810
     },
     {
       "epoch": 0.37551020408163266,
-      "grad_norm": 0.4701189398765564,
+      "grad_norm": 0.4756897985935211,
       "learning_rate": 3.676689923148056e-05,
-      "loss": 0.8534,
+      "loss": 0.8537,
       "num_input_tokens_seen": 64061440,
       "step": 7820
     },
     {
       "epoch": 0.37599039615846336,
-      "grad_norm": 0.5217798948287964,
+      "grad_norm": 0.5182291865348816,
       "learning_rate": 3.673193054373587e-05,
-      "loss": 0.9249,
+      "loss": 0.9253,
       "num_input_tokens_seen": 64143360,
       "step": 7830
     },
     {
       "epoch": 0.3764705882352941,
-      "grad_norm": 0.6249185800552368,
+      "grad_norm": 0.610496461391449,
       "learning_rate": 3.6696932395312606e-05,
-      "loss": 0.945,
+      "loss": 0.9448,
       "num_input_tokens_seen": 64225280,
       "step": 7840
     },
     {
       "epoch": 0.37695078031212487,
-      "grad_norm": 0.48508864641189575,
+      "grad_norm": 0.4692385792732239,
       "learning_rate": 3.6661904874096503e-05,
       "loss": 0.8673,
       "num_input_tokens_seen": 64307200,
@@ -6290,39 +6290,39 @@
     },
     {
       "epoch": 0.37743097238895557,
-      "grad_norm": 0.5447950959205627,
+      "grad_norm": 0.5275558233261108,
       "learning_rate": 3.662684806804704e-05,
-      "loss": 0.817,
+      "loss": 0.8175,
       "num_input_tokens_seen": 64389120,
       "step": 7860
     },
     {
       "epoch": 0.3779111644657863,
-      "grad_norm": 0.48486828804016113,
+      "grad_norm": 0.4853339195251465,
       "learning_rate": 3.659176206519724e-05,
-      "loss": 0.9525,
+      "loss": 0.9513,
       "num_input_tokens_seen": 64471040,
       "step": 7870
     },
     {
       "epoch": 0.3783913565426171,
-      "grad_norm": 0.4893752932548523,
+      "grad_norm": 0.4804559051990509,
       "learning_rate": 3.655664695365344e-05,
-      "loss": 0.8771,
+      "loss": 0.8763,
       "num_input_tokens_seen": 64552960,
       "step": 7880
     },
     {
       "epoch": 0.3788715486194478,
-      "grad_norm": 0.490164190530777,
+      "grad_norm": 0.4826927185058594,
       "learning_rate": 3.652150282159507e-05,
-      "loss": 0.895,
+      "loss": 0.8944,
       "num_input_tokens_seen": 64634880,
       "step": 7890
     },
     {
       "epoch": 0.3793517406962785,
-      "grad_norm": 0.5217410326004028,
+      "grad_norm": 0.5171544551849365,
       "learning_rate": 3.6486329757274454e-05,
       "loss": 0.8366,
       "num_input_tokens_seen": 64716800,
@@ -6330,311 +6330,311 @@
     },
     {
       "epoch": 0.3798319327731092,
-      "grad_norm": 0.49266862869262695,
+      "grad_norm": 0.49717050790786743,
       "learning_rate": 3.645112784901655e-05,
-      "loss": 0.9177,
+      "loss": 0.9178,
       "num_input_tokens_seen": 64798720,
       "step": 7910
     },
     {
       "epoch": 0.38031212484994,
-      "grad_norm": 0.47938328981399536,
+      "grad_norm": 0.47939014434814453,
       "learning_rate": 3.641589718521875e-05,
-      "loss": 0.8592,
+      "loss": 0.8597,
       "num_input_tokens_seen": 64880640,
       "step": 7920
     },
     {
       "epoch": 0.38079231692677074,
-      "grad_norm": 1.3006699085235596,
+      "grad_norm": 1.3471354246139526,
       "learning_rate": 3.6380637854350665e-05,
-      "loss": 0.9027,
+      "loss": 0.9038,
       "num_input_tokens_seen": 64962560,
       "step": 7930
     },
     {
       "epoch": 0.38127250900360143,
-      "grad_norm": 0.6962760090827942,
+      "grad_norm": 0.6851658821105957,
       "learning_rate": 3.634534994495387e-05,
-      "loss": 0.8225,
+      "loss": 0.8226,
       "num_input_tokens_seen": 65044480,
       "step": 7940
     },
     {
       "epoch": 0.3817527010804322,
-      "grad_norm": 0.5089699625968933,
+      "grad_norm": 0.5071682333946228,
       "learning_rate": 3.631003354564175e-05,
-      "loss": 0.8546,
+      "loss": 0.8539,
       "num_input_tokens_seen": 65126400,
       "step": 7950
     },
     {
       "epoch": 0.3822328931572629,
-      "grad_norm": 0.48060354590415955,
+      "grad_norm": 0.48892542719841003,
       "learning_rate": 3.6274688745099194e-05,
-      "loss": 1.0001,
+      "loss": 1.0,
       "num_input_tokens_seen": 65208320,
       "step": 7960
     },
     {
       "epoch": 0.38271308523409364,
-      "grad_norm": 0.5190022587776184,
+      "grad_norm": 0.5070081949234009,
       "learning_rate": 3.623931563208241e-05,
-      "loss": 0.9504,
+      "loss": 0.9514,
       "num_input_tokens_seen": 65290240,
       "step": 7970
     },
     {
       "epoch": 0.3831932773109244,
-      "grad_norm": 0.48244509100914,
+      "grad_norm": 0.4804380536079407,
       "learning_rate": 3.620391429541873e-05,
-      "loss": 0.8261,
+      "loss": 0.8269,
       "num_input_tokens_seen": 65372160,
       "step": 7980
     },
     {
       "epoch": 0.3836734693877551,
-      "grad_norm": 0.5255333185195923,
+      "grad_norm": 0.5210065841674805,
       "learning_rate": 3.616848482400634e-05,
-      "loss": 0.9785,
+      "loss": 0.9784,
       "num_input_tokens_seen": 65454080,
       "step": 7990
     },
     {
       "epoch": 0.38415366146458585,
-      "grad_norm": 0.507675290107727,
+      "grad_norm": 0.5196698307991028,
       "learning_rate": 3.6133027306814085e-05,
-      "loss": 0.9941,
+      "loss": 0.994,
       "num_input_tokens_seen": 65536000,
       "step": 8000
     },
     {
       "epoch": 0.38463385354141655,
-      "grad_norm": 0.5132362842559814,
+      "grad_norm": 0.5133892893791199,
       "learning_rate": 3.609754183288122e-05,
-      "loss": 1.0458,
+      "loss": 1.046,
       "num_input_tokens_seen": 65617920,
       "step": 8010
     },
     {
       "epoch": 0.3851140456182473,
-      "grad_norm": 0.5268590450286865,
+      "grad_norm": 0.5304073691368103,
       "learning_rate": 3.606202849131723e-05,
-      "loss": 0.9985,
+      "loss": 0.9983,
       "num_input_tokens_seen": 65699840,
       "step": 8020
     },
     {
       "epoch": 0.38559423769507806,
-      "grad_norm": 0.4958067834377289,
+      "grad_norm": 0.48315533995628357,
       "learning_rate": 3.6026487371301564e-05,
-      "loss": 0.9213,
+      "loss": 0.9211,
       "num_input_tokens_seen": 65781760,
       "step": 8030
     },
     {
       "epoch": 0.38607442977190876,
-      "grad_norm": 0.5100277662277222,
+      "grad_norm": 0.5015775561332703,
       "learning_rate": 3.599091856208343e-05,
-      "loss": 0.8843,
+      "loss": 0.8849,
       "num_input_tokens_seen": 65863680,
       "step": 8040
     },
     {
       "epoch": 0.3865546218487395,
-      "grad_norm": 0.49061518907546997,
+      "grad_norm": 0.48933330178260803,
       "learning_rate": 3.5955322152981575e-05,
-      "loss": 0.9195,
+      "loss": 0.9194,
       "num_input_tokens_seen": 65945600,
       "step": 8050
     },
     {
       "epoch": 0.3870348139255702,
-      "grad_norm": 0.531670093536377,
+      "grad_norm": 0.5391829013824463,
       "learning_rate": 3.5919698233384034e-05,
-      "loss": 1.0418,
+      "loss": 1.0414,
       "num_input_tokens_seen": 66027520,
       "step": 8060
     },
     {
       "epoch": 0.38751500600240096,
-      "grad_norm": 0.5200238823890686,
+      "grad_norm": 0.5336183309555054,
       "learning_rate": 3.588404689274795e-05,
-      "loss": 0.8457,
+      "loss": 0.8462,
       "num_input_tokens_seen": 66109440,
       "step": 8070
     },
     {
       "epoch": 0.3879951980792317,
-      "grad_norm": 0.5855520963668823,
+      "grad_norm": 0.5864033699035645,
       "learning_rate": 3.58483682205993e-05,
-      "loss": 0.7761,
+      "loss": 0.7765,
       "num_input_tokens_seen": 66191360,
       "step": 8080
     },
     {
       "epoch": 0.3884753901560624,
-      "grad_norm": 0.5065504908561707,
+      "grad_norm": 0.5033870339393616,
       "learning_rate": 3.581266230653271e-05,
-      "loss": 0.86,
+      "loss": 0.8597,
       "num_input_tokens_seen": 66273280,
       "step": 8090
     },
     {
       "epoch": 0.38895558223289317,
-      "grad_norm": 0.5135564804077148,
+      "grad_norm": 0.5161807537078857,
       "learning_rate": 3.5776929240211224e-05,
-      "loss": 0.8299,
+      "loss": 0.8295,
       "num_input_tokens_seen": 66355200,
       "step": 8100
     },
     {
       "epoch": 0.38943577430972387,
-      "grad_norm": 0.5159543752670288,
+      "grad_norm": 0.5269225239753723,
       "learning_rate": 3.5741169111366047e-05,
-      "loss": 1.0024,
+      "loss": 1.0028,
       "num_input_tokens_seen": 66437120,
       "step": 8110
     },
     {
       "epoch": 0.3899159663865546,
-      "grad_norm": 0.659740149974823,
+      "grad_norm": 0.5889651775360107,
       "learning_rate": 3.570538200979635e-05,
-      "loss": 1.0171,
+      "loss": 1.0164,
       "num_input_tokens_seen": 66519040,
       "step": 8120
     },
     {
       "epoch": 0.3903961584633854,
-      "grad_norm": 0.6677471995353699,
+      "grad_norm": 0.6744776964187622,
       "learning_rate": 3.566956802536904e-05,
-      "loss": 0.9044,
+      "loss": 0.9047,
       "num_input_tokens_seen": 66600960,
       "step": 8130
     },
     {
       "epoch": 0.3908763505402161,
-      "grad_norm": 0.5268581509590149,
+      "grad_norm": 0.5208677053451538,
       "learning_rate": 3.5633727248018536e-05,
-      "loss": 0.9338,
+      "loss": 0.9341,
       "num_input_tokens_seen": 66682880,
       "step": 8140
     },
     {
       "epoch": 0.39135654261704683,
-      "grad_norm": 0.7781187891960144,
+      "grad_norm": 0.7580982446670532,
       "learning_rate": 3.5597859767746524e-05,
-      "loss": 0.8946,
+      "loss": 0.8941,
       "num_input_tokens_seen": 66764800,
       "step": 8150
     },
     {
       "epoch": 0.39183673469387753,
-      "grad_norm": 0.40642163157463074,
+      "grad_norm": 0.4247629642486572,
       "learning_rate": 3.556196567462175e-05,
-      "loss": 0.8835,
+      "loss": 0.8836,
       "num_input_tokens_seen": 66846720,
       "step": 8160
     },
     {
       "epoch": 0.3923169267707083,
-      "grad_norm": 0.5072639584541321,
+      "grad_norm": 0.49652138352394104,
       "learning_rate": 3.5526045058779805e-05,
-      "loss": 0.9559,
+      "loss": 0.9574,
       "num_input_tokens_seen": 66928640,
       "step": 8170
     },
     {
       "epoch": 0.39279711884753904,
-      "grad_norm": 0.5224815011024475,
+      "grad_norm": 0.5123445391654968,
       "learning_rate": 3.549009801042286e-05,
-      "loss": 0.8609,
+      "loss": 0.8612,
       "num_input_tokens_seen": 67010560,
       "step": 8180
     },
     {
       "epoch": 0.39327731092436974,
-      "grad_norm": 0.5318178534507751,
+      "grad_norm": 0.5425747632980347,
       "learning_rate": 3.545412461981947e-05,
-      "loss": 0.9237,
+      "loss": 0.9242,
       "num_input_tokens_seen": 67092480,
       "step": 8190
     },
     {
       "epoch": 0.3937575030012005,
-      "grad_norm": 0.5438302755355835,
+      "grad_norm": 0.49756717681884766,
       "learning_rate": 3.541812497730435e-05,
-      "loss": 0.992,
+      "loss": 0.9924,
       "num_input_tokens_seen": 67174400,
       "step": 8200
     },
     {
       "epoch": 0.3942376950780312,
-      "grad_norm": 0.48217347264289856,
+      "grad_norm": 0.48790714144706726,
       "learning_rate": 3.5382099173278125e-05,
-      "loss": 0.9067,
+      "loss": 0.9064,
       "num_input_tokens_seen": 67256320,
       "step": 8210
     },
     {
       "epoch": 0.39471788715486195,
-      "grad_norm": 0.6276324391365051,
+      "grad_norm": 0.6320737600326538,
       "learning_rate": 3.5346047298207116e-05,
-      "loss": 0.891,
+      "loss": 0.8907,
       "num_input_tokens_seen": 67338240,
       "step": 8220
     },
     {
       "epoch": 0.3951980792316927,
-      "grad_norm": 0.5074647665023804,
+      "grad_norm": 0.500048816204071,
       "learning_rate": 3.530996944262312e-05,
-      "loss": 0.7989,
+      "loss": 0.7992,
       "num_input_tokens_seen": 67420160,
       "step": 8230
     },
     {
       "epoch": 0.3956782713085234,
-      "grad_norm": 0.48775187134742737,
+      "grad_norm": 0.4895772337913513,
       "learning_rate": 3.5273865697123164e-05,
-      "loss": 0.8951,
+      "loss": 0.8946,
       "num_input_tokens_seen": 67502080,
       "step": 8240
     },
     {
       "epoch": 0.39615846338535415,
-      "grad_norm": 0.5741081237792969,
+      "grad_norm": 0.5737354159355164,
       "learning_rate": 3.52377361523693e-05,
-      "loss": 1.0857,
+      "loss": 1.0861,
       "num_input_tokens_seen": 67584000,
       "step": 8250
     },
     {
       "epoch": 0.39663865546218485,
-      "grad_norm": 0.5743753910064697,
+      "grad_norm": 0.5739513039588928,
       "learning_rate": 3.520158089908836e-05,
-      "loss": 1.0151,
+      "loss": 1.0155,
       "num_input_tokens_seen": 67665920,
       "step": 8260
     },
     {
       "epoch": 0.3971188475390156,
-      "grad_norm": 0.5247855186462402,
+      "grad_norm": 0.519338846206665,
       "learning_rate": 3.516540002807174e-05,
-      "loss": 0.765,
+      "loss": 0.7652,
       "num_input_tokens_seen": 67747840,
       "step": 8270
     },
     {
       "epoch": 0.39759903961584636,
-      "grad_norm": 0.5025756359100342,
+      "grad_norm": 0.505893349647522,
       "learning_rate": 3.512919363017516e-05,
-      "loss": 0.881,
+      "loss": 0.8809,
       "num_input_tokens_seen": 67829760,
       "step": 8280
     },
     {
       "epoch": 0.39807923169267706,
-      "grad_norm": 0.46168726682662964,
+      "grad_norm": 0.47453415393829346,
       "learning_rate": 3.509296179631843e-05,
       "loss": 0.8293,
       "num_input_tokens_seen": 67911680,
@@ -6642,39 +6642,39 @@
     },
     {
       "epoch": 0.3985594237695078,
-      "grad_norm": 0.556755542755127,
+      "grad_norm": 0.545673131942749,
       "learning_rate": 3.505670461748527e-05,
-      "loss": 0.8953,
+      "loss": 0.8956,
       "num_input_tokens_seen": 67993600,
       "step": 8300
     },
     {
       "epoch": 0.3990396158463385,
-      "grad_norm": 0.5389664173126221,
+      "grad_norm": 0.5385898351669312,
       "learning_rate": 3.5020422184723e-05,
-      "loss": 1.19,
+      "loss": 1.1892,
       "num_input_tokens_seen": 68075520,
       "step": 8310
     },
     {
       "epoch": 0.39951980792316927,
-      "grad_norm": 0.5176492929458618,
+      "grad_norm": 0.5242102742195129,
       "learning_rate": 3.498411458914238e-05,
-      "loss": 0.8479,
+      "loss": 0.8476,
       "num_input_tokens_seen": 68157440,
       "step": 8320
     },
     {
       "epoch": 0.4,
-      "grad_norm": 0.5035036206245422,
+      "grad_norm": 0.5002279877662659,
       "learning_rate": 3.494778192191739e-05,
-      "loss": 0.8593,
+      "loss": 0.859,
       "num_input_tokens_seen": 68239360,
       "step": 8330
     },
     {
       "epoch": 0.4004801920768307,
-      "grad_norm": 0.49024298787117004,
+      "grad_norm": 0.4706880748271942,
       "learning_rate": 3.4911424274284886e-05,
       "loss": 0.771,
       "num_input_tokens_seen": 68321280,
@@ -6682,47 +6682,47 @@
     },
     {
       "epoch": 0.4009603841536615,
-      "grad_norm": 0.4582843482494354,
+      "grad_norm": 0.4681709408760071,
       "learning_rate": 3.4875041737544526e-05,
-      "loss": 0.9203,
+      "loss": 0.9161,
       "num_input_tokens_seen": 68403200,
       "step": 8350
     },
     {
       "epoch": 0.4014405762304922,
-      "grad_norm": 0.5551020503044128,
+      "grad_norm": 0.5591146349906921,
       "learning_rate": 3.483863440305845e-05,
-      "loss": 0.8853,
+      "loss": 0.8845,
       "num_input_tokens_seen": 68485120,
       "step": 8360
     },
     {
       "epoch": 0.40192076830732293,
-      "grad_norm": 1.1716420650482178,
+      "grad_norm": 1.1396244764328003,
       "learning_rate": 3.480220236225106e-05,
-      "loss": 0.9757,
+      "loss": 0.9752,
       "num_input_tokens_seen": 68567040,
       "step": 8370
     },
     {
       "epoch": 0.4024009603841537,
-      "grad_norm": 0.4597180485725403,
+      "grad_norm": 0.46549364924430847,
       "learning_rate": 3.476574570660879e-05,
-      "loss": 0.8719,
+      "loss": 0.8717,
       "num_input_tokens_seen": 68648960,
       "step": 8380
     },
     {
       "epoch": 0.4028811524609844,
-      "grad_norm": 0.5064864754676819,
+      "grad_norm": 0.5010836720466614,
       "learning_rate": 3.472926452767992e-05,
-      "loss": 0.7306,
+      "loss": 0.7304,
       "num_input_tokens_seen": 68730880,
       "step": 8390
     },
     {
       "epoch": 0.40336134453781514,
-      "grad_norm": 0.4826078414916992,
+      "grad_norm": 0.4840982258319855,
       "learning_rate": 3.469275891707428e-05,
       "loss": 0.8629,
       "num_input_tokens_seen": 68812800,
@@ -6730,79 +6730,79 @@
     },
     {
       "epoch": 0.40384153661464584,
-      "grad_norm": 0.4941999316215515,
+      "grad_norm": 0.5047776699066162,
       "learning_rate": 3.465622896646305e-05,
-      "loss": 0.8746,
+      "loss": 0.8738,
       "num_input_tokens_seen": 68894720,
       "step": 8410
     },
     {
       "epoch": 0.4043217286914766,
-      "grad_norm": 0.5197045207023621,
+      "grad_norm": 0.5110125541687012,
       "learning_rate": 3.461967476757857e-05,
-      "loss": 0.8477,
+      "loss": 0.8472,
       "num_input_tokens_seen": 68976640,
       "step": 8420
     },
     {
       "epoch": 0.40480192076830734,
-      "grad_norm": 0.508140504360199,
+      "grad_norm": 0.5032282471656799,
       "learning_rate": 3.4583096412214025e-05,
-      "loss": 1.0445,
+      "loss": 1.0451,
       "num_input_tokens_seen": 69058560,
       "step": 8430
     },
     {
       "epoch": 0.40528211284513804,
-      "grad_norm": 0.5302277207374573,
+      "grad_norm": 0.5277020931243896,
       "learning_rate": 3.454649399222328e-05,
-      "loss": 0.846,
+      "loss": 0.8458,
       "num_input_tokens_seen": 69140480,
       "step": 8440
     },
     {
       "epoch": 0.4057623049219688,
-      "grad_norm": 0.4965344965457916,
+      "grad_norm": 0.4992177486419678,
       "learning_rate": 3.450986759952064e-05,
-      "loss": 0.8504,
+      "loss": 0.8503,
       "num_input_tokens_seen": 69222400,
       "step": 8450
     },
     {
       "epoch": 0.4062424969987995,
-      "grad_norm": 0.6701764464378357,
+      "grad_norm": 0.6769206523895264,
       "learning_rate": 3.44732173260806e-05,
-      "loss": 0.8774,
+      "loss": 0.8767,
       "num_input_tokens_seen": 69304320,
       "step": 8460
     },
     {
       "epoch": 0.40672268907563025,
-      "grad_norm": 0.5118209719657898,
+      "grad_norm": 0.5169435143470764,
       "learning_rate": 3.4436543263937613e-05,
-      "loss": 0.9147,
+      "loss": 0.9149,
       "num_input_tokens_seen": 69386240,
       "step": 8470
     },
     {
       "epoch": 0.407202881152461,
-      "grad_norm": 0.5040360689163208,
+      "grad_norm": 0.4996021091938019,
       "learning_rate": 3.439984550518589e-05,
-      "loss": 0.8726,
+      "loss": 0.8728,
       "num_input_tokens_seen": 69468160,
       "step": 8480
     },
     {
       "epoch": 0.4076830732292917,
-      "grad_norm": 0.5248441696166992,
+      "grad_norm": 0.5270659327507019,
       "learning_rate": 3.436312414197913e-05,
-      "loss": 0.9237,
+      "loss": 0.9235,
       "num_input_tokens_seen": 69550080,
       "step": 8490
     },
     {
       "epoch": 0.40816326530612246,
-      "grad_norm": 0.47410309314727783,
+      "grad_norm": 0.4720569849014282,
       "learning_rate": 3.4326379266530314e-05,
       "loss": 1.0674,
       "num_input_tokens_seen": 69632000,
@@ -6810,167 +6810,167 @@
     },
     {
       "epoch": 0.40864345738295316,
-      "grad_norm": 0.49306410551071167,
+      "grad_norm": 0.497772753238678,
       "learning_rate": 3.428961097111146e-05,
-      "loss": 0.8397,
+      "loss": 0.8399,
       "num_input_tokens_seen": 69713920,
       "step": 8510
     },
     {
       "epoch": 0.4091236494597839,
-      "grad_norm": 0.4915395677089691,
+      "grad_norm": 0.49216228723526,
       "learning_rate": 3.4252819348053424e-05,
-      "loss": 0.909,
+      "loss": 0.9089,
       "num_input_tokens_seen": 69795840,
       "step": 8520
     },
     {
       "epoch": 0.40960384153661467,
-      "grad_norm": 0.4925106167793274,
+      "grad_norm": 0.49164268374443054,
       "learning_rate": 3.421600448974559e-05,
-      "loss": 0.9302,
+      "loss": 0.9305,
       "num_input_tokens_seen": 69877760,
       "step": 8530
     },
     {
       "epoch": 0.41008403361344536,
-      "grad_norm": 0.49705174565315247,
+      "grad_norm": 0.49229446053504944,
       "learning_rate": 3.4179166488635736e-05,
-      "loss": 0.9918,
+      "loss": 0.9921,
       "num_input_tokens_seen": 69959680,
       "step": 8540
     },
     {
       "epoch": 0.4105642256902761,
-      "grad_norm": 0.49341046810150146,
+      "grad_norm": 0.484153151512146,
       "learning_rate": 3.414230543722973e-05,
-      "loss": 0.9864,
+      "loss": 0.9862,
       "num_input_tokens_seen": 70041600,
       "step": 8550
     },
     {
       "epoch": 0.4110444177671068,
-      "grad_norm": 0.5298373699188232,
+      "grad_norm": 0.5301636457443237,
       "learning_rate": 3.410542142809134e-05,
-      "loss": 0.872,
+      "loss": 0.8721,
       "num_input_tokens_seen": 70123520,
       "step": 8560
     },
     {
       "epoch": 0.41152460984393757,
-      "grad_norm": 0.48551279306411743,
+      "grad_norm": 0.484598308801651,
       "learning_rate": 3.4068514553841965e-05,
-      "loss": 1.0904,
+      "loss": 1.0909,
       "num_input_tokens_seen": 70205440,
       "step": 8570
     },
     {
       "epoch": 0.4120048019207683,
-      "grad_norm": 0.5113745331764221,
+      "grad_norm": 0.4904525578022003,
       "learning_rate": 3.403158490716043e-05,
-      "loss": 0.8943,
+      "loss": 0.8935,
       "num_input_tokens_seen": 70287360,
       "step": 8580
     },
     {
       "epoch": 0.412484993997599,
-      "grad_norm": 0.5106720924377441,
+      "grad_norm": 0.5105434060096741,
       "learning_rate": 3.3994632580782766e-05,
-      "loss": 0.8691,
+      "loss": 0.8689,
       "num_input_tokens_seen": 70369280,
       "step": 8590
     },
     {
       "epoch": 0.4129651860744298,
-      "grad_norm": 0.5008137822151184,
+      "grad_norm": 0.5011836886405945,
       "learning_rate": 3.395765766750192e-05,
-      "loss": 0.8594,
+      "loss": 0.8591,
       "num_input_tokens_seen": 70451200,
       "step": 8600
     },
     {
       "epoch": 0.4134453781512605,
-      "grad_norm": 0.47841960191726685,
+      "grad_norm": 0.48391351103782654,
       "learning_rate": 3.392066026016757e-05,
-      "loss": 0.8132,
+      "loss": 0.8133,
       "num_input_tokens_seen": 70533120,
       "step": 8610
     },
     {
       "epoch": 0.41392557022809123,
-      "grad_norm": 0.8451586365699768,
+      "grad_norm": 0.8519997596740723,
       "learning_rate": 3.388364045168591e-05,
-      "loss": 0.8289,
+      "loss": 0.8302,
       "num_input_tokens_seen": 70615040,
       "step": 8620
     },
     {
       "epoch": 0.414405762304922,
-      "grad_norm": 0.47416937351226807,
+      "grad_norm": 0.48617443442344666,
       "learning_rate": 3.3846598335019335e-05,
-      "loss": 0.8435,
+      "loss": 0.8433,
       "num_input_tokens_seen": 70696960,
       "step": 8630
     },
     {
       "epoch": 0.4148859543817527,
-      "grad_norm": 0.8883240818977356,
+      "grad_norm": 0.8815626502037048,
       "learning_rate": 3.38095340031863e-05,
-      "loss": 0.9588,
+      "loss": 0.959,
       "num_input_tokens_seen": 70778880,
       "step": 8640
     },
     {
       "epoch": 0.41536614645858344,
-      "grad_norm": 0.5661002993583679,
+      "grad_norm": 0.5593860149383545,
       "learning_rate": 3.377244754926104e-05,
-      "loss": 1.0117,
+      "loss": 1.0122,
       "num_input_tokens_seen": 70860800,
       "step": 8650
     },
     {
       "epoch": 0.41584633853541414,
-      "grad_norm": 0.5038189888000488,
+      "grad_norm": 0.5052839517593384,
       "learning_rate": 3.3735339066373314e-05,
-      "loss": 0.8984,
+      "loss": 0.8981,
       "num_input_tokens_seen": 70942720,
       "step": 8660
     },
     {
       "epoch": 0.4163265306122449,
-      "grad_norm": 1.053067922592163,
+      "grad_norm": 1.1593550443649292,
       "learning_rate": 3.369820864770822e-05,
-      "loss": 0.852,
+      "loss": 0.8525,
       "num_input_tokens_seen": 71024640,
       "step": 8670
     },
     {
       "epoch": 0.41680672268907565,
-      "grad_norm": 0.4945598840713501,
+      "grad_norm": 0.49418941140174866,
       "learning_rate": 3.366105638650596e-05,
-      "loss": 0.8927,
+      "loss": 0.8923,
       "num_input_tokens_seen": 71106560,
       "step": 8680
     },
     {
       "epoch": 0.41728691476590635,
-      "grad_norm": 0.5233136415481567,
+      "grad_norm": 0.5242265462875366,
       "learning_rate": 3.3623882376061554e-05,
-      "loss": 0.8245,
+      "loss": 0.8246,
       "num_input_tokens_seen": 71188480,
       "step": 8690
     },
     {
       "epoch": 0.4177671068427371,
-      "grad_norm": 0.46826255321502686,
+      "grad_norm": 0.4658726155757904,
       "learning_rate": 3.358668670972465e-05,
-      "loss": 0.8433,
+      "loss": 0.8436,
       "num_input_tokens_seen": 71270400,
       "step": 8700
     },
     {
       "epoch": 0.4182472989195678,
-      "grad_norm": 0.5887758731842041,
+      "grad_norm": 0.5847034454345703,
       "learning_rate": 3.354946948089927e-05,
       "loss": 0.9426,
       "num_input_tokens_seen": 71352320,
@@ -6978,279 +6978,279 @@
     },
     {
       "epoch": 0.41872749099639855,
-      "grad_norm": 0.5028591752052307,
+      "grad_norm": 0.4962436556816101,
       "learning_rate": 3.351223078304359e-05,
-      "loss": 0.8831,
+      "loss": 0.8829,
       "num_input_tokens_seen": 71434240,
       "step": 8720
     },
     {
       "epoch": 0.4192076830732293,
-      "grad_norm": 0.5117266178131104,
+      "grad_norm": 0.5134901404380798,
       "learning_rate": 3.34749707096697e-05,
-      "loss": 0.9697,
+      "loss": 0.9692,
       "num_input_tokens_seen": 71516160,
       "step": 8730
     },
     {
       "epoch": 0.41968787515006,
-      "grad_norm": 0.5058220028877258,
+      "grad_norm": 0.5016522407531738,
       "learning_rate": 3.343768935434337e-05,
-      "loss": 0.8492,
+      "loss": 0.8491,
       "num_input_tokens_seen": 71598080,
       "step": 8740
     },
     {
       "epoch": 0.42016806722689076,
-      "grad_norm": 0.7977713346481323,
+      "grad_norm": 0.7968730330467224,
       "learning_rate": 3.34003868106838e-05,
-      "loss": 0.9056,
+      "loss": 0.905,
       "num_input_tokens_seen": 71680000,
       "step": 8750
     },
     {
       "epoch": 0.42064825930372146,
-      "grad_norm": 0.5445558428764343,
+      "grad_norm": 0.5383213758468628,
       "learning_rate": 3.3363063172363396e-05,
-      "loss": 0.9732,
+      "loss": 0.9733,
       "num_input_tokens_seen": 71761920,
       "step": 8760
     },
     {
       "epoch": 0.4211284513805522,
-      "grad_norm": 0.5837395787239075,
+      "grad_norm": 0.5921952128410339,
       "learning_rate": 3.3325718533107556e-05,
-      "loss": 1.1805,
+      "loss": 1.1804,
       "num_input_tokens_seen": 71843840,
       "step": 8770
     },
     {
       "epoch": 0.42160864345738297,
-      "grad_norm": 0.6139930486679077,
+      "grad_norm": 0.6270814538002014,
       "learning_rate": 3.3288352986694396e-05,
-      "loss": 0.9309,
+      "loss": 0.9318,
       "num_input_tokens_seen": 71925760,
       "step": 8780
     },
     {
       "epoch": 0.42208883553421367,
-      "grad_norm": 1.7015244960784912,
+      "grad_norm": 1.9248117208480835,
       "learning_rate": 3.325096662695454e-05,
-      "loss": 1.0555,
+      "loss": 1.0563,
       "num_input_tokens_seen": 72007680,
       "step": 8790
     },
     {
       "epoch": 0.4225690276110444,
-      "grad_norm": 0.49051469564437866,
+      "grad_norm": 0.491634339094162,
       "learning_rate": 3.321355954777087e-05,
-      "loss": 0.831,
+      "loss": 0.83,
       "num_input_tokens_seen": 72089600,
       "step": 8800
     },
     {
       "epoch": 0.4230492196878752,
-      "grad_norm": 0.5128316283226013,
+      "grad_norm": 0.4981280267238617,
       "learning_rate": 3.317613184307832e-05,
-      "loss": 0.9081,
+      "loss": 0.906,
       "num_input_tokens_seen": 72171520,
       "step": 8810
     },
     {
       "epoch": 0.4235294117647059,
-      "grad_norm": 0.5852062106132507,
+      "grad_norm": 0.5684440732002258,
       "learning_rate": 3.313868360686359e-05,
-      "loss": 0.9385,
+      "loss": 0.938,
       "num_input_tokens_seen": 72253440,
       "step": 8820
     },
     {
       "epoch": 0.42400960384153663,
-      "grad_norm": 0.4833785891532898,
+      "grad_norm": 0.4744153618812561,
       "learning_rate": 3.310121493316495e-05,
-      "loss": 0.8992,
+      "loss": 0.8989,
       "num_input_tokens_seen": 72335360,
       "step": 8830
     },
     {
       "epoch": 0.42448979591836733,
-      "grad_norm": 0.49650639295578003,
+      "grad_norm": 0.501400887966156,
       "learning_rate": 3.306372591607199e-05,
-      "loss": 0.8763,
+      "loss": 0.876,
       "num_input_tokens_seen": 72417280,
       "step": 8840
     },
     {
       "epoch": 0.4249699879951981,
-      "grad_norm": 0.8988882303237915,
+      "grad_norm": 0.9911435842514038,
       "learning_rate": 3.30262166497254e-05,
-      "loss": 0.8636,
+      "loss": 0.8632,
       "num_input_tokens_seen": 72499200,
       "step": 8850
     },
     {
       "epoch": 0.42545018007202884,
-      "grad_norm": 0.5129944086074829,
+      "grad_norm": 0.50812828540802,
       "learning_rate": 3.29886872283167e-05,
-      "loss": 1.1187,
+      "loss": 1.1161,
       "num_input_tokens_seen": 72581120,
       "step": 8860
     },
     {
       "epoch": 0.42593037214885954,
-      "grad_norm": 0.4534735381603241,
+      "grad_norm": 0.4761914610862732,
       "learning_rate": 3.2951137746088004e-05,
-      "loss": 0.9403,
+      "loss": 0.94,
       "num_input_tokens_seen": 72663040,
       "step": 8870
     },
     {
       "epoch": 0.4264105642256903,
-      "grad_norm": 0.48410260677337646,
+      "grad_norm": 0.4863661527633667,
       "learning_rate": 3.291356829733186e-05,
-      "loss": 0.8392,
+      "loss": 0.8393,
       "num_input_tokens_seen": 72744960,
       "step": 8880
     },
     {
       "epoch": 0.426890756302521,
-      "grad_norm": 0.9801385998725891,
+      "grad_norm": 1.054153323173523,
       "learning_rate": 3.287597897639092e-05,
-      "loss": 1.0537,
+      "loss": 1.0528,
       "num_input_tokens_seen": 72826880,
       "step": 8890
     },
     {
       "epoch": 0.42737094837935174,
-      "grad_norm": 0.7022339105606079,
+      "grad_norm": 0.677579402923584,
       "learning_rate": 3.283836987765771e-05,
-      "loss": 0.8691,
+      "loss": 0.8697,
       "num_input_tokens_seen": 72908800,
       "step": 8900
     },
     {
       "epoch": 0.4278511404561825,
-      "grad_norm": 0.4977160096168518,
+      "grad_norm": 0.48590415716171265,
       "learning_rate": 3.280074109557447e-05,
-      "loss": 0.8148,
+      "loss": 0.8158,
       "num_input_tokens_seen": 72990720,
       "step": 8910
     },
     {
       "epoch": 0.4283313325330132,
-      "grad_norm": 0.5575817823410034,
+      "grad_norm": 0.5039668679237366,
       "learning_rate": 3.2763092724632854e-05,
-      "loss": 0.8026,
+      "loss": 0.8016,
       "num_input_tokens_seen": 73072640,
       "step": 8920
     },
     {
       "epoch": 0.42881152460984395,
-      "grad_norm": 0.6442136764526367,
+      "grad_norm": 0.7368101477622986,
       "learning_rate": 3.272542485937369e-05,
-      "loss": 0.9775,
+      "loss": 0.9782,
       "num_input_tokens_seen": 73154560,
       "step": 8930
     },
     {
       "epoch": 0.42929171668667465,
-      "grad_norm": 0.4888969957828522,
+      "grad_norm": 0.4872770309448242,
       "learning_rate": 3.2687737594386766e-05,
-      "loss": 0.7946,
+      "loss": 0.7977,
       "num_input_tokens_seen": 73236480,
       "step": 8940
     },
     {
       "epoch": 0.4297719087635054,
-      "grad_norm": 0.9433819055557251,
+      "grad_norm": 1.1870628595352173,
       "learning_rate": 3.2650031024310603e-05,
-      "loss": 1.0062,
+      "loss": 1.0063,
       "num_input_tokens_seen": 73318400,
       "step": 8950
     },
     {
       "epoch": 0.43025210084033616,
-      "grad_norm": 2.151608943939209,
+      "grad_norm": 2.081188678741455,
       "learning_rate": 3.2612305243832176e-05,
-      "loss": 1.0847,
+      "loss": 1.0871,
       "num_input_tokens_seen": 73400320,
       "step": 8960
     },
     {
       "epoch": 0.43073229291716686,
-      "grad_norm": 0.5902963876724243,
+      "grad_norm": 0.5893439650535583,
       "learning_rate": 3.2574560347686725e-05,
-      "loss": 0.9699,
+      "loss": 0.9702,
       "num_input_tokens_seen": 73482240,
       "step": 8970
     },
     {
       "epoch": 0.4312124849939976,
-      "grad_norm": 0.5295826196670532,
+      "grad_norm": 0.5349632501602173,
       "learning_rate": 3.253679643065747e-05,
-      "loss": 0.9454,
+      "loss": 0.9451,
       "num_input_tokens_seen": 73564160,
       "step": 8980
     },
     {
       "epoch": 0.4316926770708283,
-      "grad_norm": 0.5086777210235596,
+      "grad_norm": 0.49540823698043823,
       "learning_rate": 3.249901358757538e-05,
-      "loss": 0.9417,
+      "loss": 0.9423,
       "num_input_tokens_seen": 73646080,
       "step": 8990
     },
     {
       "epoch": 0.43217286914765907,
-      "grad_norm": 0.5244941711425781,
+      "grad_norm": 0.5253874063491821,
       "learning_rate": 3.246121191331902e-05,
-      "loss": 0.9839,
+      "loss": 0.9845,
       "num_input_tokens_seen": 73728000,
       "step": 9000
     },
     {
       "epoch": 0.4326530612244898,
-      "grad_norm": 0.4932080805301666,
+      "grad_norm": 0.5032656788825989,
       "learning_rate": 3.242339150281417e-05,
-      "loss": 0.8661,
+      "loss": 0.8655,
       "num_input_tokens_seen": 73809920,
       "step": 9010
     },
     {
       "epoch": 0.4331332533013205,
-      "grad_norm": 0.47911617159843445,
+      "grad_norm": 0.48877930641174316,
       "learning_rate": 3.238555245103368e-05,
-      "loss": 0.8686,
+      "loss": 0.869,
       "num_input_tokens_seen": 73891840,
       "step": 9020
     },
     {
       "epoch": 0.4336134453781513,
-      "grad_norm": 0.5290552377700806,
+      "grad_norm": 0.564937949180603,
       "learning_rate": 3.234769485299724e-05,
-      "loss": 0.8504,
+      "loss": 0.8508,
       "num_input_tokens_seen": 73973760,
       "step": 9030
     },
     {
       "epoch": 0.434093637454982,
-      "grad_norm": 0.4913333058357239,
+      "grad_norm": 0.5002710819244385,
       "learning_rate": 3.230981880377107e-05,
-      "loss": 1.0346,
+      "loss": 1.035,
       "num_input_tokens_seen": 74055680,
       "step": 9040
     },
     {
       "epoch": 0.4345738295318127,
-      "grad_norm": 0.8791585564613342,
+      "grad_norm": 0.8834782242774963,
       "learning_rate": 3.227192439846775e-05,
-      "loss": 0.9299,
+      "loss": 0.9306,
       "num_input_tokens_seen": 74137600,
       "step": 9050
     },
     {
       "epoch": 0.4350540216086435,
-      "grad_norm": 0.4976731240749359,
+      "grad_norm": 0.4867245554924011,
       "learning_rate": 3.223401173224595e-05,
       "loss": 0.9369,
       "num_input_tokens_seen": 74219520,
@@ -7258,183 +7258,183 @@
     },
     {
       "epoch": 0.4355342136854742,
-      "grad_norm": 0.49430936574935913,
+      "grad_norm": 0.49409911036491394,
       "learning_rate": 3.219608090031021e-05,
-      "loss": 0.8562,
+      "loss": 0.8561,
       "num_input_tokens_seen": 74301440,
       "step": 9070
     },
     {
       "epoch": 0.43601440576230494,
-      "grad_norm": 0.4978736937046051,
+      "grad_norm": 0.4928935766220093,
       "learning_rate": 3.215813199791065e-05,
-      "loss": 1.0076,
+      "loss": 1.0078,
       "num_input_tokens_seen": 74383360,
       "step": 9080
     },
     {
       "epoch": 0.43649459783913563,
-      "grad_norm": 0.5289158225059509,
+      "grad_norm": 0.5236163139343262,
       "learning_rate": 3.212016512034279e-05,
-      "loss": 0.8394,
+      "loss": 0.8402,
       "num_input_tokens_seen": 74465280,
       "step": 9090
     },
     {
       "epoch": 0.4369747899159664,
-      "grad_norm": 0.4874771535396576,
+      "grad_norm": 0.4845563471317291,
       "learning_rate": 3.2082180362947304e-05,
-      "loss": 0.9406,
+      "loss": 0.9409,
       "num_input_tokens_seen": 74547200,
       "step": 9100
     },
     {
       "epoch": 0.43745498199279714,
-      "grad_norm": 0.5047212243080139,
+      "grad_norm": 0.5000022649765015,
       "learning_rate": 3.204417782110974e-05,
-      "loss": 1.0131,
+      "loss": 1.0121,
       "num_input_tokens_seen": 74629120,
       "step": 9110
     },
     {
       "epoch": 0.43793517406962784,
-      "grad_norm": 0.5818035006523132,
+      "grad_norm": 0.5826824307441711,
       "learning_rate": 3.200615759026031e-05,
-      "loss": 1.0524,
+      "loss": 1.0525,
       "num_input_tokens_seen": 74711040,
       "step": 9120
     },
     {
       "epoch": 0.4384153661464586,
-      "grad_norm": 0.5106756687164307,
+      "grad_norm": 0.5062296390533447,
       "learning_rate": 3.1968119765873654e-05,
-      "loss": 0.8463,
+      "loss": 0.8456,
       "num_input_tokens_seen": 74792960,
       "step": 9130
     },
     {
       "epoch": 0.4388955582232893,
-      "grad_norm": 1.4386917352676392,
+      "grad_norm": 1.3879985809326172,
       "learning_rate": 3.193006444346859e-05,
-      "loss": 0.8177,
+      "loss": 0.8187,
       "num_input_tokens_seen": 74874880,
       "step": 9140
     },
     {
       "epoch": 0.43937575030012005,
-      "grad_norm": 0.5303124785423279,
+      "grad_norm": 0.5301570892333984,
       "learning_rate": 3.189199171860787e-05,
-      "loss": 1.1675,
+      "loss": 1.1672,
       "num_input_tokens_seen": 74956800,
       "step": 9150
     },
     {
       "epoch": 0.4398559423769508,
-      "grad_norm": 0.5052891969680786,
+      "grad_norm": 0.5018822550773621,
       "learning_rate": 3.185390168689796e-05,
-      "loss": 0.7782,
+      "loss": 0.7784,
       "num_input_tokens_seen": 75038720,
       "step": 9160
     },
     {
       "epoch": 0.4403361344537815,
-      "grad_norm": 0.5097294449806213,
+      "grad_norm": 0.5099103450775146,
       "learning_rate": 3.1815794443988763e-05,
-      "loss": 0.74,
+      "loss": 0.7409,
       "num_input_tokens_seen": 75120640,
       "step": 9170
     },
     {
       "epoch": 0.44081632653061226,
-      "grad_norm": 1.9583684206008911,
+      "grad_norm": 1.997220516204834,
       "learning_rate": 3.177767008557343e-05,
-      "loss": 0.9299,
+      "loss": 0.9283,
       "num_input_tokens_seen": 75202560,
       "step": 9180
     },
     {
       "epoch": 0.44129651860744296,
-      "grad_norm": 0.5055694580078125,
+      "grad_norm": 0.4805624186992645,
       "learning_rate": 3.1739528707388066e-05,
-      "loss": 0.8375,
+      "loss": 0.8384,
       "num_input_tokens_seen": 75284480,
       "step": 9190
     },
     {
       "epoch": 0.4417767106842737,
-      "grad_norm": 0.5287858247756958,
+      "grad_norm": 0.5211551785469055,
       "learning_rate": 3.1701370405211535e-05,
-      "loss": 0.8294,
+      "loss": 0.8281,
       "num_input_tokens_seen": 75366400,
       "step": 9200
     },
     {
       "epoch": 0.44225690276110446,
-      "grad_norm": 0.5322756171226501,
+      "grad_norm": 0.5320281386375427,
       "learning_rate": 3.166319527486519e-05,
-      "loss": 0.965,
+      "loss": 0.9655,
       "num_input_tokens_seen": 75448320,
       "step": 9210
     },
     {
       "epoch": 0.44273709483793516,
-      "grad_norm": 0.475495845079422,
+      "grad_norm": 0.4732847511768341,
       "learning_rate": 3.162500341221264e-05,
-      "loss": 0.9839,
+      "loss": 0.9852,
       "num_input_tokens_seen": 75530240,
       "step": 9220
     },
     {
       "epoch": 0.4432172869147659,
-      "grad_norm": 0.5287263989448547,
+      "grad_norm": 0.5219468474388123,
       "learning_rate": 3.158679491315951e-05,
-      "loss": 0.9622,
+      "loss": 0.9623,
       "num_input_tokens_seen": 75612160,
       "step": 9230
     },
     {
       "epoch": 0.4436974789915966,
-      "grad_norm": 0.9584441184997559,
+      "grad_norm": 0.9582969546318054,
       "learning_rate": 3.154856987365322e-05,
-      "loss": 0.9561,
+      "loss": 0.9564,
       "num_input_tokens_seen": 75694080,
       "step": 9240
     },
     {
       "epoch": 0.44417767106842737,
-      "grad_norm": 0.5105222463607788,
+      "grad_norm": 0.49347057938575745,
       "learning_rate": 3.151032838968271e-05,
-      "loss": 0.794,
+      "loss": 0.7948,
       "num_input_tokens_seen": 75776000,
       "step": 9250
     },
     {
       "epoch": 0.4446578631452581,
-      "grad_norm": 0.5271983742713928,
+      "grad_norm": 0.5148050785064697,
       "learning_rate": 3.14720705572782e-05,
-      "loss": 1.0292,
+      "loss": 1.0294,
       "num_input_tokens_seen": 75857920,
       "step": 9260
     },
     {
       "epoch": 0.4451380552220888,
-      "grad_norm": 0.6865978240966797,
+      "grad_norm": 0.6848868727684021,
       "learning_rate": 3.1433796472511e-05,
-      "loss": 0.865,
+      "loss": 0.8652,
       "num_input_tokens_seen": 75939840,
       "step": 9270
     },
     {
       "epoch": 0.4456182472989196,
-      "grad_norm": 0.5002774596214294,
+      "grad_norm": 0.5023311972618103,
       "learning_rate": 3.13955062314932e-05,
-      "loss": 1.0853,
+      "loss": 1.0851,
       "num_input_tokens_seen": 76021760,
       "step": 9280
     },
     {
       "epoch": 0.4460984393757503,
-      "grad_norm": 0.5186232924461365,
+      "grad_norm": 0.5274611711502075,
       "learning_rate": 3.135719993037748e-05,
       "loss": 0.9558,
       "num_input_tokens_seen": 76103680,
@@ -7442,7 +7442,7 @@
     },
     {
       "epoch": 0.44657863145258103,
-      "grad_norm": 0.5334696769714355,
+      "grad_norm": 0.535510241985321,
       "learning_rate": 3.131887766535684e-05,
       "loss": 0.9276,
       "num_input_tokens_seen": 76185600,
@@ -7450,143 +7450,143 @@
     },
     {
       "epoch": 0.4470588235294118,
-      "grad_norm": 0.49361923336982727,
+      "grad_norm": 0.48562875390052795,
       "learning_rate": 3.1280539532664366e-05,
-      "loss": 0.8235,
+      "loss": 0.824,
       "num_input_tokens_seen": 76267520,
       "step": 9310
     },
     {
       "epoch": 0.4475390156062425,
-      "grad_norm": 0.4922276735305786,
+      "grad_norm": 0.49112045764923096,
       "learning_rate": 3.1242185628573e-05,
-      "loss": 0.8764,
+      "loss": 0.876,
       "num_input_tokens_seen": 76349440,
       "step": 9320
     },
     {
       "epoch": 0.44801920768307324,
-      "grad_norm": 0.5455789566040039,
+      "grad_norm": 0.5443177819252014,
       "learning_rate": 3.120381604939529e-05,
-      "loss": 0.8174,
+      "loss": 0.8179,
       "num_input_tokens_seen": 76431360,
       "step": 9330
     },
     {
       "epoch": 0.44849939975990394,
-      "grad_norm": 0.5186362862586975,
+      "grad_norm": 0.53428053855896,
       "learning_rate": 3.116543089148312e-05,
-      "loss": 0.9815,
+      "loss": 0.9818,
       "num_input_tokens_seen": 76513280,
       "step": 9340
     },
     {
       "epoch": 0.4489795918367347,
-      "grad_norm": 0.5068888664245605,
+      "grad_norm": 0.5094349980354309,
       "learning_rate": 3.112703025122754e-05,
-      "loss": 0.7665,
+      "loss": 0.767,
       "num_input_tokens_seen": 76595200,
       "step": 9350
     },
     {
       "epoch": 0.44945978391356545,
-      "grad_norm": 0.5032262206077576,
+      "grad_norm": 0.5008413791656494,
       "learning_rate": 3.108861422505842e-05,
-      "loss": 0.7452,
+      "loss": 0.7457,
       "num_input_tokens_seen": 76677120,
       "step": 9360
     },
     {
       "epoch": 0.44993997599039615,
-      "grad_norm": 0.4863367974758148,
+      "grad_norm": 0.4887971580028534,
       "learning_rate": 3.105018290944432e-05,
-      "loss": 0.9044,
+      "loss": 0.9048,
       "num_input_tokens_seen": 76759040,
       "step": 9370
     },
     {
       "epoch": 0.4504201680672269,
-      "grad_norm": 0.49691611528396606,
+      "grad_norm": 0.5146111845970154,
       "learning_rate": 3.1011736400892175e-05,
-      "loss": 0.8608,
+      "loss": 0.86,
       "num_input_tokens_seen": 76840960,
       "step": 9380
     },
     {
       "epoch": 0.4509003601440576,
-      "grad_norm": 0.40895968675613403,
+      "grad_norm": 0.4209975600242615,
       "learning_rate": 3.097327479594707e-05,
-      "loss": 0.7893,
+      "loss": 0.793,
       "num_input_tokens_seen": 76922880,
       "step": 9390
     },
     {
       "epoch": 0.45138055222088835,
-      "grad_norm": 0.4914085865020752,
+      "grad_norm": 0.4876270294189453,
       "learning_rate": 3.093479819119198e-05,
-      "loss": 0.935,
+      "loss": 0.9349,
       "num_input_tokens_seen": 77004800,
       "step": 9400
     },
     {
       "epoch": 0.4518607442977191,
-      "grad_norm": 0.4737802743911743,
+      "grad_norm": 0.47249627113342285,
       "learning_rate": 3.089630668324759e-05,
-      "loss": 0.92,
+      "loss": 0.9205,
       "num_input_tokens_seen": 77086720,
       "step": 9410
     },
     {
       "epoch": 0.4523409363745498,
-      "grad_norm": 0.4802420437335968,
+      "grad_norm": 0.4846927225589752,
       "learning_rate": 3.085780036877197e-05,
-      "loss": 0.8323,
+      "loss": 0.8324,
       "num_input_tokens_seen": 77168640,
       "step": 9420
     },
     {
       "epoch": 0.45282112845138056,
-      "grad_norm": 0.8194192051887512,
+      "grad_norm": 0.8228050470352173,
       "learning_rate": 3.0819279344460396e-05,
-      "loss": 0.8059,
+      "loss": 0.8051,
       "num_input_tokens_seen": 77250560,
       "step": 9430
     },
     {
       "epoch": 0.45330132052821126,
-      "grad_norm": 0.500428318977356,
+      "grad_norm": 0.5003042221069336,
       "learning_rate": 3.078074370704507e-05,
-      "loss": 0.9285,
+      "loss": 0.9294,
       "num_input_tokens_seen": 77332480,
       "step": 9440
     },
     {
       "epoch": 0.453781512605042,
-      "grad_norm": 0.6494525671005249,
+      "grad_norm": 0.6587066054344177,
       "learning_rate": 3.07421935532949e-05,
-      "loss": 0.8807,
+      "loss": 0.8818,
       "num_input_tokens_seen": 77414400,
       "step": 9450
     },
     {
       "epoch": 0.45426170468187277,
-      "grad_norm": 0.6894603967666626,
+      "grad_norm": 0.6838945746421814,
       "learning_rate": 3.0703628980015214e-05,
-      "loss": 0.9211,
+      "loss": 0.9213,
       "num_input_tokens_seen": 77496320,
       "step": 9460
     },
     {
       "epoch": 0.45474189675870347,
-      "grad_norm": 0.4977949261665344,
+      "grad_norm": 0.5055971145629883,
       "learning_rate": 3.0665050084047605e-05,
-      "loss": 0.8221,
+      "loss": 0.8218,
       "num_input_tokens_seen": 77578240,
       "step": 9470
     },
     {
       "epoch": 0.4552220888355342,
-      "grad_norm": 0.49541839957237244,
+      "grad_norm": 0.5079997181892395,
       "learning_rate": 3.062645696226959e-05,
       "loss": 0.7467,
       "num_input_tokens_seen": 77660160,
@@ -7594,135 +7594,135 @@
     },
     {
       "epoch": 0.4557022809123649,
-      "grad_norm": 0.48292815685272217,
+      "grad_norm": 0.487131267786026,
       "learning_rate": 3.0587849711594425e-05,
-      "loss": 0.982,
+      "loss": 0.9819,
       "num_input_tokens_seen": 77742080,
       "step": 9490
     },
     {
       "epoch": 0.4561824729891957,
-      "grad_norm": 0.5033644437789917,
+      "grad_norm": 0.5070083737373352,
       "learning_rate": 3.054922842897084e-05,
-      "loss": 0.863,
+      "loss": 0.8633,
       "num_input_tokens_seen": 77824000,
       "step": 9500
     },
     {
       "epoch": 0.45666266506602643,
-      "grad_norm": 0.5042487978935242,
+      "grad_norm": 0.508251965045929,
       "learning_rate": 3.051059321138281e-05,
-      "loss": 1.0322,
+      "loss": 1.0312,
       "num_input_tokens_seen": 77905920,
       "step": 9510
     },
     {
       "epoch": 0.45714285714285713,
-      "grad_norm": 0.49270498752593994,
+      "grad_norm": 0.4875255823135376,
       "learning_rate": 3.047194415584929e-05,
-      "loss": 0.9702,
+      "loss": 0.9672,
       "num_input_tokens_seen": 77987840,
       "step": 9520
     },
     {
       "epoch": 0.4576230492196879,
-      "grad_norm": 0.5053675174713135,
+      "grad_norm": 0.5202212333679199,
       "learning_rate": 3.0433281359424008e-05,
-      "loss": 0.9051,
+      "loss": 0.9048,
       "num_input_tokens_seen": 78069760,
       "step": 9530
     },
     {
       "epoch": 0.4581032412965186,
-      "grad_norm": 0.4840089976787567,
+      "grad_norm": 0.484787255525589,
       "learning_rate": 3.0394604919195156e-05,
-      "loss": 0.785,
+      "loss": 0.7858,
       "num_input_tokens_seen": 78151680,
       "step": 9540
     },
     {
       "epoch": 0.45858343337334934,
-      "grad_norm": 0.4885089099407196,
+      "grad_norm": 0.4984806776046753,
       "learning_rate": 3.0355914932285228e-05,
-      "loss": 0.9501,
+      "loss": 0.9508,
       "num_input_tokens_seen": 78233600,
       "step": 9550
     },
     {
       "epoch": 0.4590636254501801,
-      "grad_norm": 0.5052339434623718,
+      "grad_norm": 0.5085755586624146,
       "learning_rate": 3.0317211495850717e-05,
-      "loss": 0.8565,
+      "loss": 0.8558,
       "num_input_tokens_seen": 78315520,
       "step": 9560
     },
     {
       "epoch": 0.4595438175270108,
-      "grad_norm": 0.49368199706077576,
+      "grad_norm": 0.4945325553417206,
       "learning_rate": 3.02784947070819e-05,
-      "loss": 0.7823,
+      "loss": 0.7818,
       "num_input_tokens_seen": 78397440,
       "step": 9570
     },
     {
       "epoch": 0.46002400960384154,
-      "grad_norm": 0.48532119393348694,
+      "grad_norm": 0.4864473044872284,
       "learning_rate": 3.0239764663202562e-05,
-      "loss": 0.9049,
+      "loss": 0.9045,
       "num_input_tokens_seen": 78479360,
       "step": 9580
     },
     {
       "epoch": 0.46050420168067224,
-      "grad_norm": 0.48147672414779663,
+      "grad_norm": 0.4791228175163269,
       "learning_rate": 3.0201021461469803e-05,
-      "loss": 0.9584,
+      "loss": 0.9581,
       "num_input_tokens_seen": 78561280,
       "step": 9590
     },
     {
       "epoch": 0.460984393757503,
-      "grad_norm": 0.49447527527809143,
+      "grad_norm": 0.48919251561164856,
       "learning_rate": 3.0162265199173738e-05,
-      "loss": 0.8946,
+      "loss": 0.8943,
       "num_input_tokens_seen": 78643200,
       "step": 9600
     },
     {
       "epoch": 0.46146458583433375,
-      "grad_norm": 0.44538614153862,
+      "grad_norm": 0.4210091531276703,
       "learning_rate": 3.0123495973637305e-05,
-      "loss": 0.973,
+      "loss": 0.9728,
       "num_input_tokens_seen": 78725120,
       "step": 9610
     },
     {
       "epoch": 0.46194477791116445,
-      "grad_norm": 0.4778103530406952,
+      "grad_norm": 0.4873916208744049,
       "learning_rate": 3.008471388221597e-05,
-      "loss": 0.8953,
+      "loss": 0.8959,
       "num_input_tokens_seen": 78807040,
       "step": 9620
     },
     {
       "epoch": 0.4624249699879952,
-      "grad_norm": 0.5161569714546204,
+      "grad_norm": 0.5187123417854309,
       "learning_rate": 3.0045919022297524e-05,
-      "loss": 0.7189,
+      "loss": 0.7175,
       "num_input_tokens_seen": 78888960,
       "step": 9630
     },
     {
       "epoch": 0.4629051620648259,
-      "grad_norm": 0.47274622321128845,
+      "grad_norm": 0.46600180864334106,
       "learning_rate": 3.0007111491301816e-05,
-      "loss": 0.8521,
+      "loss": 0.8525,
       "num_input_tokens_seen": 78970880,
       "step": 9640
     },
     {
       "epoch": 0.46338535414165666,
-      "grad_norm": 0.4988880753517151,
+      "grad_norm": 0.5010501146316528,
       "learning_rate": 2.9968291386680503e-05,
       "loss": 0.941,
       "num_input_tokens_seen": 79052800,
@@ -7730,231 +7730,231 @@
     },
     {
       "epoch": 0.4638655462184874,
-      "grad_norm": 0.6255693435668945,
+      "grad_norm": 0.6152305006980896,
       "learning_rate": 2.9929458805916837e-05,
-      "loss": 1.0642,
+      "loss": 1.066,
       "num_input_tokens_seen": 79134720,
       "step": 9660
     },
     {
       "epoch": 0.4643457382953181,
-      "grad_norm": 0.39187878370285034,
+      "grad_norm": 0.40477296710014343,
       "learning_rate": 2.9890613846525395e-05,
-      "loss": 0.7306,
+      "loss": 0.7304,
       "num_input_tokens_seen": 79216640,
       "step": 9670
     },
     {
       "epoch": 0.46482593037214887,
-      "grad_norm": 1.9925369024276733,
+      "grad_norm": 2.6527504920959473,
       "learning_rate": 2.9851756606051817e-05,
-      "loss": 0.8926,
+      "loss": 0.8932,
       "num_input_tokens_seen": 79298560,
       "step": 9680
     },
     {
       "epoch": 0.46530612244897956,
-      "grad_norm": 0.48940789699554443,
+      "grad_norm": 0.4877612292766571,
       "learning_rate": 2.9812887182072607e-05,
-      "loss": 0.9238,
+      "loss": 0.9239,
       "num_input_tokens_seen": 79380480,
       "step": 9690
     },
     {
       "epoch": 0.4657863145258103,
-      "grad_norm": 0.5379921197891235,
+      "grad_norm": 0.5364747643470764,
       "learning_rate": 2.9774005672194854e-05,
-      "loss": 0.8351,
+      "loss": 0.8346,
       "num_input_tokens_seen": 79462400,
       "step": 9700
     },
     {
       "epoch": 0.4662665066026411,
-      "grad_norm": 0.5129654407501221,
+      "grad_norm": 0.5188019871711731,
       "learning_rate": 2.9735112174056006e-05,
-      "loss": 0.8861,
+      "loss": 0.886,
       "num_input_tokens_seen": 79544320,
       "step": 9710
     },
     {
       "epoch": 0.46674669867947177,
-      "grad_norm": 0.4713475704193115,
+      "grad_norm": 0.47986412048339844,
       "learning_rate": 2.96962067853236e-05,
-      "loss": 0.8593,
+      "loss": 0.8596,
       "num_input_tokens_seen": 79626240,
       "step": 9720
     },
     {
       "epoch": 0.4672268907563025,
-      "grad_norm": 4.391692161560059,
+      "grad_norm": 4.085470199584961,
       "learning_rate": 2.9657289603695037e-05,
-      "loss": 0.8369,
+      "loss": 0.8394,
       "num_input_tokens_seen": 79708160,
       "step": 9730
     },
     {
       "epoch": 0.4677070828331333,
-      "grad_norm": 0.5324821472167969,
+      "grad_norm": 0.5369517803192139,
       "learning_rate": 2.9618360726897344e-05,
-      "loss": 1.3,
+      "loss": 1.2984,
       "num_input_tokens_seen": 79790080,
       "step": 9740
     },
     {
       "epoch": 0.468187274909964,
-      "grad_norm": 0.5004756450653076,
+      "grad_norm": 0.4899786114692688,
       "learning_rate": 2.957942025268689e-05,
-      "loss": 0.9522,
+      "loss": 0.9511,
       "num_input_tokens_seen": 79872000,
       "step": 9750
     },
     {
       "epoch": 0.46866746698679473,
-      "grad_norm": 0.5157467722892761,
+      "grad_norm": 0.5062848329544067,
       "learning_rate": 2.9540468278849208e-05,
-      "loss": 0.8728,
+      "loss": 0.8727,
       "num_input_tokens_seen": 79953920,
       "step": 9760
     },
     {
       "epoch": 0.46914765906362543,
-      "grad_norm": 0.4873751699924469,
+      "grad_norm": 0.48712435364723206,
       "learning_rate": 2.950150490319866e-05,
-      "loss": 0.8798,
+      "loss": 0.8799,
       "num_input_tokens_seen": 80035840,
       "step": 9770
     },
     {
       "epoch": 0.4696278511404562,
-      "grad_norm": 0.9361140727996826,
+      "grad_norm": 0.9658442735671997,
       "learning_rate": 2.9462530223578273e-05,
-      "loss": 0.9698,
+      "loss": 0.9699,
       "num_input_tokens_seen": 80117760,
       "step": 9780
     },
     {
       "epoch": 0.47010804321728694,
-      "grad_norm": 0.5102428197860718,
+      "grad_norm": 0.5106319785118103,
       "learning_rate": 2.9423544337859454e-05,
-      "loss": 0.8272,
+      "loss": 0.8275,
       "num_input_tokens_seen": 80199680,
       "step": 9790
     },
     {
       "epoch": 0.47058823529411764,
-      "grad_norm": 0.5207327008247375,
+      "grad_norm": 0.519828736782074,
       "learning_rate": 2.938454734394174e-05,
-      "loss": 0.8527,
+      "loss": 0.8525,
       "num_input_tokens_seen": 80281600,
       "step": 9800
     },
     {
       "epoch": 0.4710684273709484,
-      "grad_norm": 0.5059601068496704,
+      "grad_norm": 0.5150085687637329,
       "learning_rate": 2.9345539339752575e-05,
-      "loss": 0.9777,
+      "loss": 0.9778,
       "num_input_tokens_seen": 80363520,
       "step": 9810
     },
     {
       "epoch": 0.4715486194477791,
-      "grad_norm": 0.5397816896438599,
+      "grad_norm": 0.5509699583053589,
       "learning_rate": 2.9306520423247045e-05,
-      "loss": 0.8295,
+      "loss": 0.829,
       "num_input_tokens_seen": 80445440,
       "step": 9820
     },
     {
       "epoch": 0.47202881152460985,
-      "grad_norm": 0.4623885750770569,
+      "grad_norm": 0.4511142671108246,
       "learning_rate": 2.9267490692407635e-05,
-      "loss": 0.9394,
+      "loss": 0.9411,
       "num_input_tokens_seen": 80527360,
       "step": 9830
     },
     {
       "epoch": 0.4725090036014406,
-      "grad_norm": 0.48748502135276794,
+      "grad_norm": 0.48220759630203247,
       "learning_rate": 2.9228450245243993e-05,
-      "loss": 0.9649,
+      "loss": 0.9641,
       "num_input_tokens_seen": 80609280,
       "step": 9840
     },
     {
       "epoch": 0.4729891956782713,
-      "grad_norm": 0.493064820766449,
+      "grad_norm": 0.4902176856994629,
       "learning_rate": 2.9189399179792676e-05,
-      "loss": 1.1218,
+      "loss": 1.1205,
       "num_input_tokens_seen": 80691200,
       "step": 9850
     },
     {
       "epoch": 0.47346938775510206,
-      "grad_norm": 0.5036744475364685,
+      "grad_norm": 0.49037599563598633,
       "learning_rate": 2.9150337594116904e-05,
-      "loss": 0.9087,
+      "loss": 0.9089,
       "num_input_tokens_seen": 80773120,
       "step": 9860
     },
     {
       "epoch": 0.47394957983193275,
-      "grad_norm": 0.5360898971557617,
+      "grad_norm": 0.525600016117096,
       "learning_rate": 2.9111265586306314e-05,
-      "loss": 0.849,
+      "loss": 0.8489,
       "num_input_tokens_seen": 80855040,
       "step": 9870
     },
     {
       "epoch": 0.4744297719087635,
-      "grad_norm": 0.521198034286499,
+      "grad_norm": 0.5158500075340271,
       "learning_rate": 2.9072183254476713e-05,
-      "loss": 0.8285,
+      "loss": 0.8287,
       "num_input_tokens_seen": 80936960,
       "step": 9880
     },
     {
       "epoch": 0.47490996398559426,
-      "grad_norm": 0.4821554720401764,
+      "grad_norm": 0.4754113554954529,
       "learning_rate": 2.903309069676984e-05,
-      "loss": 0.9502,
+      "loss": 0.9506,
       "num_input_tokens_seen": 81018880,
       "step": 9890
     },
     {
       "epoch": 0.47539015606242496,
-      "grad_norm": 0.483500212430954,
+      "grad_norm": 0.48330146074295044,
       "learning_rate": 2.8993988011353112e-05,
-      "loss": 0.8596,
+      "loss": 0.8601,
       "num_input_tokens_seen": 81100800,
       "step": 9900
     },
     {
       "epoch": 0.4758703481392557,
-      "grad_norm": 0.5084524750709534,
+      "grad_norm": 0.4962432086467743,
       "learning_rate": 2.8954875296419364e-05,
-      "loss": 0.8992,
+      "loss": 0.9003,
       "num_input_tokens_seen": 81182720,
       "step": 9910
     },
     {
       "epoch": 0.4763505402160864,
-      "grad_norm": 0.5362197160720825,
+      "grad_norm": 0.5256210565567017,
       "learning_rate": 2.8915752650186635e-05,
-      "loss": 0.9792,
+      "loss": 0.9797,
       "num_input_tokens_seen": 81264640,
       "step": 9920
     },
     {
       "epoch": 0.47683073229291717,
-      "grad_norm": 0.5296033024787903,
+      "grad_norm": 0.529623806476593,
       "learning_rate": 2.8876620170897895e-05,
-      "loss": 0.7938,
+      "loss": 0.7935,
       "num_input_tokens_seen": 81346560,
       "step": 9930
     },
     {
       "epoch": 0.4773109243697479,
-      "grad_norm": 0.5047898888587952,
+      "grad_norm": 0.4984219968318939,
       "learning_rate": 2.88374779568208e-05,
       "loss": 0.8632,
       "num_input_tokens_seen": 81428480,
@@ -7962,63 +7962,63 @@
     },
     {
       "epoch": 0.4777911164465786,
-      "grad_norm": 0.5180176496505737,
+      "grad_norm": 0.5229691863059998,
       "learning_rate": 2.879832610624747e-05,
-      "loss": 0.8698,
+      "loss": 0.8712,
       "num_input_tokens_seen": 81510400,
       "step": 9950
     },
     {
       "epoch": 0.4782713085234094,
-      "grad_norm": 0.4948786795139313,
+      "grad_norm": 0.49954238533973694,
       "learning_rate": 2.8759164717494202e-05,
-      "loss": 0.8159,
+      "loss": 0.8157,
       "num_input_tokens_seen": 81592320,
       "step": 9960
     },
     {
       "epoch": 0.4787515006002401,
-      "grad_norm": 0.5000596046447754,
+      "grad_norm": 0.5026626586914062,
       "learning_rate": 2.8719993888901258e-05,
-      "loss": 1.0499,
+      "loss": 1.05,
       "num_input_tokens_seen": 81674240,
       "step": 9970
     },
     {
       "epoch": 0.47923169267707083,
-      "grad_norm": 0.5099364519119263,
+      "grad_norm": 0.5073908567428589,
       "learning_rate": 2.86808137188326e-05,
-      "loss": 0.8724,
+      "loss": 0.8731,
       "num_input_tokens_seen": 81756160,
       "step": 9980
     },
     {
       "epoch": 0.4797118847539016,
-      "grad_norm": 0.48814287781715393,
+      "grad_norm": 0.48957353830337524,
       "learning_rate": 2.8641624305675657e-05,
-      "loss": 0.9881,
+      "loss": 0.9884,
       "num_input_tokens_seen": 81838080,
       "step": 9990
     },
     {
       "epoch": 0.4801920768307323,
-      "grad_norm": 0.46351805329322815,
+      "grad_norm": 0.46365344524383545,
       "learning_rate": 2.8602425747841057e-05,
-      "loss": 0.7335,
+      "loss": 0.7331,
       "num_input_tokens_seen": 81920000,
       "step": 10000
     },
     {
       "epoch": 0.48067226890756304,
-      "grad_norm": 0.4863956868648529,
+      "grad_norm": 0.4974704682826996,
       "learning_rate": 2.8563218143762383e-05,
-      "loss": 0.8356,
+      "loss": 0.8353,
       "num_input_tokens_seen": 82001920,
       "step": 10010
     },
     {
       "epoch": 0.48115246098439374,
-      "grad_norm": 0.5032678246498108,
+      "grad_norm": 0.5029409527778625,
       "learning_rate": 2.852400159189597e-05,
       "loss": 0.8458,
       "num_input_tokens_seen": 82083840,
@@ -8026,39 +8026,39 @@
     },
     {
       "epoch": 0.4816326530612245,
-      "grad_norm": 0.4915202558040619,
+      "grad_norm": 0.4853820502758026,
       "learning_rate": 2.848477619072059e-05,
-      "loss": 0.995,
+      "loss": 0.9951,
       "num_input_tokens_seen": 82165760,
       "step": 10030
     },
     {
       "epoch": 0.48211284513805525,
-      "grad_norm": 0.9962053894996643,
+      "grad_norm": 0.5146056413650513,
       "learning_rate": 2.8445542038737245e-05,
-      "loss": 1.0218,
+      "loss": 1.0225,
       "num_input_tokens_seen": 82247680,
       "step": 10040
     },
     {
       "epoch": 0.48259303721488594,
-      "grad_norm": 0.49556922912597656,
+      "grad_norm": 0.4969925284385681,
       "learning_rate": 2.8406299234468915e-05,
-      "loss": 0.9613,
+      "loss": 0.9611,
       "num_input_tokens_seen": 82329600,
       "step": 10050
     },
     {
       "epoch": 0.4830732292917167,
-      "grad_norm": 0.5691222548484802,
+      "grad_norm": 0.5693634152412415,
       "learning_rate": 2.8367047876460305e-05,
-      "loss": 1.0402,
+      "loss": 1.0389,
       "num_input_tokens_seen": 82411520,
       "step": 10060
     },
     {
       "epoch": 0.4835534213685474,
-      "grad_norm": 0.4867800176143646,
+      "grad_norm": 0.48920491337776184,
       "learning_rate": 2.8327788063277594e-05,
       "loss": 0.9294,
       "num_input_tokens_seen": 82493440,
@@ -8066,7 +8066,7 @@
     },
     {
       "epoch": 0.48403361344537815,
-      "grad_norm": 0.5041534304618835,
+      "grad_norm": 0.4980778098106384,
       "learning_rate": 2.8288519893508212e-05,
       "loss": 0.8672,
       "num_input_tokens_seen": 82575360,
@@ -8074,15 +8074,15 @@
     },
     {
       "epoch": 0.4845138055222089,
-      "grad_norm": 0.44031843543052673,
+      "grad_norm": 0.45296671986579895,
       "learning_rate": 2.8249243465760566e-05,
-      "loss": 0.8052,
+      "loss": 0.805,
       "num_input_tokens_seen": 82657280,
       "step": 10090
     },
     {
       "epoch": 0.4849939975990396,
-      "grad_norm": 0.48101985454559326,
+      "grad_norm": 0.4847956895828247,
       "learning_rate": 2.8209958878663778e-05,
       "loss": 0.8885,
       "num_input_tokens_seen": 82739200,
@@ -8090,15 +8090,15 @@
     },
     {
       "epoch": 0.48547418967587036,
-      "grad_norm": 1.5555214881896973,
+      "grad_norm": 1.503799319267273,
       "learning_rate": 2.817066623086748e-05,
-      "loss": 0.8561,
+      "loss": 0.855,
       "num_input_tokens_seen": 82821120,
       "step": 10110
     },
     {
       "epoch": 0.48595438175270106,
-      "grad_norm": 0.48857659101486206,
+      "grad_norm": 0.4808594286441803,
       "learning_rate": 2.813136562104155e-05,
       "loss": 0.8841,
       "num_input_tokens_seen": 82903040,
@@ -8106,79 +8106,79 @@
     },
     {
       "epoch": 0.4864345738295318,
-      "grad_norm": 0.49165698885917664,
+      "grad_norm": 0.4852427542209625,
       "learning_rate": 2.8092057147875856e-05,
-      "loss": 0.9224,
+      "loss": 0.9228,
       "num_input_tokens_seen": 82984960,
       "step": 10130
     },
     {
       "epoch": 0.48691476590636257,
-      "grad_norm": 0.5225831270217896,
+      "grad_norm": 0.5168618559837341,
       "learning_rate": 2.8052740910079994e-05,
-      "loss": 0.9768,
+      "loss": 0.9771,
       "num_input_tokens_seen": 83066880,
       "step": 10140
     },
     {
       "epoch": 0.48739495798319327,
-      "grad_norm": 0.5045091509819031,
+      "grad_norm": 0.4985884428024292,
       "learning_rate": 2.8013417006383076e-05,
-      "loss": 0.9255,
+      "loss": 0.9318,
       "num_input_tokens_seen": 83148800,
       "step": 10150
     },
     {
       "epoch": 0.487875150060024,
-      "grad_norm": 0.5114516615867615,
+      "grad_norm": 0.49745380878448486,
       "learning_rate": 2.7974085535533473e-05,
-      "loss": 0.8565,
+      "loss": 0.8569,
       "num_input_tokens_seen": 83230720,
       "step": 10160
     },
     {
       "epoch": 0.4883553421368547,
-      "grad_norm": 0.5040556788444519,
+      "grad_norm": 0.48418742418289185,
       "learning_rate": 2.7934746596298535e-05,
-      "loss": 1.0491,
+      "loss": 1.0495,
       "num_input_tokens_seen": 83312640,
       "step": 10170
     },
     {
       "epoch": 0.4888355342136855,
-      "grad_norm": 1.6331427097320557,
+      "grad_norm": 1.714831829071045,
       "learning_rate": 2.789540028746438e-05,
-      "loss": 0.8835,
+      "loss": 0.8839,
       "num_input_tokens_seen": 83394560,
       "step": 10180
     },
     {
       "epoch": 0.48931572629051623,
-      "grad_norm": 0.6383754014968872,
+      "grad_norm": 0.6471736431121826,
       "learning_rate": 2.785604670783563e-05,
-      "loss": 0.9588,
+      "loss": 0.9594,
       "num_input_tokens_seen": 83476480,
       "step": 10190
     },
     {
       "epoch": 0.4897959183673469,
-      "grad_norm": 0.5155805349349976,
+      "grad_norm": 0.5089517831802368,
       "learning_rate": 2.7816685956235165e-05,
-      "loss": 0.7832,
+      "loss": 0.7835,
       "num_input_tokens_seen": 83558400,
       "step": 10200
     },
     {
       "epoch": 0.4902761104441777,
-      "grad_norm": 0.5443750619888306,
+      "grad_norm": 0.5460789203643799,
       "learning_rate": 2.7777318131503873e-05,
-      "loss": 0.9374,
+      "loss": 0.9379,
       "num_input_tokens_seen": 83640320,
       "step": 10210
     },
     {
       "epoch": 0.4907563025210084,
-      "grad_norm": 0.4726339280605316,
+      "grad_norm": 0.4767736792564392,
       "learning_rate": 2.773794333250041e-05,
       "loss": 0.767,
       "num_input_tokens_seen": 83722240,
@@ -8186,15 +8186,15 @@
     },
     {
       "epoch": 0.49123649459783914,
-      "grad_norm": 0.4907895028591156,
+      "grad_norm": 0.5004221200942993,
       "learning_rate": 2.769856165810093e-05,
-      "loss": 0.9588,
+      "loss": 0.9591,
       "num_input_tokens_seen": 83804160,
       "step": 10230
     },
     {
       "epoch": 0.4917166866746699,
-      "grad_norm": 0.5048187375068665,
+      "grad_norm": 0.5016967058181763,
       "learning_rate": 2.765917320719887e-05,
       "loss": 0.929,
       "num_input_tokens_seen": 83886080,
@@ -8202,23 +8202,23 @@
     },
     {
       "epoch": 0.4921968787515006,
-      "grad_norm": 0.4887201189994812,
+      "grad_norm": 0.4898025393486023,
       "learning_rate": 2.7619778078704685e-05,
-      "loss": 0.8619,
+      "loss": 0.8609,
       "num_input_tokens_seen": 83968000,
       "step": 10250
     },
     {
       "epoch": 0.49267707082833134,
-      "grad_norm": 0.5038937926292419,
+      "grad_norm": 0.4977264106273651,
       "learning_rate": 2.7580376371545573e-05,
-      "loss": 0.7223,
+      "loss": 0.7207,
       "num_input_tokens_seen": 84049920,
       "step": 10260
     },
     {
       "epoch": 0.49315726290516204,
-      "grad_norm": 0.505540668964386,
+      "grad_norm": 0.5083197951316833,
       "learning_rate": 2.7540968184665283e-05,
       "loss": 0.8415,
       "num_input_tokens_seen": 84131840,
@@ -8226,79 +8226,79 @@
     },
     {
       "epoch": 0.4936374549819928,
-      "grad_norm": 0.4871721565723419,
+      "grad_norm": 0.4886813461780548,
       "learning_rate": 2.7501553617023816e-05,
-      "loss": 0.9001,
+      "loss": 0.9007,
       "num_input_tokens_seen": 84213760,
       "step": 10280
     },
     {
       "epoch": 0.49411764705882355,
-      "grad_norm": 0.4944019019603729,
+      "grad_norm": 0.48880535364151,
       "learning_rate": 2.7462132767597205e-05,
-      "loss": 0.8124,
+      "loss": 0.8125,
       "num_input_tokens_seen": 84295680,
       "step": 10290
     },
     {
       "epoch": 0.49459783913565425,
-      "grad_norm": 0.5085113644599915,
+      "grad_norm": 0.5156389474868774,
       "learning_rate": 2.742270573537724e-05,
-      "loss": 0.8611,
+      "loss": 0.8609,
       "num_input_tokens_seen": 84377600,
       "step": 10300
     },
     {
       "epoch": 0.495078031212485,
-      "grad_norm": 0.4947553277015686,
+      "grad_norm": 0.5027367472648621,
       "learning_rate": 2.7383272619371276e-05,
-      "loss": 0.9993,
+      "loss": 0.9997,
       "num_input_tokens_seen": 84459520,
       "step": 10310
     },
     {
       "epoch": 0.4955582232893157,
-      "grad_norm": 0.4945763945579529,
+      "grad_norm": 0.49897968769073486,
       "learning_rate": 2.7343833518601913e-05,
-      "loss": 0.8757,
+      "loss": 0.8754,
       "num_input_tokens_seen": 84541440,
       "step": 10320
     },
     {
       "epoch": 0.49603841536614646,
-      "grad_norm": 0.47624143958091736,
+      "grad_norm": 0.4910091459751129,
       "learning_rate": 2.7304388532106768e-05,
-      "loss": 1.1208,
+      "loss": 1.121,
       "num_input_tokens_seen": 84623360,
       "step": 10330
     },
     {
       "epoch": 0.4965186074429772,
-      "grad_norm": 0.5344643592834473,
+      "grad_norm": 0.5233538150787354,
       "learning_rate": 2.726493775893828e-05,
-      "loss": 0.7873,
+      "loss": 0.7872,
       "num_input_tokens_seen": 84705280,
       "step": 10340
     },
     {
       "epoch": 0.4969987995198079,
-      "grad_norm": 0.5042626261711121,
+      "grad_norm": 0.5136865377426147,
       "learning_rate": 2.7225481298163387e-05,
-      "loss": 0.9001,
+      "loss": 0.9003,
       "num_input_tokens_seen": 84787200,
       "step": 10350
     },
     {
       "epoch": 0.49747899159663866,
-      "grad_norm": 0.4883025884628296,
+      "grad_norm": 0.49212074279785156,
       "learning_rate": 2.718601924886332e-05,
-      "loss": 1.0793,
+      "loss": 1.0785,
       "num_input_tokens_seen": 84869120,
       "step": 10360
     },
     {
       "epoch": 0.49795918367346936,
-      "grad_norm": 0.512522280216217,
+      "grad_norm": 0.5100103616714478,
       "learning_rate": 2.7146551710133346e-05,
       "loss": 0.8888,
       "num_input_tokens_seen": 84951040,
@@ -8306,303 +8306,303 @@
     },
     {
       "epoch": 0.4984393757503001,
-      "grad_norm": 0.47969046235084534,
+      "grad_norm": 0.4852333962917328,
       "learning_rate": 2.7107078781082508e-05,
-      "loss": 0.8732,
+      "loss": 0.8726,
       "num_input_tokens_seen": 85032960,
       "step": 10380
     },
     {
       "epoch": 0.49891956782713087,
-      "grad_norm": 0.5086472034454346,
+      "grad_norm": 0.5148525238037109,
       "learning_rate": 2.7067600560833384e-05,
-      "loss": 0.9422,
+      "loss": 0.942,
       "num_input_tokens_seen": 85114880,
       "step": 10390
     },
     {
       "epoch": 0.49939975990396157,
-      "grad_norm": 0.5234156250953674,
+      "grad_norm": 0.5078142285346985,
       "learning_rate": 2.7028117148521863e-05,
-      "loss": 0.8577,
+      "loss": 0.8575,
       "num_input_tokens_seen": 85196800,
       "step": 10400
     },
     {
       "epoch": 0.4998799519807923,
-      "grad_norm": 0.8362540006637573,
+      "grad_norm": 0.8435948491096497,
       "learning_rate": 2.698862864329685e-05,
-      "loss": 0.9081,
+      "loss": 0.9079,
       "num_input_tokens_seen": 85278720,
       "step": 10410
     },
     {
       "epoch": 0.5003601440576231,
-      "grad_norm": 0.5078486800193787,
+      "grad_norm": 0.5046183466911316,
       "learning_rate": 2.6949135144320026e-05,
-      "loss": 0.9876,
+      "loss": 0.9877,
       "num_input_tokens_seen": 85360640,
       "step": 10420
     },
     {
       "epoch": 0.5008403361344538,
-      "grad_norm": 0.49120739102363586,
+      "grad_norm": 0.4874359369277954,
       "learning_rate": 2.6909636750765653e-05,
-      "loss": 0.8621,
+      "loss": 0.8622,
       "num_input_tokens_seen": 85442560,
       "step": 10430
     },
     {
       "epoch": 0.5013205282112845,
-      "grad_norm": 0.5183668732643127,
+      "grad_norm": 0.504699170589447,
       "learning_rate": 2.6870133561820243e-05,
-      "loss": 0.9355,
+      "loss": 0.9352,
       "num_input_tokens_seen": 85524480,
       "step": 10440
     },
     {
       "epoch": 0.5018007202881153,
-      "grad_norm": 0.8437354564666748,
+      "grad_norm": 0.8512853384017944,
       "learning_rate": 2.683062567668238e-05,
-      "loss": 1.0307,
+      "loss": 1.031,
       "num_input_tokens_seen": 85606400,
       "step": 10450
     },
     {
       "epoch": 0.502280912364946,
-      "grad_norm": 0.4748769700527191,
+      "grad_norm": 0.4725627303123474,
       "learning_rate": 2.679111319456242e-05,
-      "loss": 0.8886,
+      "loss": 0.8878,
       "num_input_tokens_seen": 85688320,
       "step": 10460
     },
     {
       "epoch": 0.5027611044417767,
-      "grad_norm": 0.4411871135234833,
+      "grad_norm": 0.4424259066581726,
       "learning_rate": 2.6751596214682278e-05,
-      "loss": 0.6568,
+      "loss": 0.6566,
       "num_input_tokens_seen": 85770240,
       "step": 10470
     },
     {
       "epoch": 0.5032412965186075,
-      "grad_norm": 0.3230482041835785,
+      "grad_norm": 0.3668357729911804,
       "learning_rate": 2.671207483627515e-05,
-      "loss": 0.9178,
+      "loss": 0.9173,
       "num_input_tokens_seen": 85852160,
       "step": 10480
     },
     {
       "epoch": 0.5037214885954382,
-      "grad_norm": 0.6780135631561279,
+      "grad_norm": 0.6287544369697571,
       "learning_rate": 2.6672549158585293e-05,
-      "loss": 0.832,
+      "loss": 0.8325,
       "num_input_tokens_seen": 85934080,
       "step": 10490
     },
     {
       "epoch": 0.5042016806722689,
-      "grad_norm": 0.47888603806495667,
+      "grad_norm": 0.4674912989139557,
       "learning_rate": 2.663301928086774e-05,
-      "loss": 0.7443,
+      "loss": 0.7438,
       "num_input_tokens_seen": 86016000,
       "step": 10500
     },
     {
       "epoch": 0.5046818727490996,
-      "grad_norm": 0.4726419448852539,
+      "grad_norm": 0.46498844027519226,
       "learning_rate": 2.6593485302388087e-05,
-      "loss": 0.8535,
+      "loss": 0.8527,
       "num_input_tokens_seen": 86097920,
       "step": 10510
     },
     {
       "epoch": 0.5051620648259304,
-      "grad_norm": 0.4687502980232239,
+      "grad_norm": 0.45717716217041016,
       "learning_rate": 2.6553947322422223e-05,
-      "loss": 0.9447,
+      "loss": 0.9448,
       "num_input_tokens_seen": 86179840,
       "step": 10520
     },
     {
       "epoch": 0.5056422569027611,
-      "grad_norm": 0.5022156834602356,
+      "grad_norm": 0.4967118501663208,
       "learning_rate": 2.6514405440256086e-05,
-      "loss": 0.8657,
+      "loss": 0.8656,
       "num_input_tokens_seen": 86261760,
       "step": 10530
     },
     {
       "epoch": 0.5061224489795918,
-      "grad_norm": 0.49966397881507874,
+      "grad_norm": 0.49645423889160156,
       "learning_rate": 2.6474859755185415e-05,
-      "loss": 0.9218,
+      "loss": 0.9212,
       "num_input_tokens_seen": 86343680,
       "step": 10540
     },
     {
       "epoch": 0.5066026410564226,
-      "grad_norm": 0.4754197597503662,
+      "grad_norm": 0.47995299100875854,
       "learning_rate": 2.6435310366515498e-05,
-      "loss": 0.9897,
+      "loss": 0.9893,
       "num_input_tokens_seen": 86425600,
       "step": 10550
     },
     {
       "epoch": 0.5070828331332533,
-      "grad_norm": 0.4898022711277008,
+      "grad_norm": 0.49421054124832153,
       "learning_rate": 2.6395757373560904e-05,
-      "loss": 1.0327,
+      "loss": 1.0323,
       "num_input_tokens_seen": 86507520,
       "step": 10560
     },
     {
       "epoch": 0.507563025210084,
-      "grad_norm": 0.49128857254981995,
+      "grad_norm": 0.48955073952674866,
       "learning_rate": 2.6356200875645287e-05,
-      "loss": 0.9704,
+      "loss": 0.9709,
       "num_input_tokens_seen": 86589440,
       "step": 10570
     },
     {
       "epoch": 0.5080432172869148,
-      "grad_norm": 0.49930843710899353,
+      "grad_norm": 0.4894655644893646,
       "learning_rate": 2.631664097210108e-05,
-      "loss": 0.7818,
+      "loss": 0.7823,
       "num_input_tokens_seen": 86671360,
       "step": 10580
     },
     {
       "epoch": 0.5085234093637455,
-      "grad_norm": 2.420948028564453,
+      "grad_norm": 2.337101697921753,
       "learning_rate": 2.6277077762269276e-05,
-      "loss": 0.717,
+      "loss": 0.7174,
       "num_input_tokens_seen": 86753280,
       "step": 10590
     },
     {
       "epoch": 0.5090036014405762,
-      "grad_norm": 0.49312740564346313,
+      "grad_norm": 0.4962919056415558,
       "learning_rate": 2.6237511345499167e-05,
-      "loss": 0.8514,
+      "loss": 0.8513,
       "num_input_tokens_seen": 86835200,
       "step": 10600
     },
     {
       "epoch": 0.5094837935174069,
-      "grad_norm": 0.5390229225158691,
+      "grad_norm": 0.5572180151939392,
       "learning_rate": 2.61979418211481e-05,
-      "loss": 0.8897,
+      "loss": 0.8895,
       "num_input_tokens_seen": 86917120,
       "step": 10610
     },
     {
       "epoch": 0.5099639855942377,
-      "grad_norm": 0.5236647129058838,
+      "grad_norm": 0.5322186350822449,
       "learning_rate": 2.615836928858122e-05,
-      "loss": 0.9499,
+      "loss": 0.9501,
       "num_input_tokens_seen": 86999040,
       "step": 10620
     },
     {
       "epoch": 0.5104441776710684,
-      "grad_norm": 0.48277348279953003,
+      "grad_norm": 0.4891637861728668,
       "learning_rate": 2.6118793847171236e-05,
-      "loss": 0.7749,
+      "loss": 0.7757,
       "num_input_tokens_seen": 87080960,
       "step": 10630
     },
     {
       "epoch": 0.5109243697478991,
-      "grad_norm": 0.5363568067550659,
+      "grad_norm": 0.5661851167678833,
       "learning_rate": 2.607921559629816e-05,
-      "loss": 1.0117,
+      "loss": 1.0115,
       "num_input_tokens_seen": 87162880,
       "step": 10640
     },
     {
       "epoch": 0.5114045618247299,
-      "grad_norm": 0.47179368138313293,
+      "grad_norm": 0.4733878970146179,
       "learning_rate": 2.6039634635349043e-05,
-      "loss": 0.9039,
+      "loss": 0.9044,
       "num_input_tokens_seen": 87244800,
       "step": 10650
     },
     {
       "epoch": 0.5118847539015606,
-      "grad_norm": 0.4801013171672821,
+      "grad_norm": 0.4792388677597046,
       "learning_rate": 2.6000051063717767e-05,
-      "loss": 0.8813,
+      "loss": 0.8815,
       "num_input_tokens_seen": 87326720,
       "step": 10660
     },
     {
       "epoch": 0.5123649459783913,
-      "grad_norm": 0.48771876096725464,
+      "grad_norm": 0.49188944697380066,
       "learning_rate": 2.596046498080475e-05,
-      "loss": 0.856,
+      "loss": 0.8555,
       "num_input_tokens_seen": 87408640,
       "step": 10670
     },
     {
       "epoch": 0.5128451380552221,
-      "grad_norm": 0.4822755455970764,
+      "grad_norm": 0.4500071108341217,
       "learning_rate": 2.5920876486016726e-05,
-      "loss": 0.9129,
+      "loss": 0.9126,
       "num_input_tokens_seen": 87490560,
       "step": 10680
     },
     {
       "epoch": 0.5133253301320528,
-      "grad_norm": 0.5060322880744934,
+      "grad_norm": 0.5068156719207764,
       "learning_rate": 2.5881285678766482e-05,
-      "loss": 0.8935,
+      "loss": 0.8932,
       "num_input_tokens_seen": 87572480,
       "step": 10690
     },
     {
       "epoch": 0.5138055222088835,
-      "grad_norm": 0.7022219300270081,
+      "grad_norm": 0.6936764121055603,
       "learning_rate": 2.5841692658472617e-05,
-      "loss": 0.8172,
+      "loss": 0.8166,
       "num_input_tokens_seen": 87654400,
       "step": 10700
     },
     {
       "epoch": 0.5142857142857142,
-      "grad_norm": 0.5739632844924927,
+      "grad_norm": 0.5786249041557312,
       "learning_rate": 2.5802097524559264e-05,
-      "loss": 1.0935,
+      "loss": 1.0928,
       "num_input_tokens_seen": 87736320,
       "step": 10710
     },
     {
       "epoch": 0.514765906362545,
-      "grad_norm": 0.478157103061676,
+      "grad_norm": 0.481042742729187,
       "learning_rate": 2.5762500376455912e-05,
-      "loss": 0.8728,
+      "loss": 0.8714,
       "num_input_tokens_seen": 87818240,
       "step": 10720
     },
     {
       "epoch": 0.5152460984393757,
-      "grad_norm": 0.5317462086677551,
+      "grad_norm": 0.5307723879814148,
       "learning_rate": 2.5722901313597052e-05,
-      "loss": 0.8416,
+      "loss": 0.8432,
       "num_input_tokens_seen": 87900160,
       "step": 10730
     },
     {
       "epoch": 0.5157262905162064,
-      "grad_norm": 0.4968595802783966,
+      "grad_norm": 0.49768760800361633,
       "learning_rate": 2.5683300435422032e-05,
-      "loss": 0.8828,
+      "loss": 0.8825,
       "num_input_tokens_seen": 87982080,
       "step": 10740
     },
     {
       "epoch": 0.5162064825930373,
-      "grad_norm": 0.5193814039230347,
+      "grad_norm": 0.5388449430465698,
       "learning_rate": 2.564369784137472e-05,
       "loss": 0.8147,
       "num_input_tokens_seen": 88064000,
@@ -8610,23 +8610,23 @@
     },
     {
       "epoch": 0.516686674669868,
-      "grad_norm": 0.5038419365882874,
+      "grad_norm": 0.5082977414131165,
       "learning_rate": 2.5604093630903307e-05,
-      "loss": 0.8245,
+      "loss": 0.8252,
       "num_input_tokens_seen": 88145920,
       "step": 10760
     },
     {
       "epoch": 0.5171668667466987,
-      "grad_norm": 0.4722188413143158,
+      "grad_norm": 0.47542208433151245,
       "learning_rate": 2.556448790346006e-05,
-      "loss": 0.7478,
+      "loss": 0.7476,
       "num_input_tokens_seen": 88227840,
       "step": 10770
     },
     {
       "epoch": 0.5176470588235295,
-      "grad_norm": 0.48475295305252075,
+      "grad_norm": 0.47928953170776367,
       "learning_rate": 2.5524880758501035e-05,
       "loss": 1.0129,
       "num_input_tokens_seen": 88309760,
@@ -8634,535 +8634,535 @@
     },
     {
       "epoch": 0.5181272509003602,
-      "grad_norm": 1.4112732410430908,
+      "grad_norm": 1.2826298475265503,
       "learning_rate": 2.5485272295485846e-05,
-      "loss": 0.9362,
+      "loss": 0.9356,
       "num_input_tokens_seen": 88391680,
       "step": 10790
     },
     {
       "epoch": 0.5186074429771909,
-      "grad_norm": 0.484065979719162,
+      "grad_norm": 0.48119160532951355,
       "learning_rate": 2.544566261387743e-05,
-      "loss": 0.756,
+      "loss": 0.7554,
       "num_input_tokens_seen": 88473600,
       "step": 10800
     },
     {
       "epoch": 0.5190876350540216,
-      "grad_norm": 0.2988108992576599,
+      "grad_norm": 0.2859084904193878,
       "learning_rate": 2.5406051813141773e-05,
-      "loss": 0.843,
+      "loss": 0.8429,
       "num_input_tokens_seen": 88555520,
       "step": 10810
     },
     {
       "epoch": 0.5195678271308524,
-      "grad_norm": 0.4882756471633911,
+      "grad_norm": 0.49243417382240295,
       "learning_rate": 2.5366439992747688e-05,
-      "loss": 0.8411,
+      "loss": 0.8418,
       "num_input_tokens_seen": 88637440,
       "step": 10820
     },
     {
       "epoch": 0.5200480192076831,
-      "grad_norm": 0.4603646695613861,
+      "grad_norm": 0.4552639424800873,
       "learning_rate": 2.5326827252166523e-05,
-      "loss": 0.8732,
+      "loss": 0.873,
       "num_input_tokens_seen": 88719360,
       "step": 10830
     },
     {
       "epoch": 0.5205282112845138,
-      "grad_norm": 0.5258098244667053,
+      "grad_norm": 0.5165627002716064,
       "learning_rate": 2.5287213690871957e-05,
-      "loss": 0.8446,
+      "loss": 0.8453,
       "num_input_tokens_seen": 88801280,
       "step": 10840
     },
     {
       "epoch": 0.5210084033613446,
-      "grad_norm": 0.5720292925834656,
+      "grad_norm": 0.5611492991447449,
       "learning_rate": 2.5247599408339723e-05,
-      "loss": 0.8941,
+      "loss": 0.8937,
       "num_input_tokens_seen": 88883200,
       "step": 10850
     },
     {
       "epoch": 0.5214885954381753,
-      "grad_norm": 0.517301619052887,
+      "grad_norm": 0.5230381488800049,
       "learning_rate": 2.5207984504047365e-05,
-      "loss": 0.8391,
+      "loss": 0.8389,
       "num_input_tokens_seen": 88965120,
       "step": 10860
     },
     {
       "epoch": 0.521968787515006,
-      "grad_norm": 0.569545567035675,
+      "grad_norm": 0.5715019702911377,
       "learning_rate": 2.5168369077474004e-05,
-      "loss": 0.7663,
+      "loss": 0.7679,
       "num_input_tokens_seen": 89047040,
       "step": 10870
     },
     {
       "epoch": 0.5224489795918368,
-      "grad_norm": 0.5130712389945984,
+      "grad_norm": 0.5099117159843445,
       "learning_rate": 2.512875322810002e-05,
-      "loss": 0.9436,
+      "loss": 0.9431,
       "num_input_tokens_seen": 89128960,
       "step": 10880
     },
     {
       "epoch": 0.5229291716686675,
-      "grad_norm": 0.531822144985199,
+      "grad_norm": 0.5225771069526672,
       "learning_rate": 2.508913705540693e-05,
-      "loss": 0.8914,
+      "loss": 0.8919,
       "num_input_tokens_seen": 89210880,
       "step": 10890
     },
     {
       "epoch": 0.5234093637454982,
-      "grad_norm": 0.6732114553451538,
+      "grad_norm": 0.6697996854782104,
       "learning_rate": 2.504952065887701e-05,
-      "loss": 0.8412,
+      "loss": 0.841,
       "num_input_tokens_seen": 89292800,
       "step": 10900
     },
     {
       "epoch": 0.5238895558223289,
-      "grad_norm": 0.5019482970237732,
+      "grad_norm": 0.5064327120780945,
       "learning_rate": 2.5009904137993106e-05,
-      "loss": 0.9038,
+      "loss": 0.904,
       "num_input_tokens_seen": 89374720,
       "step": 10910
     },
     {
       "epoch": 0.5243697478991597,
-      "grad_norm": 2.033942699432373,
+      "grad_norm": 2.08994722366333,
       "learning_rate": 2.497028759223839e-05,
-      "loss": 1.0436,
+      "loss": 1.042,
       "num_input_tokens_seen": 89456640,
       "step": 10920
     },
     {
       "epoch": 0.5248499399759904,
-      "grad_norm": 0.5160328149795532,
+      "grad_norm": 0.49806416034698486,
       "learning_rate": 2.4930671121096105e-05,
-      "loss": 0.9828,
+      "loss": 0.9832,
       "num_input_tokens_seen": 89538560,
       "step": 10930
     },
     {
       "epoch": 0.5253301320528211,
-      "grad_norm": 0.5123900175094604,
+      "grad_norm": 0.5063926577568054,
       "learning_rate": 2.4891054824049264e-05,
-      "loss": 0.7117,
+      "loss": 0.7118,
       "num_input_tokens_seen": 89620480,
       "step": 10940
     },
     {
       "epoch": 0.5258103241296519,
-      "grad_norm": 0.5120553970336914,
+      "grad_norm": 0.5132240056991577,
       "learning_rate": 2.485143880058049e-05,
-      "loss": 0.8639,
+      "loss": 0.864,
       "num_input_tokens_seen": 89702400,
       "step": 10950
     },
     {
       "epoch": 0.5262905162064826,
-      "grad_norm": 0.5253350734710693,
+      "grad_norm": 0.5221575498580933,
       "learning_rate": 2.4811823150171692e-05,
-      "loss": 0.8927,
+      "loss": 0.8928,
       "num_input_tokens_seen": 89784320,
       "step": 10960
     },
     {
       "epoch": 0.5267707082833133,
-      "grad_norm": 0.7501769661903381,
+      "grad_norm": 0.7482132911682129,
       "learning_rate": 2.477220797230385e-05,
-      "loss": 1.0442,
+      "loss": 1.0436,
       "num_input_tokens_seen": 89866240,
       "step": 10970
     },
     {
       "epoch": 0.5272509003601441,
-      "grad_norm": 1.2171481847763062,
+      "grad_norm": 1.2887660264968872,
       "learning_rate": 2.4732593366456755e-05,
-      "loss": 1.1357,
+      "loss": 1.1344,
       "num_input_tokens_seen": 89948160,
       "step": 10980
     },
     {
       "epoch": 0.5277310924369748,
-      "grad_norm": 0.5957512259483337,
+      "grad_norm": 0.6092213988304138,
       "learning_rate": 2.4692979432108777e-05,
-      "loss": 0.9067,
+      "loss": 0.9068,
       "num_input_tokens_seen": 90030080,
       "step": 10990
     },
     {
       "epoch": 0.5282112845138055,
-      "grad_norm": 0.5080304741859436,
+      "grad_norm": 0.5136516690254211,
       "learning_rate": 2.4653366268736565e-05,
-      "loss": 1.0397,
+      "loss": 1.0396,
       "num_input_tokens_seen": 90112000,
       "step": 11000
     },
     {
       "epoch": 0.5286914765906362,
-      "grad_norm": 0.46417123079299927,
+      "grad_norm": 0.4663499593734741,
       "learning_rate": 2.461375397581487e-05,
-      "loss": 0.7986,
+      "loss": 0.7987,
       "num_input_tokens_seen": 90193920,
       "step": 11010
     },
     {
       "epoch": 0.529171668667467,
-      "grad_norm": 0.5016286373138428,
+      "grad_norm": 0.4954150915145874,
       "learning_rate": 2.4574142652816238e-05,
-      "loss": 0.8494,
+      "loss": 0.8519,
       "num_input_tokens_seen": 90275840,
       "step": 11020
     },
     {
       "epoch": 0.5296518607442977,
-      "grad_norm": 0.6602762937545776,
+      "grad_norm": 0.644432544708252,
       "learning_rate": 2.453453239921077e-05,
-      "loss": 0.75,
+      "loss": 0.7499,
       "num_input_tokens_seen": 90357760,
       "step": 11030
     },
     {
       "epoch": 0.5301320528211284,
-      "grad_norm": 0.520213782787323,
+      "grad_norm": 0.5186659693717957,
       "learning_rate": 2.44949233144659e-05,
-      "loss": 0.7564,
+      "loss": 0.7552,
       "num_input_tokens_seen": 90439680,
       "step": 11040
     },
     {
       "epoch": 0.5306122448979592,
-      "grad_norm": 0.49561575055122375,
+      "grad_norm": 0.49851685762405396,
       "learning_rate": 2.4455315498046134e-05,
-      "loss": 0.8193,
+      "loss": 0.8187,
       "num_input_tokens_seen": 90521600,
       "step": 11050
     },
     {
       "epoch": 0.5310924369747899,
-      "grad_norm": 0.5517764687538147,
+      "grad_norm": 0.5538609027862549,
       "learning_rate": 2.4415709049412757e-05,
-      "loss": 0.8752,
+      "loss": 0.8756,
       "num_input_tokens_seen": 90603520,
       "step": 11060
     },
     {
       "epoch": 0.5315726290516206,
-      "grad_norm": 0.5008801817893982,
+      "grad_norm": 0.5045083165168762,
       "learning_rate": 2.437610406802365e-05,
-      "loss": 0.8828,
+      "loss": 0.8821,
       "num_input_tokens_seen": 90685440,
       "step": 11070
     },
     {
       "epoch": 0.5320528211284514,
-      "grad_norm": 0.46303442120552063,
+      "grad_norm": 0.46333587169647217,
       "learning_rate": 2.4336500653333012e-05,
-      "loss": 0.9908,
+      "loss": 0.9912,
       "num_input_tokens_seen": 90767360,
       "step": 11080
     },
     {
       "epoch": 0.5325330132052821,
-      "grad_norm": 0.49256837368011475,
+      "grad_norm": 0.48697561025619507,
       "learning_rate": 2.4296898904791076e-05,
-      "loss": 0.9681,
+      "loss": 0.9691,
       "num_input_tokens_seen": 90849280,
       "step": 11090
     },
     {
       "epoch": 0.5330132052821128,
-      "grad_norm": 0.5212172269821167,
+      "grad_norm": 0.5088743567466736,
       "learning_rate": 2.425729892184393e-05,
-      "loss": 0.8629,
+      "loss": 0.8621,
       "num_input_tokens_seen": 90931200,
       "step": 11100
     },
     {
       "epoch": 0.5334933973589436,
-      "grad_norm": 0.49792659282684326,
+      "grad_norm": 0.49558478593826294,
       "learning_rate": 2.421770080393321e-05,
-      "loss": 1.1662,
+      "loss": 1.1661,
       "num_input_tokens_seen": 91013120,
       "step": 11110
     },
     {
       "epoch": 0.5339735894357743,
-      "grad_norm": 0.5083081126213074,
+      "grad_norm": 0.5269922614097595,
       "learning_rate": 2.417810465049585e-05,
-      "loss": 0.9811,
+      "loss": 0.9819,
       "num_input_tokens_seen": 91095040,
       "step": 11120
     },
     {
       "epoch": 0.534453781512605,
-      "grad_norm": 1.1411739587783813,
+      "grad_norm": 1.1876569986343384,
       "learning_rate": 2.413851056096388e-05,
-      "loss": 0.7801,
+      "loss": 0.7812,
       "num_input_tokens_seen": 91176960,
       "step": 11130
     },
     {
       "epoch": 0.5349339735894357,
-      "grad_norm": 0.4804481863975525,
+      "grad_norm": 0.4851647615432739,
       "learning_rate": 2.4098918634764153e-05,
-      "loss": 0.8704,
+      "loss": 0.8706,
       "num_input_tokens_seen": 91258880,
       "step": 11140
     },
     {
       "epoch": 0.5354141656662665,
-      "grad_norm": 0.47054970264434814,
+      "grad_norm": 0.4763019382953644,
       "learning_rate": 2.4059328971318053e-05,
-      "loss": 1.0149,
+      "loss": 1.015,
       "num_input_tokens_seen": 91340800,
       "step": 11150
     },
     {
       "epoch": 0.5358943577430972,
-      "grad_norm": 1.7846838235855103,
+      "grad_norm": 1.7785102128982544,
       "learning_rate": 2.4019741670041305e-05,
-      "loss": 0.9053,
+      "loss": 0.9073,
       "num_input_tokens_seen": 91422720,
       "step": 11160
     },
     {
       "epoch": 0.5363745498199279,
-      "grad_norm": 0.48844608664512634,
+      "grad_norm": 0.48900002241134644,
       "learning_rate": 2.398015683034371e-05,
-      "loss": 0.966,
+      "loss": 0.9643,
       "num_input_tokens_seen": 91504640,
       "step": 11170
     },
     {
       "epoch": 0.5368547418967587,
-      "grad_norm": 0.49595731496810913,
+      "grad_norm": 0.5039075613021851,
       "learning_rate": 2.394057455162886e-05,
-      "loss": 0.696,
+      "loss": 0.6964,
       "num_input_tokens_seen": 91586560,
       "step": 11180
     },
     {
       "epoch": 0.5373349339735894,
-      "grad_norm": 0.49253153800964355,
+      "grad_norm": 0.49732545018196106,
       "learning_rate": 2.3900994933293953e-05,
-      "loss": 0.7898,
+      "loss": 0.7903,
       "num_input_tokens_seen": 91668480,
       "step": 11190
     },
     {
       "epoch": 0.5378151260504201,
-      "grad_norm": 0.5204420685768127,
+      "grad_norm": 0.525292694568634,
       "learning_rate": 2.3861418074729476e-05,
-      "loss": 0.9175,
+      "loss": 0.918,
       "num_input_tokens_seen": 91750400,
       "step": 11200
     },
     {
       "epoch": 0.538295318127251,
-      "grad_norm": 0.4966830015182495,
+      "grad_norm": 0.5006028413772583,
       "learning_rate": 2.3821844075318993e-05,
-      "loss": 0.8275,
+      "loss": 0.8276,
       "num_input_tokens_seen": 91832320,
       "step": 11210
     },
     {
       "epoch": 0.5387755102040817,
-      "grad_norm": 0.41914886236190796,
+      "grad_norm": 0.42340755462646484,
       "learning_rate": 2.378227303443889e-05,
-      "loss": 0.9653,
+      "loss": 0.9665,
       "num_input_tokens_seen": 91914240,
       "step": 11220
     },
     {
       "epoch": 0.5392557022809124,
-      "grad_norm": 0.4765452444553375,
+      "grad_norm": 0.4761850833892822,
       "learning_rate": 2.3742705051458145e-05,
-      "loss": 0.8119,
+      "loss": 0.8118,
       "num_input_tokens_seen": 91996160,
       "step": 11230
     },
     {
       "epoch": 0.539735894357743,
-      "grad_norm": 0.4974367022514343,
+      "grad_norm": 0.515305757522583,
       "learning_rate": 2.3703140225738017e-05,
-      "loss": 0.9283,
+      "loss": 0.9285,
       "num_input_tokens_seen": 92078080,
       "step": 11240
     },
     {
       "epoch": 0.5402160864345739,
-      "grad_norm": 0.48964783549308777,
+      "grad_norm": 0.4948907792568207,
       "learning_rate": 2.3663578656631858e-05,
-      "loss": 0.8144,
+      "loss": 0.814,
       "num_input_tokens_seen": 92160000,
       "step": 11250
     },
     {
       "epoch": 0.5406962785114046,
-      "grad_norm": 0.5229817628860474,
+      "grad_norm": 0.5262484550476074,
       "learning_rate": 2.362402044348486e-05,
-      "loss": 0.8041,
+      "loss": 0.8016,
       "num_input_tokens_seen": 92241920,
       "step": 11260
     },
     {
       "epoch": 0.5411764705882353,
-      "grad_norm": 0.5138763785362244,
+      "grad_norm": 0.4996432662010193,
       "learning_rate": 2.3584465685633738e-05,
-      "loss": 0.877,
+      "loss": 0.8771,
       "num_input_tokens_seen": 92323840,
       "step": 11270
     },
     {
       "epoch": 0.5416566626650661,
-      "grad_norm": 0.6102971434593201,
+      "grad_norm": 0.5967109203338623,
       "learning_rate": 2.3544914482406592e-05,
-      "loss": 0.8796,
+      "loss": 0.8791,
       "num_input_tokens_seen": 92405760,
       "step": 11280
     },
     {
       "epoch": 0.5421368547418968,
-      "grad_norm": 0.4886428713798523,
+      "grad_norm": 0.4963090419769287,
       "learning_rate": 2.350536693312255e-05,
-      "loss": 0.8823,
+      "loss": 0.8822,
       "num_input_tokens_seen": 92487680,
       "step": 11290
     },
     {
       "epoch": 0.5426170468187275,
-      "grad_norm": 0.4848209321498871,
+      "grad_norm": 0.48664769530296326,
       "learning_rate": 2.3465823137091572e-05,
-      "loss": 0.8294,
+      "loss": 0.8298,
       "num_input_tokens_seen": 92569600,
       "step": 11300
     },
     {
       "epoch": 0.5430972388955583,
-      "grad_norm": 0.4850007891654968,
+      "grad_norm": 0.48569774627685547,
       "learning_rate": 2.3426283193614208e-05,
-      "loss": 0.8574,
+      "loss": 0.8567,
       "num_input_tokens_seen": 92651520,
       "step": 11310
     },
     {
       "epoch": 0.543577430972389,
-      "grad_norm": 0.47127023339271545,
+      "grad_norm": 0.4557991623878479,
       "learning_rate": 2.3386747201981338e-05,
-      "loss": 1.143,
+      "loss": 1.1435,
       "num_input_tokens_seen": 92733440,
       "step": 11320
     },
     {
       "epoch": 0.5440576230492197,
-      "grad_norm": 0.4775993227958679,
+      "grad_norm": 0.47304612398147583,
       "learning_rate": 2.3347215261473887e-05,
-      "loss": 0.8502,
+      "loss": 0.8503,
       "num_input_tokens_seen": 92815360,
       "step": 11330
     },
     {
       "epoch": 0.5445378151260504,
-      "grad_norm": 0.4856424629688263,
+      "grad_norm": 0.4795781970024109,
       "learning_rate": 2.330768747136263e-05,
-      "loss": 0.9998,
+      "loss": 0.9992,
       "num_input_tokens_seen": 92897280,
       "step": 11340
     },
     {
       "epoch": 0.5450180072028812,
-      "grad_norm": 0.4987284541130066,
+      "grad_norm": 0.49423906207084656,
       "learning_rate": 2.3268163930907933e-05,
-      "loss": 0.8627,
+      "loss": 0.8629,
       "num_input_tokens_seen": 92979200,
       "step": 11350
     },
     {
       "epoch": 0.5454981992797119,
-      "grad_norm": 0.9060471653938293,
+      "grad_norm": 0.8439264893531799,
       "learning_rate": 2.3228644739359444e-05,
-      "loss": 0.8115,
+      "loss": 0.8107,
       "num_input_tokens_seen": 93061120,
       "step": 11360
     },
     {
       "epoch": 0.5459783913565426,
-      "grad_norm": 0.5042136907577515,
+      "grad_norm": 0.501241147518158,
       "learning_rate": 2.3189129995955943e-05,
-      "loss": 0.9262,
+      "loss": 0.9255,
       "num_input_tokens_seen": 93143040,
       "step": 11370
     },
     {
       "epoch": 0.5464585834333734,
-      "grad_norm": 0.6296837329864502,
+      "grad_norm": 0.5781614780426025,
       "learning_rate": 2.314961979992501e-05,
-      "loss": 0.9588,
+      "loss": 0.9589,
       "num_input_tokens_seen": 93224960,
       "step": 11380
     },
     {
       "epoch": 0.5469387755102041,
-      "grad_norm": 0.40785011649131775,
+      "grad_norm": 0.4057072103023529,
       "learning_rate": 2.311011425048281e-05,
-      "loss": 1.0475,
+      "loss": 1.0476,
       "num_input_tokens_seen": 93306880,
       "step": 11390
     },
     {
       "epoch": 0.5474189675870348,
-      "grad_norm": 0.5426297187805176,
+      "grad_norm": 0.5158141255378723,
       "learning_rate": 2.3070613446833842e-05,
-      "loss": 0.9962,
+      "loss": 0.9967,
       "num_input_tokens_seen": 93388800,
       "step": 11400
     },
     {
       "epoch": 0.5478991596638656,
-      "grad_norm": 0.49344369769096375,
+      "grad_norm": 0.49406811594963074,
       "learning_rate": 2.30311174881707e-05,
-      "loss": 0.9423,
+      "loss": 0.9432,
       "num_input_tokens_seen": 93470720,
       "step": 11410
     },
     {
       "epoch": 0.5483793517406963,
-      "grad_norm": 0.5111265182495117,
+      "grad_norm": 0.5071375370025635,
       "learning_rate": 2.2991626473673773e-05,
-      "loss": 0.78,
+      "loss": 0.7797,
       "num_input_tokens_seen": 93552640,
       "step": 11420
     },
     {
       "epoch": 0.548859543817527,
-      "grad_norm": 0.49911582469940186,
+      "grad_norm": 0.4987107217311859,
       "learning_rate": 2.295214050251108e-05,
-      "loss": 0.9205,
+      "loss": 0.9207,
       "num_input_tokens_seen": 93634560,
       "step": 11430
     },
     {
       "epoch": 0.5493397358943577,
-      "grad_norm": 0.5109173059463501,
+      "grad_norm": 0.5108968019485474,
       "learning_rate": 2.2912659673837965e-05,
-      "loss": 0.7834,
+      "loss": 0.7837,
       "num_input_tokens_seen": 93716480,
       "step": 11440
     },
     {
       "epoch": 0.5498199279711885,
-      "grad_norm": 0.49743011593818665,
+      "grad_norm": 0.5007959604263306,
       "learning_rate": 2.2873184086796824e-05,
       "loss": 0.8649,
       "num_input_tokens_seen": 93798400,
@@ -9170,271 +9170,271 @@
     },
     {
       "epoch": 0.5503001200480192,
-      "grad_norm": 0.5043100118637085,
+      "grad_norm": 0.49213898181915283,
       "learning_rate": 2.283371384051693e-05,
-      "loss": 1.2097,
+      "loss": 1.2096,
       "num_input_tokens_seen": 93880320,
       "step": 11460
     },
     {
       "epoch": 0.5507803121248499,
-      "grad_norm": 0.5047153830528259,
+      "grad_norm": 0.5077102780342102,
       "learning_rate": 2.2794249034114137e-05,
-      "loss": 0.9085,
+      "loss": 0.9088,
       "num_input_tokens_seen": 93962240,
       "step": 11470
     },
     {
       "epoch": 0.5512605042016807,
-      "grad_norm": 1.0863653421401978,
+      "grad_norm": 1.1485230922698975,
       "learning_rate": 2.275478976669062e-05,
-      "loss": 1.0061,
+      "loss": 1.0068,
       "num_input_tokens_seen": 94044160,
       "step": 11480
     },
     {
       "epoch": 0.5517406962785114,
-      "grad_norm": 0.5065763592720032,
+      "grad_norm": 0.5125216245651245,
       "learning_rate": 2.2715336137334657e-05,
-      "loss": 0.9816,
+      "loss": 0.9817,
       "num_input_tokens_seen": 94126080,
       "step": 11490
     },
     {
       "epoch": 0.5522208883553421,
-      "grad_norm": 0.5079005360603333,
+      "grad_norm": 0.5075448751449585,
       "learning_rate": 2.2675888245120382e-05,
-      "loss": 1.0597,
+      "loss": 1.0594,
       "num_input_tokens_seen": 94208000,
       "step": 11500
     },
     {
       "epoch": 0.5527010804321729,
-      "grad_norm": 0.5147351622581482,
+      "grad_norm": 0.5110220313072205,
       "learning_rate": 2.263644618910749e-05,
-      "loss": 0.9222,
+      "loss": 0.9223,
       "num_input_tokens_seen": 94289920,
       "step": 11510
     },
     {
       "epoch": 0.5531812725090036,
-      "grad_norm": 0.4940924048423767,
+      "grad_norm": 0.5087609887123108,
       "learning_rate": 2.2597010068341052e-05,
-      "loss": 0.6929,
+      "loss": 0.6925,
       "num_input_tokens_seen": 94371840,
       "step": 11520
     },
     {
       "epoch": 0.5536614645858343,
-      "grad_norm": 0.49539023637771606,
+      "grad_norm": 0.49689438939094543,
       "learning_rate": 2.255757998185122e-05,
-      "loss": 0.943,
+      "loss": 0.9426,
       "num_input_tokens_seen": 94453760,
       "step": 11530
     },
     {
       "epoch": 0.554141656662665,
-      "grad_norm": 0.7505642771720886,
+      "grad_norm": 0.7527957558631897,
       "learning_rate": 2.2518156028652977e-05,
-      "loss": 0.9255,
+      "loss": 0.9247,
       "num_input_tokens_seen": 94535680,
       "step": 11540
     },
     {
       "epoch": 0.5546218487394958,
-      "grad_norm": 0.2930028438568115,
+      "grad_norm": 0.30175384879112244,
       "learning_rate": 2.2478738307745938e-05,
-      "loss": 0.7929,
+      "loss": 0.7931,
       "num_input_tokens_seen": 94617600,
       "step": 11550
     },
     {
       "epoch": 0.5551020408163265,
-      "grad_norm": 0.4832445979118347,
+      "grad_norm": 0.482919305562973,
       "learning_rate": 2.243932691811405e-05,
-      "loss": 0.9169,
+      "loss": 0.9164,
       "num_input_tokens_seen": 94699520,
       "step": 11560
     },
     {
       "epoch": 0.5555822328931572,
-      "grad_norm": 0.4929827153682709,
+      "grad_norm": 0.4989522695541382,
       "learning_rate": 2.2399921958725364e-05,
-      "loss": 0.9749,
+      "loss": 0.9756,
       "num_input_tokens_seen": 94781440,
       "step": 11570
     },
     {
       "epoch": 0.556062424969988,
-      "grad_norm": 1.4206510782241821,
+      "grad_norm": 1.4328854084014893,
       "learning_rate": 2.236052352853177e-05,
-      "loss": 1.0515,
+      "loss": 1.0529,
       "num_input_tokens_seen": 94863360,
       "step": 11580
     },
     {
       "epoch": 0.5565426170468187,
-      "grad_norm": 0.5904713273048401,
+      "grad_norm": 0.554436206817627,
       "learning_rate": 2.232113172646878e-05,
-      "loss": 0.9887,
+      "loss": 0.9882,
       "num_input_tokens_seen": 94945280,
       "step": 11590
     },
     {
       "epoch": 0.5570228091236494,
-      "grad_norm": 0.49482420086860657,
+      "grad_norm": 0.49005597829818726,
       "learning_rate": 2.2281746651455272e-05,
-      "loss": 0.8613,
+      "loss": 0.8608,
       "num_input_tokens_seen": 95027200,
       "step": 11600
     },
     {
       "epoch": 0.5575030012004802,
-      "grad_norm": 0.5033355951309204,
+      "grad_norm": 0.502778947353363,
       "learning_rate": 2.2242368402393198e-05,
-      "loss": 1.061,
+      "loss": 1.0607,
       "num_input_tokens_seen": 95109120,
       "step": 11610
     },
     {
       "epoch": 0.5579831932773109,
-      "grad_norm": 1.082341194152832,
+      "grad_norm": 0.964520275592804,
       "learning_rate": 2.220299707816738e-05,
-      "loss": 0.8245,
+      "loss": 0.8243,
       "num_input_tokens_seen": 95191040,
       "step": 11620
     },
     {
       "epoch": 0.5584633853541416,
-      "grad_norm": 0.49979743361473083,
+      "grad_norm": 0.4970017969608307,
       "learning_rate": 2.2163632777645282e-05,
-      "loss": 0.8227,
+      "loss": 0.8224,
       "num_input_tokens_seen": 95272960,
       "step": 11630
     },
     {
       "epoch": 0.5589435774309723,
-      "grad_norm": 0.5040926933288574,
+      "grad_norm": 0.5044746994972229,
       "learning_rate": 2.2124275599676676e-05,
-      "loss": 0.9013,
+      "loss": 0.9011,
       "num_input_tokens_seen": 95354880,
       "step": 11640
     },
     {
       "epoch": 0.5594237695078031,
-      "grad_norm": 0.4717329144477844,
+      "grad_norm": 0.4746025800704956,
       "learning_rate": 2.20849256430935e-05,
-      "loss": 0.9034,
+      "loss": 0.9039,
       "num_input_tokens_seen": 95436800,
       "step": 11650
     },
     {
       "epoch": 0.5599039615846338,
-      "grad_norm": 0.5357513427734375,
+      "grad_norm": 0.5340638756752014,
       "learning_rate": 2.2045583006709536e-05,
-      "loss": 0.919,
+      "loss": 0.9187,
       "num_input_tokens_seen": 95518720,
       "step": 11660
     },
     {
       "epoch": 0.5603841536614645,
-      "grad_norm": 0.6526145935058594,
+      "grad_norm": 0.6348832845687866,
       "learning_rate": 2.2006247789320162e-05,
-      "loss": 0.868,
+      "loss": 0.8694,
       "num_input_tokens_seen": 95600640,
       "step": 11670
     },
     {
       "epoch": 0.5608643457382954,
-      "grad_norm": 0.527145266532898,
+      "grad_norm": 0.544661819934845,
       "learning_rate": 2.1966920089702157e-05,
-      "loss": 0.9287,
+      "loss": 0.9286,
       "num_input_tokens_seen": 95682560,
       "step": 11680
     },
     {
       "epoch": 0.561344537815126,
-      "grad_norm": 0.4794023931026459,
+      "grad_norm": 0.48837342858314514,
       "learning_rate": 2.192760000661343e-05,
-      "loss": 0.8022,
+      "loss": 0.8019,
       "num_input_tokens_seen": 95764480,
       "step": 11690
     },
     {
       "epoch": 0.5618247298919568,
-      "grad_norm": 0.4974648058414459,
+      "grad_norm": 0.5073260068893433,
       "learning_rate": 2.1888287638792722e-05,
-      "loss": 0.8273,
+      "loss": 0.8267,
       "num_input_tokens_seen": 95846400,
       "step": 11700
     },
     {
       "epoch": 0.5623049219687876,
-      "grad_norm": 0.48735299706459045,
+      "grad_norm": 0.4896014928817749,
       "learning_rate": 2.184898308495943e-05,
-      "loss": 0.8902,
+      "loss": 0.8884,
       "num_input_tokens_seen": 95928320,
       "step": 11710
     },
     {
       "epoch": 0.5627851140456183,
-      "grad_norm": 0.4717083275318146,
+      "grad_norm": 0.473751425743103,
       "learning_rate": 2.180968644381334e-05,
-      "loss": 0.9926,
+      "loss": 0.992,
       "num_input_tokens_seen": 96010240,
       "step": 11720
     },
     {
       "epoch": 0.563265306122449,
-      "grad_norm": 0.5048367977142334,
+      "grad_norm": 0.5091336965560913,
       "learning_rate": 2.1770397814034315e-05,
-      "loss": 0.8164,
+      "loss": 0.8166,
       "num_input_tokens_seen": 96092160,
       "step": 11730
     },
     {
       "epoch": 0.5637454981992797,
-      "grad_norm": 0.5213897228240967,
+      "grad_norm": 0.5212090015411377,
       "learning_rate": 2.1731117294282166e-05,
-      "loss": 0.9192,
+      "loss": 0.9194,
       "num_input_tokens_seen": 96174080,
       "step": 11740
     },
     {
       "epoch": 0.5642256902761105,
-      "grad_norm": 0.47659391164779663,
+      "grad_norm": 0.47561123967170715,
       "learning_rate": 2.16918449831963e-05,
-      "loss": 0.841,
+      "loss": 0.8408,
       "num_input_tokens_seen": 96256000,
       "step": 11750
     },
     {
       "epoch": 0.5647058823529412,
-      "grad_norm": 0.49390923976898193,
+      "grad_norm": 0.49135398864746094,
       "learning_rate": 2.165258097939551e-05,
-      "loss": 0.8303,
+      "loss": 0.8308,
       "num_input_tokens_seen": 96337920,
       "step": 11760
     },
     {
       "epoch": 0.5651860744297719,
-      "grad_norm": 0.4540766775608063,
+      "grad_norm": 0.4511423110961914,
       "learning_rate": 2.1613325381477744e-05,
-      "loss": 0.6811,
+      "loss": 0.6813,
       "num_input_tokens_seen": 96419840,
       "step": 11770
     },
     {
       "epoch": 0.5656662665066027,
-      "grad_norm": 0.5054943561553955,
+      "grad_norm": 0.5014552474021912,
       "learning_rate": 2.1574078288019846e-05,
-      "loss": 0.9841,
+      "loss": 0.9842,
       "num_input_tokens_seen": 96501760,
       "step": 11780
     },
     {
       "epoch": 0.5661464585834334,
-      "grad_norm": 0.5268092751502991,
+      "grad_norm": 0.5226877927780151,
       "learning_rate": 2.1534839797577268e-05,
       "loss": 1.0156,
       "num_input_tokens_seen": 96583680,
@@ -9442,135 +9442,135 @@
     },
     {
       "epoch": 0.5666266506602641,
-      "grad_norm": 1.0047301054000854,
+      "grad_norm": 1.0016844272613525,
       "learning_rate": 2.14956100086839e-05,
-      "loss": 0.813,
+      "loss": 0.8119,
       "num_input_tokens_seen": 96665600,
       "step": 11800
     },
     {
       "epoch": 0.5671068427370949,
-      "grad_norm": 0.489765465259552,
+      "grad_norm": 0.487522155046463,
       "learning_rate": 2.1456389019851762e-05,
-      "loss": 0.8036,
+      "loss": 0.8039,
       "num_input_tokens_seen": 96747520,
       "step": 11810
     },
     {
       "epoch": 0.5675870348139256,
-      "grad_norm": 0.4869178831577301,
+      "grad_norm": 0.4897004961967468,
       "learning_rate": 2.1417176929570768e-05,
-      "loss": 0.8429,
+      "loss": 0.8425,
       "num_input_tokens_seen": 96829440,
       "step": 11820
     },
     {
       "epoch": 0.5680672268907563,
-      "grad_norm": 0.6066574454307556,
+      "grad_norm": 0.6142635941505432,
       "learning_rate": 2.137797383630851e-05,
-      "loss": 0.8213,
+      "loss": 0.8214,
       "num_input_tokens_seen": 96911360,
       "step": 11830
     },
     {
       "epoch": 0.568547418967587,
-      "grad_norm": 0.48497194051742554,
+      "grad_norm": 0.48310551047325134,
       "learning_rate": 2.1338779838509965e-05,
-      "loss": 0.8338,
+      "loss": 0.8337,
       "num_input_tokens_seen": 96993280,
       "step": 11840
     },
     {
       "epoch": 0.5690276110444178,
-      "grad_norm": 0.49360740184783936,
+      "grad_norm": 0.4890468418598175,
       "learning_rate": 2.129959503459728e-05,
-      "loss": 1.0187,
+      "loss": 1.018,
       "num_input_tokens_seen": 97075200,
       "step": 11850
     },
     {
       "epoch": 0.5695078031212485,
-      "grad_norm": 0.4896343946456909,
+      "grad_norm": 0.49132439494132996,
       "learning_rate": 2.126041952296951e-05,
-      "loss": 0.8479,
+      "loss": 0.8476,
       "num_input_tokens_seen": 97157120,
       "step": 11860
     },
     {
       "epoch": 0.5699879951980792,
-      "grad_norm": 0.626423180103302,
+      "grad_norm": 0.6127716302871704,
       "learning_rate": 2.122125340200239e-05,
-      "loss": 1.0873,
+      "loss": 1.086,
       "num_input_tokens_seen": 97239040,
       "step": 11870
     },
     {
       "epoch": 0.57046818727491,
-      "grad_norm": 0.5536409616470337,
+      "grad_norm": 0.5488608479499817,
       "learning_rate": 2.1182096770048045e-05,
-      "loss": 1.1087,
+      "loss": 1.1085,
       "num_input_tokens_seen": 97320960,
       "step": 11880
     },
     {
       "epoch": 0.5709483793517407,
-      "grad_norm": 0.4912538528442383,
+      "grad_norm": 0.4914509356021881,
       "learning_rate": 2.11429497254348e-05,
-      "loss": 0.872,
+      "loss": 0.8716,
       "num_input_tokens_seen": 97402880,
       "step": 11890
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 0.5036265254020691,
+      "grad_norm": 0.5030828714370728,
       "learning_rate": 2.1103812366466896e-05,
-      "loss": 0.9375,
+      "loss": 0.9373,
       "num_input_tokens_seen": 97484800,
       "step": 11900
     },
     {
       "epoch": 0.5719087635054022,
-      "grad_norm": 0.47547534108161926,
+      "grad_norm": 0.4771164059638977,
       "learning_rate": 2.1064684791424236e-05,
-      "loss": 0.9402,
+      "loss": 0.9405,
       "num_input_tokens_seen": 97566720,
       "step": 11910
     },
     {
       "epoch": 0.5723889555822329,
-      "grad_norm": 0.6462324857711792,
+      "grad_norm": 0.8517586588859558,
       "learning_rate": 2.1025567098562177e-05,
-      "loss": 0.83,
+      "loss": 0.8314,
       "num_input_tokens_seen": 97648640,
       "step": 11920
     },
     {
       "epoch": 0.5728691476590636,
-      "grad_norm": 0.6137294769287109,
+      "grad_norm": 0.6191906332969666,
       "learning_rate": 2.0986459386111256e-05,
-      "loss": 0.8851,
+      "loss": 0.8852,
       "num_input_tokens_seen": 97730560,
       "step": 11930
     },
     {
       "epoch": 0.5733493397358943,
-      "grad_norm": 0.474915087223053,
+      "grad_norm": 0.4799683690071106,
       "learning_rate": 2.0947361752276935e-05,
-      "loss": 0.8404,
+      "loss": 0.8412,
       "num_input_tokens_seen": 97812480,
       "step": 11940
     },
     {
       "epoch": 0.5738295318127251,
-      "grad_norm": 0.5123166441917419,
+      "grad_norm": 0.5109384059906006,
       "learning_rate": 2.0908274295239365e-05,
-      "loss": 1.0064,
+      "loss": 1.007,
       "num_input_tokens_seen": 97894400,
       "step": 11950
     },
     {
       "epoch": 0.5743097238895558,
-      "grad_norm": 1.154371738433838,
+      "grad_norm": 1.1497302055358887,
       "learning_rate": 2.0869197113153175e-05,
       "loss": 0.882,
       "num_input_tokens_seen": 97976320,
@@ -9578,15 +9578,15 @@
     },
     {
       "epoch": 0.5747899159663865,
-      "grad_norm": 0.48168784379959106,
+      "grad_norm": 0.4946820139884949,
       "learning_rate": 2.083013030414714e-05,
-      "loss": 0.7295,
+      "loss": 0.7298,
       "num_input_tokens_seen": 98058240,
       "step": 11970
     },
     {
       "epoch": 0.5752701080432173,
-      "grad_norm": 0.5147429704666138,
+      "grad_norm": 0.5023751258850098,
       "learning_rate": 2.0791073966324037e-05,
       "loss": 0.808,
       "num_input_tokens_seen": 98140160,
@@ -9594,55 +9594,55 @@
     },
     {
       "epoch": 0.575750300120048,
-      "grad_norm": 0.47852545976638794,
+      "grad_norm": 0.47996094822883606,
       "learning_rate": 2.0752028197760323e-05,
-      "loss": 0.7292,
+      "loss": 0.729,
       "num_input_tokens_seen": 98222080,
       "step": 11990
     },
     {
       "epoch": 0.5762304921968787,
-      "grad_norm": 0.48357313871383667,
+      "grad_norm": 0.482053279876709,
       "learning_rate": 2.07129930965059e-05,
-      "loss": 0.9465,
+      "loss": 0.9467,
       "num_input_tokens_seen": 98304000,
       "step": 12000
     },
     {
       "epoch": 0.5767106842737095,
-      "grad_norm": 1.0818113088607788,
+      "grad_norm": 1.1011282205581665,
       "learning_rate": 2.0673968760583912e-05,
-      "loss": 0.8273,
+      "loss": 0.8261,
       "num_input_tokens_seen": 98385920,
       "step": 12010
     },
     {
       "epoch": 0.5771908763505402,
-      "grad_norm": 0.551762580871582,
+      "grad_norm": 0.5430877804756165,
       "learning_rate": 2.0634955287990465e-05,
-      "loss": 1.0037,
+      "loss": 1.0051,
       "num_input_tokens_seen": 98467840,
       "step": 12020
     },
     {
       "epoch": 0.5776710684273709,
-      "grad_norm": 0.5415941476821899,
+      "grad_norm": 0.5249661207199097,
       "learning_rate": 2.059595277669436e-05,
-      "loss": 0.7684,
+      "loss": 0.7679,
       "num_input_tokens_seen": 98549760,
       "step": 12030
     },
     {
       "epoch": 0.5781512605042017,
-      "grad_norm": 0.5173623561859131,
+      "grad_norm": 0.5203903317451477,
       "learning_rate": 2.0556961324636903e-05,
-      "loss": 0.9755,
+      "loss": 0.9756,
       "num_input_tokens_seen": 98631680,
       "step": 12040
     },
     {
       "epoch": 0.5786314525810324,
-      "grad_norm": 0.4724039137363434,
+      "grad_norm": 0.47762370109558105,
       "learning_rate": 2.0517981029731616e-05,
       "loss": 0.9699,
       "num_input_tokens_seen": 98713600,
@@ -9650,119 +9650,119 @@
     },
     {
       "epoch": 0.5791116446578631,
-      "grad_norm": 0.5577803254127502,
+      "grad_norm": 0.5500503182411194,
       "learning_rate": 2.0479011989863988e-05,
-      "loss": 1.0823,
+      "loss": 1.0828,
       "num_input_tokens_seen": 98795520,
       "step": 12060
     },
     {
       "epoch": 0.5795918367346938,
-      "grad_norm": 0.9037268757820129,
+      "grad_norm": 0.8895681500434875,
       "learning_rate": 2.0440054302891276e-05,
-      "loss": 0.8871,
+      "loss": 0.8873,
       "num_input_tokens_seen": 98877440,
       "step": 12070
     },
     {
       "epoch": 0.5800720288115246,
-      "grad_norm": 0.4992835223674774,
+      "grad_norm": 0.49974870681762695,
       "learning_rate": 2.0401108066642217e-05,
-      "loss": 0.7901,
+      "loss": 0.7899,
       "num_input_tokens_seen": 98959360,
       "step": 12080
     },
     {
       "epoch": 0.5805522208883553,
-      "grad_norm": 0.5037127733230591,
+      "grad_norm": 0.5034114122390747,
       "learning_rate": 2.0362173378916763e-05,
-      "loss": 0.844,
+      "loss": 0.843,
       "num_input_tokens_seen": 99041280,
       "step": 12090
     },
     {
       "epoch": 0.581032412965186,
-      "grad_norm": 0.5936850905418396,
+      "grad_norm": 0.5820488929748535,
       "learning_rate": 2.032325033748591e-05,
-      "loss": 0.8851,
+      "loss": 0.8847,
       "num_input_tokens_seen": 99123200,
       "step": 12100
     },
     {
       "epoch": 0.5815126050420169,
-      "grad_norm": 0.6421457529067993,
+      "grad_norm": 0.6207419037818909,
       "learning_rate": 2.0284339040091403e-05,
-      "loss": 0.9621,
+      "loss": 0.9604,
       "num_input_tokens_seen": 99205120,
       "step": 12110
     },
     {
       "epoch": 0.5819927971188475,
-      "grad_norm": 0.49321863055229187,
+      "grad_norm": 0.49734240770339966,
       "learning_rate": 2.0245439584445457e-05,
-      "loss": 0.7169,
+      "loss": 0.716,
       "num_input_tokens_seen": 99287040,
       "step": 12120
     },
     {
       "epoch": 0.5824729891956782,
-      "grad_norm": 0.4810725152492523,
+      "grad_norm": 0.48361942172050476,
       "learning_rate": 2.0206552068230587e-05,
-      "loss": 0.9376,
+      "loss": 0.9379,
       "num_input_tokens_seen": 99368960,
       "step": 12130
     },
     {
       "epoch": 0.5829531812725091,
-      "grad_norm": 0.4972326159477234,
+      "grad_norm": 0.49501538276672363,
       "learning_rate": 2.0167676589099324e-05,
-      "loss": 0.9773,
+      "loss": 0.9769,
       "num_input_tokens_seen": 99450880,
       "step": 12140
     },
     {
       "epoch": 0.5834333733493398,
-      "grad_norm": 0.4891018569469452,
+      "grad_norm": 0.4937068819999695,
       "learning_rate": 2.0128813244673946e-05,
-      "loss": 0.8787,
+      "loss": 0.8786,
       "num_input_tokens_seen": 99532800,
       "step": 12150
     },
     {
       "epoch": 0.5839135654261705,
-      "grad_norm": 0.562864363193512,
+      "grad_norm": 0.5697944164276123,
       "learning_rate": 2.0089962132546296e-05,
-      "loss": 0.9362,
+      "loss": 0.9365,
       "num_input_tokens_seen": 99614720,
       "step": 12160
     },
     {
       "epoch": 0.5843937575030012,
-      "grad_norm": 0.5169605016708374,
+      "grad_norm": 0.5267488360404968,
       "learning_rate": 2.0051123350277477e-05,
-      "loss": 0.8472,
+      "loss": 0.8467,
       "num_input_tokens_seen": 99696640,
       "step": 12170
     },
     {
       "epoch": 0.584873949579832,
-      "grad_norm": 0.47625747323036194,
+      "grad_norm": 0.4827499985694885,
       "learning_rate": 2.0012296995397613e-05,
-      "loss": 0.6606,
+      "loss": 0.6612,
       "num_input_tokens_seen": 99778560,
       "step": 12180
     },
     {
       "epoch": 0.5853541416566627,
-      "grad_norm": 0.34957268834114075,
+      "grad_norm": 0.34603819251060486,
       "learning_rate": 1.997348316540566e-05,
-      "loss": 0.7587,
+      "loss": 0.7583,
       "num_input_tokens_seen": 99860480,
       "step": 12190
     },
     {
       "epoch": 0.5858343337334934,
-      "grad_norm": 0.5261297821998596,
+      "grad_norm": 0.5291000604629517,
       "learning_rate": 1.9934681957769107e-05,
       "loss": 0.7988,
       "num_input_tokens_seen": 99942400,
@@ -9770,87 +9770,87 @@
     },
     {
       "epoch": 0.5863145258103242,
-      "grad_norm": 0.5246568322181702,
+      "grad_norm": 0.5321553349494934,
       "learning_rate": 1.9895893469923736e-05,
-      "loss": 0.9584,
+      "loss": 0.9576,
       "num_input_tokens_seen": 100024320,
       "step": 12210
     },
     {
       "epoch": 0.5867947178871549,
-      "grad_norm": 0.45888885855674744,
+      "grad_norm": 0.4535011947154999,
       "learning_rate": 1.985711779927339e-05,
-      "loss": 1.1023,
+      "loss": 1.1025,
       "num_input_tokens_seen": 100106240,
       "step": 12220
     },
     {
       "epoch": 0.5872749099639856,
-      "grad_norm": 0.6073553562164307,
+      "grad_norm": 0.600887656211853,
       "learning_rate": 1.9818355043189732e-05,
-      "loss": 0.8485,
+      "loss": 0.8477,
       "num_input_tokens_seen": 100188160,
       "step": 12230
     },
     {
       "epoch": 0.5877551020408164,
-      "grad_norm": 0.8071925640106201,
+      "grad_norm": 0.8248658180236816,
       "learning_rate": 1.9779605299012005e-05,
-      "loss": 0.9277,
+      "loss": 0.9275,
       "num_input_tokens_seen": 100270080,
       "step": 12240
     },
     {
       "epoch": 0.5882352941176471,
-      "grad_norm": 0.5476227402687073,
+      "grad_norm": 0.5473348498344421,
       "learning_rate": 1.974086866404675e-05,
-      "loss": 0.9011,
+      "loss": 0.9013,
       "num_input_tokens_seen": 100352000,
       "step": 12250
     },
     {
       "epoch": 0.5887154861944778,
-      "grad_norm": 0.4784226417541504,
+      "grad_norm": 0.47622087597846985,
       "learning_rate": 1.970214523556761e-05,
-      "loss": 1.0077,
+      "loss": 1.0079,
       "num_input_tokens_seen": 100433920,
       "step": 12260
     },
     {
       "epoch": 0.5891956782713085,
-      "grad_norm": 0.49654924869537354,
+      "grad_norm": 0.493226021528244,
       "learning_rate": 1.9663435110815065e-05,
-      "loss": 0.9088,
+      "loss": 0.909,
       "num_input_tokens_seen": 100515840,
       "step": 12270
     },
     {
       "epoch": 0.5896758703481393,
-      "grad_norm": 0.4656108617782593,
+      "grad_norm": 0.46058568358421326,
       "learning_rate": 1.9624738386996163e-05,
-      "loss": 0.8294,
+      "loss": 0.8296,
       "num_input_tokens_seen": 100597760,
       "step": 12280
     },
     {
       "epoch": 0.59015606242497,
-      "grad_norm": 0.5135471820831299,
+      "grad_norm": 0.5092949867248535,
       "learning_rate": 1.9586055161284332e-05,
-      "loss": 0.914,
+      "loss": 0.9134,
       "num_input_tokens_seen": 100679680,
       "step": 12290
     },
     {
       "epoch": 0.5906362545018007,
-      "grad_norm": 0.5345959663391113,
+      "grad_norm": 0.536564826965332,
       "learning_rate": 1.954738553081909e-05,
-      "loss": 1.1152,
+      "loss": 1.115,
       "num_input_tokens_seen": 100761600,
       "step": 12300
     },
     {
       "epoch": 0.5911164465786315,
-      "grad_norm": 0.5068685412406921,
+      "grad_norm": 0.5003234148025513,
       "learning_rate": 1.950872959270581e-05,
       "loss": 0.8391,
       "num_input_tokens_seen": 100843520,
@@ -9858,135 +9858,135 @@
     },
     {
       "epoch": 0.5915966386554622,
-      "grad_norm": 0.5310745239257812,
+      "grad_norm": 0.529944658279419,
       "learning_rate": 1.9470087444015484e-05,
-      "loss": 0.8974,
+      "loss": 0.897,
       "num_input_tokens_seen": 100925440,
       "step": 12320
     },
     {
       "epoch": 0.5920768307322929,
-      "grad_norm": 0.4799753725528717,
+      "grad_norm": 0.48205164074897766,
       "learning_rate": 1.9431459181784495e-05,
-      "loss": 1.1511,
+      "loss": 1.1514,
       "num_input_tokens_seen": 101007360,
       "step": 12330
     },
     {
       "epoch": 0.5925570228091237,
-      "grad_norm": 0.51557856798172,
+      "grad_norm": 0.5071780681610107,
       "learning_rate": 1.939284490301432e-05,
-      "loss": 0.8286,
+      "loss": 0.8289,
       "num_input_tokens_seen": 101089280,
       "step": 12340
     },
     {
       "epoch": 0.5930372148859544,
-      "grad_norm": 0.49833816289901733,
+      "grad_norm": 0.49877625703811646,
       "learning_rate": 1.935424470467135e-05,
-      "loss": 0.7323,
+      "loss": 0.7324,
       "num_input_tokens_seen": 101171200,
       "step": 12350
     },
     {
       "epoch": 0.5935174069627851,
-      "grad_norm": 0.5022624731063843,
+      "grad_norm": 0.5038672089576721,
       "learning_rate": 1.9315658683686615e-05,
-      "loss": 0.9878,
+      "loss": 0.9871,
       "num_input_tokens_seen": 101253120,
       "step": 12360
     },
     {
       "epoch": 0.5939975990396158,
-      "grad_norm": 0.5500155091285706,
+      "grad_norm": 0.5478134155273438,
       "learning_rate": 1.9277086936955517e-05,
-      "loss": 0.8682,
+      "loss": 0.8688,
       "num_input_tokens_seen": 101335040,
       "step": 12370
     },
     {
       "epoch": 0.5944777911164466,
-      "grad_norm": 0.4885292053222656,
+      "grad_norm": 0.4918793737888336,
       "learning_rate": 1.9238529561337646e-05,
-      "loss": 0.8724,
+      "loss": 0.8722,
       "num_input_tokens_seen": 101416960,
       "step": 12380
     },
     {
       "epoch": 0.5949579831932773,
-      "grad_norm": 0.4912605583667755,
+      "grad_norm": 0.48617425560951233,
       "learning_rate": 1.9199986653656502e-05,
-      "loss": 0.9832,
+      "loss": 0.9837,
       "num_input_tokens_seen": 101498880,
       "step": 12390
     },
     {
       "epoch": 0.595438175270108,
-      "grad_norm": 0.6492443084716797,
+      "grad_norm": 0.6774265766143799,
       "learning_rate": 1.9161458310699225e-05,
-      "loss": 0.8047,
+      "loss": 0.8041,
       "num_input_tokens_seen": 101580800,
       "step": 12400
     },
     {
       "epoch": 0.5959183673469388,
-      "grad_norm": 0.5103248953819275,
+      "grad_norm": 0.5218595266342163,
       "learning_rate": 1.9122944629216402e-05,
-      "loss": 0.9766,
+      "loss": 0.9768,
       "num_input_tokens_seen": 101662720,
       "step": 12410
     },
     {
       "epoch": 0.5963985594237695,
-      "grad_norm": 0.4565691351890564,
+      "grad_norm": 0.4586848020553589,
       "learning_rate": 1.9084445705921815e-05,
-      "loss": 0.8339,
+      "loss": 0.8343,
       "num_input_tokens_seen": 101744640,
       "step": 12420
     },
     {
       "epoch": 0.5968787515006002,
-      "grad_norm": 0.47968733310699463,
+      "grad_norm": 0.47861307859420776,
       "learning_rate": 1.9045961637492145e-05,
-      "loss": 1.0445,
+      "loss": 1.0443,
       "num_input_tokens_seen": 101826560,
       "step": 12430
     },
     {
       "epoch": 0.597358943577431,
-      "grad_norm": 0.49476364254951477,
+      "grad_norm": 0.4952894151210785,
       "learning_rate": 1.9007492520566814e-05,
-      "loss": 0.9611,
+      "loss": 0.9608,
       "num_input_tokens_seen": 101908480,
       "step": 12440
     },
     {
       "epoch": 0.5978391356542617,
-      "grad_norm": 0.5539293885231018,
+      "grad_norm": 0.5489120483398438,
       "learning_rate": 1.896903845174768e-05,
-      "loss": 0.9063,
+      "loss": 0.9062,
       "num_input_tokens_seen": 101990400,
       "step": 12450
     },
     {
       "epoch": 0.5983193277310924,
-      "grad_norm": 0.49969589710235596,
+      "grad_norm": 0.49838995933532715,
       "learning_rate": 1.8930599527598797e-05,
-      "loss": 0.8149,
+      "loss": 0.815,
       "num_input_tokens_seen": 102072320,
       "step": 12460
     },
     {
       "epoch": 0.5987995198079231,
-      "grad_norm": 0.6476652026176453,
+      "grad_norm": 0.6173511743545532,
       "learning_rate": 1.8892175844646215e-05,
-      "loss": 1.05,
+      "loss": 1.0498,
       "num_input_tokens_seen": 102154240,
       "step": 12470
     },
     {
       "epoch": 0.5992797118847539,
-      "grad_norm": 0.485588401556015,
+      "grad_norm": 0.48503515124320984,
       "learning_rate": 1.8853767499377712e-05,
       "loss": 0.7995,
       "num_input_tokens_seen": 102236160,
@@ -9994,47 +9994,47 @@
     },
     {
       "epoch": 0.5997599039615846,
-      "grad_norm": 0.5099250674247742,
+      "grad_norm": 0.5097170472145081,
       "learning_rate": 1.8815374588242523e-05,
-      "loss": 0.7582,
+      "loss": 0.7581,
       "num_input_tokens_seen": 102318080,
       "step": 12490
     },
     {
       "epoch": 0.6002400960384153,
-      "grad_norm": 0.49604761600494385,
+      "grad_norm": 0.5017048716545105,
       "learning_rate": 1.877699720765114e-05,
-      "loss": 0.8494,
+      "loss": 0.8486,
       "num_input_tokens_seen": 102400000,
       "step": 12500
     },
     {
       "epoch": 0.6007202881152461,
-      "grad_norm": 0.5131434798240662,
+      "grad_norm": 0.5360931158065796,
       "learning_rate": 1.873863545397507e-05,
-      "loss": 0.7685,
+      "loss": 0.7697,
       "num_input_tokens_seen": 102481920,
       "step": 12510
     },
     {
       "epoch": 0.6012004801920768,
-      "grad_norm": 0.45130103826522827,
+      "grad_norm": 0.45245033502578735,
       "learning_rate": 1.870028942354655e-05,
-      "loss": 0.9137,
+      "loss": 0.9144,
       "num_input_tokens_seen": 102563840,
       "step": 12520
     },
     {
       "epoch": 0.6016806722689075,
-      "grad_norm": 0.4828559160232544,
+      "grad_norm": 0.48523399233818054,
       "learning_rate": 1.8661959212658365e-05,
-      "loss": 0.8186,
+      "loss": 0.8182,
       "num_input_tokens_seen": 102645760,
       "step": 12530
     },
     {
       "epoch": 0.6021608643457383,
-      "grad_norm": 0.5113935470581055,
+      "grad_norm": 0.5199438333511353,
       "learning_rate": 1.862364491756355e-05,
       "loss": 0.8923,
       "num_input_tokens_seen": 102727680,
@@ -10042,31 +10042,31 @@
     },
     {
       "epoch": 0.602641056422569,
-      "grad_norm": 0.5698245167732239,
+      "grad_norm": 0.5562946200370789,
       "learning_rate": 1.8585346634475175e-05,
-      "loss": 0.9799,
+      "loss": 0.9803,
       "num_input_tokens_seen": 102809600,
       "step": 12550
     },
     {
       "epoch": 0.6031212484993997,
-      "grad_norm": 0.4895305335521698,
+      "grad_norm": 0.48669108748435974,
       "learning_rate": 1.8547064459566117e-05,
-      "loss": 0.9472,
+      "loss": 0.9476,
       "num_input_tokens_seen": 102891520,
       "step": 12560
     },
     {
       "epoch": 0.6036014405762304,
-      "grad_norm": 0.4899882376194,
+      "grad_norm": 0.4891805052757263,
       "learning_rate": 1.8508798488968803e-05,
-      "loss": 0.7548,
+      "loss": 0.7547,
       "num_input_tokens_seen": 102973440,
       "step": 12570
     },
     {
       "epoch": 0.6040816326530613,
-      "grad_norm": 0.48758333921432495,
+      "grad_norm": 0.4860639274120331,
       "learning_rate": 1.8470548818774942e-05,
       "loss": 0.9321,
       "num_input_tokens_seen": 103055360,
@@ -10074,87 +10074,87 @@
     },
     {
       "epoch": 0.604561824729892,
-      "grad_norm": 0.4785870313644409,
+      "grad_norm": 0.47964203357696533,
       "learning_rate": 1.8432315545035328e-05,
-      "loss": 0.8554,
+      "loss": 0.8547,
       "num_input_tokens_seen": 103137280,
       "step": 12590
     },
     {
       "epoch": 0.6050420168067226,
-      "grad_norm": 0.46478036046028137,
+      "grad_norm": 0.47002238035202026,
       "learning_rate": 1.83940987637596e-05,
-      "loss": 0.9034,
+      "loss": 0.9036,
       "num_input_tokens_seen": 103219200,
       "step": 12600
     },
     {
       "epoch": 0.6055222088835535,
-      "grad_norm": 0.5116038918495178,
+      "grad_norm": 0.5217668414115906,
       "learning_rate": 1.8355898570915937e-05,
-      "loss": 1.0032,
+      "loss": 1.0021,
       "num_input_tokens_seen": 103301120,
       "step": 12610
     },
     {
       "epoch": 0.6060024009603842,
-      "grad_norm": 0.4858674108982086,
+      "grad_norm": 0.48160508275032043,
       "learning_rate": 1.8317715062430902e-05,
-      "loss": 0.9037,
+      "loss": 0.9036,
       "num_input_tokens_seen": 103383040,
       "step": 12620
     },
     {
       "epoch": 0.6064825930372149,
-      "grad_norm": 0.49994757771492004,
+      "grad_norm": 0.4867957532405853,
       "learning_rate": 1.8279548334189146e-05,
-      "loss": 0.9669,
+      "loss": 0.9671,
       "num_input_tokens_seen": 103464960,
       "step": 12630
     },
     {
       "epoch": 0.6069627851140457,
-      "grad_norm": 0.5256997346878052,
+      "grad_norm": 0.5135723948478699,
       "learning_rate": 1.8241398482033185e-05,
-      "loss": 0.9426,
+      "loss": 0.943,
       "num_input_tokens_seen": 103546880,
       "step": 12640
     },
     {
       "epoch": 0.6074429771908764,
-      "grad_norm": 0.48835039138793945,
+      "grad_norm": 0.48528727889060974,
       "learning_rate": 1.8203265601763136e-05,
-      "loss": 0.9299,
+      "loss": 0.9297,
       "num_input_tokens_seen": 103628800,
       "step": 12650
     },
     {
       "epoch": 0.6079231692677071,
-      "grad_norm": 0.4843783676624298,
+      "grad_norm": 0.4840488135814667,
       "learning_rate": 1.816514978913655e-05,
-      "loss": 0.8059,
+      "loss": 0.8064,
       "num_input_tokens_seen": 103710720,
       "step": 12660
     },
     {
       "epoch": 0.6084033613445378,
-      "grad_norm": 0.6175418496131897,
+      "grad_norm": 0.615665853023529,
       "learning_rate": 1.8127051139868044e-05,
-      "loss": 0.963,
+      "loss": 0.9639,
       "num_input_tokens_seen": 103792640,
       "step": 12670
     },
     {
       "epoch": 0.6088835534213686,
-      "grad_norm": 0.503555178642273,
+      "grad_norm": 0.5075173377990723,
       "learning_rate": 1.8088969749629197e-05,
-      "loss": 0.9894,
+      "loss": 0.9895,
       "num_input_tokens_seen": 103874560,
       "step": 12680
     },
     {
       "epoch": 0.6093637454981993,
-      "grad_norm": 0.48050084710121155,
+      "grad_norm": 0.47513309121131897,
       "learning_rate": 1.8050905714048233e-05,
       "loss": 1.0683,
       "num_input_tokens_seen": 103956480,
@@ -10162,231 +10162,231 @@
     },
     {
       "epoch": 0.60984393757503,
-      "grad_norm": 0.5039294362068176,
+      "grad_norm": 0.5024633407592773,
       "learning_rate": 1.8012859128709766e-05,
-      "loss": 0.7611,
+      "loss": 0.7629,
       "num_input_tokens_seen": 104038400,
       "step": 12700
     },
     {
       "epoch": 0.6103241296518608,
-      "grad_norm": 0.6034661531448364,
+      "grad_norm": 0.5909437537193298,
       "learning_rate": 1.7974830089154624e-05,
-      "loss": 0.9293,
+      "loss": 0.9292,
       "num_input_tokens_seen": 104120320,
       "step": 12710
     },
     {
       "epoch": 0.6108043217286915,
-      "grad_norm": 0.5101216435432434,
+      "grad_norm": 0.5102412104606628,
       "learning_rate": 1.7936818690879574e-05,
-      "loss": 0.9849,
+      "loss": 0.9859,
       "num_input_tokens_seen": 104202240,
       "step": 12720
     },
     {
       "epoch": 0.6112845138055222,
-      "grad_norm": 0.49056023359298706,
+      "grad_norm": 0.4822942018508911,
       "learning_rate": 1.7898825029337054e-05,
-      "loss": 0.9917,
+      "loss": 0.9921,
       "num_input_tokens_seen": 104284160,
       "step": 12730
     },
     {
       "epoch": 0.611764705882353,
-      "grad_norm": 0.5179559588432312,
+      "grad_norm": 0.5105669498443604,
       "learning_rate": 1.7860849199934983e-05,
-      "loss": 0.8664,
+      "loss": 0.8653,
       "num_input_tokens_seen": 104366080,
       "step": 12740
     },
     {
       "epoch": 0.6122448979591837,
-      "grad_norm": 0.5554147958755493,
+      "grad_norm": 0.5741724967956543,
       "learning_rate": 1.7822891298036515e-05,
-      "loss": 0.9739,
+      "loss": 0.9741,
       "num_input_tokens_seen": 104448000,
       "step": 12750
     },
     {
       "epoch": 0.6127250900360144,
-      "grad_norm": 0.5510497689247131,
+      "grad_norm": 0.5551435351371765,
       "learning_rate": 1.7784951418959747e-05,
-      "loss": 0.9441,
+      "loss": 0.9431,
       "num_input_tokens_seen": 104529920,
       "step": 12760
     },
     {
       "epoch": 0.6132052821128451,
-      "grad_norm": 0.5032241344451904,
+      "grad_norm": 0.49893906712532043,
       "learning_rate": 1.7747029657977556e-05,
-      "loss": 0.9443,
+      "loss": 0.9441,
       "num_input_tokens_seen": 104611840,
       "step": 12770
     },
     {
       "epoch": 0.6136854741896759,
-      "grad_norm": 0.48411932587623596,
+      "grad_norm": 0.4885694682598114,
       "learning_rate": 1.7709126110317297e-05,
-      "loss": 1.0327,
+      "loss": 1.0325,
       "num_input_tokens_seen": 104693760,
       "step": 12780
     },
     {
       "epoch": 0.6141656662665066,
-      "grad_norm": 0.4949854016304016,
+      "grad_norm": 0.49348217248916626,
       "learning_rate": 1.7671240871160593e-05,
-      "loss": 0.9103,
+      "loss": 0.9102,
       "num_input_tokens_seen": 104775680,
       "step": 12790
     },
     {
       "epoch": 0.6146458583433373,
-      "grad_norm": 0.47453951835632324,
+      "grad_norm": 0.4734002351760864,
       "learning_rate": 1.76333740356431e-05,
-      "loss": 0.8165,
+      "loss": 0.817,
       "num_input_tokens_seen": 104857600,
       "step": 12800
     },
     {
       "epoch": 0.6151260504201681,
-      "grad_norm": 0.4928891658782959,
+      "grad_norm": 0.49383237957954407,
       "learning_rate": 1.7595525698854263e-05,
-      "loss": 0.8398,
+      "loss": 0.8403,
       "num_input_tokens_seen": 104939520,
       "step": 12810
     },
     {
       "epoch": 0.6156062424969988,
-      "grad_norm": 0.6075853109359741,
+      "grad_norm": 0.603259265422821,
       "learning_rate": 1.7557695955837063e-05,
-      "loss": 0.8922,
+      "loss": 0.8923,
       "num_input_tokens_seen": 105021440,
       "step": 12820
     },
     {
       "epoch": 0.6160864345738295,
-      "grad_norm": 0.6649598479270935,
+      "grad_norm": 0.6551387310028076,
       "learning_rate": 1.7519884901587772e-05,
-      "loss": 0.8345,
+      "loss": 0.8347,
       "num_input_tokens_seen": 105103360,
       "step": 12830
     },
     {
       "epoch": 0.6165666266506603,
-      "grad_norm": 0.5287781357765198,
+      "grad_norm": 0.5301949977874756,
       "learning_rate": 1.748209263105577e-05,
-      "loss": 1.0156,
+      "loss": 1.0152,
       "num_input_tokens_seen": 105185280,
       "step": 12840
     },
     {
       "epoch": 0.617046818727491,
-      "grad_norm": 0.48797744512557983,
+      "grad_norm": 0.4785909354686737,
       "learning_rate": 1.744431923914326e-05,
-      "loss": 0.9688,
+      "loss": 0.9691,
       "num_input_tokens_seen": 105267200,
       "step": 12850
     },
     {
       "epoch": 0.6175270108043217,
-      "grad_norm": 0.4974718391895294,
+      "grad_norm": 0.4949907660484314,
       "learning_rate": 1.7406564820705e-05,
-      "loss": 0.9532,
+      "loss": 0.9528,
       "num_input_tokens_seen": 105349120,
       "step": 12860
     },
     {
       "epoch": 0.6180072028811524,
-      "grad_norm": 1.6541121006011963,
+      "grad_norm": 1.5191551446914673,
       "learning_rate": 1.736882947054815e-05,
-      "loss": 0.818,
+      "loss": 0.8165,
       "num_input_tokens_seen": 105431040,
       "step": 12870
     },
     {
       "epoch": 0.6184873949579832,
-      "grad_norm": 0.5184734463691711,
+      "grad_norm": 0.5097967386245728,
       "learning_rate": 1.7331113283431966e-05,
-      "loss": 0.994,
+      "loss": 0.9941,
       "num_input_tokens_seen": 105512960,
       "step": 12880
     },
     {
       "epoch": 0.6189675870348139,
-      "grad_norm": 0.5024128556251526,
+      "grad_norm": 0.5027289390563965,
       "learning_rate": 1.729341635406757e-05,
-      "loss": 0.7879,
+      "loss": 0.7875,
       "num_input_tokens_seen": 105594880,
       "step": 12890
     },
     {
       "epoch": 0.6194477791116446,
-      "grad_norm": 0.48197275400161743,
+      "grad_norm": 0.4796588718891144,
       "learning_rate": 1.725573877711776e-05,
-      "loss": 0.9374,
+      "loss": 0.9376,
       "num_input_tokens_seen": 105676800,
       "step": 12900
     },
     {
       "epoch": 0.6199279711884754,
-      "grad_norm": 0.5016830563545227,
+      "grad_norm": 0.4965690076351166,
       "learning_rate": 1.7218080647196698e-05,
-      "loss": 0.9342,
+      "loss": 0.9336,
       "num_input_tokens_seen": 105758720,
       "step": 12910
     },
     {
       "epoch": 0.6204081632653061,
-      "grad_norm": 0.48292943835258484,
+      "grad_norm": 0.4842563271522522,
       "learning_rate": 1.7180442058869732e-05,
-      "loss": 0.8143,
+      "loss": 0.8139,
       "num_input_tokens_seen": 105840640,
       "step": 12920
     },
     {
       "epoch": 0.6208883553421368,
-      "grad_norm": 0.4773729145526886,
+      "grad_norm": 0.4824802577495575,
       "learning_rate": 1.7142823106653135e-05,
-      "loss": 0.8115,
+      "loss": 0.8121,
       "num_input_tokens_seen": 105922560,
       "step": 12930
     },
     {
       "epoch": 0.6213685474189676,
-      "grad_norm": 0.5274072885513306,
+      "grad_norm": 0.5284069776535034,
       "learning_rate": 1.7105223885013884e-05,
-      "loss": 0.8874,
+      "loss": 0.8867,
       "num_input_tokens_seen": 106004480,
       "step": 12940
     },
     {
       "epoch": 0.6218487394957983,
-      "grad_norm": 1.052327036857605,
+      "grad_norm": 1.0092517137527466,
       "learning_rate": 1.706764448836938e-05,
-      "loss": 0.9164,
+      "loss": 0.9163,
       "num_input_tokens_seen": 106086400,
       "step": 12950
     },
     {
       "epoch": 0.622328931572629,
-      "grad_norm": 0.5385840535163879,
+      "grad_norm": 0.5435160994529724,
       "learning_rate": 1.703008501108726e-05,
-      "loss": 0.9694,
+      "loss": 0.969,
       "num_input_tokens_seen": 106168320,
       "step": 12960
     },
     {
       "epoch": 0.6228091236494598,
-      "grad_norm": 0.4652009904384613,
+      "grad_norm": 0.463956356048584,
       "learning_rate": 1.699254554748515e-05,
-      "loss": 0.9424,
+      "loss": 0.9426,
       "num_input_tokens_seen": 106250240,
       "step": 12970
     },
     {
       "epoch": 0.6232893157262905,
-      "grad_norm": 0.5203952193260193,
+      "grad_norm": 0.5222662687301636,
       "learning_rate": 1.6955026191830385e-05,
       "loss": 0.8038,
       "num_input_tokens_seen": 106332160,
@@ -10394,175 +10394,175 @@
     },
     {
       "epoch": 0.6237695078031212,
-      "grad_norm": 0.47074759006500244,
+      "grad_norm": 0.47449785470962524,
       "learning_rate": 1.691752703833984e-05,
-      "loss": 0.9074,
+      "loss": 0.9071,
       "num_input_tokens_seen": 106414080,
       "step": 12990
     },
     {
       "epoch": 0.6242496998799519,
-      "grad_norm": 0.5040807723999023,
+      "grad_norm": 0.5060712099075317,
       "learning_rate": 1.6880048181179652e-05,
-      "loss": 0.8315,
+      "loss": 0.8317,
       "num_input_tokens_seen": 106496000,
       "step": 13000
     },
     {
       "epoch": 0.6247298919567827,
-      "grad_norm": 0.5343396663665771,
+      "grad_norm": 0.5310168266296387,
       "learning_rate": 1.684258971446497e-05,
-      "loss": 0.9378,
+      "loss": 0.938,
       "num_input_tokens_seen": 106577920,
       "step": 13010
     },
     {
       "epoch": 0.6252100840336134,
-      "grad_norm": 0.48654720187187195,
+      "grad_norm": 0.4854814410209656,
       "learning_rate": 1.6805151732259755e-05,
-      "loss": 0.8081,
+      "loss": 0.8077,
       "num_input_tokens_seen": 106659840,
       "step": 13020
     },
     {
       "epoch": 0.6256902761104441,
-      "grad_norm": 0.4875001907348633,
+      "grad_norm": 0.4907275140285492,
       "learning_rate": 1.6767734328576544e-05,
-      "loss": 0.8395,
+      "loss": 0.8399,
       "num_input_tokens_seen": 106741760,
       "step": 13030
     },
     {
       "epoch": 0.626170468187275,
-      "grad_norm": 0.4956626296043396,
+      "grad_norm": 0.5009474754333496,
       "learning_rate": 1.6730337597376165e-05,
-      "loss": 0.8904,
+      "loss": 0.8906,
       "num_input_tokens_seen": 106823680,
       "step": 13040
     },
     {
       "epoch": 0.6266506602641057,
-      "grad_norm": 0.49269092082977295,
+      "grad_norm": 0.4950611889362335,
       "learning_rate": 1.669296163256755e-05,
-      "loss": 0.7173,
+      "loss": 0.7171,
       "num_input_tokens_seen": 106905600,
       "step": 13050
     },
     {
       "epoch": 0.6271308523409364,
-      "grad_norm": 0.7931904196739197,
+      "grad_norm": 0.7820413708686829,
       "learning_rate": 1.6655606528007505e-05,
-      "loss": 1.0009,
+      "loss": 1.0005,
       "num_input_tokens_seen": 106987520,
       "step": 13060
     },
     {
       "epoch": 0.6276110444177672,
-      "grad_norm": 0.5141395926475525,
+      "grad_norm": 0.5212737917900085,
       "learning_rate": 1.6618272377500406e-05,
-      "loss": 0.9796,
+      "loss": 0.9795,
       "num_input_tokens_seen": 107069440,
       "step": 13070
     },
     {
       "epoch": 0.6280912364945979,
-      "grad_norm": 0.49574312567710876,
+      "grad_norm": 0.494011253118515,
       "learning_rate": 1.658095927479805e-05,
-      "loss": 0.7448,
+      "loss": 0.7443,
       "num_input_tokens_seen": 107151360,
       "step": 13080
     },
     {
       "epoch": 0.6285714285714286,
-      "grad_norm": 0.548512876033783,
+      "grad_norm": 0.5520787835121155,
       "learning_rate": 1.6543667313599365e-05,
-      "loss": 1.1406,
+      "loss": 1.141,
       "num_input_tokens_seen": 107233280,
       "step": 13090
     },
     {
       "epoch": 0.6290516206482593,
-      "grad_norm": 0.49996235966682434,
+      "grad_norm": 0.5004755854606628,
       "learning_rate": 1.6506396587550188e-05,
-      "loss": 0.8865,
+      "loss": 0.8862,
       "num_input_tokens_seen": 107315200,
       "step": 13100
     },
     {
       "epoch": 0.6295318127250901,
-      "grad_norm": 0.5021194815635681,
+      "grad_norm": 0.5015995502471924,
       "learning_rate": 1.6469147190243027e-05,
-      "loss": 0.8561,
+      "loss": 0.8555,
       "num_input_tokens_seen": 107397120,
       "step": 13110
     },
     {
       "epoch": 0.6300120048019208,
-      "grad_norm": 0.4927319884300232,
+      "grad_norm": 0.49403059482574463,
       "learning_rate": 1.6431919215216862e-05,
-      "loss": 0.8364,
+      "loss": 0.8369,
       "num_input_tokens_seen": 107479040,
       "step": 13120
     },
     {
       "epoch": 0.6304921968787515,
-      "grad_norm": 0.5446639060974121,
+      "grad_norm": 0.6032249927520752,
       "learning_rate": 1.6394712755956817e-05,
-      "loss": 0.996,
+      "loss": 0.9955,
       "num_input_tokens_seen": 107560960,
       "step": 13130
     },
     {
       "epoch": 0.6309723889555823,
-      "grad_norm": 0.5035328269004822,
+      "grad_norm": 0.5043690204620361,
       "learning_rate": 1.635752790589405e-05,
-      "loss": 0.8958,
+      "loss": 0.8961,
       "num_input_tokens_seen": 107642880,
       "step": 13140
     },
     {
       "epoch": 0.631452581032413,
-      "grad_norm": 0.4967860281467438,
+      "grad_norm": 0.49665483832359314,
       "learning_rate": 1.6320364758405422e-05,
-      "loss": 0.9864,
+      "loss": 0.987,
       "num_input_tokens_seen": 107724800,
       "step": 13150
     },
     {
       "epoch": 0.6319327731092437,
-      "grad_norm": 0.554397463798523,
+      "grad_norm": 0.5483250617980957,
       "learning_rate": 1.628322340681329e-05,
-      "loss": 0.9916,
+      "loss": 0.9919,
       "num_input_tokens_seen": 107806720,
       "step": 13160
     },
     {
       "epoch": 0.6324129651860745,
-      "grad_norm": 0.5113590955734253,
+      "grad_norm": 0.5074891448020935,
       "learning_rate": 1.6246103944385295e-05,
-      "loss": 1.0281,
+      "loss": 1.0279,
       "num_input_tokens_seen": 107888640,
       "step": 13170
     },
     {
       "epoch": 0.6328931572629052,
-      "grad_norm": 0.49097153544425964,
+      "grad_norm": 0.4931067228317261,
       "learning_rate": 1.620900646433412e-05,
-      "loss": 0.8902,
+      "loss": 0.8903,
       "num_input_tokens_seen": 107970560,
       "step": 13180
     },
     {
       "epoch": 0.6333733493397359,
-      "grad_norm": 0.523250937461853,
+      "grad_norm": 0.5287465453147888,
       "learning_rate": 1.6171931059817214e-05,
-      "loss": 0.9968,
+      "loss": 0.9969,
       "num_input_tokens_seen": 108052480,
       "step": 13190
     },
     {
       "epoch": 0.6338535414165666,
-      "grad_norm": 0.4726887345314026,
+      "grad_norm": 0.47930577397346497,
       "learning_rate": 1.613487782393661e-05,
       "loss": 0.6298,
       "num_input_tokens_seen": 108134400,
@@ -10570,7 +10570,7 @@
     },
     {
       "epoch": 0.6343337334933974,
-      "grad_norm": 0.5005057454109192,
+      "grad_norm": 0.5008885860443115,
       "learning_rate": 1.6097846849738685e-05,
       "loss": 0.7394,
       "num_input_tokens_seen": 108216320,
@@ -10578,127 +10578,127 @@
     },
     {
       "epoch": 0.6348139255702281,
-      "grad_norm": 0.4948345124721527,
+      "grad_norm": 0.4981120228767395,
       "learning_rate": 1.6060838230213883e-05,
-      "loss": 0.9741,
+      "loss": 0.9739,
       "num_input_tokens_seen": 108298240,
       "step": 13220
     },
     {
       "epoch": 0.6352941176470588,
-      "grad_norm": 0.6280222535133362,
+      "grad_norm": 0.6545652151107788,
       "learning_rate": 1.6023852058296544e-05,
-      "loss": 1.0429,
+      "loss": 1.0424,
       "num_input_tokens_seen": 108380160,
       "step": 13230
     },
     {
       "epoch": 0.6357743097238896,
-      "grad_norm": 0.553558349609375,
+      "grad_norm": 0.5533519387245178,
       "learning_rate": 1.5986888426864617e-05,
-      "loss": 0.8975,
+      "loss": 0.897,
       "num_input_tokens_seen": 108462080,
       "step": 13240
     },
     {
       "epoch": 0.6362545018007203,
-      "grad_norm": 0.47756895422935486,
+      "grad_norm": 0.47119852900505066,
       "learning_rate": 1.5949947428739448e-05,
-      "loss": 0.8076,
+      "loss": 0.8078,
       "num_input_tokens_seen": 108544000,
       "step": 13250
     },
     {
       "epoch": 0.636734693877551,
-      "grad_norm": 0.8511344194412231,
+      "grad_norm": 0.8498918414115906,
       "learning_rate": 1.591302915668556e-05,
-      "loss": 0.747,
+      "loss": 0.7465,
       "num_input_tokens_seen": 108625920,
       "step": 13260
     },
     {
       "epoch": 0.6372148859543818,
-      "grad_norm": 0.4981195032596588,
+      "grad_norm": 0.49918386340141296,
       "learning_rate": 1.5876133703410412e-05,
-      "loss": 0.9432,
+      "loss": 0.9428,
       "num_input_tokens_seen": 108707840,
       "step": 13270
     },
     {
       "epoch": 0.6376950780312125,
-      "grad_norm": 0.4754570722579956,
+      "grad_norm": 0.48045846819877625,
       "learning_rate": 1.5839261161564138e-05,
-      "loss": 1.0365,
+      "loss": 1.0362,
       "num_input_tokens_seen": 108789760,
       "step": 13280
     },
     {
       "epoch": 0.6381752701080432,
-      "grad_norm": 1.2110140323638916,
+      "grad_norm": 1.4066901206970215,
       "learning_rate": 1.5802411623739345e-05,
-      "loss": 0.9743,
+      "loss": 0.9749,
       "num_input_tokens_seen": 108871680,
       "step": 13290
     },
     {
       "epoch": 0.6386554621848739,
-      "grad_norm": 0.4123174250125885,
+      "grad_norm": 0.4242941737174988,
       "learning_rate": 1.57655851824709e-05,
-      "loss": 0.8376,
+      "loss": 0.8378,
       "num_input_tokens_seen": 108953600,
       "step": 13300
     },
     {
       "epoch": 0.6391356542617047,
-      "grad_norm": 0.5139141082763672,
+      "grad_norm": 0.5066592693328857,
       "learning_rate": 1.5728781930235627e-05,
-      "loss": 0.9858,
+      "loss": 0.9856,
       "num_input_tokens_seen": 109035520,
       "step": 13310
     },
     {
       "epoch": 0.6396158463385354,
-      "grad_norm": 0.49317967891693115,
+      "grad_norm": 0.4904208183288574,
       "learning_rate": 1.5692001959452164e-05,
-      "loss": 1.0486,
+      "loss": 1.0489,
       "num_input_tokens_seen": 109117440,
       "step": 13320
     },
     {
       "epoch": 0.6400960384153661,
-      "grad_norm": 0.48599720001220703,
+      "grad_norm": 0.48512765765190125,
       "learning_rate": 1.5655245362480654e-05,
-      "loss": 0.8965,
+      "loss": 0.8971,
       "num_input_tokens_seen": 109199360,
       "step": 13330
     },
     {
       "epoch": 0.6405762304921969,
-      "grad_norm": 0.4471222162246704,
+      "grad_norm": 0.43509966135025024,
       "learning_rate": 1.561851223162254e-05,
-      "loss": 0.8839,
+      "loss": 0.8842,
       "num_input_tokens_seen": 109281280,
       "step": 13340
     },
     {
       "epoch": 0.6410564225690276,
-      "grad_norm": 0.4929245412349701,
+      "grad_norm": 0.49229124188423157,
       "learning_rate": 1.558180265912037e-05,
-      "loss": 0.9363,
+      "loss": 0.9354,
       "num_input_tokens_seen": 109363200,
       "step": 13350
     },
     {
       "epoch": 0.6415366146458583,
-      "grad_norm": 0.4899829626083374,
+      "grad_norm": 0.492304265499115,
       "learning_rate": 1.5545116737157522e-05,
-      "loss": 0.9724,
+      "loss": 0.9725,
       "num_input_tokens_seen": 109445120,
       "step": 13360
     },
     {
       "epoch": 0.6420168067226891,
-      "grad_norm": 0.4964812994003296,
+      "grad_norm": 0.4966655373573303,
       "learning_rate": 1.5508454557857966e-05,
       "loss": 0.7562,
       "num_input_tokens_seen": 109527040,
@@ -10706,103 +10706,103 @@
     },
     {
       "epoch": 0.6424969987995198,
-      "grad_norm": 0.44805675745010376,
+      "grad_norm": 0.44081437587738037,
       "learning_rate": 1.5471816213286054e-05,
-      "loss": 0.8147,
+      "loss": 0.8151,
       "num_input_tokens_seen": 109608960,
       "step": 13380
     },
     {
       "epoch": 0.6429771908763505,
-      "grad_norm": 0.8485963940620422,
+      "grad_norm": 0.8634435534477234,
       "learning_rate": 1.5435201795446317e-05,
-      "loss": 0.9489,
+      "loss": 0.9497,
       "num_input_tokens_seen": 109690880,
       "step": 13390
     },
     {
       "epoch": 0.6434573829531812,
-      "grad_norm": 0.4873722195625305,
+      "grad_norm": 0.4802958071231842,
       "learning_rate": 1.5398611396283153e-05,
-      "loss": 0.8655,
+      "loss": 0.8651,
       "num_input_tokens_seen": 109772800,
       "step": 13400
     },
     {
       "epoch": 0.643937575030012,
-      "grad_norm": 0.5201349258422852,
+      "grad_norm": 0.5202709436416626,
       "learning_rate": 1.536204510768069e-05,
-      "loss": 0.8621,
+      "loss": 0.8622,
       "num_input_tokens_seen": 109854720,
       "step": 13410
     },
     {
       "epoch": 0.6444177671068427,
-      "grad_norm": 0.4951586425304413,
+      "grad_norm": 0.4949398934841156,
       "learning_rate": 1.532550302146249e-05,
-      "loss": 0.8371,
+      "loss": 0.8377,
       "num_input_tokens_seen": 109936640,
       "step": 13420
     },
     {
       "epoch": 0.6448979591836734,
-      "grad_norm": 0.4836975336074829,
+      "grad_norm": 0.4846876859664917,
       "learning_rate": 1.528898522939133e-05,
-      "loss": 0.7762,
+      "loss": 0.7773,
       "num_input_tokens_seen": 110018560,
       "step": 13430
     },
     {
       "epoch": 0.6453781512605042,
-      "grad_norm": 1.549204707145691,
+      "grad_norm": 1.569688081741333,
       "learning_rate": 1.5252491823168994e-05,
-      "loss": 0.9059,
+      "loss": 0.9058,
       "num_input_tokens_seen": 110100480,
       "step": 13440
     },
     {
       "epoch": 0.6458583433373349,
-      "grad_norm": 0.39099404215812683,
+      "grad_norm": 0.39199429750442505,
       "learning_rate": 1.5216022894436043e-05,
-      "loss": 0.9104,
+      "loss": 0.9108,
       "num_input_tokens_seen": 110182400,
       "step": 13450
     },
     {
       "epoch": 0.6463385354141656,
-      "grad_norm": 0.4727102816104889,
+      "grad_norm": 0.47286975383758545,
       "learning_rate": 1.517957853477154e-05,
-      "loss": 0.9764,
+      "loss": 0.9771,
       "num_input_tokens_seen": 110264320,
       "step": 13460
     },
     {
       "epoch": 0.6468187274909964,
-      "grad_norm": 0.5066577792167664,
+      "grad_norm": 0.5029356479644775,
       "learning_rate": 1.5143158835692866e-05,
-      "loss": 0.8985,
+      "loss": 0.8984,
       "num_input_tokens_seen": 110346240,
       "step": 13470
     },
     {
       "epoch": 0.6472989195678271,
-      "grad_norm": 0.5499137043952942,
+      "grad_norm": 0.5453057289123535,
       "learning_rate": 1.5106763888655478e-05,
-      "loss": 1.1936,
+      "loss": 1.1943,
       "num_input_tokens_seen": 110428160,
       "step": 13480
     },
     {
       "epoch": 0.6477791116446578,
-      "grad_norm": 0.4742192029953003,
+      "grad_norm": 0.47629615664482117,
       "learning_rate": 1.5070393785052695e-05,
-      "loss": 0.9241,
+      "loss": 0.9233,
       "num_input_tokens_seen": 110510080,
       "step": 13490
     },
     {
       "epoch": 0.6482593037214885,
-      "grad_norm": 0.4930437207221985,
+      "grad_norm": 0.49545878171920776,
       "learning_rate": 1.5034048616215402e-05,
       "loss": 0.7714,
       "num_input_tokens_seen": 110592000,
@@ -10810,31 +10810,31 @@
     },
     {
       "epoch": 0.6487394957983194,
-      "grad_norm": 0.48428675532341003,
+      "grad_norm": 0.4852381646633148,
       "learning_rate": 1.4997728473411903e-05,
-      "loss": 1.0863,
+      "loss": 1.0867,
       "num_input_tokens_seen": 110673920,
       "step": 13510
     },
     {
       "epoch": 0.64921968787515,
-      "grad_norm": 0.49991798400878906,
+      "grad_norm": 0.4980316758155823,
       "learning_rate": 1.4961433447847672e-05,
-      "loss": 0.8677,
+      "loss": 0.8682,
       "num_input_tokens_seen": 110755840,
       "step": 13520
     },
     {
       "epoch": 0.6496998799519808,
-      "grad_norm": 0.4799818694591522,
+      "grad_norm": 0.48034679889678955,
       "learning_rate": 1.4925163630665065e-05,
-      "loss": 0.8387,
+      "loss": 0.8386,
       "num_input_tokens_seen": 110837760,
       "step": 13530
     },
     {
       "epoch": 0.6501800720288116,
-      "grad_norm": 0.4553060531616211,
+      "grad_norm": 0.45721235871315,
       "learning_rate": 1.4888919112943173e-05,
       "loss": 0.9952,
       "num_input_tokens_seen": 110919680,
@@ -10842,23 +10842,23 @@
     },
     {
       "epoch": 0.6506602641056423,
-      "grad_norm": 0.8867954611778259,
+      "grad_norm": 0.8776998519897461,
       "learning_rate": 1.4852699985697546e-05,
-      "loss": 0.8378,
+      "loss": 0.8368,
       "num_input_tokens_seen": 111001600,
       "step": 13550
     },
     {
       "epoch": 0.651140456182473,
-      "grad_norm": 0.49487099051475525,
+      "grad_norm": 0.4943079352378845,
       "learning_rate": 1.4816506339879965e-05,
-      "loss": 0.8531,
+      "loss": 0.8539,
       "num_input_tokens_seen": 111083520,
       "step": 13560
     },
     {
       "epoch": 0.6516206482593038,
-      "grad_norm": 0.5083696246147156,
+      "grad_norm": 0.49719420075416565,
       "learning_rate": 1.4780338266378232e-05,
       "loss": 0.8573,
       "num_input_tokens_seen": 111165440,
@@ -10866,23 +10866,23 @@
     },
     {
       "epoch": 0.6521008403361345,
-      "grad_norm": 0.5299503207206726,
+      "grad_norm": 0.5202630758285522,
       "learning_rate": 1.4744195856015947e-05,
-      "loss": 0.6841,
+      "loss": 0.6835,
       "num_input_tokens_seen": 111247360,
       "step": 13580
     },
     {
       "epoch": 0.6525810324129652,
-      "grad_norm": 0.49826177954673767,
+      "grad_norm": 0.4931369423866272,
       "learning_rate": 1.4708079199552221e-05,
-      "loss": 0.8537,
+      "loss": 0.8532,
       "num_input_tokens_seen": 111329280,
       "step": 13590
     },
     {
       "epoch": 0.6530612244897959,
-      "grad_norm": 0.5107062458992004,
+      "grad_norm": 0.5132030248641968,
       "learning_rate": 1.4671988387681549e-05,
       "loss": 1.038,
       "num_input_tokens_seen": 111411200,
@@ -10890,79 +10890,79 @@
     },
     {
       "epoch": 0.6535414165666267,
-      "grad_norm": 0.4928327202796936,
+      "grad_norm": 0.4891904592514038,
       "learning_rate": 1.4635923511033494e-05,
-      "loss": 0.9946,
+      "loss": 0.9945,
       "num_input_tokens_seen": 111493120,
       "step": 13610
     },
     {
       "epoch": 0.6540216086434574,
-      "grad_norm": 0.4964877963066101,
+      "grad_norm": 0.503991425037384,
       "learning_rate": 1.4599884660172485e-05,
-      "loss": 1.8003,
+      "loss": 1.7508,
       "num_input_tokens_seen": 111575040,
       "step": 13620
     },
     {
       "epoch": 0.6545018007202881,
-      "grad_norm": 0.530273973941803,
+      "grad_norm": 0.5447900891304016,
       "learning_rate": 1.4563871925597622e-05,
-      "loss": 0.8789,
+      "loss": 0.8786,
       "num_input_tokens_seen": 111656960,
       "step": 13630
     },
     {
       "epoch": 0.6549819927971189,
-      "grad_norm": 0.48265501856803894,
+      "grad_norm": 0.49095600843429565,
       "learning_rate": 1.452788539774241e-05,
-      "loss": 0.6229,
+      "loss": 0.6227,
       "num_input_tokens_seen": 111738880,
       "step": 13640
     },
     {
       "epoch": 0.6554621848739496,
-      "grad_norm": 0.4895990788936615,
+      "grad_norm": 0.49171656370162964,
       "learning_rate": 1.4491925166974532e-05,
-      "loss": 0.8647,
+      "loss": 0.8651,
       "num_input_tokens_seen": 111820800,
       "step": 13650
     },
     {
       "epoch": 0.6559423769507803,
-      "grad_norm": 0.534498929977417,
+      "grad_norm": 0.5414605736732483,
       "learning_rate": 1.4455991323595655e-05,
-      "loss": 0.8938,
+      "loss": 0.8945,
       "num_input_tokens_seen": 111902720,
       "step": 13660
     },
     {
       "epoch": 0.6564225690276111,
-      "grad_norm": 0.48878028988838196,
+      "grad_norm": 0.49501684308052063,
       "learning_rate": 1.4420083957841185e-05,
-      "loss": 0.8419,
+      "loss": 0.8423,
       "num_input_tokens_seen": 111984640,
       "step": 13670
     },
     {
       "epoch": 0.6569027611044418,
-      "grad_norm": 0.49631527066230774,
+      "grad_norm": 0.4984915256500244,
       "learning_rate": 1.4384203159880017e-05,
-      "loss": 0.8683,
+      "loss": 0.8682,
       "num_input_tokens_seen": 112066560,
       "step": 13680
     },
     {
       "epoch": 0.6573829531812725,
-      "grad_norm": 0.5460849404335022,
+      "grad_norm": 0.5721986293792725,
       "learning_rate": 1.4348349019814344e-05,
-      "loss": 1.1068,
+      "loss": 1.1063,
       "num_input_tokens_seen": 112148480,
       "step": 13690
     },
     {
       "epoch": 0.6578631452581032,
-      "grad_norm": 0.4952901601791382,
+      "grad_norm": 0.5022817850112915,
       "learning_rate": 1.4312521627679428e-05,
       "loss": 0.9632,
       "num_input_tokens_seen": 112230400,
@@ -10970,47 +10970,47 @@
     },
     {
       "epoch": 0.658343337334934,
-      "grad_norm": 0.5365872979164124,
+      "grad_norm": 0.5426187515258789,
       "learning_rate": 1.4276721073443344e-05,
-      "loss": 0.8043,
+      "loss": 0.804,
       "num_input_tokens_seen": 112312320,
       "step": 13710
     },
     {
       "epoch": 0.6588235294117647,
-      "grad_norm": 0.5443843007087708,
+      "grad_norm": 0.5450629591941833,
       "learning_rate": 1.4240947447006764e-05,
-      "loss": 1.025,
+      "loss": 1.0239,
       "num_input_tokens_seen": 112394240,
       "step": 13720
     },
     {
       "epoch": 0.6593037214885954,
-      "grad_norm": 0.49245166778564453,
+      "grad_norm": 0.4963448643684387,
       "learning_rate": 1.4205200838202782e-05,
-      "loss": 1.0484,
+      "loss": 1.049,
       "num_input_tokens_seen": 112476160,
       "step": 13730
     },
     {
       "epoch": 0.6597839135654262,
-      "grad_norm": 0.5047618746757507,
+      "grad_norm": 0.5081272125244141,
       "learning_rate": 1.4169481336796597e-05,
-      "loss": 1.0862,
+      "loss": 1.0863,
       "num_input_tokens_seen": 112558080,
       "step": 13740
     },
     {
       "epoch": 0.6602641056422569,
-      "grad_norm": 0.5078234076499939,
+      "grad_norm": 0.5098510384559631,
       "learning_rate": 1.4133789032485367e-05,
-      "loss": 0.9006,
+      "loss": 0.9013,
       "num_input_tokens_seen": 112640000,
       "step": 13750
     },
     {
       "epoch": 0.6607442977190876,
-      "grad_norm": 1.923171877861023,
+      "grad_norm": 1.9367784261703491,
       "learning_rate": 1.4098124014897961e-05,
       "loss": 0.8773,
       "num_input_tokens_seen": 112721920,
@@ -11018,119 +11018,119 @@
     },
     {
       "epoch": 0.6612244897959184,
-      "grad_norm": 0.5198791027069092,
+      "grad_norm": 0.514941930770874,
       "learning_rate": 1.4062486373594694e-05,
-      "loss": 0.802,
+      "loss": 0.7995,
       "num_input_tokens_seen": 112803840,
       "step": 13770
     },
     {
       "epoch": 0.6617046818727491,
-      "grad_norm": 0.4909871518611908,
+      "grad_norm": 0.48932793736457825,
       "learning_rate": 1.4026876198067163e-05,
-      "loss": 1.0097,
+      "loss": 1.0093,
       "num_input_tokens_seen": 112885760,
       "step": 13780
     },
     {
       "epoch": 0.6621848739495798,
-      "grad_norm": 0.4735969305038452,
+      "grad_norm": 0.4777870774269104,
       "learning_rate": 1.399129357773799e-05,
-      "loss": 0.9333,
+      "loss": 0.9332,
       "num_input_tokens_seen": 112967680,
       "step": 13790
     },
     {
       "epoch": 0.6626650660264105,
-      "grad_norm": 0.52108234167099,
+      "grad_norm": 0.5338255763053894,
       "learning_rate": 1.3955738601960588e-05,
-      "loss": 0.8832,
+      "loss": 0.8836,
       "num_input_tokens_seen": 113049600,
       "step": 13800
     },
     {
       "epoch": 0.6631452581032413,
-      "grad_norm": 0.49739062786102295,
+      "grad_norm": 0.5129244923591614,
       "learning_rate": 1.392021136001897e-05,
-      "loss": 0.7867,
+      "loss": 0.7864,
       "num_input_tokens_seen": 113131520,
       "step": 13810
     },
     {
       "epoch": 0.663625450180072,
-      "grad_norm": 0.5152223706245422,
+      "grad_norm": 0.502555251121521,
       "learning_rate": 1.3884711941127487e-05,
-      "loss": 0.9924,
+      "loss": 0.9925,
       "num_input_tokens_seen": 113213440,
       "step": 13820
     },
     {
       "epoch": 0.6641056422569027,
-      "grad_norm": 0.4832291901111603,
+      "grad_norm": 0.4870724678039551,
       "learning_rate": 1.384924043443062e-05,
-      "loss": 0.7811,
+      "loss": 0.7822,
       "num_input_tokens_seen": 113295360,
       "step": 13830
     },
     {
       "epoch": 0.6645858343337335,
-      "grad_norm": 0.49378257989883423,
+      "grad_norm": 0.49208691716194153,
       "learning_rate": 1.3813796929002779e-05,
-      "loss": 0.7269,
+      "loss": 0.7278,
       "num_input_tokens_seen": 113377280,
       "step": 13840
     },
     {
       "epoch": 0.6650660264105642,
-      "grad_norm": 0.4792208969593048,
+      "grad_norm": 0.4737597107887268,
       "learning_rate": 1.3778381513848055e-05,
-      "loss": 0.7797,
+      "loss": 0.7792,
       "num_input_tokens_seen": 113459200,
       "step": 13850
     },
     {
       "epoch": 0.6655462184873949,
-      "grad_norm": 0.5114631056785583,
+      "grad_norm": 0.5034937858581543,
       "learning_rate": 1.3742994277899967e-05,
-      "loss": 0.7852,
+      "loss": 0.7843,
       "num_input_tokens_seen": 113541120,
       "step": 13860
     },
     {
       "epoch": 0.6660264105642257,
-      "grad_norm": 0.49299582839012146,
+      "grad_norm": 0.49161359667778015,
       "learning_rate": 1.370763531002132e-05,
-      "loss": 1.2018,
+      "loss": 1.2019,
       "num_input_tokens_seen": 113623040,
       "step": 13870
     },
     {
       "epoch": 0.6665066026410564,
-      "grad_norm": 0.48771587014198303,
+      "grad_norm": 0.487804114818573,
       "learning_rate": 1.3672304699003908e-05,
-      "loss": 0.7667,
+      "loss": 0.7672,
       "num_input_tokens_seen": 113704960,
       "step": 13880
     },
     {
       "epoch": 0.6669867947178871,
-      "grad_norm": 0.6052321791648865,
+      "grad_norm": 0.6117998957633972,
       "learning_rate": 1.3637002533568302e-05,
-      "loss": 0.8471,
+      "loss": 0.847,
       "num_input_tokens_seen": 113786880,
       "step": 13890
     },
     {
       "epoch": 0.6674669867947179,
-      "grad_norm": 0.47869783639907837,
+      "grad_norm": 0.47483766078948975,
       "learning_rate": 1.3601728902363681e-05,
-      "loss": 1.1894,
+      "loss": 1.1898,
       "num_input_tokens_seen": 113868800,
       "step": 13900
     },
     {
       "epoch": 0.6679471788715486,
-      "grad_norm": 0.4907095432281494,
+      "grad_norm": 0.48828792572021484,
       "learning_rate": 1.356648389396754e-05,
       "loss": 0.9134,
       "num_input_tokens_seen": 113950720,
@@ -11138,119 +11138,119 @@
     },
     {
       "epoch": 0.6684273709483793,
-      "grad_norm": 0.4225086569786072,
+      "grad_norm": 0.4222254455089569,
       "learning_rate": 1.3531267596885488e-05,
-      "loss": 1.1423,
+      "loss": 1.1422,
       "num_input_tokens_seen": 114032640,
       "step": 13920
     },
     {
       "epoch": 0.66890756302521,
-      "grad_norm": 0.4796990156173706,
+      "grad_norm": 0.47872963547706604,
       "learning_rate": 1.349608009955107e-05,
-      "loss": 0.7921,
+      "loss": 0.7922,
       "num_input_tokens_seen": 114114560,
       "step": 13930
     },
     {
       "epoch": 0.6693877551020408,
-      "grad_norm": 0.4856274724006653,
+      "grad_norm": 0.48117169737815857,
       "learning_rate": 1.34609214903255e-05,
-      "loss": 0.8029,
+      "loss": 0.8027,
       "num_input_tokens_seen": 114196480,
       "step": 13940
     },
     {
       "epoch": 0.6698679471788715,
-      "grad_norm": 0.4970095753669739,
+      "grad_norm": 0.49767807126045227,
       "learning_rate": 1.3425791857497422e-05,
-      "loss": 0.9638,
+      "loss": 0.9642,
       "num_input_tokens_seen": 114278400,
       "step": 13950
     },
     {
       "epoch": 0.6703481392557022,
-      "grad_norm": 0.47488856315612793,
+      "grad_norm": 0.4757895767688751,
       "learning_rate": 1.3390691289282754e-05,
-      "loss": 0.8415,
+      "loss": 0.8414,
       "num_input_tokens_seen": 114360320,
       "step": 13960
     },
     {
       "epoch": 0.6708283313325331,
-      "grad_norm": 0.4611085057258606,
+      "grad_norm": 0.4622432291507721,
       "learning_rate": 1.335561987382441e-05,
-      "loss": 0.9274,
+      "loss": 0.9272,
       "num_input_tokens_seen": 114442240,
       "step": 13970
     },
     {
       "epoch": 0.6713085234093638,
-      "grad_norm": 0.4784678518772125,
+      "grad_norm": 0.465688019990921,
       "learning_rate": 1.3320577699192086e-05,
-      "loss": 0.7943,
+      "loss": 0.7937,
       "num_input_tokens_seen": 114524160,
       "step": 13980
     },
     {
       "epoch": 0.6717887154861945,
-      "grad_norm": 0.49473994970321655,
+      "grad_norm": 0.4986606240272522,
       "learning_rate": 1.3285564853382076e-05,
-      "loss": 0.9644,
+      "loss": 0.9645,
       "num_input_tokens_seen": 114606080,
       "step": 13990
     },
     {
       "epoch": 0.6722689075630253,
-      "grad_norm": 0.45741796493530273,
+      "grad_norm": 0.46979689598083496,
       "learning_rate": 1.325058142431701e-05,
-      "loss": 0.8024,
+      "loss": 0.8027,
       "num_input_tokens_seen": 114688000,
       "step": 14000
     },
     {
       "epoch": 0.672749099639856,
-      "grad_norm": 0.4712899625301361,
+      "grad_norm": 0.47202202677726746,
       "learning_rate": 1.321562749984563e-05,
-      "loss": 0.7327,
+      "loss": 0.7335,
       "num_input_tokens_seen": 114769920,
       "step": 14010
     },
     {
       "epoch": 0.6732292917166867,
-      "grad_norm": 0.4954095780849457,
+      "grad_norm": 0.49672776460647583,
       "learning_rate": 1.318070316774262e-05,
-      "loss": 0.9788,
+      "loss": 0.9804,
       "num_input_tokens_seen": 114851840,
       "step": 14020
     },
     {
       "epoch": 0.6737094837935174,
-      "grad_norm": 0.5219026207923889,
+      "grad_norm": 0.5184418559074402,
       "learning_rate": 1.3145808515708347e-05,
-      "loss": 0.7022,
+      "loss": 0.7016,
       "num_input_tokens_seen": 114933760,
       "step": 14030
     },
     {
       "epoch": 0.6741896758703482,
-      "grad_norm": 0.4924544394016266,
+      "grad_norm": 0.491964727640152,
       "learning_rate": 1.3110943631368616e-05,
-      "loss": 0.9829,
+      "loss": 0.9825,
       "num_input_tokens_seen": 115015680,
       "step": 14040
     },
     {
       "epoch": 0.6746698679471789,
-      "grad_norm": 0.49134406447410583,
+      "grad_norm": 0.4879056215286255,
       "learning_rate": 1.3076108602274522e-05,
-      "loss": 0.7022,
+      "loss": 0.7027,
       "num_input_tokens_seen": 115097600,
       "step": 14050
     },
     {
       "epoch": 0.6751500600240096,
-      "grad_norm": 0.4494248926639557,
+      "grad_norm": 0.4461091160774231,
       "learning_rate": 1.3041303515902179e-05,
       "loss": 0.8251,
       "num_input_tokens_seen": 115179520,
@@ -11258,15 +11258,15 @@
     },
     {
       "epoch": 0.6756302521008404,
-      "grad_norm": 0.46173524856567383,
+      "grad_norm": 0.45616379380226135,
       "learning_rate": 1.3006528459652476e-05,
-      "loss": 1.0901,
+      "loss": 1.0897,
       "num_input_tokens_seen": 115261440,
       "step": 14070
     },
     {
       "epoch": 0.6761104441776711,
-      "grad_norm": 0.5183840394020081,
+      "grad_norm": 0.5146764516830444,
       "learning_rate": 1.2971783520850939e-05,
       "loss": 0.9864,
       "num_input_tokens_seen": 115343360,
@@ -11274,151 +11274,151 @@
     },
     {
       "epoch": 0.6765906362545018,
-      "grad_norm": 0.4756566286087036,
+      "grad_norm": 0.4778116047382355,
       "learning_rate": 1.2937068786747438e-05,
-      "loss": 0.9465,
+      "loss": 0.9467,
       "num_input_tokens_seen": 115425280,
       "step": 14090
     },
     {
       "epoch": 0.6770708283313326,
-      "grad_norm": 0.47903919219970703,
+      "grad_norm": 0.4811628758907318,
       "learning_rate": 1.2902384344515986e-05,
-      "loss": 0.9708,
+      "loss": 0.971,
       "num_input_tokens_seen": 115507200,
       "step": 14100
     },
     {
       "epoch": 0.6775510204081633,
-      "grad_norm": 0.6484615206718445,
+      "grad_norm": 0.6663933396339417,
       "learning_rate": 1.286773028125455e-05,
-      "loss": 0.7833,
+      "loss": 0.7835,
       "num_input_tokens_seen": 115589120,
       "step": 14110
     },
     {
       "epoch": 0.678031212484994,
-      "grad_norm": 0.48011910915374756,
+      "grad_norm": 0.47772663831710815,
       "learning_rate": 1.2833106683984808e-05,
-      "loss": 0.9278,
+      "loss": 0.9291,
       "num_input_tokens_seen": 115671040,
       "step": 14120
     },
     {
       "epoch": 0.6785114045618247,
-      "grad_norm": 0.4931375980377197,
+      "grad_norm": 0.48866212368011475,
       "learning_rate": 1.279851363965193e-05,
-      "loss": 0.8275,
+      "loss": 0.8278,
       "num_input_tokens_seen": 115752960,
       "step": 14130
     },
     {
       "epoch": 0.6789915966386555,
-      "grad_norm": 0.5052535533905029,
+      "grad_norm": 0.504115879535675,
       "learning_rate": 1.2763951235124346e-05,
-      "loss": 0.8764,
+      "loss": 0.8757,
       "num_input_tokens_seen": 115834880,
       "step": 14140
     },
     {
       "epoch": 0.6794717887154862,
-      "grad_norm": 0.49252551794052124,
+      "grad_norm": 0.4960152804851532,
       "learning_rate": 1.2729419557193573e-05,
-      "loss": 0.9236,
+      "loss": 0.9235,
       "num_input_tokens_seen": 115916800,
       "step": 14150
     },
     {
       "epoch": 0.6799519807923169,
-      "grad_norm": 0.45454323291778564,
+      "grad_norm": 0.4555470049381256,
       "learning_rate": 1.2694918692573954e-05,
-      "loss": 0.7963,
+      "loss": 0.7965,
       "num_input_tokens_seen": 115998720,
       "step": 14160
     },
     {
       "epoch": 0.6804321728691477,
-      "grad_norm": 0.47249314188957214,
+      "grad_norm": 0.475454181432724,
       "learning_rate": 1.2660448727902457e-05,
-      "loss": 0.7971,
+      "loss": 0.7966,
       "num_input_tokens_seen": 116080640,
       "step": 14170
     },
     {
       "epoch": 0.6809123649459784,
-      "grad_norm": 0.4945741891860962,
+      "grad_norm": 0.49502047896385193,
       "learning_rate": 1.2626009749738444e-05,
-      "loss": 0.7678,
+      "loss": 0.7681,
       "num_input_tokens_seen": 116162560,
       "step": 14180
     },
     {
       "epoch": 0.6813925570228091,
-      "grad_norm": 0.4679727852344513,
+      "grad_norm": 0.4328559339046478,
       "learning_rate": 1.2591601844563488e-05,
-      "loss": 0.8935,
+      "loss": 0.8936,
       "num_input_tokens_seen": 116244480,
       "step": 14190
     },
     {
       "epoch": 0.6818727490996399,
-      "grad_norm": 0.49965664744377136,
+      "grad_norm": 0.5078925490379333,
       "learning_rate": 1.2557225098781105e-05,
-      "loss": 0.9236,
+      "loss": 0.9239,
       "num_input_tokens_seen": 116326400,
       "step": 14200
     },
     {
       "epoch": 0.6823529411764706,
-      "grad_norm": 0.46757179498672485,
+      "grad_norm": 0.46798259019851685,
       "learning_rate": 1.2522879598716595e-05,
-      "loss": 0.8667,
+      "loss": 0.8666,
       "num_input_tokens_seen": 116408320,
       "step": 14210
     },
     {
       "epoch": 0.6828331332533013,
-      "grad_norm": 0.6531784534454346,
+      "grad_norm": 0.6590942740440369,
       "learning_rate": 1.2488565430616785e-05,
-      "loss": 0.8728,
+      "loss": 0.873,
       "num_input_tokens_seen": 116490240,
       "step": 14220
     },
     {
       "epoch": 0.683313325330132,
-      "grad_norm": 0.4773021340370178,
+      "grad_norm": 0.4791242778301239,
       "learning_rate": 1.2454282680649804e-05,
-      "loss": 0.6747,
+      "loss": 0.6743,
       "num_input_tokens_seen": 116572160,
       "step": 14230
     },
     {
       "epoch": 0.6837935174069628,
-      "grad_norm": 0.6445793509483337,
+      "grad_norm": 0.656548023223877,
       "learning_rate": 1.2420031434904906e-05,
-      "loss": 0.864,
+      "loss": 0.8642,
       "num_input_tokens_seen": 116654080,
       "step": 14240
     },
     {
       "epoch": 0.6842737094837935,
-      "grad_norm": 0.479495644569397,
+      "grad_norm": 0.4806419909000397,
       "learning_rate": 1.2385811779392236e-05,
-      "loss": 0.8979,
+      "loss": 0.8987,
       "num_input_tokens_seen": 116736000,
       "step": 14250
     },
     {
       "epoch": 0.6847539015606242,
-      "grad_norm": 0.8227359652519226,
+      "grad_norm": 0.8275937438011169,
       "learning_rate": 1.2351623800042587e-05,
-      "loss": 0.9371,
+      "loss": 0.9379,
       "num_input_tokens_seen": 116817920,
       "step": 14260
     },
     {
       "epoch": 0.685234093637455,
-      "grad_norm": 0.41954296827316284,
+      "grad_norm": 0.44159382581710815,
       "learning_rate": 1.2317467582707238e-05,
       "loss": 0.9432,
       "num_input_tokens_seen": 116899840,
@@ -11426,79 +11426,79 @@
     },
     {
       "epoch": 0.6857142857142857,
-      "grad_norm": 0.5001243948936462,
+      "grad_norm": 0.5025272965431213,
       "learning_rate": 1.2283343213157688e-05,
-      "loss": 0.8542,
+      "loss": 0.854,
       "num_input_tokens_seen": 116981760,
       "step": 14280
     },
     {
       "epoch": 0.6861944777911164,
-      "grad_norm": 0.5058388113975525,
+      "grad_norm": 0.5092597603797913,
       "learning_rate": 1.2249250777085456e-05,
-      "loss": 0.897,
+      "loss": 0.8973,
       "num_input_tokens_seen": 117063680,
       "step": 14290
     },
     {
       "epoch": 0.6866746698679472,
-      "grad_norm": 0.5120940804481506,
+      "grad_norm": 0.47980546951293945,
       "learning_rate": 1.221519036010189e-05,
-      "loss": 0.9898,
+      "loss": 0.9893,
       "num_input_tokens_seen": 117145600,
       "step": 14300
     },
     {
       "epoch": 0.6871548619447779,
-      "grad_norm": 0.5334838032722473,
+      "grad_norm": 0.5349389910697937,
       "learning_rate": 1.2181162047737942e-05,
-      "loss": 0.7581,
+      "loss": 0.7582,
       "num_input_tokens_seen": 117227520,
       "step": 14310
     },
     {
       "epoch": 0.6876350540216086,
-      "grad_norm": 0.4675463140010834,
+      "grad_norm": 0.467484712600708,
       "learning_rate": 1.2147165925443904e-05,
-      "loss": 1.0274,
+      "loss": 1.0281,
       "num_input_tokens_seen": 117309440,
       "step": 14320
     },
     {
       "epoch": 0.6881152460984393,
-      "grad_norm": 0.7059617638587952,
+      "grad_norm": 0.6963463425636292,
       "learning_rate": 1.2113202078589267e-05,
-      "loss": 0.8403,
+      "loss": 0.84,
       "num_input_tokens_seen": 117391360,
       "step": 14330
     },
     {
       "epoch": 0.6885954381752701,
-      "grad_norm": 0.5068497061729431,
+      "grad_norm": 0.507810652256012,
       "learning_rate": 1.2079270592462475e-05,
-      "loss": 0.8582,
+      "loss": 0.8581,
       "num_input_tokens_seen": 117473280,
       "step": 14340
     },
     {
       "epoch": 0.6890756302521008,
-      "grad_norm": 0.5074156522750854,
+      "grad_norm": 0.5044135451316833,
       "learning_rate": 1.204537155227068e-05,
-      "loss": 0.876,
+      "loss": 0.8766,
       "num_input_tokens_seen": 117555200,
       "step": 14350
     },
     {
       "epoch": 0.6895558223289315,
-      "grad_norm": 0.5661312341690063,
+      "grad_norm": 0.5582197904586792,
       "learning_rate": 1.201150504313959e-05,
-      "loss": 0.9847,
+      "loss": 0.985,
       "num_input_tokens_seen": 117637120,
       "step": 14360
     },
     {
       "epoch": 0.6900360144057623,
-      "grad_norm": 0.48914408683776855,
+      "grad_norm": 0.4891442358493805,
       "learning_rate": 1.1977671150113206e-05,
       "loss": 0.8981,
       "num_input_tokens_seen": 117719040,
@@ -11506,23 +11506,23 @@
     },
     {
       "epoch": 0.690516206482593,
-      "grad_norm": 0.5188919305801392,
+      "grad_norm": 0.5189415812492371,
       "learning_rate": 1.1943869958153613e-05,
-      "loss": 0.929,
+      "loss": 0.9291,
       "num_input_tokens_seen": 117800960,
       "step": 14380
     },
     {
       "epoch": 0.6909963985594237,
-      "grad_norm": 0.4931916892528534,
+      "grad_norm": 0.49603569507598877,
       "learning_rate": 1.1910101552140806e-05,
-      "loss": 0.9613,
+      "loss": 0.9612,
       "num_input_tokens_seen": 117882880,
       "step": 14390
     },
     {
       "epoch": 0.6914765906362546,
-      "grad_norm": 0.5177183747291565,
+      "grad_norm": 0.5161275267601013,
       "learning_rate": 1.1876366016872445e-05,
       "loss": 0.9258,
       "num_input_tokens_seen": 117964800,
@@ -11530,151 +11530,151 @@
     },
     {
       "epoch": 0.6919567827130852,
-      "grad_norm": 0.48315808176994324,
+      "grad_norm": 0.48330655694007874,
       "learning_rate": 1.1842663437063613e-05,
-      "loss": 0.7828,
+      "loss": 0.783,
       "num_input_tokens_seen": 118046720,
       "step": 14410
     },
     {
       "epoch": 0.692436974789916,
-      "grad_norm": 0.4908153712749481,
+      "grad_norm": 0.4942614436149597,
       "learning_rate": 1.180899389734668e-05,
-      "loss": 0.9457,
+      "loss": 0.9465,
       "num_input_tokens_seen": 118128640,
       "step": 14420
     },
     {
       "epoch": 0.6929171668667466,
-      "grad_norm": 0.47637316584587097,
+      "grad_norm": 0.47883716225624084,
       "learning_rate": 1.1775357482271032e-05,
-      "loss": 0.872,
+      "loss": 0.8717,
       "num_input_tokens_seen": 118210560,
       "step": 14430
     },
     {
       "epoch": 0.6933973589435775,
-      "grad_norm": 0.5091587901115417,
+      "grad_norm": 0.5108613967895508,
       "learning_rate": 1.1741754276302851e-05,
-      "loss": 0.8554,
+      "loss": 0.8557,
       "num_input_tokens_seen": 118292480,
       "step": 14440
     },
     {
       "epoch": 0.6938775510204082,
-      "grad_norm": 0.49093228578567505,
+      "grad_norm": 0.4875372350215912,
       "learning_rate": 1.170818436382497e-05,
-      "loss": 0.9142,
+      "loss": 0.9129,
       "num_input_tokens_seen": 118374400,
       "step": 14450
     },
     {
       "epoch": 0.6943577430972389,
-      "grad_norm": 0.47567296028137207,
+      "grad_norm": 0.4819384217262268,
       "learning_rate": 1.1674647829136581e-05,
-      "loss": 0.9356,
+      "loss": 0.9353,
       "num_input_tokens_seen": 118456320,
       "step": 14460
     },
     {
       "epoch": 0.6948379351740697,
-      "grad_norm": 0.49377357959747314,
+      "grad_norm": 0.49324291944503784,
       "learning_rate": 1.164114475645306e-05,
-      "loss": 0.9423,
+      "loss": 0.9419,
       "num_input_tokens_seen": 118538240,
       "step": 14470
     },
     {
       "epoch": 0.6953181272509004,
-      "grad_norm": 0.49795305728912354,
+      "grad_norm": 0.49606454372406006,
       "learning_rate": 1.1607675229905776e-05,
-      "loss": 0.8267,
+      "loss": 0.8269,
       "num_input_tokens_seen": 118620160,
       "step": 14480
     },
     {
       "epoch": 0.6957983193277311,
-      "grad_norm": 0.42667868733406067,
+      "grad_norm": 0.4274294674396515,
       "learning_rate": 1.1574239333541856e-05,
-      "loss": 0.8272,
+      "loss": 0.8265,
       "num_input_tokens_seen": 118702080,
       "step": 14490
     },
     {
       "epoch": 0.6962785114045619,
-      "grad_norm": 0.4863702356815338,
+      "grad_norm": 0.4905083179473877,
       "learning_rate": 1.1540837151323951e-05,
-      "loss": 1.0219,
+      "loss": 1.0225,
       "num_input_tokens_seen": 118784000,
       "step": 14500
     },
     {
       "epoch": 0.6967587034813926,
-      "grad_norm": 0.42491790652275085,
+      "grad_norm": 0.42654213309288025,
       "learning_rate": 1.150746876713008e-05,
-      "loss": 1.0269,
+      "loss": 1.0267,
       "num_input_tokens_seen": 118865920,
       "step": 14510
     },
     {
       "epoch": 0.6972388955582233,
-      "grad_norm": 1.2376126050949097,
+      "grad_norm": 1.1795427799224854,
       "learning_rate": 1.1474134264753384e-05,
-      "loss": 0.9424,
+      "loss": 0.9422,
       "num_input_tokens_seen": 118947840,
       "step": 14520
     },
     {
       "epoch": 0.697719087635054,
-      "grad_norm": 0.4999983310699463,
+      "grad_norm": 0.4983981251716614,
       "learning_rate": 1.1440833727901894e-05,
-      "loss": 0.8436,
+      "loss": 0.8435,
       "num_input_tokens_seen": 119029760,
       "step": 14530
     },
     {
       "epoch": 0.6981992797118848,
-      "grad_norm": 1.1919395923614502,
+      "grad_norm": 1.133950114250183,
       "learning_rate": 1.1407567240198397e-05,
-      "loss": 0.7876,
+      "loss": 0.7872,
       "num_input_tokens_seen": 119111680,
       "step": 14540
     },
     {
       "epoch": 0.6986794717887155,
-      "grad_norm": 0.4805420935153961,
+      "grad_norm": 0.4824429750442505,
       "learning_rate": 1.1374334885180135e-05,
-      "loss": 0.8579,
+      "loss": 0.8584,
       "num_input_tokens_seen": 119193600,
       "step": 14550
     },
     {
       "epoch": 0.6991596638655462,
-      "grad_norm": 0.505160391330719,
+      "grad_norm": 0.5034725666046143,
       "learning_rate": 1.1341136746298647e-05,
-      "loss": 0.9189,
+      "loss": 0.9188,
       "num_input_tokens_seen": 119275520,
       "step": 14560
     },
     {
       "epoch": 0.699639855942377,
-      "grad_norm": 0.369730681180954,
+      "grad_norm": 0.3501124978065491,
       "learning_rate": 1.1307972906919562e-05,
-      "loss": 0.834,
+      "loss": 0.8331,
       "num_input_tokens_seen": 119357440,
       "step": 14570
     },
     {
       "epoch": 0.7001200480192077,
-      "grad_norm": 0.493691086769104,
+      "grad_norm": 0.4934571087360382,
       "learning_rate": 1.1274843450322381e-05,
-      "loss": 0.8542,
+      "loss": 0.8539,
       "num_input_tokens_seen": 119439360,
       "step": 14580
     },
     {
       "epoch": 0.7006002400960384,
-      "grad_norm": 0.49764561653137207,
+      "grad_norm": 0.49862003326416016,
       "learning_rate": 1.1241748459700241e-05,
       "loss": 0.8792,
       "num_input_tokens_seen": 119521280,
@@ -11682,63 +11682,63 @@
     },
     {
       "epoch": 0.7010804321728692,
-      "grad_norm": 0.6448412537574768,
+      "grad_norm": 0.6504396796226501,
       "learning_rate": 1.1208688018159746e-05,
-      "loss": 0.9641,
+      "loss": 0.9642,
       "num_input_tokens_seen": 119603200,
       "step": 14600
     },
     {
       "epoch": 0.7015606242496999,
-      "grad_norm": 0.499411016702652,
+      "grad_norm": 0.4994623363018036,
       "learning_rate": 1.1175662208720758e-05,
-      "loss": 1.0869,
+      "loss": 1.0876,
       "num_input_tokens_seen": 119685120,
       "step": 14610
     },
     {
       "epoch": 0.7020408163265306,
-      "grad_norm": 0.5112809538841248,
+      "grad_norm": 0.5138194561004639,
       "learning_rate": 1.1142671114316127e-05,
-      "loss": 0.9409,
+      "loss": 0.941,
       "num_input_tokens_seen": 119767040,
       "step": 14620
     },
     {
       "epoch": 0.7025210084033613,
-      "grad_norm": 0.5502108931541443,
+      "grad_norm": 0.5315556526184082,
       "learning_rate": 1.1109714817791584e-05,
-      "loss": 0.9631,
+      "loss": 0.9637,
       "num_input_tokens_seen": 119848960,
       "step": 14630
     },
     {
       "epoch": 0.7030012004801921,
-      "grad_norm": 0.4769909381866455,
+      "grad_norm": 0.48073628544807434,
       "learning_rate": 1.1076793401905419e-05,
-      "loss": 0.8025,
+      "loss": 0.8024,
       "num_input_tokens_seen": 119930880,
       "step": 14640
     },
     {
       "epoch": 0.7034813925570228,
-      "grad_norm": 0.5704587697982788,
+      "grad_norm": 0.5759013295173645,
       "learning_rate": 1.1043906949328387e-05,
-      "loss": 0.8847,
+      "loss": 0.8844,
       "num_input_tokens_seen": 120012800,
       "step": 14650
     },
     {
       "epoch": 0.7039615846338535,
-      "grad_norm": 0.49126431345939636,
+      "grad_norm": 0.4914475977420807,
       "learning_rate": 1.1011055542643398e-05,
-      "loss": 0.8116,
+      "loss": 0.8125,
       "num_input_tokens_seen": 120094720,
       "step": 14660
     },
     {
       "epoch": 0.7044417767106843,
-      "grad_norm": 0.5111186504364014,
+      "grad_norm": 0.5116990208625793,
       "learning_rate": 1.0978239264345397e-05,
       "loss": 0.9473,
       "num_input_tokens_seen": 120176640,
@@ -11746,71 +11746,71 @@
     },
     {
       "epoch": 0.704921968787515,
-      "grad_norm": 0.48010018467903137,
+      "grad_norm": 0.48121222853660583,
       "learning_rate": 1.0945458196841078e-05,
-      "loss": 0.9452,
+      "loss": 0.9443,
       "num_input_tokens_seen": 120258560,
       "step": 14680
     },
     {
       "epoch": 0.7054021608643457,
-      "grad_norm": 0.4862194061279297,
+      "grad_norm": 0.48448559641838074,
       "learning_rate": 1.0912712422448737e-05,
-      "loss": 1.089,
+      "loss": 1.0888,
       "num_input_tokens_seen": 120340480,
       "step": 14690
     },
     {
       "epoch": 0.7058823529411765,
-      "grad_norm": 0.5020564198493958,
+      "grad_norm": 0.49979427456855774,
       "learning_rate": 1.0880002023398058e-05,
-      "loss": 0.8222,
+      "loss": 0.8219,
       "num_input_tokens_seen": 120422400,
       "step": 14700
     },
     {
       "epoch": 0.7063625450180072,
-      "grad_norm": 0.5132585167884827,
+      "grad_norm": 0.5122401118278503,
       "learning_rate": 1.0847327081829853e-05,
-      "loss": 0.8349,
+      "loss": 0.8348,
       "num_input_tokens_seen": 120504320,
       "step": 14710
     },
     {
       "epoch": 0.7068427370948379,
-      "grad_norm": 0.5931466221809387,
+      "grad_norm": 0.6076194643974304,
       "learning_rate": 1.0814687679795924e-05,
-      "loss": 1.4205,
+      "loss": 1.4222,
       "num_input_tokens_seen": 120586240,
       "step": 14720
     },
     {
       "epoch": 0.7073229291716686,
-      "grad_norm": 0.529240608215332,
+      "grad_norm": 0.5290340185165405,
       "learning_rate": 1.0782083899258827e-05,
-      "loss": 0.9358,
+      "loss": 0.9349,
       "num_input_tokens_seen": 120668160,
       "step": 14730
     },
     {
       "epoch": 0.7078031212484994,
-      "grad_norm": 0.5132381319999695,
+      "grad_norm": 0.510225772857666,
       "learning_rate": 1.0749515822091658e-05,
-      "loss": 0.8712,
+      "loss": 0.871,
       "num_input_tokens_seen": 120750080,
       "step": 14740
     },
     {
       "epoch": 0.7082833133253301,
-      "grad_norm": 0.5172890424728394,
+      "grad_norm": 0.5205603241920471,
       "learning_rate": 1.0716983530077843e-05,
-      "loss": 1.2485,
+      "loss": 1.2482,
       "num_input_tokens_seen": 120832000,
       "step": 14750
     },
     {
       "epoch": 0.7087635054021608,
-      "grad_norm": 0.4927189350128174,
+      "grad_norm": 0.4906553030014038,
       "learning_rate": 1.0684487104910974e-05,
       "loss": 0.7965,
       "num_input_tokens_seen": 120913920,
@@ -11818,7 +11818,7 @@
     },
     {
       "epoch": 0.7092436974789916,
-      "grad_norm": 0.3909131586551666,
+      "grad_norm": 0.4048673212528229,
       "learning_rate": 1.0652026628194567e-05,
       "loss": 0.8365,
       "num_input_tokens_seen": 120995840,
@@ -11826,111 +11826,111 @@
     },
     {
       "epoch": 0.7097238895558223,
-      "grad_norm": 0.5766429901123047,
+      "grad_norm": 0.5668584108352661,
       "learning_rate": 1.0619602181441848e-05,
-      "loss": 0.8742,
+      "loss": 0.8743,
       "num_input_tokens_seen": 121077760,
       "step": 14780
     },
     {
       "epoch": 0.710204081632653,
-      "grad_norm": 0.5812046527862549,
+      "grad_norm": 0.5875443816184998,
       "learning_rate": 1.0587213846075591e-05,
-      "loss": 0.7952,
+      "loss": 0.7954,
       "num_input_tokens_seen": 121159680,
       "step": 14790
     },
     {
       "epoch": 0.7106842737094838,
-      "grad_norm": 0.5952445864677429,
+      "grad_norm": 0.5778575539588928,
       "learning_rate": 1.0554861703427884e-05,
-      "loss": 0.9441,
+      "loss": 0.9451,
       "num_input_tokens_seen": 121241600,
       "step": 14800
     },
     {
       "epoch": 0.7111644657863145,
-      "grad_norm": 1.0399448871612549,
+      "grad_norm": 1.0493206977844238,
       "learning_rate": 1.0522545834739908e-05,
-      "loss": 1.0587,
+      "loss": 1.0584,
       "num_input_tokens_seen": 121323520,
       "step": 14810
     },
     {
       "epoch": 0.7116446578631452,
-      "grad_norm": 0.519305944442749,
+      "grad_norm": 0.5271633267402649,
       "learning_rate": 1.0490266321161785e-05,
-      "loss": 0.9414,
+      "loss": 0.9405,
       "num_input_tokens_seen": 121405440,
       "step": 14820
     },
     {
       "epoch": 0.712124849939976,
-      "grad_norm": 0.517493724822998,
+      "grad_norm": 0.5183375477790833,
       "learning_rate": 1.0458023243752321e-05,
-      "loss": 0.9523,
+      "loss": 0.9526,
       "num_input_tokens_seen": 121487360,
       "step": 14830
     },
     {
       "epoch": 0.7126050420168067,
-      "grad_norm": 0.4983735978603363,
+      "grad_norm": 0.4991316497325897,
       "learning_rate": 1.0425816683478823e-05,
-      "loss": 1.145,
+      "loss": 1.1452,
       "num_input_tokens_seen": 121569280,
       "step": 14840
     },
     {
       "epoch": 0.7130852340936374,
-      "grad_norm": 1.5348645448684692,
+      "grad_norm": 1.4794950485229492,
       "learning_rate": 1.039364672121692e-05,
-      "loss": 0.9529,
+      "loss": 0.9537,
       "num_input_tokens_seen": 121651200,
       "step": 14850
     },
     {
       "epoch": 0.7135654261704681,
-      "grad_norm": 0.44925153255462646,
+      "grad_norm": 0.4501538574695587,
       "learning_rate": 1.0361513437750333e-05,
-      "loss": 1.0087,
+      "loss": 1.0094,
       "num_input_tokens_seen": 121733120,
       "step": 14860
     },
     {
       "epoch": 0.714045618247299,
-      "grad_norm": 0.28816720843315125,
+      "grad_norm": 0.31150001287460327,
       "learning_rate": 1.0329416913770651e-05,
-      "loss": 0.8125,
+      "loss": 0.8121,
       "num_input_tokens_seen": 121815040,
       "step": 14870
     },
     {
       "epoch": 0.7145258103241297,
-      "grad_norm": 0.5006898045539856,
+      "grad_norm": 0.5036375522613525,
       "learning_rate": 1.0297357229877183e-05,
-      "loss": 0.8591,
+      "loss": 0.8583,
       "num_input_tokens_seen": 121896960,
       "step": 14880
     },
     {
       "epoch": 0.7150060024009603,
-      "grad_norm": 0.4817348122596741,
+      "grad_norm": 0.4911898374557495,
       "learning_rate": 1.0265334466576723e-05,
-      "loss": 0.9232,
+      "loss": 0.9234,
       "num_input_tokens_seen": 121978880,
       "step": 14890
     },
     {
       "epoch": 0.7154861944777912,
-      "grad_norm": 0.511573851108551,
+      "grad_norm": 0.5137236714363098,
       "learning_rate": 1.0233348704283332e-05,
-      "loss": 0.7655,
+      "loss": 0.7661,
       "num_input_tokens_seen": 122060800,
       "step": 14900
     },
     {
       "epoch": 0.7159663865546219,
-      "grad_norm": 0.49658405780792236,
+      "grad_norm": 0.49499648809432983,
       "learning_rate": 1.0201400023318184e-05,
       "loss": 0.8836,
       "num_input_tokens_seen": 122142720,
@@ -11938,79 +11938,79 @@
     },
     {
       "epoch": 0.7164465786314526,
-      "grad_norm": 0.49688369035720825,
+      "grad_norm": 0.49660295248031616,
       "learning_rate": 1.0169488503909313e-05,
-      "loss": 0.9663,
+      "loss": 0.9656,
       "num_input_tokens_seen": 122224640,
       "step": 14920
     },
     {
       "epoch": 0.7169267707082834,
-      "grad_norm": 0.4753476679325104,
+      "grad_norm": 0.47622841596603394,
       "learning_rate": 1.0137614226191434e-05,
-      "loss": 1.1426,
+      "loss": 1.1411,
       "num_input_tokens_seen": 122306560,
       "step": 14930
     },
     {
       "epoch": 0.7174069627851141,
-      "grad_norm": 0.496444433927536,
+      "grad_norm": 0.5041408538818359,
       "learning_rate": 1.010577727020576e-05,
-      "loss": 1.0298,
+      "loss": 1.0296,
       "num_input_tokens_seen": 122388480,
       "step": 14940
     },
     {
       "epoch": 0.7178871548619448,
-      "grad_norm": 0.5547921657562256,
+      "grad_norm": 0.5578935146331787,
       "learning_rate": 1.0073977715899785e-05,
-      "loss": 1.0564,
+      "loss": 1.0557,
       "num_input_tokens_seen": 122470400,
       "step": 14950
     },
     {
       "epoch": 0.7183673469387755,
-      "grad_norm": 0.49256107211112976,
+      "grad_norm": 0.4937760829925537,
       "learning_rate": 1.0042215643127051e-05,
-      "loss": 0.8569,
+      "loss": 0.857,
       "num_input_tokens_seen": 122552320,
       "step": 14960
     },
     {
       "epoch": 0.7188475390156063,
-      "grad_norm": 0.48709073662757874,
+      "grad_norm": 0.48638951778411865,
       "learning_rate": 1.0010491131647013e-05,
-      "loss": 0.9653,
+      "loss": 0.9649,
       "num_input_tokens_seen": 122634240,
       "step": 14970
     },
     {
       "epoch": 0.719327731092437,
-      "grad_norm": 0.9637261629104614,
+      "grad_norm": 1.094740390777588,
       "learning_rate": 9.978804261124792e-06,
-      "loss": 1.0368,
+      "loss": 1.0374,
       "num_input_tokens_seen": 122716160,
       "step": 14980
     },
     {
       "epoch": 0.7198079231692677,
-      "grad_norm": 0.5114941000938416,
+      "grad_norm": 0.5187696814537048,
       "learning_rate": 9.947155111130969e-06,
-      "loss": 0.8155,
+      "loss": 0.8151,
       "num_input_tokens_seen": 122798080,
       "step": 14990
     },
     {
       "epoch": 0.7202881152460985,
-      "grad_norm": 0.4979882538318634,
+      "grad_norm": 0.5019155144691467,
       "learning_rate": 9.915543761141432e-06,
-      "loss": 0.9462,
+      "loss": 0.9461,
       "num_input_tokens_seen": 122880000,
       "step": 15000
     },
     {
       "epoch": 0.7207683073229292,
-      "grad_norm": 0.49206554889678955,
+      "grad_norm": 0.49210861325263977,
       "learning_rate": 9.883970290537134e-06,
       "loss": 0.7584,
       "num_input_tokens_seen": 122961920,
@@ -12018,79 +12018,79 @@
     },
     {
       "epoch": 0.7212484993997599,
-      "grad_norm": 0.5425018668174744,
+      "grad_norm": 0.541955828666687,
       "learning_rate": 9.852434778603888e-06,
-      "loss": 0.9354,
+      "loss": 0.9357,
       "num_input_tokens_seen": 123043840,
       "step": 15020
     },
     {
       "epoch": 0.7217286914765907,
-      "grad_norm": 0.311642587184906,
+      "grad_norm": 0.2484675943851471,
       "learning_rate": 9.820937304532221e-06,
-      "loss": 0.8208,
+      "loss": 0.8206,
       "num_input_tokens_seen": 123125760,
       "step": 15030
     },
     {
       "epoch": 0.7222088835534214,
-      "grad_norm": 0.5740408301353455,
+      "grad_norm": 0.5742604732513428,
       "learning_rate": 9.789477947417131e-06,
-      "loss": 1.0232,
+      "loss": 1.0253,
       "num_input_tokens_seen": 123207680,
       "step": 15040
     },
     {
       "epoch": 0.7226890756302521,
-      "grad_norm": 0.48182716965675354,
+      "grad_norm": 0.48535361886024475,
       "learning_rate": 9.758056786257874e-06,
-      "loss": 0.8119,
+      "loss": 0.8122,
       "num_input_tokens_seen": 123289600,
       "step": 15050
     },
     {
       "epoch": 0.7231692677070828,
-      "grad_norm": 0.6770291328430176,
+      "grad_norm": 0.7094045281410217,
       "learning_rate": 9.726673899957823e-06,
-      "loss": 0.9681,
+      "loss": 0.9684,
       "num_input_tokens_seen": 123371520,
       "step": 15060
     },
     {
       "epoch": 0.7236494597839136,
-      "grad_norm": 0.5068685412406921,
+      "grad_norm": 0.5078888535499573,
       "learning_rate": 9.695329367324226e-06,
-      "loss": 0.831,
+      "loss": 0.8309,
       "num_input_tokens_seen": 123453440,
       "step": 15070
     },
     {
       "epoch": 0.7241296518607443,
-      "grad_norm": 0.5525927543640137,
+      "grad_norm": 0.5580435395240784,
       "learning_rate": 9.664023267068007e-06,
-      "loss": 0.7507,
+      "loss": 0.7509,
       "num_input_tokens_seen": 123535360,
       "step": 15080
     },
     {
       "epoch": 0.724609843937575,
-      "grad_norm": 0.5521044731140137,
+      "grad_norm": 0.5548977255821228,
       "learning_rate": 9.632755677803595e-06,
-      "loss": 0.9068,
+      "loss": 0.9072,
       "num_input_tokens_seen": 123617280,
       "step": 15090
     },
     {
       "epoch": 0.7250900360144058,
-      "grad_norm": 0.628960907459259,
+      "grad_norm": 0.633141815662384,
       "learning_rate": 9.601526678048701e-06,
-      "loss": 0.9601,
+      "loss": 0.9596,
       "num_input_tokens_seen": 123699200,
       "step": 15100
     },
     {
       "epoch": 0.7255702280912365,
-      "grad_norm": 0.49019432067871094,
+      "grad_norm": 0.49129536747932434,
       "learning_rate": 9.570336346224145e-06,
       "loss": 0.8658,
       "num_input_tokens_seen": 123781120,
@@ -12098,127 +12098,127 @@
     },
     {
       "epoch": 0.7260504201680672,
-      "grad_norm": 0.49973979592323303,
+      "grad_norm": 0.49976715445518494,
       "learning_rate": 9.53918476065363e-06,
-      "loss": 0.9606,
+      "loss": 0.9609,
       "num_input_tokens_seen": 123863040,
       "step": 15120
     },
     {
       "epoch": 0.726530612244898,
-      "grad_norm": 0.5277695655822754,
+      "grad_norm": 0.5346314311027527,
       "learning_rate": 9.508071999563578e-06,
-      "loss": 0.8749,
+      "loss": 0.8759,
       "num_input_tokens_seen": 123944960,
       "step": 15130
     },
     {
       "epoch": 0.7270108043217287,
-      "grad_norm": 0.6320548057556152,
+      "grad_norm": 0.6391932368278503,
       "learning_rate": 9.476998141082896e-06,
-      "loss": 0.8534,
+      "loss": 0.8535,
       "num_input_tokens_seen": 124026880,
       "step": 15140
     },
     {
       "epoch": 0.7274909963985594,
-      "grad_norm": 0.4708552658557892,
+      "grad_norm": 0.4688946306705475,
       "learning_rate": 9.445963263242822e-06,
-      "loss": 0.8115,
+      "loss": 0.8117,
       "num_input_tokens_seen": 124108800,
       "step": 15150
     },
     {
       "epoch": 0.7279711884753901,
-      "grad_norm": 0.49595674872398376,
+      "grad_norm": 0.4958951771259308,
       "learning_rate": 9.414967443976705e-06,
-      "loss": 0.8173,
+      "loss": 0.8172,
       "num_input_tokens_seen": 124190720,
       "step": 15160
     },
     {
       "epoch": 0.7284513805522209,
-      "grad_norm": 0.49820470809936523,
+      "grad_norm": 0.4968552887439728,
       "learning_rate": 9.384010761119787e-06,
-      "loss": 0.7552,
+      "loss": 0.7557,
       "num_input_tokens_seen": 124272640,
       "step": 15170
     },
     {
       "epoch": 0.7289315726290516,
-      "grad_norm": 0.47693219780921936,
+      "grad_norm": 0.4845529794692993,
       "learning_rate": 9.353093292409063e-06,
-      "loss": 1.0986,
+      "loss": 1.0988,
       "num_input_tokens_seen": 124354560,
       "step": 15180
     },
     {
       "epoch": 0.7294117647058823,
-      "grad_norm": 0.5159751176834106,
+      "grad_norm": 0.5118656158447266,
       "learning_rate": 9.322215115483049e-06,
-      "loss": 0.9024,
+      "loss": 0.9023,
       "num_input_tokens_seen": 124436480,
       "step": 15190
     },
     {
       "epoch": 0.7298919567827131,
-      "grad_norm": 0.4883824586868286,
+      "grad_norm": 0.4848543703556061,
       "learning_rate": 9.291376307881577e-06,
-      "loss": 0.7029,
+      "loss": 0.7027,
       "num_input_tokens_seen": 124518400,
       "step": 15200
     },
     {
       "epoch": 0.7303721488595438,
-      "grad_norm": 0.4944133162498474,
+      "grad_norm": 0.49403074383735657,
       "learning_rate": 9.260576947045624e-06,
-      "loss": 0.7564,
+      "loss": 0.7578,
       "num_input_tokens_seen": 124600320,
       "step": 15210
     },
     {
       "epoch": 0.7308523409363745,
-      "grad_norm": 0.5287705063819885,
+      "grad_norm": 0.5141822099685669,
       "learning_rate": 9.229817110317126e-06,
-      "loss": 0.8904,
+      "loss": 0.8909,
       "num_input_tokens_seen": 124682240,
       "step": 15220
     },
     {
       "epoch": 0.7313325330132053,
-      "grad_norm": 0.5056361556053162,
+      "grad_norm": 0.5016796588897705,
       "learning_rate": 9.19909687493874e-06,
-      "loss": 0.8753,
+      "loss": 0.8752,
       "num_input_tokens_seen": 124764160,
       "step": 15230
     },
     {
       "epoch": 0.731812725090036,
-      "grad_norm": 0.49271291494369507,
+      "grad_norm": 0.48871156573295593,
       "learning_rate": 9.168416318053701e-06,
-      "loss": 0.7468,
+      "loss": 0.7467,
       "num_input_tokens_seen": 124846080,
       "step": 15240
     },
     {
       "epoch": 0.7322929171668667,
-      "grad_norm": 0.4896814823150635,
+      "grad_norm": 0.4882807731628418,
       "learning_rate": 9.137775516705604e-06,
-      "loss": 0.7828,
+      "loss": 0.7833,
       "num_input_tokens_seen": 124928000,
       "step": 15250
     },
     {
       "epoch": 0.7327731092436974,
-      "grad_norm": 0.5445641279220581,
+      "grad_norm": 2.4790802001953125,
       "learning_rate": 9.107174547838188e-06,
-      "loss": 0.978,
+      "loss": 0.9774,
       "num_input_tokens_seen": 125009920,
       "step": 15260
     },
     {
       "epoch": 0.7332533013205282,
-      "grad_norm": 0.4939377009868622,
+      "grad_norm": 0.5495137572288513,
       "learning_rate": 9.076613488295193e-06,
       "loss": 0.9886,
       "num_input_tokens_seen": 125091840,
@@ -12226,159 +12226,159 @@
     },
     {
       "epoch": 0.7337334933973589,
-      "grad_norm": 0.5052213668823242,
+      "grad_norm": 0.5097706913948059,
       "learning_rate": 9.04609241482014e-06,
-      "loss": 0.9165,
+      "loss": 0.9161,
       "num_input_tokens_seen": 125173760,
       "step": 15280
     },
     {
       "epoch": 0.7342136854741896,
-      "grad_norm": 0.5136955976486206,
+      "grad_norm": 0.5205005407333374,
       "learning_rate": 9.015611404056121e-06,
-      "loss": 0.9454,
+      "loss": 0.9453,
       "num_input_tokens_seen": 125255680,
       "step": 15290
     },
     {
       "epoch": 0.7346938775510204,
-      "grad_norm": 0.49934035539627075,
+      "grad_norm": 0.5136666893959045,
       "learning_rate": 8.985170532545622e-06,
-      "loss": 0.9743,
+      "loss": 0.9736,
       "num_input_tokens_seen": 125337600,
       "step": 15300
     },
     {
       "epoch": 0.7351740696278511,
-      "grad_norm": 0.47998034954071045,
+      "grad_norm": 0.48384732007980347,
       "learning_rate": 8.954769876730368e-06,
-      "loss": 1.0288,
+      "loss": 1.0289,
       "num_input_tokens_seen": 125419520,
       "step": 15310
     },
     {
       "epoch": 0.7356542617046818,
-      "grad_norm": 0.512691080570221,
+      "grad_norm": 0.5137535929679871,
       "learning_rate": 8.924409512951045e-06,
-      "loss": 0.8448,
+      "loss": 0.8446,
       "num_input_tokens_seen": 125501440,
       "step": 15320
     },
     {
       "epoch": 0.7361344537815127,
-      "grad_norm": 0.5149726271629333,
+      "grad_norm": 0.5144986510276794,
       "learning_rate": 8.894089517447206e-06,
-      "loss": 1.0112,
+      "loss": 1.0116,
       "num_input_tokens_seen": 125583360,
       "step": 15330
     },
     {
       "epoch": 0.7366146458583434,
-      "grad_norm": 0.47288787364959717,
+      "grad_norm": 0.474317729473114,
       "learning_rate": 8.863809966357017e-06,
-      "loss": 0.9288,
+      "loss": 0.9291,
       "num_input_tokens_seen": 125665280,
       "step": 15340
     },
     {
       "epoch": 0.737094837935174,
-      "grad_norm": 0.4858817756175995,
+      "grad_norm": 0.4877590835094452,
       "learning_rate": 8.833570935717064e-06,
-      "loss": 0.8124,
+      "loss": 0.8133,
       "num_input_tokens_seen": 125747200,
       "step": 15350
     },
     {
       "epoch": 0.7375750300120048,
-      "grad_norm": 0.496822714805603,
+      "grad_norm": 0.49466702342033386,
       "learning_rate": 8.803372501462203e-06,
-      "loss": 0.8461,
+      "loss": 0.845,
       "num_input_tokens_seen": 125829120,
       "step": 15360
     },
     {
       "epoch": 0.7380552220888356,
-      "grad_norm": 0.510150134563446,
+      "grad_norm": 0.5105023384094238,
       "learning_rate": 8.773214739425346e-06,
-      "loss": 0.7163,
+      "loss": 0.7169,
       "num_input_tokens_seen": 125911040,
       "step": 15370
     },
     {
       "epoch": 0.7385354141656663,
-      "grad_norm": 0.4825844466686249,
+      "grad_norm": 0.47850966453552246,
       "learning_rate": 8.743097725337255e-06,
-      "loss": 0.5987,
+      "loss": 0.599,
       "num_input_tokens_seen": 125992960,
       "step": 15380
     },
     {
       "epoch": 0.739015606242497,
-      "grad_norm": 0.47801604866981506,
+      "grad_norm": 0.47564321756362915,
       "learning_rate": 8.713021534826366e-06,
-      "loss": 0.8509,
+      "loss": 0.8508,
       "num_input_tokens_seen": 126074880,
       "step": 15390
     },
     {
       "epoch": 0.7394957983193278,
-      "grad_norm": 0.5060662031173706,
+      "grad_norm": 0.5005056262016296,
       "learning_rate": 8.68298624341862e-06,
-      "loss": 1.0267,
+      "loss": 1.0262,
       "num_input_tokens_seen": 126156800,
       "step": 15400
     },
     {
       "epoch": 0.7399759903961585,
-      "grad_norm": 0.46689069271087646,
+      "grad_norm": 0.46413904428482056,
       "learning_rate": 8.652991926537254e-06,
-      "loss": 0.913,
+      "loss": 0.9126,
       "num_input_tokens_seen": 126238720,
       "step": 15410
     },
     {
       "epoch": 0.7404561824729892,
-      "grad_norm": 0.4775540828704834,
+      "grad_norm": 0.47946876287460327,
       "learning_rate": 8.623038659502583e-06,
-      "loss": 0.9146,
+      "loss": 0.9147,
       "num_input_tokens_seen": 126320640,
       "step": 15420
     },
     {
       "epoch": 0.74093637454982,
-      "grad_norm": 0.5078407526016235,
+      "grad_norm": 0.5045376420021057,
       "learning_rate": 8.59312651753187e-06,
-      "loss": 0.7079,
+      "loss": 0.7083,
       "num_input_tokens_seen": 126402560,
       "step": 15430
     },
     {
       "epoch": 0.7414165666266507,
-      "grad_norm": 0.6031984090805054,
+      "grad_norm": 0.5812355875968933,
       "learning_rate": 8.5632555757391e-06,
-      "loss": 0.9282,
+      "loss": 0.9287,
       "num_input_tokens_seen": 126484480,
       "step": 15440
     },
     {
       "epoch": 0.7418967587034814,
-      "grad_norm": 0.48151716589927673,
+      "grad_norm": 0.4817546606063843,
       "learning_rate": 8.53342590913478e-06,
-      "loss": 0.8451,
+      "loss": 0.845,
       "num_input_tokens_seen": 126566400,
       "step": 15450
     },
     {
       "epoch": 0.7423769507803121,
-      "grad_norm": 0.4922950863838196,
+      "grad_norm": 0.4910758137702942,
       "learning_rate": 8.503637592625796e-06,
-      "loss": 0.8644,
+      "loss": 0.8641,
       "num_input_tokens_seen": 126648320,
       "step": 15460
     },
     {
       "epoch": 0.7428571428571429,
-      "grad_norm": 0.48294711112976074,
+      "grad_norm": 0.47712764143943787,
       "learning_rate": 8.473890701015177e-06,
       "loss": 0.9529,
       "num_input_tokens_seen": 126730240,
@@ -12386,79 +12386,79 @@
     },
     {
       "epoch": 0.7433373349339736,
-      "grad_norm": 1.1124995946884155,
+      "grad_norm": 1.1171008348464966,
       "learning_rate": 8.444185309001926e-06,
-      "loss": 1.0753,
+      "loss": 1.0765,
       "num_input_tokens_seen": 126812160,
       "step": 15480
     },
     {
       "epoch": 0.7438175270108043,
-      "grad_norm": 0.6511945128440857,
+      "grad_norm": 0.635892927646637,
       "learning_rate": 8.41452149118085e-06,
-      "loss": 0.8009,
+      "loss": 0.7988,
       "num_input_tokens_seen": 126894080,
       "step": 15490
     },
     {
       "epoch": 0.7442977190876351,
-      "grad_norm": 0.47648462653160095,
+      "grad_norm": 0.46706971526145935,
       "learning_rate": 8.384899322042356e-06,
-      "loss": 0.8622,
+      "loss": 0.8623,
       "num_input_tokens_seen": 126976000,
       "step": 15500
     },
     {
       "epoch": 0.7447779111644658,
-      "grad_norm": 0.48163020610809326,
+      "grad_norm": 0.4782085418701172,
       "learning_rate": 8.355318875972243e-06,
-      "loss": 0.8215,
+      "loss": 0.8205,
       "num_input_tokens_seen": 127057920,
       "step": 15510
     },
     {
       "epoch": 0.7452581032412965,
-      "grad_norm": 0.4855590760707855,
+      "grad_norm": 0.4798336625099182,
       "learning_rate": 8.325780227251562e-06,
-      "loss": 0.7382,
+      "loss": 0.7383,
       "num_input_tokens_seen": 127139840,
       "step": 15520
     },
     {
       "epoch": 0.7457382953181273,
-      "grad_norm": 0.4971849024295807,
+      "grad_norm": 0.49591222405433655,
       "learning_rate": 8.2962834500564e-06,
-      "loss": 0.9809,
+      "loss": 0.9802,
       "num_input_tokens_seen": 127221760,
       "step": 15530
     },
     {
       "epoch": 0.746218487394958,
-      "grad_norm": 0.4910465478897095,
+      "grad_norm": 0.4897187054157257,
       "learning_rate": 8.266828618457678e-06,
-      "loss": 0.9193,
+      "loss": 0.9196,
       "num_input_tokens_seen": 127303680,
       "step": 15540
     },
     {
       "epoch": 0.7466986794717887,
-      "grad_norm": 0.5851929187774658,
+      "grad_norm": 0.5829177498817444,
       "learning_rate": 8.237415806421015e-06,
-      "loss": 1.1018,
+      "loss": 1.1014,
       "num_input_tokens_seen": 127385600,
       "step": 15550
     },
     {
       "epoch": 0.7471788715486194,
-      "grad_norm": 0.5044286847114563,
+      "grad_norm": 0.5015119910240173,
       "learning_rate": 8.20804508780648e-06,
-      "loss": 0.7888,
+      "loss": 0.7883,
       "num_input_tokens_seen": 127467520,
       "step": 15560
     },
     {
       "epoch": 0.7476590636254502,
-      "grad_norm": 0.48159295320510864,
+      "grad_norm": 0.48306459188461304,
       "learning_rate": 8.178716536368475e-06,
       "loss": 1.0093,
       "num_input_tokens_seen": 127549440,
@@ -12466,39 +12466,39 @@
     },
     {
       "epoch": 0.7481392557022809,
-      "grad_norm": 0.5040597319602966,
+      "grad_norm": 0.5044358372688293,
       "learning_rate": 8.149430225755476e-06,
-      "loss": 0.9785,
+      "loss": 0.978,
       "num_input_tokens_seen": 127631360,
       "step": 15580
     },
     {
       "epoch": 0.7486194477791116,
-      "grad_norm": 0.4986848831176758,
+      "grad_norm": 0.5006290674209595,
       "learning_rate": 8.120186229509922e-06,
-      "loss": 0.8937,
+      "loss": 0.8938,
       "num_input_tokens_seen": 127713280,
       "step": 15590
     },
     {
       "epoch": 0.7490996398559424,
-      "grad_norm": 0.5762357711791992,
+      "grad_norm": 0.5820693969726562,
       "learning_rate": 8.090984621067963e-06,
-      "loss": 1.0423,
+      "loss": 1.0422,
       "num_input_tokens_seen": 127795200,
       "step": 15600
     },
     {
       "epoch": 0.7495798319327731,
-      "grad_norm": 0.4688274562358856,
+      "grad_norm": 0.471348375082016,
       "learning_rate": 8.061825473759324e-06,
-      "loss": 0.8387,
+      "loss": 0.8388,
       "num_input_tokens_seen": 127877120,
       "step": 15610
     },
     {
       "epoch": 0.7500600240096038,
-      "grad_norm": 0.49585649371147156,
+      "grad_norm": 0.49735504388809204,
       "learning_rate": 8.032708860807111e-06,
       "loss": 0.8558,
       "num_input_tokens_seen": 127959040,
@@ -12506,175 +12506,175 @@
     },
     {
       "epoch": 0.7505402160864346,
-      "grad_norm": 0.501162052154541,
+      "grad_norm": 0.49957790970802307,
       "learning_rate": 8.003634855327594e-06,
-      "loss": 0.7608,
+      "loss": 0.7606,
       "num_input_tokens_seen": 128040960,
       "step": 15630
     },
     {
       "epoch": 0.7510204081632653,
-      "grad_norm": 0.8193572759628296,
+      "grad_norm": 0.8058281540870667,
       "learning_rate": 7.974603530330069e-06,
-      "loss": 0.7416,
+      "loss": 0.7422,
       "num_input_tokens_seen": 128122880,
       "step": 15640
     },
     {
       "epoch": 0.751500600240096,
-      "grad_norm": 0.5035424828529358,
+      "grad_norm": 0.49757471680641174,
       "learning_rate": 7.945614958716658e-06,
-      "loss": 0.7773,
+      "loss": 0.7774,
       "num_input_tokens_seen": 128204800,
       "step": 15650
     },
     {
       "epoch": 0.7519807923169267,
-      "grad_norm": 0.48581522703170776,
+      "grad_norm": 0.48207026720046997,
       "learning_rate": 7.916669213282107e-06,
-      "loss": 0.8304,
+      "loss": 0.831,
       "num_input_tokens_seen": 128286720,
       "step": 15660
     },
     {
       "epoch": 0.7524609843937575,
-      "grad_norm": 0.5436628460884094,
+      "grad_norm": 0.5513833165168762,
       "learning_rate": 7.88776636671362e-06,
-      "loss": 1.1004,
+      "loss": 1.1006,
       "num_input_tokens_seen": 128368640,
       "step": 15670
     },
     {
       "epoch": 0.7529411764705882,
-      "grad_norm": 0.4647444784641266,
+      "grad_norm": 0.46379712224006653,
       "learning_rate": 7.858906491590697e-06,
-      "loss": 0.9029,
+      "loss": 0.903,
       "num_input_tokens_seen": 128450560,
       "step": 15680
     },
     {
       "epoch": 0.7534213685474189,
-      "grad_norm": 0.4686352014541626,
+      "grad_norm": 0.4677363336086273,
       "learning_rate": 7.830089660384895e-06,
-      "loss": 0.8456,
+      "loss": 0.8458,
       "num_input_tokens_seen": 128532480,
       "step": 15690
     },
     {
       "epoch": 0.7539015606242497,
-      "grad_norm": 0.4896450340747833,
+      "grad_norm": 0.49112623929977417,
       "learning_rate": 7.801315945459714e-06,
-      "loss": 1.3139,
+      "loss": 1.3148,
       "num_input_tokens_seen": 128614400,
       "step": 15700
     },
     {
       "epoch": 0.7543817527010804,
-      "grad_norm": 0.5118237137794495,
+      "grad_norm": 0.5155120491981506,
       "learning_rate": 7.772585419070374e-06,
-      "loss": 0.931,
+      "loss": 0.9315,
       "num_input_tokens_seen": 128696320,
       "step": 15710
     },
     {
       "epoch": 0.7548619447779111,
-      "grad_norm": 0.4990551471710205,
+      "grad_norm": 0.5028974413871765,
       "learning_rate": 7.743898153363625e-06,
-      "loss": 0.7754,
+      "loss": 0.7756,
       "num_input_tokens_seen": 128778240,
       "step": 15720
     },
     {
       "epoch": 0.7553421368547419,
-      "grad_norm": 0.47710755467414856,
+      "grad_norm": 0.3549368679523468,
       "learning_rate": 7.715254220377596e-06,
-      "loss": 0.8358,
+      "loss": 0.8351,
       "num_input_tokens_seen": 128860160,
       "step": 15730
     },
     {
       "epoch": 0.7558223289315726,
-      "grad_norm": 1.072224497795105,
+      "grad_norm": 1.1180391311645508,
       "learning_rate": 7.686653692041615e-06,
-      "loss": 0.7976,
+      "loss": 0.7978,
       "num_input_tokens_seen": 128942080,
       "step": 15740
     },
     {
       "epoch": 0.7563025210084033,
-      "grad_norm": 0.48185160756111145,
+      "grad_norm": 0.48073238134384155,
       "learning_rate": 7.658096640175988e-06,
-      "loss": 0.834,
+      "loss": 0.8335,
       "num_input_tokens_seen": 129024000,
       "step": 15750
     },
     {
       "epoch": 0.7567827130852341,
-      "grad_norm": 0.5147521495819092,
+      "grad_norm": 0.5103764533996582,
       "learning_rate": 7.629583136491844e-06,
-      "loss": 0.78,
+      "loss": 0.7803,
       "num_input_tokens_seen": 129105920,
       "step": 15760
     },
     {
       "epoch": 0.7572629051620648,
-      "grad_norm": 0.5172405242919922,
+      "grad_norm": 0.5050859451293945,
       "learning_rate": 7.601113252590991e-06,
-      "loss": 0.855,
+      "loss": 0.8551,
       "num_input_tokens_seen": 129187840,
       "step": 15770
     },
     {
       "epoch": 0.7577430972388955,
-      "grad_norm": 0.5000312328338623,
+      "grad_norm": 0.4922633469104767,
       "learning_rate": 7.572687059965661e-06,
-      "loss": 1.0542,
+      "loss": 1.0544,
       "num_input_tokens_seen": 129269760,
       "step": 15780
     },
     {
       "epoch": 0.7582232893157262,
-      "grad_norm": 0.5540167689323425,
+      "grad_norm": 0.5366300344467163,
       "learning_rate": 7.544304629998389e-06,
-      "loss": 0.9999,
+      "loss": 0.9997,
       "num_input_tokens_seen": 129351680,
       "step": 15790
     },
     {
       "epoch": 0.758703481392557,
-      "grad_norm": 0.4820740222930908,
+      "grad_norm": 0.4798520505428314,
       "learning_rate": 7.51596603396183e-06,
-      "loss": 0.8808,
+      "loss": 0.8806,
       "num_input_tokens_seen": 129433600,
       "step": 15800
     },
     {
       "epoch": 0.7591836734693878,
-      "grad_norm": 0.5026702880859375,
+      "grad_norm": 0.49928078055381775,
       "learning_rate": 7.4876713430185265e-06,
-      "loss": 0.9648,
+      "loss": 0.9649,
       "num_input_tokens_seen": 129515520,
       "step": 15810
     },
     {
       "epoch": 0.7596638655462185,
-      "grad_norm": 0.4794592261314392,
+      "grad_norm": 0.48000165820121765,
       "learning_rate": 7.4594206282208e-06,
-      "loss": 0.9263,
+      "loss": 0.926,
       "num_input_tokens_seen": 129597440,
       "step": 15820
     },
     {
       "epoch": 0.7601440576230493,
-      "grad_norm": 0.4958498477935791,
+      "grad_norm": 0.49403661489486694,
       "learning_rate": 7.431213960510544e-06,
-      "loss": 0.8364,
+      "loss": 0.8361,
       "num_input_tokens_seen": 129679360,
       "step": 15830
     },
     {
       "epoch": 0.76062424969988,
-      "grad_norm": 0.4976142644882202,
+      "grad_norm": 0.49990224838256836,
       "learning_rate": 7.40305141071902e-06,
       "loss": 1.068,
       "num_input_tokens_seen": 129761280,
@@ -12682,15 +12682,15 @@
     },
     {
       "epoch": 0.7611044417767107,
-      "grad_norm": 0.4213819205760956,
+      "grad_norm": 0.4160490930080414,
       "learning_rate": 7.374933049566704e-06,
-      "loss": 0.7245,
+      "loss": 0.724,
       "num_input_tokens_seen": 129843200,
       "step": 15850
     },
     {
       "epoch": 0.7615846338535415,
-      "grad_norm": 0.49244338274002075,
+      "grad_norm": 0.49283355474472046,
       "learning_rate": 7.346858947663138e-06,
       "loss": 0.8833,
       "num_input_tokens_seen": 129925120,
@@ -12698,87 +12698,87 @@
     },
     {
       "epoch": 0.7620648259303722,
-      "grad_norm": 0.385324090719223,
+      "grad_norm": 0.39053088426589966,
       "learning_rate": 7.318829175506684e-06,
-      "loss": 1.0079,
+      "loss": 1.0074,
       "num_input_tokens_seen": 130007040,
       "step": 15870
     },
     {
       "epoch": 0.7625450180072029,
-      "grad_norm": 0.6321762800216675,
+      "grad_norm": 0.6311637163162231,
       "learning_rate": 7.290843803484409e-06,
-      "loss": 0.9743,
+      "loss": 0.9735,
       "num_input_tokens_seen": 130088960,
       "step": 15880
     },
     {
       "epoch": 0.7630252100840336,
-      "grad_norm": 0.5179278254508972,
+      "grad_norm": 0.5157874226570129,
       "learning_rate": 7.262902901871885e-06,
-      "loss": 0.923,
+      "loss": 0.9231,
       "num_input_tokens_seen": 130170880,
       "step": 15890
     },
     {
       "epoch": 0.7635054021608644,
-      "grad_norm": 0.48879900574684143,
+      "grad_norm": 0.4867972135543823,
       "learning_rate": 7.235006540832995e-06,
-      "loss": 0.947,
+      "loss": 0.9472,
       "num_input_tokens_seen": 130252800,
       "step": 15900
     },
     {
       "epoch": 0.7639855942376951,
-      "grad_norm": 0.4765695631504059,
+      "grad_norm": 0.4728543758392334,
       "learning_rate": 7.207154790419784e-06,
-      "loss": 0.8526,
+      "loss": 0.8525,
       "num_input_tokens_seen": 130334720,
       "step": 15910
     },
     {
       "epoch": 0.7644657863145258,
-      "grad_norm": 0.5195481181144714,
+      "grad_norm": 0.5194888114929199,
       "learning_rate": 7.179347720572288e-06,
-      "loss": 1.0454,
+      "loss": 1.0461,
       "num_input_tokens_seen": 130416640,
       "step": 15920
     },
     {
       "epoch": 0.7649459783913566,
-      "grad_norm": 0.48631834983825684,
+      "grad_norm": 0.48327717185020447,
       "learning_rate": 7.151585401118316e-06,
-      "loss": 0.7704,
+      "loss": 0.77,
       "num_input_tokens_seen": 130498560,
       "step": 15930
     },
     {
       "epoch": 0.7654261704681873,
-      "grad_norm": 0.4827914237976074,
+      "grad_norm": 0.4789976179599762,
       "learning_rate": 7.12386790177331e-06,
-      "loss": 0.8247,
+      "loss": 0.8252,
       "num_input_tokens_seen": 130580480,
       "step": 15940
     },
     {
       "epoch": 0.765906362545018,
-      "grad_norm": 0.21842628717422485,
+      "grad_norm": 0.2211165875196457,
       "learning_rate": 7.096195292140173e-06,
-      "loss": 0.8725,
+      "loss": 0.8723,
       "num_input_tokens_seen": 130662400,
       "step": 15950
     },
     {
       "epoch": 0.7663865546218488,
-      "grad_norm": 0.46142032742500305,
+      "grad_norm": 0.45980551838874817,
       "learning_rate": 7.06856764170907e-06,
-      "loss": 0.7432,
+      "loss": 0.7436,
       "num_input_tokens_seen": 130744320,
       "step": 15960
     },
     {
       "epoch": 0.7668667466986795,
-      "grad_norm": 0.5066096186637878,
+      "grad_norm": 0.5044180154800415,
       "learning_rate": 7.040985019857274e-06,
       "loss": 0.959,
       "num_input_tokens_seen": 130826240,
@@ -12786,23 +12786,23 @@
     },
     {
       "epoch": 0.7673469387755102,
-      "grad_norm": 0.4842053949832916,
+      "grad_norm": 0.48258551955223083,
       "learning_rate": 7.013447495848996e-06,
-      "loss": 1.1146,
+      "loss": 1.1143,
       "num_input_tokens_seen": 130908160,
       "step": 15980
     },
     {
       "epoch": 0.7678271308523409,
-      "grad_norm": 0.5671928524971008,
+      "grad_norm": 0.5670498609542847,
       "learning_rate": 6.985955138835162e-06,
-      "loss": 0.9096,
+      "loss": 0.9102,
       "num_input_tokens_seen": 130990080,
       "step": 15990
     },
     {
       "epoch": 0.7683073229291717,
-      "grad_norm": 0.4887046813964844,
+      "grad_norm": 0.48566895723342896,
       "learning_rate": 6.958508017853319e-06,
       "loss": 0.9424,
       "num_input_tokens_seen": 131072000,
@@ -12810,7 +12810,7 @@
     },
     {
       "epoch": 0.7687875150060024,
-      "grad_norm": 0.5106115341186523,
+      "grad_norm": 0.5094760656356812,
       "learning_rate": 6.931106201827397e-06,
       "loss": 0.905,
       "num_input_tokens_seen": 131153920,
@@ -12818,7 +12818,7 @@
     },
     {
       "epoch": 0.7692677070828331,
-      "grad_norm": 0.4807291626930237,
+      "grad_norm": 0.48024067282676697,
       "learning_rate": 6.903749759567557e-06,
       "loss": 0.8972,
       "num_input_tokens_seen": 131235840,
@@ -12826,15 +12826,15 @@
     },
     {
       "epoch": 0.7697478991596639,
-      "grad_norm": 0.4951104521751404,
+      "grad_norm": 0.4950743019580841,
       "learning_rate": 6.876438759770037e-06,
-      "loss": 0.8672,
+      "loss": 0.8669,
       "num_input_tokens_seen": 131317760,
       "step": 16030
     },
     {
       "epoch": 0.7702280912364946,
-      "grad_norm": 0.5364571213722229,
+      "grad_norm": 0.5339250564575195,
       "learning_rate": 6.8491732710169344e-06,
       "loss": 1.0239,
       "num_input_tokens_seen": 131399680,
@@ -12842,47 +12842,47 @@
     },
     {
       "epoch": 0.7707082833133253,
-      "grad_norm": 0.4739181399345398,
+      "grad_norm": 0.4749922454357147,
       "learning_rate": 6.821953361776093e-06,
-      "loss": 0.7971,
+      "loss": 0.7977,
       "num_input_tokens_seen": 131481600,
       "step": 16050
     },
     {
       "epoch": 0.7711884753901561,
-      "grad_norm": 1.0601301193237305,
+      "grad_norm": 1.0639270544052124,
       "learning_rate": 6.7947791004008665e-06,
-      "loss": 0.8697,
+      "loss": 0.8704,
       "num_input_tokens_seen": 131563520,
       "step": 16060
     },
     {
       "epoch": 0.7716686674669868,
-      "grad_norm": 0.3752616047859192,
+      "grad_norm": 0.3700399100780487,
       "learning_rate": 6.767650555130009e-06,
-      "loss": 0.8022,
+      "loss": 0.8018,
       "num_input_tokens_seen": 131645440,
       "step": 16070
     },
     {
       "epoch": 0.7721488595438175,
-      "grad_norm": 0.5150649547576904,
+      "grad_norm": 0.5122857689857483,
       "learning_rate": 6.740567794087463e-06,
-      "loss": 0.9051,
+      "loss": 0.9043,
       "num_input_tokens_seen": 131727360,
       "step": 16080
     },
     {
       "epoch": 0.7726290516206482,
-      "grad_norm": 0.5848947763442993,
+      "grad_norm": 0.5689642429351807,
       "learning_rate": 6.713530885282188e-06,
-      "loss": 1.1109,
+      "loss": 1.1105,
       "num_input_tokens_seen": 131809280,
       "step": 16090
     },
     {
       "epoch": 0.773109243697479,
-      "grad_norm": 0.5849442481994629,
+      "grad_norm": 0.5890742540359497,
       "learning_rate": 6.686539896608016e-06,
       "loss": 0.781,
       "num_input_tokens_seen": 131891200,
@@ -12890,63 +12890,63 @@
     },
     {
       "epoch": 0.7735894357743097,
-      "grad_norm": 0.47524645924568176,
+      "grad_norm": 0.4772266149520874,
       "learning_rate": 6.659594895843477e-06,
-      "loss": 0.7884,
+      "loss": 0.7886,
       "num_input_tokens_seen": 131973120,
       "step": 16110
     },
     {
       "epoch": 0.7740696278511404,
-      "grad_norm": 0.46588289737701416,
+      "grad_norm": 0.4759899377822876,
       "learning_rate": 6.632695950651594e-06,
-      "loss": 0.8852,
+      "loss": 0.8849,
       "num_input_tokens_seen": 132055040,
       "step": 16120
     },
     {
       "epoch": 0.7745498199279712,
-      "grad_norm": 0.6264036893844604,
+      "grad_norm": 0.6416568160057068,
       "learning_rate": 6.605843128579739e-06,
-      "loss": 0.9598,
+      "loss": 0.9594,
       "num_input_tokens_seen": 132136960,
       "step": 16130
     },
     {
       "epoch": 0.7750300120048019,
-      "grad_norm": 0.47402629256248474,
+      "grad_norm": 0.47457703948020935,
       "learning_rate": 6.579036497059482e-06,
-      "loss": 0.8386,
+      "loss": 0.8384,
       "num_input_tokens_seen": 132218880,
       "step": 16140
     },
     {
       "epoch": 0.7755102040816326,
-      "grad_norm": 0.4720049500465393,
+      "grad_norm": 0.47239863872528076,
       "learning_rate": 6.552276123406384e-06,
-      "loss": 0.8359,
+      "loss": 0.8364,
       "num_input_tokens_seen": 132300800,
       "step": 16150
     },
     {
       "epoch": 0.7759903961584634,
-      "grad_norm": 0.4943426847457886,
+      "grad_norm": 0.4923716187477112,
       "learning_rate": 6.525562074819852e-06,
-      "loss": 1.043,
+      "loss": 1.0425,
       "num_input_tokens_seen": 132382720,
       "step": 16160
     },
     {
       "epoch": 0.7764705882352941,
-      "grad_norm": 0.5186018347740173,
+      "grad_norm": 0.5167580246925354,
       "learning_rate": 6.4988944183829695e-06,
-      "loss": 0.9455,
+      "loss": 0.9449,
       "num_input_tokens_seen": 132464640,
       "step": 16170
     },
     {
       "epoch": 0.7769507803121248,
-      "grad_norm": 0.5227720141410828,
+      "grad_norm": 0.5179154276847839,
       "learning_rate": 6.472273221062305e-06,
       "loss": 1.1211,
       "num_input_tokens_seen": 132546560,
@@ -12954,423 +12954,423 @@
     },
     {
       "epoch": 0.7774309723889555,
-      "grad_norm": 0.491787850856781,
+      "grad_norm": 0.49237915873527527,
       "learning_rate": 6.445698549707776e-06,
-      "loss": 1.0617,
+      "loss": 1.0615,
       "num_input_tokens_seen": 132628480,
       "step": 16190
     },
     {
       "epoch": 0.7779111644657863,
-      "grad_norm": 0.5019481778144836,
+      "grad_norm": 0.5016152262687683,
       "learning_rate": 6.419170471052472e-06,
-      "loss": 0.7963,
+      "loss": 0.7958,
       "num_input_tokens_seen": 132710400,
       "step": 16200
     },
     {
       "epoch": 0.778391356542617,
-      "grad_norm": 0.5009409189224243,
+      "grad_norm": 0.5024312734603882,
       "learning_rate": 6.392689051712458e-06,
-      "loss": 0.9802,
+      "loss": 0.9789,
       "num_input_tokens_seen": 132792320,
       "step": 16210
     },
     {
       "epoch": 0.7788715486194477,
-      "grad_norm": 0.5145575404167175,
+      "grad_norm": 0.5140400528907776,
       "learning_rate": 6.3662543581866405e-06,
-      "loss": 1.0569,
+      "loss": 1.0557,
       "num_input_tokens_seen": 132874240,
       "step": 16220
     },
     {
       "epoch": 0.7793517406962785,
-      "grad_norm": 0.4958726167678833,
+      "grad_norm": 0.49644625186920166,
       "learning_rate": 6.339866456856608e-06,
-      "loss": 0.7891,
+      "loss": 0.7892,
       "num_input_tokens_seen": 132956160,
       "step": 16230
     },
     {
       "epoch": 0.7798319327731092,
-      "grad_norm": 0.49058905243873596,
+      "grad_norm": 0.4914844036102295,
       "learning_rate": 6.313525413986415e-06,
-      "loss": 0.9846,
+      "loss": 0.9853,
       "num_input_tokens_seen": 133038080,
       "step": 16240
     },
     {
       "epoch": 0.78031212484994,
-      "grad_norm": 0.48392897844314575,
+      "grad_norm": 0.48480865359306335,
       "learning_rate": 6.28723129572247e-06,
-      "loss": 0.8808,
+      "loss": 0.8806,
       "num_input_tokens_seen": 133120000,
       "step": 16250
     },
     {
       "epoch": 0.7807923169267708,
-      "grad_norm": 0.5041011571884155,
+      "grad_norm": 0.504016637802124,
       "learning_rate": 6.260984168093353e-06,
-      "loss": 0.8577,
+      "loss": 0.8576,
       "num_input_tokens_seen": 133201920,
       "step": 16260
     },
     {
       "epoch": 0.7812725090036015,
-      "grad_norm": 0.49976569414138794,
+      "grad_norm": 0.5014110803604126,
       "learning_rate": 6.234784097009608e-06,
-      "loss": 0.9177,
+      "loss": 0.9178,
       "num_input_tokens_seen": 133283840,
       "step": 16270
     },
     {
       "epoch": 0.7817527010804322,
-      "grad_norm": 1.3650128841400146,
+      "grad_norm": 1.3052635192871094,
       "learning_rate": 6.208631148263649e-06,
-      "loss": 0.8475,
+      "loss": 0.8479,
       "num_input_tokens_seen": 133365760,
       "step": 16280
     },
     {
       "epoch": 0.7822328931572629,
-      "grad_norm": 1.7946758270263672,
+      "grad_norm": 1.8097127676010132,
       "learning_rate": 6.18252538752955e-06,
-      "loss": 0.6443,
+      "loss": 0.6448,
       "num_input_tokens_seen": 133447680,
       "step": 16290
     },
     {
       "epoch": 0.7827130852340937,
-      "grad_norm": 0.47387173771858215,
+      "grad_norm": 0.5451312065124512,
       "learning_rate": 6.156466880362877e-06,
-      "loss": 0.7558,
+      "loss": 0.7555,
       "num_input_tokens_seen": 133529600,
       "step": 16300
     },
     {
       "epoch": 0.7831932773109244,
-      "grad_norm": 0.47302481532096863,
+      "grad_norm": 0.47235605120658875,
       "learning_rate": 6.1304556922005315e-06,
-      "loss": 0.8476,
+      "loss": 0.8473,
       "num_input_tokens_seen": 133611520,
       "step": 16310
     },
     {
       "epoch": 0.7836734693877551,
-      "grad_norm": 0.6125491261482239,
+      "grad_norm": 0.5912697911262512,
       "learning_rate": 6.1044918883606225e-06,
-      "loss": 0.8751,
+      "loss": 0.8754,
       "num_input_tokens_seen": 133693440,
       "step": 16320
     },
     {
       "epoch": 0.7841536614645859,
-      "grad_norm": 0.6477288007736206,
+      "grad_norm": 0.629851222038269,
       "learning_rate": 6.078575534042222e-06,
-      "loss": 0.9734,
+      "loss": 0.9731,
       "num_input_tokens_seen": 133775360,
       "step": 16330
     },
     {
       "epoch": 0.7846338535414166,
-      "grad_norm": 1.2066618204116821,
+      "grad_norm": 1.2804758548736572,
       "learning_rate": 6.052706694325292e-06,
-      "loss": 0.9515,
+      "loss": 0.952,
       "num_input_tokens_seen": 133857280,
       "step": 16340
     },
     {
       "epoch": 0.7851140456182473,
-      "grad_norm": 0.5165063142776489,
+      "grad_norm": 0.5203343629837036,
       "learning_rate": 6.026885434170457e-06,
-      "loss": 0.9599,
+      "loss": 0.9592,
       "num_input_tokens_seen": 133939200,
       "step": 16350
     },
     {
       "epoch": 0.7855942376950781,
-      "grad_norm": 1.0127323865890503,
+      "grad_norm": 1.0266387462615967,
       "learning_rate": 6.001111818418859e-06,
-      "loss": 0.975,
+      "loss": 0.9753,
       "num_input_tokens_seen": 134021120,
       "step": 16360
     },
     {
       "epoch": 0.7860744297719088,
-      "grad_norm": 0.5036996603012085,
+      "grad_norm": 0.5004794001579285,
       "learning_rate": 5.975385911792006e-06,
-      "loss": 0.9928,
+      "loss": 0.9931,
       "num_input_tokens_seen": 134103040,
       "step": 16370
     },
     {
       "epoch": 0.7865546218487395,
-      "grad_norm": 0.49564144015312195,
+      "grad_norm": 0.4966083765029907,
       "learning_rate": 5.9497077788916055e-06,
-      "loss": 1.0484,
+      "loss": 1.0476,
       "num_input_tokens_seen": 134184960,
       "step": 16380
     },
     {
       "epoch": 0.7870348139255702,
-      "grad_norm": 1.0678465366363525,
+      "grad_norm": 1.0581490993499756,
       "learning_rate": 5.924077484199389e-06,
-      "loss": 0.832,
+      "loss": 0.8325,
       "num_input_tokens_seen": 134266880,
       "step": 16390
     },
     {
       "epoch": 0.787515006002401,
-      "grad_norm": 0.4710373878479004,
+      "grad_norm": 0.4584285616874695,
       "learning_rate": 5.89849509207695e-06,
-      "loss": 0.7669,
+      "loss": 0.7674,
       "num_input_tokens_seen": 134348800,
       "step": 16400
     },
     {
       "epoch": 0.7879951980792317,
-      "grad_norm": 0.48212987184524536,
+      "grad_norm": 0.48210394382476807,
       "learning_rate": 5.872960666765618e-06,
-      "loss": 0.9089,
+      "loss": 0.909,
       "num_input_tokens_seen": 134430720,
       "step": 16410
     },
     {
       "epoch": 0.7884753901560624,
-      "grad_norm": 1.6406066417694092,
+      "grad_norm": 1.591680884361267,
       "learning_rate": 5.847474272386239e-06,
-      "loss": 1.5337,
+      "loss": 1.5853,
       "num_input_tokens_seen": 134512640,
       "step": 16420
     },
     {
       "epoch": 0.7889555822328932,
-      "grad_norm": 0.5122446417808533,
+      "grad_norm": 0.5121796727180481,
       "learning_rate": 5.822035972939069e-06,
-      "loss": 0.8218,
+      "loss": 0.8213,
       "num_input_tokens_seen": 134594560,
       "step": 16430
     },
     {
       "epoch": 0.7894357743097239,
-      "grad_norm": 0.4671303927898407,
+      "grad_norm": 0.4680531322956085,
       "learning_rate": 5.79664583230359e-06,
-      "loss": 0.8714,
+      "loss": 0.8711,
       "num_input_tokens_seen": 134676480,
       "step": 16440
     },
     {
       "epoch": 0.7899159663865546,
-      "grad_norm": 0.5007370710372925,
+      "grad_norm": 0.5005565285682678,
       "learning_rate": 5.771303914238333e-06,
-      "loss": 0.7859,
+      "loss": 0.7858,
       "num_input_tokens_seen": 134758400,
       "step": 16450
     },
     {
       "epoch": 0.7903961584633854,
-      "grad_norm": 0.4834369122982025,
+      "grad_norm": 0.48560014367103577,
       "learning_rate": 5.746010282380745e-06,
-      "loss": 0.8862,
+      "loss": 0.8857,
       "num_input_tokens_seen": 134840320,
       "step": 16460
     },
     {
       "epoch": 0.7908763505402161,
-      "grad_norm": 0.3713008761405945,
+      "grad_norm": 0.3735353946685791,
       "learning_rate": 5.7207650002470274e-06,
-      "loss": 0.7945,
+      "loss": 0.7948,
       "num_input_tokens_seen": 134922240,
       "step": 16470
     },
     {
       "epoch": 0.7913565426170468,
-      "grad_norm": 0.6575520634651184,
+      "grad_norm": 0.6581428050994873,
       "learning_rate": 5.695568131231949e-06,
-      "loss": 0.8114,
+      "loss": 0.8112,
       "num_input_tokens_seen": 135004160,
       "step": 16480
     },
     {
       "epoch": 0.7918367346938775,
-      "grad_norm": 0.5816577672958374,
+      "grad_norm": 0.5894924402236938,
       "learning_rate": 5.670419738608723e-06,
-      "loss": 0.7961,
+      "loss": 0.7962,
       "num_input_tokens_seen": 135086080,
       "step": 16490
     },
     {
       "epoch": 0.7923169267707083,
-      "grad_norm": 0.7607868313789368,
+      "grad_norm": 0.7610780596733093,
       "learning_rate": 5.645319885528824e-06,
-      "loss": 0.867,
+      "loss": 0.8665,
       "num_input_tokens_seen": 135168000,
       "step": 16500
     },
     {
       "epoch": 0.792797118847539,
-      "grad_norm": 0.49401038885116577,
+      "grad_norm": 0.4935015141963959,
       "learning_rate": 5.620268635021825e-06,
-      "loss": 0.8139,
+      "loss": 0.8142,
       "num_input_tokens_seen": 135249920,
       "step": 16510
     },
     {
       "epoch": 0.7932773109243697,
-      "grad_norm": 0.5071017742156982,
+      "grad_norm": 0.5041617155075073,
       "learning_rate": 5.595266049995268e-06,
-      "loss": 0.9577,
+      "loss": 0.9578,
       "num_input_tokens_seen": 135331840,
       "step": 16520
     },
     {
       "epoch": 0.7937575030012005,
-      "grad_norm": 0.5186456441879272,
+      "grad_norm": 0.524131715297699,
       "learning_rate": 5.5703121932344896e-06,
-      "loss": 0.9578,
+      "loss": 0.9576,
       "num_input_tokens_seen": 135413760,
       "step": 16530
     },
     {
       "epoch": 0.7942376950780312,
-      "grad_norm": 0.5309847593307495,
+      "grad_norm": 0.5360773205757141,
       "learning_rate": 5.5454071274024436e-06,
-      "loss": 0.8885,
+      "loss": 0.8884,
       "num_input_tokens_seen": 135495680,
       "step": 16540
     },
     {
       "epoch": 0.7947178871548619,
-      "grad_norm": 0.5009894967079163,
+      "grad_norm": 0.5002300143241882,
       "learning_rate": 5.520550915039579e-06,
-      "loss": 0.9609,
+      "loss": 0.961,
       "num_input_tokens_seen": 135577600,
       "step": 16550
     },
     {
       "epoch": 0.7951980792316927,
-      "grad_norm": 0.6387041211128235,
+      "grad_norm": 0.5231886506080627,
       "learning_rate": 5.495743618563668e-06,
-      "loss": 0.8386,
+      "loss": 0.838,
       "num_input_tokens_seen": 135659520,
       "step": 16560
     },
     {
       "epoch": 0.7956782713085234,
-      "grad_norm": 0.4808179438114166,
+      "grad_norm": 0.4792463779449463,
       "learning_rate": 5.4709853002696236e-06,
-      "loss": 0.8747,
+      "loss": 0.8735,
       "num_input_tokens_seen": 135741440,
       "step": 16570
     },
     {
       "epoch": 0.7961584633853541,
-      "grad_norm": 0.49355682730674744,
+      "grad_norm": 0.4909564256668091,
       "learning_rate": 5.4462760223294e-06,
-      "loss": 0.8467,
+      "loss": 0.8468,
       "num_input_tokens_seen": 135823360,
       "step": 16580
     },
     {
       "epoch": 0.7966386554621848,
-      "grad_norm": 0.4921320378780365,
+      "grad_norm": 0.4909401535987854,
       "learning_rate": 5.42161584679178e-06,
-      "loss": 0.8295,
+      "loss": 0.8294,
       "num_input_tokens_seen": 135905280,
       "step": 16590
     },
     {
       "epoch": 0.7971188475390156,
-      "grad_norm": 0.4814390242099762,
+      "grad_norm": 0.4839801490306854,
       "learning_rate": 5.397004835582242e-06,
-      "loss": 0.8218,
+      "loss": 0.8217,
       "num_input_tokens_seen": 135987200,
       "step": 16600
     },
     {
       "epoch": 0.7975990396158463,
-      "grad_norm": 0.5022986531257629,
+      "grad_norm": 0.5041458606719971,
       "learning_rate": 5.372443050502823e-06,
-      "loss": 0.8522,
+      "loss": 0.8524,
       "num_input_tokens_seen": 136069120,
       "step": 16610
     },
     {
       "epoch": 0.798079231692677,
-      "grad_norm": 0.48835310339927673,
+      "grad_norm": 0.4912269711494446,
       "learning_rate": 5.347930553231942e-06,
-      "loss": 0.8879,
+      "loss": 0.8883,
       "num_input_tokens_seen": 136151040,
       "step": 16620
     },
     {
       "epoch": 0.7985594237695078,
-      "grad_norm": 0.48353898525238037,
+      "grad_norm": 0.48488718271255493,
       "learning_rate": 5.323467405324226e-06,
-      "loss": 1.0158,
+      "loss": 1.0163,
       "num_input_tokens_seen": 136232960,
       "step": 16630
     },
     {
       "epoch": 0.7990396158463385,
-      "grad_norm": 0.5025900602340698,
+      "grad_norm": 0.4990125596523285,
       "learning_rate": 5.299053668210402e-06,
-      "loss": 1.0689,
+      "loss": 1.069,
       "num_input_tokens_seen": 136314880,
       "step": 16640
     },
     {
       "epoch": 0.7995198079231692,
-      "grad_norm": 0.4869900345802307,
+      "grad_norm": 0.4874297082424164,
       "learning_rate": 5.274689403197119e-06,
-      "loss": 0.7811,
+      "loss": 0.7804,
       "num_input_tokens_seen": 136396800,
       "step": 16650
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.5168715119361877,
+      "grad_norm": 0.5153248310089111,
       "learning_rate": 5.250374671466776e-06,
-      "loss": 0.8692,
+      "loss": 0.8694,
       "num_input_tokens_seen": 136478720,
       "step": 16660
     },
     {
       "epoch": 0.8004801920768307,
-      "grad_norm": 0.5069206357002258,
+      "grad_norm": 0.5061455965042114,
       "learning_rate": 5.2261095340774085e-06,
-      "loss": 1.1452,
+      "loss": 1.145,
       "num_input_tokens_seen": 136560640,
       "step": 16670
     },
     {
       "epoch": 0.8009603841536614,
-      "grad_norm": 0.4853163957595825,
+      "grad_norm": 0.4891180098056793,
       "learning_rate": 5.201894051962486e-06,
-      "loss": 1.0006,
+      "loss": 0.9997,
       "num_input_tokens_seen": 136642560,
       "step": 16680
     },
     {
       "epoch": 0.8014405762304923,
-      "grad_norm": 0.4831332862377167,
+      "grad_norm": 0.48567453026771545,
       "learning_rate": 5.177728285930816e-06,
-      "loss": 0.897,
+      "loss": 0.8972,
       "num_input_tokens_seen": 136724480,
       "step": 16690
     },
     {
       "epoch": 0.801920768307323,
-      "grad_norm": 0.5170623660087585,
+      "grad_norm": 0.5174875259399414,
       "learning_rate": 5.153612296666335e-06,
-      "loss": 0.7989,
+      "loss": 0.7987,
       "num_input_tokens_seen": 136806400,
       "step": 16700
     },
     {
       "epoch": 0.8024009603841536,
-      "grad_norm": 0.47998228669166565,
+      "grad_norm": 0.4791352152824402,
       "learning_rate": 5.129546144727998e-06,
       "loss": 0.9656,
       "num_input_tokens_seen": 136888320,
@@ -13378,47 +13378,47 @@
     },
     {
       "epoch": 0.8028811524609843,
-      "grad_norm": 0.4863240420818329,
+      "grad_norm": 0.4892936050891876,
       "learning_rate": 5.105529890549618e-06,
-      "loss": 0.8726,
+      "loss": 0.873,
       "num_input_tokens_seen": 136970240,
       "step": 16720
     },
     {
       "epoch": 0.8033613445378152,
-      "grad_norm": 0.4388894736766815,
+      "grad_norm": 0.4297828674316406,
       "learning_rate": 5.081563594439676e-06,
-      "loss": 0.7868,
+      "loss": 0.7871,
       "num_input_tokens_seen": 137052160,
       "step": 16730
     },
     {
       "epoch": 0.8038415366146459,
-      "grad_norm": 0.4876529276371002,
+      "grad_norm": 0.4883136749267578,
       "learning_rate": 5.057647316581232e-06,
-      "loss": 1.0533,
+      "loss": 1.0537,
       "num_input_tokens_seen": 137134080,
       "step": 16740
     },
     {
       "epoch": 0.8043217286914766,
-      "grad_norm": 0.47916775941848755,
+      "grad_norm": 0.4795672297477722,
       "learning_rate": 5.033781117031738e-06,
-      "loss": 0.9322,
+      "loss": 0.9321,
       "num_input_tokens_seen": 137216000,
       "step": 16750
     },
     {
       "epoch": 0.8048019207683074,
-      "grad_norm": 0.47834011912345886,
+      "grad_norm": 0.4753381013870239,
       "learning_rate": 5.0099650557228785e-06,
-      "loss": 1.0616,
+      "loss": 1.0611,
       "num_input_tokens_seen": 137297920,
       "step": 16760
     },
     {
       "epoch": 0.8052821128451381,
-      "grad_norm": 0.5335723757743835,
+      "grad_norm": 0.5331737399101257,
       "learning_rate": 4.986199192460428e-06,
       "loss": 0.8815,
       "num_input_tokens_seen": 137379840,
@@ -13426,47 +13426,47 @@
     },
     {
       "epoch": 0.8057623049219688,
-      "grad_norm": 0.5083367228507996,
+      "grad_norm": 0.5075027346611023,
       "learning_rate": 4.962483586924136e-06,
-      "loss": 0.7737,
+      "loss": 0.7744,
       "num_input_tokens_seen": 137461760,
       "step": 16780
     },
     {
       "epoch": 0.8062424969987996,
-      "grad_norm": 0.4924291968345642,
+      "grad_norm": 0.490078866481781,
       "learning_rate": 4.93881829866751e-06,
-      "loss": 0.8263,
+      "loss": 0.8282,
       "num_input_tokens_seen": 137543680,
       "step": 16790
     },
     {
       "epoch": 0.8067226890756303,
-      "grad_norm": 0.5008875131607056,
+      "grad_norm": 0.5007233023643494,
       "learning_rate": 4.915203387117736e-06,
-      "loss": 0.9976,
+      "loss": 0.9975,
       "num_input_tokens_seen": 137625600,
       "step": 16800
     },
     {
       "epoch": 0.807202881152461,
-      "grad_norm": 0.5046831369400024,
+      "grad_norm": 0.5057447552680969,
       "learning_rate": 4.891638911575483e-06,
-      "loss": 0.8821,
+      "loss": 0.882,
       "num_input_tokens_seen": 137707520,
       "step": 16810
     },
     {
       "epoch": 0.8076830732292917,
-      "grad_norm": 0.5597212910652161,
+      "grad_norm": 0.5599965453147888,
       "learning_rate": 4.868124931214752e-06,
-      "loss": 0.9083,
+      "loss": 0.9084,
       "num_input_tokens_seen": 137789440,
       "step": 16820
     },
     {
       "epoch": 0.8081632653061225,
-      "grad_norm": 0.5007030367851257,
+      "grad_norm": 0.5017980933189392,
       "learning_rate": 4.844661505082768e-06,
       "loss": 0.8356,
       "num_input_tokens_seen": 137871360,
@@ -13474,23 +13474,23 @@
     },
     {
       "epoch": 0.8086434573829532,
-      "grad_norm": 0.3486262857913971,
+      "grad_norm": 0.3630382716655731,
       "learning_rate": 4.8212486920998005e-06,
-      "loss": 0.7684,
+      "loss": 0.7691,
       "num_input_tokens_seen": 137953280,
       "step": 16840
     },
     {
       "epoch": 0.8091236494597839,
-      "grad_norm": 0.4673330783843994,
+      "grad_norm": 0.46930795907974243,
       "learning_rate": 4.797886551059011e-06,
-      "loss": 0.9145,
+      "loss": 0.915,
       "num_input_tokens_seen": 138035200,
       "step": 16850
     },
     {
       "epoch": 0.8096038415366147,
-      "grad_norm": 0.4685009717941284,
+      "grad_norm": 0.46570199728012085,
       "learning_rate": 4.7745751406263165e-06,
       "loss": 0.9692,
       "num_input_tokens_seen": 138117120,
@@ -13498,207 +13498,207 @@
     },
     {
       "epoch": 0.8100840336134454,
-      "grad_norm": 0.4861032962799072,
+      "grad_norm": 0.48233890533447266,
       "learning_rate": 4.751314519340258e-06,
-      "loss": 0.8577,
+      "loss": 0.8575,
       "num_input_tokens_seen": 138199040,
       "step": 16870
     },
     {
       "epoch": 0.8105642256902761,
-      "grad_norm": 0.5741161108016968,
+      "grad_norm": 0.5723303556442261,
       "learning_rate": 4.728104745611814e-06,
-      "loss": 0.8875,
+      "loss": 0.8863,
       "num_input_tokens_seen": 138280960,
       "step": 16880
     },
     {
       "epoch": 0.8110444177671069,
-      "grad_norm": 0.5122044682502747,
+      "grad_norm": 0.5114380121231079,
       "learning_rate": 4.704945877724295e-06,
-      "loss": 1.0989,
+      "loss": 1.099,
       "num_input_tokens_seen": 138362880,
       "step": 16890
     },
     {
       "epoch": 0.8115246098439376,
-      "grad_norm": 0.8650029897689819,
+      "grad_norm": 0.8760389089584351,
       "learning_rate": 4.681837973833181e-06,
-      "loss": 0.9019,
+      "loss": 0.9003,
       "num_input_tokens_seen": 138444800,
       "step": 16900
     },
     {
       "epoch": 0.8120048019207683,
-      "grad_norm": 0.49754172563552856,
+      "grad_norm": 0.49562937021255493,
       "learning_rate": 4.658781091965955e-06,
-      "loss": 0.8082,
+      "loss": 0.8087,
       "num_input_tokens_seen": 138526720,
       "step": 16910
     },
     {
       "epoch": 0.812484993997599,
-      "grad_norm": 0.45871591567993164,
+      "grad_norm": 0.4572142958641052,
       "learning_rate": 4.635775290021988e-06,
-      "loss": 0.8978,
+      "loss": 0.8977,
       "num_input_tokens_seen": 138608640,
       "step": 16920
     },
     {
       "epoch": 0.8129651860744298,
-      "grad_norm": 0.3866370916366577,
+      "grad_norm": 0.39572256803512573,
       "learning_rate": 4.612820625772391e-06,
-      "loss": 0.6356,
+      "loss": 0.6357,
       "num_input_tokens_seen": 138690560,
       "step": 16930
     },
     {
       "epoch": 0.8134453781512605,
-      "grad_norm": 0.6949254870414734,
+      "grad_norm": 0.6943681240081787,
       "learning_rate": 4.589917156859838e-06,
-      "loss": 0.8645,
+      "loss": 0.8636,
       "num_input_tokens_seen": 138772480,
       "step": 16940
     },
     {
       "epoch": 0.8139255702280912,
-      "grad_norm": 0.6086997985839844,
+      "grad_norm": 0.6021678447723389,
       "learning_rate": 4.5670649407984625e-06,
-      "loss": 0.8896,
+      "loss": 0.8897,
       "num_input_tokens_seen": 138854400,
       "step": 16950
     },
     {
       "epoch": 0.814405762304922,
-      "grad_norm": 0.4606131613254547,
+      "grad_norm": 0.4605492651462555,
       "learning_rate": 4.544264034973686e-06,
-      "loss": 0.7794,
+      "loss": 0.7806,
       "num_input_tokens_seen": 138936320,
       "step": 16960
     },
     {
       "epoch": 0.8148859543817527,
-      "grad_norm": 0.5047131180763245,
+      "grad_norm": 0.5034934282302856,
       "learning_rate": 4.521514496642074e-06,
-      "loss": 0.9631,
+      "loss": 0.9632,
       "num_input_tokens_seen": 139018240,
       "step": 16970
     },
     {
       "epoch": 0.8153661464585834,
-      "grad_norm": 0.5257802605628967,
+      "grad_norm": 0.5086869597434998,
       "learning_rate": 4.498816382931217e-06,
-      "loss": 0.9246,
+      "loss": 0.9245,
       "num_input_tokens_seen": 139100160,
       "step": 16980
     },
     {
       "epoch": 0.8158463385354142,
-      "grad_norm": 0.5585389733314514,
+      "grad_norm": 0.5553309321403503,
       "learning_rate": 4.476169750839571e-06,
-      "loss": 1.0151,
+      "loss": 1.015,
       "num_input_tokens_seen": 139182080,
       "step": 16990
     },
     {
       "epoch": 0.8163265306122449,
-      "grad_norm": 0.47912609577178955,
+      "grad_norm": 0.4763419032096863,
       "learning_rate": 4.45357465723629e-06,
-      "loss": 0.8939,
+      "loss": 0.8941,
       "num_input_tokens_seen": 139264000,
       "step": 17000
     },
     {
       "epoch": 0.8168067226890756,
-      "grad_norm": 0.4979766309261322,
+      "grad_norm": 0.49482813477516174,
       "learning_rate": 4.4310311588611294e-06,
-      "loss": 0.8846,
+      "loss": 0.8841,
       "num_input_tokens_seen": 139345920,
       "step": 17010
     },
     {
       "epoch": 0.8172869147659063,
-      "grad_norm": 0.4919578731060028,
+      "grad_norm": 0.4898833930492401,
       "learning_rate": 4.408539312324281e-06,
-      "loss": 0.8446,
+      "loss": 0.8452,
       "num_input_tokens_seen": 139427840,
       "step": 17020
     },
     {
       "epoch": 0.8177671068427371,
-      "grad_norm": 0.510263204574585,
+      "grad_norm": 0.5128147006034851,
       "learning_rate": 4.386099174106212e-06,
-      "loss": 0.8381,
+      "loss": 0.8382,
       "num_input_tokens_seen": 139509760,
       "step": 17030
     },
     {
       "epoch": 0.8182472989195678,
-      "grad_norm": 0.5191063284873962,
+      "grad_norm": 0.5220133066177368,
       "learning_rate": 4.363710800557566e-06,
-      "loss": 0.9471,
+      "loss": 0.9468,
       "num_input_tokens_seen": 139591680,
       "step": 17040
     },
     {
       "epoch": 0.8187274909963985,
-      "grad_norm": 0.5044844150543213,
+      "grad_norm": 0.5058349967002869,
       "learning_rate": 4.341374247898983e-06,
-      "loss": 0.7787,
+      "loss": 0.7798,
       "num_input_tokens_seen": 139673600,
       "step": 17050
     },
     {
       "epoch": 0.8192076830732293,
-      "grad_norm": 0.48283785581588745,
+      "grad_norm": 0.4795903265476227,
       "learning_rate": 4.3190895722209635e-06,
-      "loss": 0.9487,
+      "loss": 0.949,
       "num_input_tokens_seen": 139755520,
       "step": 17060
     },
     {
       "epoch": 0.81968787515006,
-      "grad_norm": 0.36586451530456543,
+      "grad_norm": 0.3831540048122406,
       "learning_rate": 4.296856829483759e-06,
-      "loss": 0.884,
+      "loss": 0.8836,
       "num_input_tokens_seen": 139837440,
       "step": 17070
     },
     {
       "epoch": 0.8201680672268907,
-      "grad_norm": 0.49161645770072937,
+      "grad_norm": 0.4896925687789917,
       "learning_rate": 4.274676075517206e-06,
-      "loss": 0.741,
+      "loss": 0.7409,
       "num_input_tokens_seen": 139919360,
       "step": 17080
     },
     {
       "epoch": 0.8206482593037215,
-      "grad_norm": 0.47819775342941284,
+      "grad_norm": 0.47725582122802734,
       "learning_rate": 4.252547366020568e-06,
-      "loss": 0.8349,
+      "loss": 0.8346,
       "num_input_tokens_seen": 140001280,
       "step": 17090
     },
     {
       "epoch": 0.8211284513805522,
-      "grad_norm": 0.48540183901786804,
+      "grad_norm": 0.4832172989845276,
       "learning_rate": 4.230470756562438e-06,
-      "loss": 0.9832,
+      "loss": 0.983,
       "num_input_tokens_seen": 140083200,
       "step": 17100
     },
     {
       "epoch": 0.8216086434573829,
-      "grad_norm": 0.7266910672187805,
+      "grad_norm": 0.7230331897735596,
       "learning_rate": 4.208446302580582e-06,
-      "loss": 0.7601,
+      "loss": 0.7598,
       "num_input_tokens_seen": 140165120,
       "step": 17110
     },
     {
       "epoch": 0.8220888355342136,
-      "grad_norm": 0.48397764563560486,
+      "grad_norm": 0.48454123735427856,
       "learning_rate": 4.186474059381768e-06,
       "loss": 0.7198,
       "num_input_tokens_seen": 140247040,
@@ -13706,23 +13706,23 @@
     },
     {
       "epoch": 0.8225690276110444,
-      "grad_norm": 0.5694714188575745,
+      "grad_norm": 0.5540962815284729,
       "learning_rate": 4.164554082141683e-06,
-      "loss": 0.8547,
+      "loss": 0.8544,
       "num_input_tokens_seen": 140328960,
       "step": 17130
     },
     {
       "epoch": 0.8230492196878751,
-      "grad_norm": 1.3133982419967651,
+      "grad_norm": 1.901149034500122,
       "learning_rate": 4.142686425904752e-06,
-      "loss": 0.8511,
+      "loss": 0.8519,
       "num_input_tokens_seen": 140410880,
       "step": 17140
     },
     {
       "epoch": 0.8235294117647058,
-      "grad_norm": 0.48804813623428345,
+      "grad_norm": 0.48827600479125977,
       "learning_rate": 4.12087114558401e-06,
       "loss": 0.8076,
       "num_input_tokens_seen": 140492800,
@@ -13730,23 +13730,23 @@
     },
     {
       "epoch": 0.8240096038415367,
-      "grad_norm": 0.4581037163734436,
+      "grad_norm": 0.4572823941707611,
       "learning_rate": 4.099108295960977e-06,
-      "loss": 0.7348,
+      "loss": 0.7344,
       "num_input_tokens_seen": 140574720,
       "step": 17160
     },
     {
       "epoch": 0.8244897959183674,
-      "grad_norm": 0.48801615834236145,
+      "grad_norm": 0.4875430464744568,
       "learning_rate": 4.077397931685523e-06,
-      "loss": 0.8413,
+      "loss": 0.8416,
       "num_input_tokens_seen": 140656640,
       "step": 17170
     },
     {
       "epoch": 0.824969987995198,
-      "grad_norm": 0.46194446086883545,
+      "grad_norm": 0.46101564168930054,
       "learning_rate": 4.055740107275685e-06,
       "loss": 1.1131,
       "num_input_tokens_seen": 140738560,
@@ -13754,23 +13754,23 @@
     },
     {
       "epoch": 0.8254501800720289,
-      "grad_norm": 0.5051575899124146,
+      "grad_norm": 0.5040237903594971,
       "learning_rate": 4.0341348771175955e-06,
-      "loss": 0.9822,
+      "loss": 0.9831,
       "num_input_tokens_seen": 140820480,
       "step": 17190
     },
     {
       "epoch": 0.8259303721488596,
-      "grad_norm": 0.4934830367565155,
+      "grad_norm": 0.492987722158432,
       "learning_rate": 4.012582295465308e-06,
-      "loss": 0.834,
+      "loss": 0.8343,
       "num_input_tokens_seen": 140902400,
       "step": 17200
     },
     {
       "epoch": 0.8264105642256903,
-      "grad_norm": 0.5021478533744812,
+      "grad_norm": 0.499546080827713,
       "learning_rate": 3.991082416440656e-06,
       "loss": 0.8526,
       "num_input_tokens_seen": 140984320,
@@ -13778,39 +13778,39 @@
     },
     {
       "epoch": 0.826890756302521,
-      "grad_norm": 0.4906045198440552,
+      "grad_norm": 0.4928174316883087,
       "learning_rate": 3.969635294033144e-06,
-      "loss": 0.794,
+      "loss": 0.7942,
       "num_input_tokens_seen": 141066240,
       "step": 17220
     },
     {
       "epoch": 0.8273709483793518,
-      "grad_norm": 0.49712374806404114,
+      "grad_norm": 0.49705326557159424,
       "learning_rate": 3.9482409820997826e-06,
-      "loss": 0.8676,
+      "loss": 0.8681,
       "num_input_tokens_seen": 141148160,
       "step": 17230
     },
     {
       "epoch": 0.8278511404561825,
-      "grad_norm": 0.4694906771183014,
+      "grad_norm": 0.4672960937023163,
       "learning_rate": 3.926899534364969e-06,
-      "loss": 0.9085,
+      "loss": 0.9081,
       "num_input_tokens_seen": 141230080,
       "step": 17240
     },
     {
       "epoch": 0.8283313325330132,
-      "grad_norm": 0.4903012216091156,
+      "grad_norm": 0.4908684492111206,
       "learning_rate": 3.90561100442036e-06,
-      "loss": 0.8688,
+      "loss": 0.8689,
       "num_input_tokens_seen": 141312000,
       "step": 17250
     },
     {
       "epoch": 0.828811524609844,
-      "grad_norm": 0.5216225385665894,
+      "grad_norm": 0.5210193395614624,
       "learning_rate": 3.8843754457247275e-06,
       "loss": 1.0517,
       "num_input_tokens_seen": 141393920,
@@ -13818,7 +13818,7 @@
     },
     {
       "epoch": 0.8292917166866747,
-      "grad_norm": 0.48557716608047485,
+      "grad_norm": 0.48606058955192566,
       "learning_rate": 3.863192911603808e-06,
       "loss": 0.9196,
       "num_input_tokens_seen": 141475840,
@@ -13826,95 +13826,95 @@
     },
     {
       "epoch": 0.8297719087635054,
-      "grad_norm": 0.4983423352241516,
+      "grad_norm": 0.49873122572898865,
       "learning_rate": 3.842063455250203e-06,
-      "loss": 0.9045,
+      "loss": 0.9042,
       "num_input_tokens_seen": 141557760,
       "step": 17280
     },
     {
       "epoch": 0.8302521008403362,
-      "grad_norm": 0.5038923621177673,
+      "grad_norm": 0.5023019313812256,
       "learning_rate": 3.820987129723228e-06,
-      "loss": 0.8666,
+      "loss": 0.867,
       "num_input_tokens_seen": 141639680,
       "step": 17290
     },
     {
       "epoch": 0.8307322929171669,
-      "grad_norm": 0.4999052584171295,
+      "grad_norm": 0.4990661144256592,
       "learning_rate": 3.799963987948757e-06,
-      "loss": 0.788,
+      "loss": 0.7882,
       "num_input_tokens_seen": 141721600,
       "step": 17300
     },
     {
       "epoch": 0.8312124849939976,
-      "grad_norm": 0.5127232074737549,
+      "grad_norm": 0.5133486986160278,
       "learning_rate": 3.7789940827191395e-06,
-      "loss": 0.8889,
+      "loss": 0.8897,
       "num_input_tokens_seen": 141803520,
       "step": 17310
     },
     {
       "epoch": 0.8316926770708283,
-      "grad_norm": 0.5200673937797546,
+      "grad_norm": 0.5203906297683716,
       "learning_rate": 3.7580774666930134e-06,
-      "loss": 0.8877,
+      "loss": 0.8876,
       "num_input_tokens_seen": 141885440,
       "step": 17320
     },
     {
       "epoch": 0.8321728691476591,
-      "grad_norm": 0.4389039874076843,
+      "grad_norm": 0.46600568294525146,
       "learning_rate": 3.737214192395225e-06,
-      "loss": 0.8935,
+      "loss": 0.8931,
       "num_input_tokens_seen": 141967360,
       "step": 17330
     },
     {
       "epoch": 0.8326530612244898,
-      "grad_norm": 0.481927752494812,
+      "grad_norm": 0.48274463415145874,
       "learning_rate": 3.7164043122166508e-06,
-      "loss": 1.0336,
+      "loss": 1.0343,
       "num_input_tokens_seen": 142049280,
       "step": 17340
     },
     {
       "epoch": 0.8331332533013205,
-      "grad_norm": 0.5148651599884033,
+      "grad_norm": 0.5128066539764404,
       "learning_rate": 3.6956478784140937e-06,
-      "loss": 1.0014,
+      "loss": 1.0005,
       "num_input_tokens_seen": 142131200,
       "step": 17350
     },
     {
       "epoch": 0.8336134453781513,
-      "grad_norm": 0.48852846026420593,
+      "grad_norm": 0.48960596323013306,
       "learning_rate": 3.674944943110156e-06,
-      "loss": 0.9075,
+      "loss": 0.9079,
       "num_input_tokens_seen": 142213120,
       "step": 17360
     },
     {
       "epoch": 0.834093637454982,
-      "grad_norm": 0.4977552592754364,
+      "grad_norm": 0.49447503685951233,
       "learning_rate": 3.6542955582930748e-06,
-      "loss": 0.7461,
+      "loss": 0.7459,
       "num_input_tokens_seen": 142295040,
       "step": 17370
     },
     {
       "epoch": 0.8345738295318127,
-      "grad_norm": 0.4385221600532532,
+      "grad_norm": 0.43031996488571167,
       "learning_rate": 3.6336997758166263e-06,
-      "loss": 0.9869,
+      "loss": 0.9866,
       "num_input_tokens_seen": 142376960,
       "step": 17380
     },
     {
       "epoch": 0.8350540216086435,
-      "grad_norm": 0.4688054919242859,
+      "grad_norm": 0.46145930886268616,
       "learning_rate": 3.6131576473999924e-06,
       "loss": 0.775,
       "num_input_tokens_seen": 142458880,
@@ -13922,7 +13922,7 @@
     },
     {
       "epoch": 0.8355342136854742,
-      "grad_norm": 0.4902810752391815,
+      "grad_norm": 0.49166035652160645,
       "learning_rate": 3.592669224627601e-06,
       "loss": 0.916,
       "num_input_tokens_seen": 142540800,
@@ -13930,63 +13930,63 @@
     },
     {
       "epoch": 0.8360144057623049,
-      "grad_norm": 0.49315813183784485,
+      "grad_norm": 0.4930817782878876,
       "learning_rate": 3.5722345589490306e-06,
-      "loss": 1.0457,
+      "loss": 1.046,
       "num_input_tokens_seen": 142622720,
       "step": 17410
     },
     {
       "epoch": 0.8364945978391356,
-      "grad_norm": 0.8642867207527161,
+      "grad_norm": 0.7804808616638184,
       "learning_rate": 3.5518537016788646e-06,
-      "loss": 1.0179,
+      "loss": 1.0183,
       "num_input_tokens_seen": 142704640,
       "step": 17420
     },
     {
       "epoch": 0.8369747899159664,
-      "grad_norm": 0.487728476524353,
+      "grad_norm": 0.48990410566329956,
       "learning_rate": 3.531526703996557e-06,
-      "loss": 0.8232,
+      "loss": 0.8231,
       "num_input_tokens_seen": 142786560,
       "step": 17430
     },
     {
       "epoch": 0.8374549819927971,
-      "grad_norm": 0.48335549235343933,
+      "grad_norm": 0.4804694354534149,
       "learning_rate": 3.511253616946325e-06,
-      "loss": 0.9279,
+      "loss": 0.9277,
       "num_input_tokens_seen": 142868480,
       "step": 17440
     },
     {
       "epoch": 0.8379351740696278,
-      "grad_norm": 0.5654048323631287,
+      "grad_norm": 0.5617939829826355,
       "learning_rate": 3.4910344914370093e-06,
-      "loss": 0.8721,
+      "loss": 0.8723,
       "num_input_tokens_seen": 142950400,
       "step": 17450
     },
     {
       "epoch": 0.8384153661464586,
-      "grad_norm": 0.6163344383239746,
+      "grad_norm": 0.6097043752670288,
       "learning_rate": 3.4708693782419225e-06,
-      "loss": 0.8941,
+      "loss": 0.8936,
       "num_input_tokens_seen": 143032320,
       "step": 17460
     },
     {
       "epoch": 0.8388955582232893,
-      "grad_norm": 0.5046043992042542,
+      "grad_norm": 0.5014716982841492,
       "learning_rate": 3.450758327998768e-06,
-      "loss": 0.8403,
+      "loss": 0.8404,
       "num_input_tokens_seen": 143114240,
       "step": 17470
     },
     {
       "epoch": 0.83937575030012,
-      "grad_norm": 0.44665244221687317,
+      "grad_norm": 0.4464024603366852,
       "learning_rate": 3.4307013912094845e-06,
       "loss": 0.9405,
       "num_input_tokens_seen": 143196160,
@@ -13994,63 +13994,63 @@
     },
     {
       "epoch": 0.8398559423769508,
-      "grad_norm": 0.5446330904960632,
+      "grad_norm": 0.5696042776107788,
       "learning_rate": 3.41069861824011e-06,
-      "loss": 0.6972,
+      "loss": 0.6978,
       "num_input_tokens_seen": 143278080,
       "step": 17490
     },
     {
       "epoch": 0.8403361344537815,
-      "grad_norm": 0.5121775269508362,
+      "grad_norm": 0.510014533996582,
       "learning_rate": 3.390750059320688e-06,
-      "loss": 0.9488,
+      "loss": 0.9487,
       "num_input_tokens_seen": 143360000,
       "step": 17500
     },
     {
       "epoch": 0.8408163265306122,
-      "grad_norm": 0.42379680275917053,
+      "grad_norm": 0.41205736994743347,
       "learning_rate": 3.3708557645451053e-06,
-      "loss": 0.7979,
+      "loss": 0.7978,
       "num_input_tokens_seen": 143441920,
       "step": 17510
     },
     {
       "epoch": 0.8412965186074429,
-      "grad_norm": 0.915107250213623,
+      "grad_norm": 0.8583078980445862,
       "learning_rate": 3.3510157838709895e-06,
-      "loss": 0.9223,
+      "loss": 0.9229,
       "num_input_tokens_seen": 143523840,
       "step": 17520
     },
     {
       "epoch": 0.8417767106842737,
-      "grad_norm": 0.4954970180988312,
+      "grad_norm": 0.4940420687198639,
       "learning_rate": 3.3312301671195784e-06,
-      "loss": 1.0832,
+      "loss": 1.0837,
       "num_input_tokens_seen": 143605760,
       "step": 17530
     },
     {
       "epoch": 0.8422569027611044,
-      "grad_norm": 0.37159988284111023,
+      "grad_norm": 0.3699828088283539,
       "learning_rate": 3.3114989639755983e-06,
-      "loss": 0.8365,
+      "loss": 0.8367,
       "num_input_tokens_seen": 143687680,
       "step": 17540
     },
     {
       "epoch": 0.8427370948379351,
-      "grad_norm": 0.4709157943725586,
+      "grad_norm": 0.47020527720451355,
       "learning_rate": 3.2918222239871206e-06,
-      "loss": 0.8637,
+      "loss": 0.863,
       "num_input_tokens_seen": 143769600,
       "step": 17550
     },
     {
       "epoch": 0.8432172869147659,
-      "grad_norm": 0.4602372646331787,
+      "grad_norm": 0.4599441885948181,
       "learning_rate": 3.272199996565464e-06,
       "loss": 0.8599,
       "num_input_tokens_seen": 143851520,
@@ -14058,15 +14058,15 @@
     },
     {
       "epoch": 0.8436974789915966,
-      "grad_norm": 0.4694511890411377,
+      "grad_norm": 0.46887126564979553,
       "learning_rate": 3.252632330985059e-06,
-      "loss": 0.8833,
+      "loss": 0.8835,
       "num_input_tokens_seen": 143933440,
       "step": 17570
     },
     {
       "epoch": 0.8441776710684273,
-      "grad_norm": 0.4874444603919983,
+      "grad_norm": 0.4868754744529724,
       "learning_rate": 3.233119276383309e-06,
       "loss": 0.8098,
       "num_input_tokens_seen": 144015360,
@@ -14074,71 +14074,71 @@
     },
     {
       "epoch": 0.8446578631452581,
-      "grad_norm": 0.7956154346466064,
+      "grad_norm": 0.7933685183525085,
       "learning_rate": 3.2136608817604998e-06,
-      "loss": 0.8137,
+      "loss": 0.8136,
       "num_input_tokens_seen": 144097280,
       "step": 17590
     },
     {
       "epoch": 0.8451380552220888,
-      "grad_norm": 0.5142585039138794,
+      "grad_norm": 0.5126639008522034,
       "learning_rate": 3.1942571959796414e-06,
-      "loss": 0.9244,
+      "loss": 0.9242,
       "num_input_tokens_seen": 144179200,
       "step": 17600
     },
     {
       "epoch": 0.8456182472989195,
-      "grad_norm": 0.49832209944725037,
+      "grad_norm": 0.49685657024383545,
       "learning_rate": 3.1749082677663606e-06,
-      "loss": 0.9445,
+      "loss": 0.9447,
       "num_input_tokens_seen": 144261120,
       "step": 17610
     },
     {
       "epoch": 0.8460984393757504,
-      "grad_norm": 0.49295270442962646,
+      "grad_norm": 0.493713915348053,
       "learning_rate": 3.1556141457087932e-06,
-      "loss": 1.0138,
+      "loss": 1.0145,
       "num_input_tokens_seen": 144343040,
       "step": 17620
     },
     {
       "epoch": 0.846578631452581,
-      "grad_norm": 0.49073633551597595,
+      "grad_norm": 0.4875103533267975,
       "learning_rate": 3.1363748782574475e-06,
-      "loss": 0.8728,
+      "loss": 0.8731,
       "num_input_tokens_seen": 144424960,
       "step": 17630
     },
     {
       "epoch": 0.8470588235294118,
-      "grad_norm": 0.48808997869491577,
+      "grad_norm": 0.48699912428855896,
       "learning_rate": 3.1171905137250655e-06,
-      "loss": 0.8881,
+      "loss": 0.8883,
       "num_input_tokens_seen": 144506880,
       "step": 17640
     },
     {
       "epoch": 0.8475390156062425,
-      "grad_norm": 0.5085486769676208,
+      "grad_norm": 0.5091184973716736,
       "learning_rate": 3.098061100286537e-06,
-      "loss": 0.8107,
+      "loss": 0.8102,
       "num_input_tokens_seen": 144588800,
       "step": 17650
     },
     {
       "epoch": 0.8480192076830733,
-      "grad_norm": 0.4130302965641022,
+      "grad_norm": 0.41908127069473267,
       "learning_rate": 3.078986685978763e-06,
-      "loss": 1.0014,
+      "loss": 1.0013,
       "num_input_tokens_seen": 144670720,
       "step": 17660
     },
     {
       "epoch": 0.848499399759904,
-      "grad_norm": 0.4925101697444916,
+      "grad_norm": 0.49007028341293335,
       "learning_rate": 3.059967318700513e-06,
       "loss": 0.7933,
       "num_input_tokens_seen": 144752640,
@@ -14146,71 +14146,71 @@
     },
     {
       "epoch": 0.8489795918367347,
-      "grad_norm": 0.5227538347244263,
+      "grad_norm": 0.5263189077377319,
       "learning_rate": 3.0410030462123486e-06,
-      "loss": 0.8734,
+      "loss": 0.8733,
       "num_input_tokens_seen": 144834560,
       "step": 17680
     },
     {
       "epoch": 0.8494597839135655,
-      "grad_norm": 0.48874667286872864,
+      "grad_norm": 0.4886823892593384,
       "learning_rate": 3.022093916136465e-06,
-      "loss": 1.1692,
+      "loss": 1.1693,
       "num_input_tokens_seen": 144916480,
       "step": 17690
     },
     {
       "epoch": 0.8499399759903962,
-      "grad_norm": 0.5069953799247742,
+      "grad_norm": 0.507895290851593,
       "learning_rate": 3.0032399759565845e-06,
-      "loss": 0.8841,
+      "loss": 0.8838,
       "num_input_tokens_seen": 144998400,
       "step": 17700
     },
     {
       "epoch": 0.8504201680672269,
-      "grad_norm": 0.49537745118141174,
+      "grad_norm": 0.49439266324043274,
       "learning_rate": 2.9844412730178515e-06,
-      "loss": 0.9305,
+      "loss": 0.93,
       "num_input_tokens_seen": 145080320,
       "step": 17710
     },
     {
       "epoch": 0.8509003601440577,
-      "grad_norm": 0.48820367455482483,
+      "grad_norm": 0.48874273896217346,
       "learning_rate": 2.9656978545267002e-06,
-      "loss": 0.9277,
+      "loss": 0.9282,
       "num_input_tokens_seen": 145162240,
       "step": 17720
     },
     {
       "epoch": 0.8513805522208884,
-      "grad_norm": 0.5144213438034058,
+      "grad_norm": 0.49839329719543457,
       "learning_rate": 2.947009767550718e-06,
-      "loss": 0.9099,
+      "loss": 0.9096,
       "num_input_tokens_seen": 145244160,
       "step": 17730
     },
     {
       "epoch": 0.8518607442977191,
-      "grad_norm": 0.5397959351539612,
+      "grad_norm": 0.5408912301063538,
       "learning_rate": 2.9283770590185696e-06,
-      "loss": 1.0301,
+      "loss": 1.0297,
       "num_input_tokens_seen": 145326080,
       "step": 17740
     },
     {
       "epoch": 0.8523409363745498,
-      "grad_norm": 0.8778235912322998,
+      "grad_norm": 0.8705450892448425,
       "learning_rate": 2.9097997757198516e-06,
-      "loss": 1.1378,
+      "loss": 1.1381,
       "num_input_tokens_seen": 145408000,
       "step": 17750
     },
     {
       "epoch": 0.8528211284513806,
-      "grad_norm": 0.4958626329898834,
+      "grad_norm": 0.4960118532180786,
       "learning_rate": 2.891277964304959e-06,
       "loss": 0.8219,
       "num_input_tokens_seen": 145489920,
@@ -14218,111 +14218,111 @@
     },
     {
       "epoch": 0.8533013205282113,
-      "grad_norm": 0.5561241507530212,
+      "grad_norm": 0.5603978037834167,
       "learning_rate": 2.8728116712850193e-06,
-      "loss": 0.8583,
+      "loss": 0.8588,
       "num_input_tokens_seen": 145571840,
       "step": 17770
     },
     {
       "epoch": 0.853781512605042,
-      "grad_norm": 0.5028048753738403,
+      "grad_norm": 0.5049429535865784,
       "learning_rate": 2.8544009430317153e-06,
-      "loss": 0.9945,
+      "loss": 0.9943,
       "num_input_tokens_seen": 145653760,
       "step": 17780
     },
     {
       "epoch": 0.8542617046818728,
-      "grad_norm": 0.4903087615966797,
+      "grad_norm": 0.49399763345718384,
       "learning_rate": 2.8360458257772228e-06,
-      "loss": 0.9056,
+      "loss": 0.9051,
       "num_input_tokens_seen": 145735680,
       "step": 17790
     },
     {
       "epoch": 0.8547418967587035,
-      "grad_norm": 0.4912867844104767,
+      "grad_norm": 0.4904550015926361,
       "learning_rate": 2.817746365614049e-06,
-      "loss": 0.8998,
+      "loss": 0.8999,
       "num_input_tokens_seen": 145817600,
       "step": 17800
     },
     {
       "epoch": 0.8552220888355342,
-      "grad_norm": 0.5079895853996277,
+      "grad_norm": 0.5089582800865173,
       "learning_rate": 2.7995026084949584e-06,
-      "loss": 0.931,
+      "loss": 0.9312,
       "num_input_tokens_seen": 145899520,
       "step": 17810
     },
     {
       "epoch": 0.855702280912365,
-      "grad_norm": 0.547659695148468,
+      "grad_norm": 0.540856659412384,
       "learning_rate": 2.781314600232815e-06,
-      "loss": 1.0527,
+      "loss": 1.0525,
       "num_input_tokens_seen": 145981440,
       "step": 17820
     },
     {
       "epoch": 0.8561824729891957,
-      "grad_norm": 0.4735318124294281,
+      "grad_norm": 0.4742536246776581,
       "learning_rate": 2.763182386500504e-06,
-      "loss": 0.8695,
+      "loss": 0.8697,
       "num_input_tokens_seen": 146063360,
       "step": 17830
     },
     {
       "epoch": 0.8566626650660264,
-      "grad_norm": 0.4886067509651184,
+      "grad_norm": 0.4883480966091156,
       "learning_rate": 2.745106012830806e-06,
-      "loss": 0.9396,
+      "loss": 0.9403,
       "num_input_tokens_seen": 146145280,
       "step": 17840
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.48998895287513733,
+      "grad_norm": 0.4900871515274048,
       "learning_rate": 2.7270855246162547e-06,
-      "loss": 1.0026,
+      "loss": 1.0035,
       "num_input_tokens_seen": 146227200,
       "step": 17850
     },
     {
       "epoch": 0.8576230492196879,
-      "grad_norm": 0.4987630844116211,
+      "grad_norm": 0.4983166456222534,
       "learning_rate": 2.7091209671090715e-06,
-      "loss": 0.9145,
+      "loss": 0.9146,
       "num_input_tokens_seen": 146309120,
       "step": 17860
     },
     {
       "epoch": 0.8581032412965186,
-      "grad_norm": 0.5029290318489075,
+      "grad_norm": 0.5034369826316833,
       "learning_rate": 2.6912123854210212e-06,
-      "loss": 0.8782,
+      "loss": 0.8785,
       "num_input_tokens_seen": 146391040,
       "step": 17870
     },
     {
       "epoch": 0.8585834333733493,
-      "grad_norm": 0.5902127027511597,
+      "grad_norm": 0.5906243324279785,
       "learning_rate": 2.673359824523297e-06,
-      "loss": 0.9368,
+      "loss": 0.9372,
       "num_input_tokens_seen": 146472960,
       "step": 17880
     },
     {
       "epoch": 0.8590636254501801,
-      "grad_norm": 0.6036652326583862,
+      "grad_norm": 0.6103810667991638,
       "learning_rate": 2.655563329246413e-06,
-      "loss": 0.946,
+      "loss": 0.9457,
       "num_input_tokens_seen": 146554880,
       "step": 17890
     },
     {
       "epoch": 0.8595438175270108,
-      "grad_norm": 0.4699823558330536,
+      "grad_norm": 0.469939261674881,
       "learning_rate": 2.637822944280116e-06,
       "loss": 0.8075,
       "num_input_tokens_seen": 146636800,
@@ -14330,7 +14330,7 @@
     },
     {
       "epoch": 0.8600240096038415,
-      "grad_norm": 0.5656408071517944,
+      "grad_norm": 0.567516028881073,
       "learning_rate": 2.6201387141732205e-06,
       "loss": 0.9612,
       "num_input_tokens_seen": 146718720,
@@ -14338,151 +14338,151 @@
     },
     {
       "epoch": 0.8605042016806723,
-      "grad_norm": 0.4844336211681366,
+      "grad_norm": 0.4764900803565979,
       "learning_rate": 2.6025106833335505e-06,
-      "loss": 1.0511,
+      "loss": 1.0508,
       "num_input_tokens_seen": 146800640,
       "step": 17920
     },
     {
       "epoch": 0.860984393757503,
-      "grad_norm": 0.7261025309562683,
+      "grad_norm": 0.7253336906433105,
       "learning_rate": 2.5849388960277997e-06,
-      "loss": 0.7721,
+      "loss": 0.7723,
       "num_input_tokens_seen": 146882560,
       "step": 17930
     },
     {
       "epoch": 0.8614645858343337,
-      "grad_norm": 0.5041067600250244,
+      "grad_norm": 0.5117282867431641,
       "learning_rate": 2.567423396381419e-06,
-      "loss": 0.7826,
+      "loss": 0.783,
       "num_input_tokens_seen": 146964480,
       "step": 17940
     },
     {
       "epoch": 0.8619447779111644,
-      "grad_norm": 0.28633078932762146,
+      "grad_norm": 0.28951549530029297,
       "learning_rate": 2.549964228378518e-06,
-      "loss": 0.7408,
+      "loss": 0.7405,
       "num_input_tokens_seen": 147046400,
       "step": 17950
     },
     {
       "epoch": 0.8624249699879952,
-      "grad_norm": 0.8320537209510803,
+      "grad_norm": 0.8366937637329102,
       "learning_rate": 2.532561435861755e-06,
-      "loss": 0.906,
+      "loss": 0.9065,
       "num_input_tokens_seen": 147128320,
       "step": 17960
     },
     {
       "epoch": 0.8629051620648259,
-      "grad_norm": 0.4853600561618805,
+      "grad_norm": 0.4839094579219818,
       "learning_rate": 2.515215062532206e-06,
-      "loss": 0.8907,
+      "loss": 0.8906,
       "num_input_tokens_seen": 147210240,
       "step": 17970
     },
     {
       "epoch": 0.8633853541416566,
-      "grad_norm": 0.5016003251075745,
+      "grad_norm": 0.5016359090805054,
       "learning_rate": 2.497925151949271e-06,
-      "loss": 0.9341,
+      "loss": 0.9354,
       "num_input_tokens_seen": 147292160,
       "step": 17980
     },
     {
       "epoch": 0.8638655462184874,
-      "grad_norm": 0.4873434603214264,
+      "grad_norm": 0.4866850674152374,
       "learning_rate": 2.4806917475305806e-06,
-      "loss": 0.9855,
+      "loss": 0.985,
       "num_input_tokens_seen": 147374080,
       "step": 17990
     },
     {
       "epoch": 0.8643457382953181,
-      "grad_norm": 0.4814073443412781,
+      "grad_norm": 0.4811359643936157,
       "learning_rate": 2.4635148925518577e-06,
-      "loss": 0.8596,
+      "loss": 0.8599,
       "num_input_tokens_seen": 147456000,
       "step": 18000
     },
     {
       "epoch": 0.8648259303721488,
-      "grad_norm": 0.4790041446685791,
+      "grad_norm": 0.476534903049469,
       "learning_rate": 2.4463946301468143e-06,
-      "loss": 0.8384,
+      "loss": 0.8387,
       "num_input_tokens_seen": 147537920,
       "step": 18010
     },
     {
       "epoch": 0.8653061224489796,
-      "grad_norm": 0.4911400377750397,
+      "grad_norm": 0.4927884340286255,
       "learning_rate": 2.4293310033070614e-06,
-      "loss": 0.9121,
+      "loss": 0.9122,
       "num_input_tokens_seen": 147619840,
       "step": 18020
     },
     {
       "epoch": 0.8657863145258103,
-      "grad_norm": 0.5386258363723755,
+      "grad_norm": 0.539406955242157,
       "learning_rate": 2.4123240548819955e-06,
-      "loss": 0.9519,
+      "loss": 0.9507,
       "num_input_tokens_seen": 147701760,
       "step": 18030
     },
     {
       "epoch": 0.866266506602641,
-      "grad_norm": 0.4828698933124542,
+      "grad_norm": 0.48301905393600464,
       "learning_rate": 2.3953738275786565e-06,
-      "loss": 0.7976,
+      "loss": 0.7972,
       "num_input_tokens_seen": 147783680,
       "step": 18040
     },
     {
       "epoch": 0.8667466986794717,
-      "grad_norm": 0.4941945970058441,
+      "grad_norm": 0.49446576833724976,
       "learning_rate": 2.3784803639616854e-06,
-      "loss": 0.8773,
+      "loss": 0.8777,
       "num_input_tokens_seen": 147865600,
       "step": 18050
     },
     {
       "epoch": 0.8672268907563025,
-      "grad_norm": 0.4941422939300537,
+      "grad_norm": 0.49398505687713623,
       "learning_rate": 2.361643706453151e-06,
-      "loss": 0.8026,
+      "loss": 0.8028,
       "num_input_tokens_seen": 147947520,
       "step": 18060
     },
     {
       "epoch": 0.8677070828331332,
-      "grad_norm": 0.5104964971542358,
+      "grad_norm": 0.5100568532943726,
       "learning_rate": 2.3448638973324833e-06,
-      "loss": 0.8714,
+      "loss": 0.872,
       "num_input_tokens_seen": 148029440,
       "step": 18070
     },
     {
       "epoch": 0.868187274909964,
-      "grad_norm": 0.4910539984703064,
+      "grad_norm": 0.4940645694732666,
       "learning_rate": 2.328140978736365e-06,
-      "loss": 1.224,
+      "loss": 1.2244,
       "num_input_tokens_seen": 148111360,
       "step": 18080
     },
     {
       "epoch": 0.8686674669867948,
-      "grad_norm": 0.6451656222343445,
+      "grad_norm": 0.6448796391487122,
       "learning_rate": 2.311474992658613e-06,
-      "loss": 0.8576,
+      "loss": 0.8573,
       "num_input_tokens_seen": 148193280,
       "step": 18090
     },
     {
       "epoch": 0.8691476590636255,
-      "grad_norm": 0.47667035460472107,
+      "grad_norm": 0.4770622253417969,
       "learning_rate": 2.29486598095007e-06,
       "loss": 0.7746,
       "num_input_tokens_seen": 148275200,
@@ -14490,103 +14490,103 @@
     },
     {
       "epoch": 0.8696278511404562,
-      "grad_norm": 0.4710484743118286,
+      "grad_norm": 0.47341033816337585,
       "learning_rate": 2.278313985318517e-06,
-      "loss": 0.9388,
+      "loss": 0.9389,
       "num_input_tokens_seen": 148357120,
       "step": 18110
     },
     {
       "epoch": 0.870108043217287,
-      "grad_norm": 0.5291144251823425,
+      "grad_norm": 0.5365563631057739,
       "learning_rate": 2.261819047328562e-06,
-      "loss": 0.8536,
+      "loss": 0.8541,
       "num_input_tokens_seen": 148439040,
       "step": 18120
     },
     {
       "epoch": 0.8705882352941177,
-      "grad_norm": 0.5111077427864075,
+      "grad_norm": 0.5126563906669617,
       "learning_rate": 2.2453812084015175e-06,
-      "loss": 0.9051,
+      "loss": 0.9053,
       "num_input_tokens_seen": 148520960,
       "step": 18130
     },
     {
       "epoch": 0.8710684273709484,
-      "grad_norm": 0.4752892553806305,
+      "grad_norm": 0.4737769365310669,
       "learning_rate": 2.2290005098153296e-06,
-      "loss": 1.2451,
+      "loss": 1.2452,
       "num_input_tokens_seen": 148602880,
       "step": 18140
     },
     {
       "epoch": 0.8715486194477791,
-      "grad_norm": 0.501112163066864,
+      "grad_norm": 0.5032677054405212,
       "learning_rate": 2.212676992704435e-06,
-      "loss": 0.9271,
+      "loss": 0.9268,
       "num_input_tokens_seen": 148684800,
       "step": 18150
     },
     {
       "epoch": 0.8720288115246099,
-      "grad_norm": 0.46904444694519043,
+      "grad_norm": 0.4705372750759125,
       "learning_rate": 2.1964106980597034e-06,
-      "loss": 0.6553,
+      "loss": 0.6561,
       "num_input_tokens_seen": 148766720,
       "step": 18160
     },
     {
       "epoch": 0.8725090036014406,
-      "grad_norm": 0.5374292135238647,
+      "grad_norm": 0.5352415442466736,
       "learning_rate": 2.1802016667282847e-06,
-      "loss": 0.8742,
+      "loss": 0.8747,
       "num_input_tokens_seen": 148848640,
       "step": 18170
     },
     {
       "epoch": 0.8729891956782713,
-      "grad_norm": 0.5054563283920288,
+      "grad_norm": 0.5062108635902405,
       "learning_rate": 2.1640499394135595e-06,
-      "loss": 0.8704,
+      "loss": 0.8711,
       "num_input_tokens_seen": 148930560,
       "step": 18180
     },
     {
       "epoch": 0.8734693877551021,
-      "grad_norm": 0.48471224308013916,
+      "grad_norm": 0.48544827103614807,
       "learning_rate": 2.1479555566749825e-06,
-      "loss": 0.9272,
+      "loss": 0.9279,
       "num_input_tokens_seen": 149012480,
       "step": 18190
     },
     {
       "epoch": 0.8739495798319328,
-      "grad_norm": 0.5023584961891174,
+      "grad_norm": 0.5034095048904419,
       "learning_rate": 2.131918558928023e-06,
-      "loss": 0.9053,
+      "loss": 0.9047,
       "num_input_tokens_seen": 149094400,
       "step": 18200
     },
     {
       "epoch": 0.8744297719087635,
-      "grad_norm": 0.46324488520622253,
+      "grad_norm": 0.4514198899269104,
       "learning_rate": 2.1159389864440495e-06,
-      "loss": 0.8871,
+      "loss": 0.8876,
       "num_input_tokens_seen": 149176320,
       "step": 18210
     },
     {
       "epoch": 0.8749099639855943,
-      "grad_norm": 0.5278415083885193,
+      "grad_norm": 0.5257483124732971,
       "learning_rate": 2.100016879350214e-06,
-      "loss": 0.7591,
+      "loss": 0.7592,
       "num_input_tokens_seen": 149258240,
       "step": 18220
     },
     {
       "epoch": 0.875390156062425,
-      "grad_norm": 0.4872019290924072,
+      "grad_norm": 0.48762884736061096,
       "learning_rate": 2.0841522776293725e-06,
       "loss": 0.8473,
       "num_input_tokens_seen": 149340160,
@@ -14594,55 +14594,55 @@
     },
     {
       "epoch": 0.8758703481392557,
-      "grad_norm": 0.5031896233558655,
+      "grad_norm": 0.5084396600723267,
       "learning_rate": 2.0683452211199854e-06,
-      "loss": 1.118,
+      "loss": 1.1182,
       "num_input_tokens_seen": 149422080,
       "step": 18240
     },
     {
       "epoch": 0.8763505402160864,
-      "grad_norm": 0.48833465576171875,
+      "grad_norm": 0.48876795172691345,
       "learning_rate": 2.052595749515987e-06,
-      "loss": 0.8301,
+      "loss": 0.8305,
       "num_input_tokens_seen": 149504000,
       "step": 18250
     },
     {
       "epoch": 0.8768307322929172,
-      "grad_norm": 0.5221076607704163,
+      "grad_norm": 0.5128353238105774,
       "learning_rate": 2.0369039023667215e-06,
-      "loss": 0.9347,
+      "loss": 0.9346,
       "num_input_tokens_seen": 149585920,
       "step": 18260
     },
     {
       "epoch": 0.8773109243697479,
-      "grad_norm": 0.476140558719635,
+      "grad_norm": 0.4771345555782318,
       "learning_rate": 2.0212697190768263e-06,
-      "loss": 0.9084,
+      "loss": 0.9082,
       "num_input_tokens_seen": 149667840,
       "step": 18270
     },
     {
       "epoch": 0.8777911164465786,
-      "grad_norm": 0.49334678053855896,
+      "grad_norm": 0.4934534430503845,
       "learning_rate": 2.0056932389061338e-06,
-      "loss": 0.8173,
+      "loss": 0.8169,
       "num_input_tokens_seen": 149749760,
       "step": 18280
     },
     {
       "epoch": 0.8782713085234094,
-      "grad_norm": 0.8516491055488586,
+      "grad_norm": 0.8390567302703857,
       "learning_rate": 1.9901745009695773e-06,
-      "loss": 0.9435,
+      "loss": 0.9433,
       "num_input_tokens_seen": 149831680,
       "step": 18290
     },
     {
       "epoch": 0.8787515006002401,
-      "grad_norm": 0.5442166328430176,
+      "grad_norm": 0.5534536242485046,
       "learning_rate": 1.9747135442370946e-06,
       "loss": 0.8976,
       "num_input_tokens_seen": 149913600,
@@ -14650,55 +14650,55 @@
     },
     {
       "epoch": 0.8792316926770708,
-      "grad_norm": 0.3513561189174652,
+      "grad_norm": 0.35009366273880005,
       "learning_rate": 1.9593104075335158e-06,
-      "loss": 0.762,
+      "loss": 0.7622,
       "num_input_tokens_seen": 149995520,
       "step": 18310
     },
     {
       "epoch": 0.8797118847539016,
-      "grad_norm": 0.48064127564430237,
+      "grad_norm": 0.47988003492355347,
       "learning_rate": 1.943965129538483e-06,
-      "loss": 0.8791,
+      "loss": 0.8789,
       "num_input_tokens_seen": 150077440,
       "step": 18320
     },
     {
       "epoch": 0.8801920768307323,
-      "grad_norm": 0.48252299427986145,
+      "grad_norm": 0.48413175344467163,
       "learning_rate": 1.9286777487863477e-06,
-      "loss": 0.8332,
+      "loss": 0.8331,
       "num_input_tokens_seen": 150159360,
       "step": 18330
     },
     {
       "epoch": 0.880672268907563,
-      "grad_norm": 0.45519593358039856,
+      "grad_norm": 0.46263861656188965,
       "learning_rate": 1.913448303666071e-06,
-      "loss": 0.999,
+      "loss": 0.9994,
       "num_input_tokens_seen": 150241280,
       "step": 18340
     },
     {
       "epoch": 0.8811524609843937,
-      "grad_norm": 0.47915107011795044,
+      "grad_norm": 0.4799620509147644,
       "learning_rate": 1.8982768324211197e-06,
-      "loss": 0.9294,
+      "loss": 0.9293,
       "num_input_tokens_seen": 150323200,
       "step": 18350
     },
     {
       "epoch": 0.8816326530612245,
-      "grad_norm": 0.5818130373954773,
+      "grad_norm": 0.583249568939209,
       "learning_rate": 1.8831633731493963e-06,
-      "loss": 0.9785,
+      "loss": 0.9782,
       "num_input_tokens_seen": 150405120,
       "step": 18360
     },
     {
       "epoch": 0.8821128451380552,
-      "grad_norm": 0.4887988865375519,
+      "grad_norm": 0.4899525046348572,
       "learning_rate": 1.8681079638031062e-06,
       "loss": 0.8177,
       "num_input_tokens_seen": 150487040,
@@ -14706,191 +14706,191 @@
     },
     {
       "epoch": 0.8825930372148859,
-      "grad_norm": 0.6993192434310913,
+      "grad_norm": 0.696358859539032,
       "learning_rate": 1.8531106421887017e-06,
-      "loss": 0.8862,
+      "loss": 0.8863,
       "num_input_tokens_seen": 150568960,
       "step": 18380
     },
     {
       "epoch": 0.8830732292917167,
-      "grad_norm": 0.486240416765213,
+      "grad_norm": 0.4858860969543457,
       "learning_rate": 1.8381714459667603e-06,
-      "loss": 0.7221,
+      "loss": 0.7228,
       "num_input_tokens_seen": 150650880,
       "step": 18390
     },
     {
       "epoch": 0.8835534213685474,
-      "grad_norm": 0.7017090916633606,
+      "grad_norm": 0.695509135723114,
       "learning_rate": 1.823290412651893e-06,
-      "loss": 1.0007,
+      "loss": 1.0011,
       "num_input_tokens_seen": 150732800,
       "step": 18400
     },
     {
       "epoch": 0.8840336134453781,
-      "grad_norm": 0.46545252203941345,
+      "grad_norm": 0.46307533979415894,
       "learning_rate": 1.8084675796126576e-06,
-      "loss": 1.1891,
+      "loss": 1.1894,
       "num_input_tokens_seen": 150814720,
       "step": 18410
     },
     {
       "epoch": 0.8845138055222089,
-      "grad_norm": 0.5387531518936157,
+      "grad_norm": 0.5278788208961487,
       "learning_rate": 1.7937029840714715e-06,
-      "loss": 1.0123,
+      "loss": 1.0122,
       "num_input_tokens_seen": 150896640,
       "step": 18420
     },
     {
       "epoch": 0.8849939975990396,
-      "grad_norm": 0.4813983142375946,
+      "grad_norm": 0.48216572403907776,
       "learning_rate": 1.778996663104493e-06,
-      "loss": 0.8358,
+      "loss": 0.836,
       "num_input_tokens_seen": 150978560,
       "step": 18430
     },
     {
       "epoch": 0.8854741896758703,
-      "grad_norm": 0.4829707443714142,
+      "grad_norm": 0.4845614433288574,
       "learning_rate": 1.7643486536415537e-06,
-      "loss": 0.9354,
+      "loss": 0.9355,
       "num_input_tokens_seen": 151060480,
       "step": 18440
     },
     {
       "epoch": 0.885954381752701,
-      "grad_norm": 0.4897553622722626,
+      "grad_norm": 0.4892086088657379,
       "learning_rate": 1.7497589924660552e-06,
-      "loss": 0.8802,
+      "loss": 0.8795,
       "num_input_tokens_seen": 151142400,
       "step": 18450
     },
     {
       "epoch": 0.8864345738295318,
-      "grad_norm": 0.826809823513031,
+      "grad_norm": 0.8257756233215332,
       "learning_rate": 1.7352277162148712e-06,
-      "loss": 0.8142,
+      "loss": 0.814,
       "num_input_tokens_seen": 151224320,
       "step": 18460
     },
     {
       "epoch": 0.8869147659063625,
-      "grad_norm": 0.4941288232803345,
+      "grad_norm": 0.49343836307525635,
       "learning_rate": 1.7207548613782709e-06,
-      "loss": 0.8752,
+      "loss": 0.8757,
       "num_input_tokens_seen": 151306240,
       "step": 18470
     },
     {
       "epoch": 0.8873949579831932,
-      "grad_norm": 0.4736701548099518,
+      "grad_norm": 0.4691942632198334,
       "learning_rate": 1.7063404642998186e-06,
-      "loss": 1.0158,
+      "loss": 1.0152,
       "num_input_tokens_seen": 151388160,
       "step": 18480
     },
     {
       "epoch": 0.887875150060024,
-      "grad_norm": 0.5565969944000244,
+      "grad_norm": 0.549484133720398,
       "learning_rate": 1.6919845611762714e-06,
-      "loss": 0.9463,
+      "loss": 0.9454,
       "num_input_tokens_seen": 151470080,
       "step": 18490
     },
     {
       "epoch": 0.8883553421368547,
-      "grad_norm": 0.523489773273468,
+      "grad_norm": 0.522466242313385,
       "learning_rate": 1.6776871880575084e-06,
-      "loss": 1.0312,
+      "loss": 1.0313,
       "num_input_tokens_seen": 151552000,
       "step": 18500
     },
     {
       "epoch": 0.8888355342136854,
-      "grad_norm": 0.5504932403564453,
+      "grad_norm": 0.5475363731384277,
       "learning_rate": 1.663448380846433e-06,
-      "loss": 0.841,
+      "loss": 0.8417,
       "num_input_tokens_seen": 151633920,
       "step": 18510
     },
     {
       "epoch": 0.8893157262905163,
-      "grad_norm": 2.449960470199585,
+      "grad_norm": 2.4417154788970947,
       "learning_rate": 1.649268175298868e-06,
-      "loss": 0.819,
+      "loss": 0.8187,
       "num_input_tokens_seen": 151715840,
       "step": 18520
     },
     {
       "epoch": 0.889795918367347,
-      "grad_norm": 0.6405714750289917,
+      "grad_norm": 0.643785297870636,
       "learning_rate": 1.6351466070234882e-06,
-      "loss": 1.1654,
+      "loss": 1.1659,
       "num_input_tokens_seen": 151797760,
       "step": 18530
     },
     {
       "epoch": 0.8902761104441776,
-      "grad_norm": 0.5734189748764038,
+      "grad_norm": 0.5771599411964417,
       "learning_rate": 1.6210837114817272e-06,
-      "loss": 0.8883,
+      "loss": 0.8884,
       "num_input_tokens_seen": 151879680,
       "step": 18540
     },
     {
       "epoch": 0.8907563025210085,
-      "grad_norm": 0.4893549382686615,
+      "grad_norm": 0.490934282541275,
       "learning_rate": 1.6070795239876618e-06,
-      "loss": 0.785,
+      "loss": 0.7856,
       "num_input_tokens_seen": 151961600,
       "step": 18550
     },
     {
       "epoch": 0.8912364945978392,
-      "grad_norm": 0.48380154371261597,
+      "grad_norm": 0.4851411283016205,
       "learning_rate": 1.5931340797079613e-06,
-      "loss": 0.7937,
+      "loss": 0.7931,
       "num_input_tokens_seen": 152043520,
       "step": 18560
     },
     {
       "epoch": 0.8917166866746699,
-      "grad_norm": 0.44979286193847656,
+      "grad_norm": 0.4503311514854431,
       "learning_rate": 1.5792474136617858e-06,
-      "loss": 0.8338,
+      "loss": 0.8341,
       "num_input_tokens_seen": 152125440,
       "step": 18570
     },
     {
       "epoch": 0.8921968787515006,
-      "grad_norm": 0.48284098505973816,
+      "grad_norm": 0.4835943281650543,
       "learning_rate": 1.5654195607206712e-06,
-      "loss": 0.8318,
+      "loss": 0.8317,
       "num_input_tokens_seen": 152207360,
       "step": 18580
     },
     {
       "epoch": 0.8926770708283314,
-      "grad_norm": 0.5251838564872742,
+      "grad_norm": 0.5173797011375427,
       "learning_rate": 1.5516505556084888e-06,
-      "loss": 0.8587,
+      "loss": 0.859,
       "num_input_tokens_seen": 152289280,
       "step": 18590
     },
     {
       "epoch": 0.8931572629051621,
-      "grad_norm": 0.6006345152854919,
+      "grad_norm": 0.5888597369194031,
       "learning_rate": 1.5379404329013246e-06,
-      "loss": 1.0141,
+      "loss": 1.0143,
       "num_input_tokens_seen": 152371200,
       "step": 18600
     },
     {
       "epoch": 0.8936374549819928,
-      "grad_norm": 0.48716527223587036,
+      "grad_norm": 0.46301090717315674,
       "learning_rate": 1.5242892270273951e-06,
       "loss": 0.8301,
       "num_input_tokens_seen": 152453120,
@@ -14898,87 +14898,87 @@
     },
     {
       "epoch": 0.8941176470588236,
-      "grad_norm": 0.4787551462650299,
+      "grad_norm": 0.479010671377182,
       "learning_rate": 1.5106969722669812e-06,
-      "loss": 0.9144,
+      "loss": 0.9141,
       "num_input_tokens_seen": 152535040,
       "step": 18620
     },
     {
       "epoch": 0.8945978391356543,
-      "grad_norm": 0.48084962368011475,
+      "grad_norm": 0.47869157791137695,
       "learning_rate": 1.4971637027523106e-06,
-      "loss": 1.1305,
+      "loss": 1.1308,
       "num_input_tokens_seen": 152616960,
       "step": 18630
     },
     {
       "epoch": 0.895078031212485,
-      "grad_norm": 0.5110411643981934,
+      "grad_norm": 0.5103761553764343,
       "learning_rate": 1.4836894524675126e-06,
-      "loss": 0.9489,
+      "loss": 0.9492,
       "num_input_tokens_seen": 152698880,
       "step": 18640
     },
     {
       "epoch": 0.8955582232893158,
-      "grad_norm": 0.5295985341072083,
+      "grad_norm": 0.5287526249885559,
       "learning_rate": 1.4702742552484884e-06,
-      "loss": 1.1124,
+      "loss": 1.1131,
       "num_input_tokens_seen": 152780800,
       "step": 18650
     },
     {
       "epoch": 0.8960384153661465,
-      "grad_norm": 0.47781914472579956,
+      "grad_norm": 0.4778541922569275,
       "learning_rate": 1.4569181447828623e-06,
-      "loss": 0.836,
+      "loss": 0.8355,
       "num_input_tokens_seen": 152862720,
       "step": 18660
     },
     {
       "epoch": 0.8965186074429772,
-      "grad_norm": 0.523032546043396,
+      "grad_norm": 0.5224360823631287,
       "learning_rate": 1.4436211546098782e-06,
-      "loss": 1.0544,
+      "loss": 1.055,
       "num_input_tokens_seen": 152944640,
       "step": 18670
     },
     {
       "epoch": 0.8969987995198079,
-      "grad_norm": 0.5448631644248962,
+      "grad_norm": 0.5333329439163208,
       "learning_rate": 1.430383318120318e-06,
-      "loss": 0.9155,
+      "loss": 0.9159,
       "num_input_tokens_seen": 153026560,
       "step": 18680
     },
     {
       "epoch": 0.8974789915966387,
-      "grad_norm": 0.4606798589229584,
+      "grad_norm": 0.4066697061061859,
       "learning_rate": 1.4172046685564212e-06,
-      "loss": 0.862,
+      "loss": 0.8634,
       "num_input_tokens_seen": 153108480,
       "step": 18690
     },
     {
       "epoch": 0.8979591836734694,
-      "grad_norm": 0.44669726490974426,
+      "grad_norm": 0.4478163719177246,
       "learning_rate": 1.4040852390118042e-06,
-      "loss": 0.9023,
+      "loss": 0.9027,
       "num_input_tokens_seen": 153190400,
       "step": 18700
     },
     {
       "epoch": 0.8984393757503001,
-      "grad_norm": 0.4901637136936188,
+      "grad_norm": 0.48882734775543213,
       "learning_rate": 1.3910250624313642e-06,
-      "loss": 1.0013,
+      "loss": 1.0014,
       "num_input_tokens_seen": 153272320,
       "step": 18710
     },
     {
       "epoch": 0.8989195678271309,
-      "grad_norm": 0.4382149875164032,
+      "grad_norm": 0.4382336139678955,
       "learning_rate": 1.3780241716112057e-06,
       "loss": 0.8953,
       "num_input_tokens_seen": 153354240,
@@ -14986,159 +14986,159 @@
     },
     {
       "epoch": 0.8993997599039616,
-      "grad_norm": 0.5043530464172363,
+      "grad_norm": 0.5193009972572327,
       "learning_rate": 1.3650825991985722e-06,
-      "loss": 0.8403,
+      "loss": 0.8398,
       "num_input_tokens_seen": 153436160,
       "step": 18730
     },
     {
       "epoch": 0.8998799519807923,
-      "grad_norm": 0.8724451065063477,
+      "grad_norm": 0.8743091225624084,
       "learning_rate": 1.3522003776917285e-06,
-      "loss": 0.7954,
+      "loss": 0.7956,
       "num_input_tokens_seen": 153518080,
       "step": 18740
     },
     {
       "epoch": 0.9003601440576231,
-      "grad_norm": 0.4843882620334625,
+      "grad_norm": 0.4847816228866577,
       "learning_rate": 1.3393775394399123e-06,
-      "loss": 0.8046,
+      "loss": 0.8044,
       "num_input_tokens_seen": 153600000,
       "step": 18750
     },
     {
       "epoch": 0.9008403361344538,
-      "grad_norm": 0.5170858502388,
+      "grad_norm": 0.4940853714942932,
       "learning_rate": 1.326614116643246e-06,
-      "loss": 0.712,
+      "loss": 0.7121,
       "num_input_tokens_seen": 153681920,
       "step": 18760
     },
     {
       "epoch": 0.9013205282112845,
-      "grad_norm": 0.45620083808898926,
+      "grad_norm": 0.45956119894981384,
       "learning_rate": 1.3139101413526339e-06,
-      "loss": 0.8778,
+      "loss": 0.8775,
       "num_input_tokens_seen": 153763840,
       "step": 18770
     },
     {
       "epoch": 0.9018007202881152,
-      "grad_norm": 0.710594654083252,
+      "grad_norm": 0.7106777429580688,
       "learning_rate": 1.3012656454697125e-06,
-      "loss": 1.0348,
+      "loss": 1.0345,
       "num_input_tokens_seen": 153845760,
       "step": 18780
     },
     {
       "epoch": 0.902280912364946,
-      "grad_norm": 0.5861712098121643,
+      "grad_norm": 0.5692814588546753,
       "learning_rate": 1.2886806607467578e-06,
-      "loss": 0.9589,
+      "loss": 0.9592,
       "num_input_tokens_seen": 153927680,
       "step": 18790
     },
     {
       "epoch": 0.9027611044417767,
-      "grad_norm": 0.48728203773498535,
+      "grad_norm": 0.4884362816810608,
       "learning_rate": 1.2761552187865899e-06,
-      "loss": 0.9268,
+      "loss": 0.927,
       "num_input_tokens_seen": 154009600,
       "step": 18800
     },
     {
       "epoch": 0.9032412965186074,
-      "grad_norm": 0.49788719415664673,
+      "grad_norm": 0.4982493817806244,
       "learning_rate": 1.2636893510425186e-06,
-      "loss": 0.8947,
+      "loss": 0.8942,
       "num_input_tokens_seen": 154091520,
       "step": 18810
     },
     {
       "epoch": 0.9037214885954382,
-      "grad_norm": 0.5064159035682678,
+      "grad_norm": 0.5100504159927368,
       "learning_rate": 1.2512830888182531e-06,
-      "loss": 0.9701,
+      "loss": 0.97,
       "num_input_tokens_seen": 154173440,
       "step": 18820
     },
     {
       "epoch": 0.9042016806722689,
-      "grad_norm": 0.5245488286018372,
+      "grad_norm": 0.5252644419670105,
       "learning_rate": 1.23893646326782e-06,
-      "loss": 0.9968,
+      "loss": 0.9964,
       "num_input_tokens_seen": 154255360,
       "step": 18830
     },
     {
       "epoch": 0.9046818727490996,
-      "grad_norm": 0.49203112721443176,
+      "grad_norm": 0.4914807975292206,
       "learning_rate": 1.2266495053954913e-06,
-      "loss": 0.7056,
+      "loss": 0.7055,
       "num_input_tokens_seen": 154337280,
       "step": 18840
     },
     {
       "epoch": 0.9051620648259304,
-      "grad_norm": 0.47168827056884766,
+      "grad_norm": 0.47101646661758423,
       "learning_rate": 1.2144222460557074e-06,
-      "loss": 0.8708,
+      "loss": 0.8709,
       "num_input_tokens_seen": 154419200,
       "step": 18850
     },
     {
       "epoch": 0.9056422569027611,
-      "grad_norm": 0.5028705596923828,
+      "grad_norm": 0.502715528011322,
       "learning_rate": 1.2022547159529911e-06,
-      "loss": 1.0057,
+      "loss": 1.0051,
       "num_input_tokens_seen": 154501120,
       "step": 18860
     },
     {
       "epoch": 0.9061224489795918,
-      "grad_norm": 0.4962250888347626,
+      "grad_norm": 0.4965539276599884,
       "learning_rate": 1.190146945641879e-06,
-      "loss": 0.8827,
+      "loss": 0.8824,
       "num_input_tokens_seen": 154583040,
       "step": 18870
     },
     {
       "epoch": 0.9066026410564225,
-      "grad_norm": 0.5061412453651428,
+      "grad_norm": 0.5059521794319153,
       "learning_rate": 1.1780989655268415e-06,
-      "loss": 0.9975,
+      "loss": 0.997,
       "num_input_tokens_seen": 154664960,
       "step": 18880
     },
     {
       "epoch": 0.9070828331332533,
-      "grad_norm": 0.4892423152923584,
+      "grad_norm": 0.48782065510749817,
       "learning_rate": 1.1661108058622082e-06,
-      "loss": 0.8246,
+      "loss": 0.824,
       "num_input_tokens_seen": 154746880,
       "step": 18890
     },
     {
       "epoch": 0.907563025210084,
-      "grad_norm": 0.48785221576690674,
+      "grad_norm": 0.49019843339920044,
       "learning_rate": 1.154182496752082e-06,
-      "loss": 0.7727,
+      "loss": 0.7726,
       "num_input_tokens_seen": 154828800,
       "step": 18900
     },
     {
       "epoch": 0.9080432172869147,
-      "grad_norm": 0.5420461297035217,
+      "grad_norm": 0.5387908816337585,
       "learning_rate": 1.142314068150288e-06,
-      "loss": 1.0012,
+      "loss": 1.003,
       "num_input_tokens_seen": 154910720,
       "step": 18910
     },
     {
       "epoch": 0.9085234093637455,
-      "grad_norm": 0.4156251549720764,
+      "grad_norm": 0.4145044684410095,
       "learning_rate": 1.1305055498602584e-06,
       "loss": 0.8584,
       "num_input_tokens_seen": 154992640,
@@ -15146,103 +15146,103 @@
     },
     {
       "epoch": 0.9090036014405762,
-      "grad_norm": 0.4756897985935211,
+      "grad_norm": 0.44272181391716003,
       "learning_rate": 1.1187569715350066e-06,
-      "loss": 0.8211,
+      "loss": 0.8205,
       "num_input_tokens_seen": 155074560,
       "step": 18930
     },
     {
       "epoch": 0.9094837935174069,
-      "grad_norm": 0.47108086943626404,
+      "grad_norm": 0.4699978828430176,
       "learning_rate": 1.1070683626770162e-06,
-      "loss": 0.7993,
+      "loss": 0.7995,
       "num_input_tokens_seen": 155156480,
       "step": 18940
     },
     {
       "epoch": 0.9099639855942377,
-      "grad_norm": 0.4969809949398041,
+      "grad_norm": 0.49818235635757446,
       "learning_rate": 1.0954397526381694e-06,
-      "loss": 0.88,
+      "loss": 0.8801,
       "num_input_tokens_seen": 155238400,
       "step": 18950
     },
     {
       "epoch": 0.9104441776710684,
-      "grad_norm": 1.2800873517990112,
+      "grad_norm": 1.2911748886108398,
       "learning_rate": 1.0838711706196992e-06,
-      "loss": 0.9003,
+      "loss": 0.9004,
       "num_input_tokens_seen": 155320320,
       "step": 18960
     },
     {
       "epoch": 0.9109243697478991,
-      "grad_norm": 0.5365675687789917,
+      "grad_norm": 0.5381184220314026,
       "learning_rate": 1.0723626456720925e-06,
-      "loss": 0.9935,
+      "loss": 0.9954,
       "num_input_tokens_seen": 155402240,
       "step": 18970
     },
     {
       "epoch": 0.9114045618247298,
-      "grad_norm": 0.516943097114563,
+      "grad_norm": 0.5187402963638306,
       "learning_rate": 1.0609142066950157e-06,
-      "loss": 1.0318,
+      "loss": 1.032,
       "num_input_tokens_seen": 155484160,
       "step": 18980
     },
     {
       "epoch": 0.9118847539015607,
-      "grad_norm": 0.5084674954414368,
+      "grad_norm": 0.5067307949066162,
       "learning_rate": 1.0495258824372578e-06,
-      "loss": 0.837,
+      "loss": 0.8363,
       "num_input_tokens_seen": 155566080,
       "step": 18990
     },
     {
       "epoch": 0.9123649459783914,
-      "grad_norm": 0.5022624135017395,
+      "grad_norm": 0.5005121827125549,
       "learning_rate": 1.0381977014966543e-06,
-      "loss": 0.7732,
+      "loss": 0.7728,
       "num_input_tokens_seen": 155648000,
       "step": 19000
     },
     {
       "epoch": 0.912845138055222,
-      "grad_norm": 0.47421613335609436,
+      "grad_norm": 0.4743783473968506,
       "learning_rate": 1.0269296923199972e-06,
-      "loss": 1.0176,
+      "loss": 1.018,
       "num_input_tokens_seen": 155729920,
       "step": 19010
     },
     {
       "epoch": 0.9133253301320529,
-      "grad_norm": 0.4957895278930664,
+      "grad_norm": 0.49608567357063293,
       "learning_rate": 1.0157218832029969e-06,
-      "loss": 1.012,
+      "loss": 1.0119,
       "num_input_tokens_seen": 155811840,
       "step": 19020
     },
     {
       "epoch": 0.9138055222088836,
-      "grad_norm": 0.5938453078269958,
+      "grad_norm": 0.5876844525337219,
       "learning_rate": 1.0045743022901787e-06,
-      "loss": 1.009,
+      "loss": 1.0094,
       "num_input_tokens_seen": 155893760,
       "step": 19030
     },
     {
       "epoch": 0.9142857142857143,
-      "grad_norm": 0.46902185678482056,
+      "grad_norm": 0.4514719247817993,
       "learning_rate": 9.934869775748258e-07,
-      "loss": 0.7293,
+      "loss": 0.7286,
       "num_input_tokens_seen": 155975680,
       "step": 19040
     },
     {
       "epoch": 0.9147659063625451,
-      "grad_norm": 0.4982973337173462,
+      "grad_norm": 0.4981219470500946,
       "learning_rate": 9.824599368989163e-07,
       "loss": 0.8924,
       "num_input_tokens_seen": 156057600,
@@ -15250,103 +15250,103 @@
     },
     {
       "epoch": 0.9152460984393758,
-      "grad_norm": 0.47380825877189636,
+      "grad_norm": 0.4736618101596832,
       "learning_rate": 9.714932079530476e-07,
-      "loss": 0.8136,
+      "loss": 0.814,
       "num_input_tokens_seen": 156139520,
       "step": 19060
     },
     {
       "epoch": 0.9157262905162065,
-      "grad_norm": 0.43160074949264526,
+      "grad_norm": 0.4343636929988861,
       "learning_rate": 9.60586818276349e-07,
-      "loss": 0.7444,
+      "loss": 0.7436,
       "num_input_tokens_seen": 156221440,
       "step": 19070
     },
     {
       "epoch": 0.9162064825930372,
-      "grad_norm": 0.5274019837379456,
+      "grad_norm": 0.37545332312583923,
       "learning_rate": 9.497407952564485e-07,
-      "loss": 0.7502,
+      "loss": 0.7512,
       "num_input_tokens_seen": 156303360,
       "step": 19080
     },
     {
       "epoch": 0.916686674669868,
-      "grad_norm": 0.5028293132781982,
+      "grad_norm": 0.5027362108230591,
       "learning_rate": 9.389551661293683e-07,
-      "loss": 0.9223,
+      "loss": 0.9225,
       "num_input_tokens_seen": 156385280,
       "step": 19090
     },
     {
       "epoch": 0.9171668667466987,
-      "grad_norm": 0.4965846538543701,
+      "grad_norm": 0.4961453676223755,
       "learning_rate": 9.282299579794789e-07,
-      "loss": 0.8431,
+      "loss": 0.8426,
       "num_input_tokens_seen": 156467200,
       "step": 19100
     },
     {
       "epoch": 0.9176470588235294,
-      "grad_norm": 1.1068954467773438,
+      "grad_norm": 1.0757991075515747,
       "learning_rate": 9.175651977394284e-07,
-      "loss": 0.8672,
+      "loss": 0.867,
       "num_input_tokens_seen": 156549120,
       "step": 19110
     },
     {
       "epoch": 0.9181272509003602,
-      "grad_norm": 0.4944973886013031,
+      "grad_norm": 0.49529650807380676,
       "learning_rate": 9.069609121900663e-07,
-      "loss": 0.8328,
+      "loss": 0.8332,
       "num_input_tokens_seen": 156631040,
       "step": 19120
     },
     {
       "epoch": 0.9186074429771909,
-      "grad_norm": 0.47309666872024536,
+      "grad_norm": 0.4741530120372772,
       "learning_rate": 8.964171279603778e-07,
-      "loss": 0.7688,
+      "loss": 0.7689,
       "num_input_tokens_seen": 156712960,
       "step": 19130
     },
     {
       "epoch": 0.9190876350540216,
-      "grad_norm": 0.4812490940093994,
+      "grad_norm": 0.48308661580085754,
       "learning_rate": 8.859338715274279e-07,
-      "loss": 0.7694,
+      "loss": 0.7697,
       "num_input_tokens_seen": 156794880,
       "step": 19140
     },
     {
       "epoch": 0.9195678271308524,
-      "grad_norm": 0.7439998984336853,
+      "grad_norm": 0.7624333500862122,
       "learning_rate": 8.755111692162837e-07,
-      "loss": 0.9128,
+      "loss": 0.9131,
       "num_input_tokens_seen": 156876800,
       "step": 19150
     },
     {
       "epoch": 0.9200480192076831,
-      "grad_norm": 1.6029683351516724,
+      "grad_norm": 1.5886449813842773,
       "learning_rate": 8.651490471999424e-07,
-      "loss": 0.9739,
+      "loss": 0.9745,
       "num_input_tokens_seen": 156958720,
       "step": 19160
     },
     {
       "epoch": 0.9205282112845138,
-      "grad_norm": 0.5646688938140869,
+      "grad_norm": 0.569049596786499,
       "learning_rate": 8.548475314992949e-07,
-      "loss": 0.9776,
+      "loss": 0.9774,
       "num_input_tokens_seen": 157040640,
       "step": 19170
     },
     {
       "epoch": 0.9210084033613445,
-      "grad_norm": 0.4853648841381073,
+      "grad_norm": 0.4898834824562073,
       "learning_rate": 8.446066479830206e-07,
       "loss": 0.9048,
       "num_input_tokens_seen": 157122560,
@@ -15354,63 +15354,63 @@
     },
     {
       "epoch": 0.9214885954381753,
-      "grad_norm": 0.5098863840103149,
+      "grad_norm": 0.5096047520637512,
       "learning_rate": 8.344264223675485e-07,
-      "loss": 0.8928,
+      "loss": 0.8926,
       "num_input_tokens_seen": 157204480,
       "step": 19190
     },
     {
       "epoch": 0.921968787515006,
-      "grad_norm": 0.4745608866214752,
+      "grad_norm": 0.46778738498687744,
       "learning_rate": 8.243068802169906e-07,
-      "loss": 0.8414,
+      "loss": 0.8408,
       "num_input_tokens_seen": 157286400,
       "step": 19200
     },
     {
       "epoch": 0.9224489795918367,
-      "grad_norm": 0.555088996887207,
+      "grad_norm": 0.5187940001487732,
       "learning_rate": 8.14248046943078e-07,
-      "loss": 0.9802,
+      "loss": 0.9799,
       "num_input_tokens_seen": 157368320,
       "step": 19210
     },
     {
       "epoch": 0.9229291716686675,
-      "grad_norm": 0.5188843011856079,
+      "grad_norm": 0.5219123959541321,
       "learning_rate": 8.042499478050719e-07,
-      "loss": 0.7226,
+      "loss": 0.7224,
       "num_input_tokens_seen": 157450240,
       "step": 19220
     },
     {
       "epoch": 0.9234093637454982,
-      "grad_norm": 0.4765109121799469,
+      "grad_norm": 0.47635209560394287,
       "learning_rate": 7.943126079097418e-07,
-      "loss": 0.8491,
+      "loss": 0.8495,
       "num_input_tokens_seen": 157532160,
       "step": 19230
     },
     {
       "epoch": 0.9238895558223289,
-      "grad_norm": 1.0354602336883545,
+      "grad_norm": 1.093017339706421,
       "learning_rate": 7.844360522112737e-07,
-      "loss": 0.9703,
+      "loss": 0.9701,
       "num_input_tokens_seen": 157614080,
       "step": 19240
     },
     {
       "epoch": 0.9243697478991597,
-      "grad_norm": 0.48967981338500977,
+      "grad_norm": 0.4886617660522461,
       "learning_rate": 7.746203055112145e-07,
-      "loss": 0.8683,
+      "loss": 0.8682,
       "num_input_tokens_seen": 157696000,
       "step": 19250
     },
     {
       "epoch": 0.9248499399759904,
-      "grad_norm": 0.553525984287262,
+      "grad_norm": 0.5579150319099426,
       "learning_rate": 7.648653924584137e-07,
       "loss": 0.9395,
       "num_input_tokens_seen": 157777920,
@@ -15418,7 +15418,7 @@
     },
     {
       "epoch": 0.9253301320528211,
-      "grad_norm": 0.47739177942276,
+      "grad_norm": 0.458242803812027,
       "learning_rate": 7.55171337548946e-07,
       "loss": 0.84,
       "num_input_tokens_seen": 157859840,
@@ -15426,7 +15426,7 @@
     },
     {
       "epoch": 0.9258103241296518,
-      "grad_norm": 0.47741973400115967,
+      "grad_norm": 0.475402295589447,
       "learning_rate": 7.455381651260807e-07,
       "loss": 0.7814,
       "num_input_tokens_seen": 157941760,
@@ -15434,15 +15434,15 @@
     },
     {
       "epoch": 0.9262905162064826,
-      "grad_norm": 0.49380743503570557,
+      "grad_norm": 0.4926086962223053,
       "learning_rate": 7.359658993801894e-07,
-      "loss": 0.954,
+      "loss": 0.9542,
       "num_input_tokens_seen": 158023680,
       "step": 19290
     },
     {
       "epoch": 0.9267707082833133,
-      "grad_norm": 0.5180758833885193,
+      "grad_norm": 0.5198042392730713,
       "learning_rate": 7.264545643486997e-07,
       "loss": 0.9109,
       "num_input_tokens_seen": 158105600,
@@ -15450,95 +15450,95 @@
     },
     {
       "epoch": 0.927250900360144,
-      "grad_norm": 0.7896413803100586,
+      "grad_norm": 0.7973329424858093,
       "learning_rate": 7.170041839160368e-07,
-      "loss": 0.9779,
+      "loss": 0.9783,
       "num_input_tokens_seen": 158187520,
       "step": 19310
     },
     {
       "epoch": 0.9277310924369748,
-      "grad_norm": 0.48943135142326355,
+      "grad_norm": 0.4901650547981262,
       "learning_rate": 7.076147818135537e-07,
-      "loss": 0.8145,
+      "loss": 0.8147,
       "num_input_tokens_seen": 158269440,
       "step": 19320
     },
     {
       "epoch": 0.9282112845138055,
-      "grad_norm": 1.7075966596603394,
+      "grad_norm": 1.725703477859497,
       "learning_rate": 6.982863816194812e-07,
-      "loss": 0.7036,
+      "loss": 0.7043,
       "num_input_tokens_seen": 158351360,
       "step": 19330
     },
     {
       "epoch": 0.9286914765906362,
-      "grad_norm": 0.48135149478912354,
+      "grad_norm": 0.46620166301727295,
       "learning_rate": 6.890190067588648e-07,
-      "loss": 0.8395,
+      "loss": 0.8402,
       "num_input_tokens_seen": 158433280,
       "step": 19340
     },
     {
       "epoch": 0.929171668667467,
-      "grad_norm": 0.43532514572143555,
+      "grad_norm": 0.4360329806804657,
       "learning_rate": 6.798126805035082e-07,
-      "loss": 0.9633,
+      "loss": 0.9634,
       "num_input_tokens_seen": 158515200,
       "step": 19350
     },
     {
       "epoch": 0.9296518607442977,
-      "grad_norm": 0.5122432708740234,
+      "grad_norm": 0.5116370916366577,
       "learning_rate": 6.706674259719048e-07,
-      "loss": 0.9551,
+      "loss": 0.9554,
       "num_input_tokens_seen": 158597120,
       "step": 19360
     },
     {
       "epoch": 0.9301320528211284,
-      "grad_norm": 0.4708029627799988,
+      "grad_norm": 0.4693267345428467,
       "learning_rate": 6.615832661291954e-07,
-      "loss": 0.895,
+      "loss": 0.8942,
       "num_input_tokens_seen": 158679040,
       "step": 19370
     },
     {
       "epoch": 0.9306122448979591,
-      "grad_norm": 0.5076829195022583,
+      "grad_norm": 0.5132554769515991,
       "learning_rate": 6.525602237870993e-07,
-      "loss": 0.7502,
+      "loss": 0.7503,
       "num_input_tokens_seen": 158760960,
       "step": 19380
     },
     {
       "epoch": 0.9310924369747899,
-      "grad_norm": 0.5408411622047424,
+      "grad_norm": 0.5355455279350281,
       "learning_rate": 6.435983216038583e-07,
-      "loss": 0.693,
+      "loss": 0.6928,
       "num_input_tokens_seen": 158842880,
       "step": 19390
     },
     {
       "epoch": 0.9315726290516206,
-      "grad_norm": 0.5284355282783508,
+      "grad_norm": 0.5428586006164551,
       "learning_rate": 6.346975820841927e-07,
-      "loss": 0.8507,
+      "loss": 0.8504,
       "num_input_tokens_seen": 158924800,
       "step": 19400
     },
     {
       "epoch": 0.9320528211284513,
-      "grad_norm": 0.47476431727409363,
+      "grad_norm": 0.474017471075058,
       "learning_rate": 6.258580275792153e-07,
-      "loss": 0.9174,
+      "loss": 0.9175,
       "num_input_tokens_seen": 159006720,
       "step": 19410
     },
     {
       "epoch": 0.9325330132052821,
-      "grad_norm": 0.4776346981525421,
+      "grad_norm": 0.47815394401550293,
       "learning_rate": 6.170796802864115e-07,
       "loss": 0.7601,
       "num_input_tokens_seen": 159088640,
@@ -15546,183 +15546,183 @@
     },
     {
       "epoch": 0.9330132052821128,
-      "grad_norm": 0.5064809322357178,
+      "grad_norm": 0.5085130333900452,
       "learning_rate": 6.083625622495565e-07,
-      "loss": 0.8922,
+      "loss": 0.8923,
       "num_input_tokens_seen": 159170560,
       "step": 19430
     },
     {
       "epoch": 0.9334933973589435,
-      "grad_norm": 0.48573869466781616,
+      "grad_norm": 0.4867092967033386,
       "learning_rate": 5.997066953586761e-07,
-      "loss": 0.8506,
+      "loss": 0.8514,
       "num_input_tokens_seen": 159252480,
       "step": 19440
     },
     {
       "epoch": 0.9339735894357744,
-      "grad_norm": 0.48475074768066406,
+      "grad_norm": 0.48415231704711914,
       "learning_rate": 5.911121013499721e-07,
-      "loss": 0.801,
+      "loss": 0.8006,
       "num_input_tokens_seen": 159334400,
       "step": 19450
     },
     {
       "epoch": 0.934453781512605,
-      "grad_norm": 0.5202433466911316,
+      "grad_norm": 0.4926071763038635,
       "learning_rate": 5.825788018057971e-07,
-      "loss": 0.8358,
+      "loss": 0.8361,
       "num_input_tokens_seen": 159416320,
       "step": 19460
     },
     {
       "epoch": 0.9349339735894358,
-      "grad_norm": 0.48488256335258484,
+      "grad_norm": 0.485619455575943,
       "learning_rate": 5.741068181545684e-07,
-      "loss": 1.0555,
+      "loss": 1.0557,
       "num_input_tokens_seen": 159498240,
       "step": 19470
     },
     {
       "epoch": 0.9354141656662666,
-      "grad_norm": 0.467579185962677,
+      "grad_norm": 0.4691498875617981,
       "learning_rate": 5.656961716707459e-07,
-      "loss": 1.184,
+      "loss": 1.1845,
       "num_input_tokens_seen": 159580160,
       "step": 19480
     },
     {
       "epoch": 0.9358943577430973,
-      "grad_norm": 0.47693246603012085,
+      "grad_norm": 0.48349183797836304,
       "learning_rate": 5.57346883474752e-07,
-      "loss": 0.8181,
+      "loss": 0.818,
       "num_input_tokens_seen": 159662080,
       "step": 19490
     },
     {
       "epoch": 0.936374549819928,
-      "grad_norm": 0.3317853808403015,
+      "grad_norm": 0.3223581314086914,
       "learning_rate": 5.490589745329261e-07,
-      "loss": 0.9217,
+      "loss": 0.9216,
       "num_input_tokens_seen": 159744000,
       "step": 19500
     },
     {
       "epoch": 0.9368547418967587,
-      "grad_norm": 0.48513829708099365,
+      "grad_norm": 0.4837208688259125,
       "learning_rate": 5.408324656574842e-07,
-      "loss": 1.0929,
+      "loss": 1.0933,
       "num_input_tokens_seen": 159825920,
       "step": 19510
     },
     {
       "epoch": 0.9373349339735895,
-      "grad_norm": 0.5054202675819397,
+      "grad_norm": 0.5018269419670105,
       "learning_rate": 5.326673775064545e-07,
-      "loss": 0.9711,
+      "loss": 0.9709,
       "num_input_tokens_seen": 159907840,
       "step": 19520
     },
     {
       "epoch": 0.9378151260504202,
-      "grad_norm": 0.48606500029563904,
+      "grad_norm": 0.4863048791885376,
       "learning_rate": 5.245637305836243e-07,
-      "loss": 0.872,
+      "loss": 0.8722,
       "num_input_tokens_seen": 159989760,
       "step": 19530
     },
     {
       "epoch": 0.9382953181272509,
-      "grad_norm": 0.47018781304359436,
+      "grad_norm": 0.4706777036190033,
       "learning_rate": 5.165215452384936e-07,
-      "loss": 0.8849,
+      "loss": 0.8854,
       "num_input_tokens_seen": 160071680,
       "step": 19540
     },
     {
       "epoch": 0.9387755102040817,
-      "grad_norm": 0.5236411094665527,
+      "grad_norm": 0.5222617983818054,
       "learning_rate": 5.085408416662274e-07,
-      "loss": 0.8509,
+      "loss": 0.8518,
       "num_input_tokens_seen": 160153600,
       "step": 19550
     },
     {
       "epoch": 0.9392557022809124,
-      "grad_norm": 0.45992419123649597,
+      "grad_norm": 0.46015259623527527,
       "learning_rate": 5.006216399075947e-07,
-      "loss": 0.9798,
+      "loss": 0.9801,
       "num_input_tokens_seen": 160235520,
       "step": 19560
     },
     {
       "epoch": 0.9397358943577431,
-      "grad_norm": 0.5223960280418396,
+      "grad_norm": 0.522504448890686,
       "learning_rate": 4.92763959848927e-07,
-      "loss": 0.8332,
+      "loss": 0.8329,
       "num_input_tokens_seen": 160317440,
       "step": 19570
     },
     {
       "epoch": 0.9402160864345739,
-      "grad_norm": 0.4571020305156708,
+      "grad_norm": 0.45878833532333374,
       "learning_rate": 4.849678212220682e-07,
-      "loss": 0.9557,
+      "loss": 0.9552,
       "num_input_tokens_seen": 160399360,
       "step": 19580
     },
     {
       "epoch": 0.9406962785114046,
-      "grad_norm": 0.4470067620277405,
+      "grad_norm": 0.44788938760757446,
       "learning_rate": 4.772332436043165e-07,
-      "loss": 0.7306,
+      "loss": 0.7305,
       "num_input_tokens_seen": 160481280,
       "step": 19590
     },
     {
       "epoch": 0.9411764705882353,
-      "grad_norm": 0.43297266960144043,
+      "grad_norm": 0.4315415322780609,
       "learning_rate": 4.6956024641838237e-07,
-      "loss": 0.9514,
+      "loss": 0.9516,
       "num_input_tokens_seen": 160563200,
       "step": 19600
     },
     {
       "epoch": 0.941656662665066,
-      "grad_norm": 0.49218934774398804,
+      "grad_norm": 0.49236106872558594,
       "learning_rate": 4.619488489323448e-07,
-      "loss": 0.9364,
+      "loss": 0.9362,
       "num_input_tokens_seen": 160645120,
       "step": 19610
     },
     {
       "epoch": 0.9421368547418968,
-      "grad_norm": 0.440889447927475,
+      "grad_norm": 0.44332748651504517,
       "learning_rate": 4.5439907025958405e-07,
-      "loss": 1.0067,
+      "loss": 1.006,
       "num_input_tokens_seen": 160727040,
       "step": 19620
     },
     {
       "epoch": 0.9426170468187275,
-      "grad_norm": 0.4658232033252716,
+      "grad_norm": 0.46615105867385864,
       "learning_rate": 4.4691092935876256e-07,
-      "loss": 0.985,
+      "loss": 0.9851,
       "num_input_tokens_seen": 160808960,
       "step": 19630
     },
     {
       "epoch": 0.9430972388955582,
-      "grad_norm": 0.4865157902240753,
+      "grad_norm": 0.48708418011665344,
       "learning_rate": 4.394844450337443e-07,
-      "loss": 1.1212,
+      "loss": 1.1214,
       "num_input_tokens_seen": 160890880,
       "step": 19640
     },
     {
       "epoch": 0.943577430972389,
-      "grad_norm": 0.782720685005188,
+      "grad_norm": 0.7900726199150085,
       "learning_rate": 4.3211963593357275e-07,
       "loss": 1.0098,
       "num_input_tokens_seen": 160972800,
@@ -15730,143 +15730,143 @@
     },
     {
       "epoch": 0.9440576230492197,
-      "grad_norm": 0.4879094660282135,
+      "grad_norm": 0.48821115493774414,
       "learning_rate": 4.248165205524152e-07,
-      "loss": 0.8137,
+      "loss": 0.8139,
       "num_input_tokens_seen": 161054720,
       "step": 19660
     },
     {
       "epoch": 0.9445378151260504,
-      "grad_norm": 0.45209428668022156,
+      "grad_norm": 0.44911932945251465,
       "learning_rate": 4.175751172295156e-07,
-      "loss": 0.9042,
+      "loss": 0.9041,
       "num_input_tokens_seen": 161136640,
       "step": 19670
     },
     {
       "epoch": 0.9450180072028812,
-      "grad_norm": 0.4872719347476959,
+      "grad_norm": 0.4880870580673218,
       "learning_rate": 4.1039544414914753e-07,
-      "loss": 0.8619,
+      "loss": 0.8617,
       "num_input_tokens_seen": 161218560,
       "step": 19680
     },
     {
       "epoch": 0.9454981992797119,
-      "grad_norm": 0.7091497182846069,
+      "grad_norm": 0.7131670117378235,
       "learning_rate": 4.032775193405752e-07,
-      "loss": 0.8915,
+      "loss": 0.8918,
       "num_input_tokens_seen": 161300480,
       "step": 19690
     },
     {
       "epoch": 0.9459783913565426,
-      "grad_norm": 0.45862463116645813,
+      "grad_norm": 0.4578578770160675,
       "learning_rate": 3.96221360677998e-07,
-      "loss": 0.815,
+      "loss": 0.8151,
       "num_input_tokens_seen": 161382400,
       "step": 19700
     },
     {
       "epoch": 0.9464585834333733,
-      "grad_norm": 0.5041652321815491,
+      "grad_norm": 0.5045084953308105,
       "learning_rate": 3.892269858805142e-07,
-      "loss": 0.9076,
+      "loss": 0.9074,
       "num_input_tokens_seen": 161464320,
       "step": 19710
     },
     {
       "epoch": 0.9469387755102041,
-      "grad_norm": 0.4937167763710022,
+      "grad_norm": 0.4939689040184021,
       "learning_rate": 3.82294412512077e-07,
-      "loss": 0.8012,
+      "loss": 0.8005,
       "num_input_tokens_seen": 161546240,
       "step": 19720
     },
     {
       "epoch": 0.9474189675870348,
-      "grad_norm": 0.44221341609954834,
+      "grad_norm": 0.44345298409461975,
       "learning_rate": 3.7542365798143573e-07,
-      "loss": 0.8382,
+      "loss": 0.8381,
       "num_input_tokens_seen": 161628160,
       "step": 19730
     },
     {
       "epoch": 0.9478991596638655,
-      "grad_norm": 0.4632039964199066,
+      "grad_norm": 0.4660447835922241,
       "learning_rate": 3.6861473954210855e-07,
-      "loss": 1.0418,
+      "loss": 1.0417,
       "num_input_tokens_seen": 161710080,
       "step": 19740
     },
     {
       "epoch": 0.9483793517406963,
-      "grad_norm": 0.4950004518032074,
+      "grad_norm": 0.49638691544532776,
       "learning_rate": 3.6186767429234323e-07,
-      "loss": 0.7393,
+      "loss": 0.7395,
       "num_input_tokens_seen": 161792000,
       "step": 19750
     },
     {
       "epoch": 0.948859543817527,
-      "grad_norm": 0.4772696793079376,
+      "grad_norm": 0.4776581823825836,
       "learning_rate": 3.5518247917505077e-07,
-      "loss": 0.8615,
+      "loss": 0.8611,
       "num_input_tokens_seen": 161873920,
       "step": 19760
     },
     {
       "epoch": 0.9493397358943577,
-      "grad_norm": 0.4546825885772705,
+      "grad_norm": 0.4587015211582184,
       "learning_rate": 3.485591709777802e-07,
-      "loss": 1.041,
+      "loss": 1.0409,
       "num_input_tokens_seen": 161955840,
       "step": 19770
     },
     {
       "epoch": 0.9498199279711885,
-      "grad_norm": 0.4924812912940979,
+      "grad_norm": 0.4919114112854004,
       "learning_rate": 3.419977663326801e-07,
-      "loss": 0.813,
+      "loss": 0.8133,
       "num_input_tokens_seen": 162037760,
       "step": 19780
     },
     {
       "epoch": 0.9503001200480192,
-      "grad_norm": 0.5705674290657043,
+      "grad_norm": 0.571787416934967,
       "learning_rate": 3.3549828171644537e-07,
-      "loss": 1.2698,
+      "loss": 1.2694,
       "num_input_tokens_seen": 162119680,
       "step": 19790
     },
     {
       "epoch": 0.9507803121248499,
-      "grad_norm": 0.48878228664398193,
+      "grad_norm": 0.48567476868629456,
       "learning_rate": 3.29060733450276e-07,
-      "loss": 0.8896,
+      "loss": 0.8892,
       "num_input_tokens_seen": 162201600,
       "step": 19800
     },
     {
       "epoch": 0.9512605042016806,
-      "grad_norm": 0.4292643070220947,
+      "grad_norm": 0.42919716238975525,
       "learning_rate": 3.2268513769984634e-07,
-      "loss": 1.0284,
+      "loss": 1.028,
       "num_input_tokens_seen": 162283520,
       "step": 19810
     },
     {
       "epoch": 0.9517406962785114,
-      "grad_norm": 0.715688169002533,
+      "grad_norm": 0.7574768662452698,
       "learning_rate": 3.163715104752524e-07,
-      "loss": 0.9543,
+      "loss": 0.9548,
       "num_input_tokens_seen": 162365440,
       "step": 19820
     },
     {
       "epoch": 0.9522208883553421,
-      "grad_norm": 0.46790245175361633,
+      "grad_norm": 0.46974584460258484,
       "learning_rate": 3.101198676309841e-07,
       "loss": 0.9059,
       "num_input_tokens_seen": 162447360,
@@ -15874,87 +15874,87 @@
     },
     {
       "epoch": 0.9527010804321728,
-      "grad_norm": 0.48006242513656616,
+      "grad_norm": 0.48091816902160645,
       "learning_rate": 3.039302248658754e-07,
-      "loss": 0.9259,
+      "loss": 0.9254,
       "num_input_tokens_seen": 162529280,
       "step": 19840
     },
     {
       "epoch": 0.9531812725090036,
-      "grad_norm": 0.3433704376220703,
+      "grad_norm": 0.35266348719596863,
       "learning_rate": 2.978025977230736e-07,
-      "loss": 1.0082,
+      "loss": 1.0085,
       "num_input_tokens_seen": 162611200,
       "step": 19850
     },
     {
       "epoch": 0.9536614645858343,
-      "grad_norm": 0.4852867126464844,
+      "grad_norm": 0.48670434951782227,
       "learning_rate": 2.91737001589984e-07,
-      "loss": 0.9153,
+      "loss": 0.9163,
       "num_input_tokens_seen": 162693120,
       "step": 19860
     },
     {
       "epoch": 0.954141656662665,
-      "grad_norm": 0.5154809951782227,
+      "grad_norm": 0.5381135940551758,
       "learning_rate": 2.8573345169825296e-07,
-      "loss": 0.9446,
+      "loss": 0.9455,
       "num_input_tokens_seen": 162775040,
       "step": 19870
     },
     {
       "epoch": 0.9546218487394958,
-      "grad_norm": 0.707761287689209,
+      "grad_norm": 0.7383513450622559,
       "learning_rate": 2.797919631237156e-07,
-      "loss": 0.7736,
+      "loss": 0.774,
       "num_input_tokens_seen": 162856960,
       "step": 19880
     },
     {
       "epoch": 0.9551020408163265,
-      "grad_norm": 0.3873317837715149,
+      "grad_norm": 0.37961041927337646,
       "learning_rate": 2.739125507863649e-07,
-      "loss": 0.9343,
+      "loss": 0.935,
       "num_input_tokens_seen": 162938880,
       "step": 19890
     },
     {
       "epoch": 0.9555822328931572,
-      "grad_norm": 0.48126840591430664,
+      "grad_norm": 0.48009783029556274,
       "learning_rate": 2.680952294503075e-07,
-      "loss": 0.9554,
+      "loss": 0.9552,
       "num_input_tokens_seen": 163020800,
       "step": 19900
     },
     {
       "epoch": 0.956062424969988,
-      "grad_norm": 0.4585091471672058,
+      "grad_norm": 0.45567837357521057,
       "learning_rate": 2.6234001372372194e-07,
-      "loss": 0.9323,
+      "loss": 0.9321,
       "num_input_tokens_seen": 163102720,
       "step": 19910
     },
     {
       "epoch": 0.9565426170468188,
-      "grad_norm": 0.5132349729537964,
+      "grad_norm": 0.5141881108283997,
       "learning_rate": 2.5664691805884767e-07,
-      "loss": 0.9947,
+      "loss": 0.9948,
       "num_input_tokens_seen": 163184640,
       "step": 19920
     },
     {
       "epoch": 0.9570228091236495,
-      "grad_norm": 0.5150689482688904,
+      "grad_norm": 0.5147581100463867,
       "learning_rate": 2.5101595675191827e-07,
-      "loss": 0.9797,
+      "loss": 0.9796,
       "num_input_tokens_seen": 163266560,
       "step": 19930
     },
     {
       "epoch": 0.9575030012004802,
-      "grad_norm": 0.46773561835289,
+      "grad_norm": 0.4667784571647644,
       "learning_rate": 2.4544714394314215e-07,
       "loss": 0.9892,
       "num_input_tokens_seen": 163348480,
@@ -15962,15 +15962,15 @@
     },
     {
       "epoch": 0.957983193277311,
-      "grad_norm": 0.7742196917533875,
+      "grad_norm": 0.7636064887046814,
       "learning_rate": 2.399404936166638e-07,
-      "loss": 1.171,
+      "loss": 1.1706,
       "num_input_tokens_seen": 163430400,
       "step": 19950
     },
     {
       "epoch": 0.9584633853541417,
-      "grad_norm": 0.4795765280723572,
+      "grad_norm": 0.4796493947505951,
       "learning_rate": 2.3449601960052746e-07,
       "loss": 0.7551,
       "num_input_tokens_seen": 163512320,
@@ -15978,15 +15978,15 @@
     },
     {
       "epoch": 0.9589435774309724,
-      "grad_norm": 0.46741360425949097,
+      "grad_norm": 0.46821102499961853,
       "learning_rate": 2.2911373556664118e-07,
-      "loss": 0.7974,
+      "loss": 0.7976,
       "num_input_tokens_seen": 163594240,
       "step": 19970
     },
     {
       "epoch": 0.9594237695078032,
-      "grad_norm": 0.47377344965934753,
+      "grad_norm": 0.4732038378715515,
       "learning_rate": 2.23793655030749e-07,
       "loss": 0.8963,
       "num_input_tokens_seen": 163676160,
@@ -15994,39 +15994,39 @@
     },
     {
       "epoch": 0.9599039615846339,
-      "grad_norm": 0.5991731286048889,
+      "grad_norm": 0.5945155620574951,
       "learning_rate": 2.1853579135238667e-07,
-      "loss": 0.9377,
+      "loss": 0.9375,
       "num_input_tokens_seen": 163758080,
       "step": 19990
     },
     {
       "epoch": 0.9603841536614646,
-      "grad_norm": 0.4813549816608429,
+      "grad_norm": 0.48151063919067383,
       "learning_rate": 2.1334015773486203e-07,
-      "loss": 1.0792,
+      "loss": 1.0805,
       "num_input_tokens_seen": 163840000,
       "step": 20000
     },
     {
       "epoch": 0.9608643457382953,
-      "grad_norm": 0.5007798075675964,
+      "grad_norm": 0.5012336373329163,
       "learning_rate": 2.0820676722520526e-07,
-      "loss": 0.7725,
+      "loss": 0.7723,
       "num_input_tokens_seen": 163921920,
       "step": 20010
     },
     {
       "epoch": 0.9613445378151261,
-      "grad_norm": 0.49032649397850037,
+      "grad_norm": 0.489702433347702,
       "learning_rate": 2.0313563271414927e-07,
-      "loss": 0.8982,
+      "loss": 0.8981,
       "num_input_tokens_seen": 164003840,
       "step": 20020
     },
     {
       "epoch": 0.9618247298919568,
-      "grad_norm": 0.49168848991394043,
+      "grad_norm": 0.4924655854701996,
       "learning_rate": 1.9812676693608812e-07,
       "loss": 0.7376,
       "num_input_tokens_seen": 164085760,
@@ -16034,47 +16034,47 @@
     },
     {
       "epoch": 0.9623049219687875,
-      "grad_norm": 0.47681140899658203,
+      "grad_norm": 0.47769734263420105,
       "learning_rate": 1.9318018246905488e-07,
-      "loss": 0.8378,
+      "loss": 0.8382,
       "num_input_tokens_seen": 164167680,
       "step": 20040
     },
     {
       "epoch": 0.9627851140456183,
-      "grad_norm": 0.4583689272403717,
+      "grad_norm": 0.4559698700904846,
       "learning_rate": 1.8829589173468552e-07,
-      "loss": 0.8883,
+      "loss": 0.8877,
       "num_input_tokens_seen": 164249600,
       "step": 20050
     },
     {
       "epoch": 0.963265306122449,
-      "grad_norm": 0.3779040575027466,
+      "grad_norm": 0.36575424671173096,
       "learning_rate": 1.8347390699817724e-07,
-      "loss": 0.7442,
+      "loss": 0.7448,
       "num_input_tokens_seen": 164331520,
       "step": 20060
     },
     {
       "epoch": 0.9637454981992797,
-      "grad_norm": 0.571790874004364,
+      "grad_norm": 0.5725572109222412,
       "learning_rate": 1.7871424036828288e-07,
-      "loss": 0.9274,
+      "loss": 0.9267,
       "num_input_tokens_seen": 164413440,
       "step": 20070
     },
     {
       "epoch": 0.9642256902761105,
-      "grad_norm": 0.5354510545730591,
+      "grad_norm": 0.5294426083564758,
       "learning_rate": 1.7401690379724722e-07,
-      "loss": 0.9755,
+      "loss": 0.9753,
       "num_input_tokens_seen": 164495360,
       "step": 20080
     },
     {
       "epoch": 0.9647058823529412,
-      "grad_norm": 0.5626926422119141,
+      "grad_norm": 0.5594549179077148,
       "learning_rate": 1.6938190908080688e-07,
       "loss": 0.8417,
       "num_input_tokens_seen": 164577280,
@@ -16082,31 +16082,31 @@
     },
     {
       "epoch": 0.9651860744297719,
-      "grad_norm": 0.5134274363517761,
+      "grad_norm": 0.5328065752983093,
       "learning_rate": 1.6480926785814866e-07,
-      "loss": 0.7835,
+      "loss": 0.7834,
       "num_input_tokens_seen": 164659200,
       "step": 20100
     },
     {
       "epoch": 0.9656662665066026,
-      "grad_norm": 0.5222331881523132,
+      "grad_norm": 0.5204812288284302,
       "learning_rate": 1.6029899161187079e-07,
-      "loss": 1.0117,
+      "loss": 1.0118,
       "num_input_tokens_seen": 164741120,
       "step": 20110
     },
     {
       "epoch": 0.9661464585834334,
-      "grad_norm": 0.5277162194252014,
+      "grad_norm": 0.5157402753829956,
       "learning_rate": 1.5585109166796896e-07,
-      "loss": 0.7976,
+      "loss": 0.7969,
       "num_input_tokens_seen": 164823040,
       "step": 20120
     },
     {
       "epoch": 0.9666266506602641,
-      "grad_norm": 0.5840759873390198,
+      "grad_norm": 0.5858813524246216,
       "learning_rate": 1.5146557919581138e-07,
       "loss": 0.9373,
       "num_input_tokens_seen": 164904960,
@@ -16114,7 +16114,7 @@
     },
     {
       "epoch": 0.9671068427370948,
-      "grad_norm": 0.4930008053779602,
+      "grad_norm": 0.4909372627735138,
       "learning_rate": 1.4714246520808328e-07,
       "loss": 0.717,
       "num_input_tokens_seen": 164986880,
@@ -16122,31 +16122,31 @@
     },
     {
       "epoch": 0.9675870348139256,
-      "grad_norm": 0.49782538414001465,
+      "grad_norm": 0.4969266653060913,
       "learning_rate": 1.4288176056079238e-07,
-      "loss": 0.8526,
+      "loss": 0.8529,
       "num_input_tokens_seen": 165068800,
       "step": 20150
     },
     {
       "epoch": 0.9680672268907563,
-      "grad_norm": 0.48981910943984985,
+      "grad_norm": 0.4890909194946289,
       "learning_rate": 1.3868347595322184e-07,
-      "loss": 0.8959,
+      "loss": 0.8958,
       "num_input_tokens_seen": 165150720,
       "step": 20160
     },
     {
       "epoch": 0.968547418967587,
-      "grad_norm": 0.46395954489707947,
+      "grad_norm": 0.4637242257595062,
       "learning_rate": 1.3454762192790794e-07,
-      "loss": 1.1636,
+      "loss": 1.1627,
       "num_input_tokens_seen": 165232640,
       "step": 20170
     },
     {
       "epoch": 0.9690276110444178,
-      "grad_norm": 0.4878246784210205,
+      "grad_norm": 0.48529329895973206,
       "learning_rate": 1.3047420887061513e-07,
       "loss": 1.0033,
       "num_input_tokens_seen": 165314560,
@@ -16154,79 +16154,79 @@
     },
     {
       "epoch": 0.9695078031212485,
-      "grad_norm": 0.45499855279922485,
+      "grad_norm": 0.4544067084789276,
       "learning_rate": 1.264632470103111e-07,
-      "loss": 0.8926,
+      "loss": 0.8922,
       "num_input_tokens_seen": 165396480,
       "step": 20190
     },
     {
       "epoch": 0.9699879951980792,
-      "grad_norm": 0.4934712052345276,
+      "grad_norm": 0.49211665987968445,
       "learning_rate": 1.225147464191334e-07,
-      "loss": 0.9251,
+      "loss": 0.9253,
       "num_input_tokens_seen": 165478400,
       "step": 20200
     },
     {
       "epoch": 0.9704681872749099,
-      "grad_norm": 0.48829373717308044,
+      "grad_norm": 0.4904029667377472,
       "learning_rate": 1.1862871701237288e-07,
-      "loss": 0.754,
+      "loss": 0.7544,
       "num_input_tokens_seen": 165560320,
       "step": 20210
     },
     {
       "epoch": 0.9709483793517407,
-      "grad_norm": 0.4913751184940338,
+      "grad_norm": 0.490373432636261,
       "learning_rate": 1.1480516854844858e-07,
-      "loss": 0.799,
+      "loss": 0.7983,
       "num_input_tokens_seen": 165642240,
       "step": 20220
     },
     {
       "epoch": 0.9714285714285714,
-      "grad_norm": 0.5303522944450378,
+      "grad_norm": 0.5297425985336304,
       "learning_rate": 1.1104411062887732e-07,
-      "loss": 0.797,
+      "loss": 0.7969,
       "num_input_tokens_seen": 165724160,
       "step": 20230
     },
     {
       "epoch": 0.9719087635054021,
-      "grad_norm": 0.4915572702884674,
+      "grad_norm": 0.49064329266548157,
       "learning_rate": 1.0734555269825141e-07,
-      "loss": 0.7708,
+      "loss": 0.7713,
       "num_input_tokens_seen": 165806080,
       "step": 20240
     },
     {
       "epoch": 0.9723889555822329,
-      "grad_norm": 0.5113864541053772,
+      "grad_norm": 0.5096157789230347,
       "learning_rate": 1.0370950404421931e-07,
-      "loss": 0.7448,
+      "loss": 0.7452,
       "num_input_tokens_seen": 165888000,
       "step": 20250
     },
     {
       "epoch": 0.9728691476590636,
-      "grad_norm": 0.49266868829727173,
+      "grad_norm": 0.4939180314540863,
       "learning_rate": 1.0013597379745776e-07,
-      "loss": 0.8983,
+      "loss": 0.8992,
       "num_input_tokens_seen": 165969920,
       "step": 20260
     },
     {
       "epoch": 0.9733493397358943,
-      "grad_norm": 0.9062735438346863,
+      "grad_norm": 0.9096865653991699,
       "learning_rate": 9.662497093164691e-08,
-      "loss": 0.7648,
+      "loss": 0.7645,
       "num_input_tokens_seen": 166051840,
       "step": 20270
     },
     {
       "epoch": 0.9738295318127251,
-      "grad_norm": 0.5166149139404297,
+      "grad_norm": 0.5157402157783508,
       "learning_rate": 9.317650426345637e-08,
       "loss": 0.9752,
       "num_input_tokens_seen": 166133760,
@@ -16234,39 +16234,39 @@
     },
     {
       "epoch": 0.9743097238895558,
-      "grad_norm": 0.4873685836791992,
+      "grad_norm": 0.4877472221851349,
       "learning_rate": 8.979058245251193e-08,
-      "loss": 0.8632,
+      "loss": 0.8634,
       "num_input_tokens_seen": 166215680,
       "step": 20290
     },
     {
       "epoch": 0.9747899159663865,
-      "grad_norm": 0.48087936639785767,
+      "grad_norm": 0.4800505042076111,
       "learning_rate": 8.646721400138724e-08,
-      "loss": 1.0177,
+      "loss": 1.0172,
       "num_input_tokens_seen": 166297600,
       "step": 20300
     },
     {
       "epoch": 0.9752701080432172,
-      "grad_norm": 0.4865896999835968,
+      "grad_norm": 0.4879034161567688,
       "learning_rate": 8.320640725556773e-08,
-      "loss": 0.9075,
+      "loss": 0.9079,
       "num_input_tokens_seen": 166379520,
       "step": 20310
     },
     {
       "epoch": 0.975750300120048,
-      "grad_norm": 0.4811105728149414,
+      "grad_norm": 0.4818404018878937,
       "learning_rate": 8.000817040344222e-08,
-      "loss": 0.9434,
+      "loss": 0.9436,
       "num_input_tokens_seen": 166461440,
       "step": 20320
     },
     {
       "epoch": 0.9762304921968787,
-      "grad_norm": 0.47913140058517456,
+      "grad_norm": 0.4712151288986206,
       "learning_rate": 7.687251147627251e-08,
       "loss": 0.8307,
       "num_input_tokens_seen": 166543360,
@@ -16274,127 +16274,127 @@
     },
     {
       "epoch": 0.9767106842737094,
-      "grad_norm": 0.4741958677768707,
+      "grad_norm": 0.4749000072479248,
       "learning_rate": 7.379943834818214e-08,
-      "loss": 0.7942,
+      "loss": 0.7941,
       "num_input_tokens_seen": 166625280,
       "step": 20340
     },
     {
       "epoch": 0.9771908763505402,
-      "grad_norm": 0.492371529340744,
+      "grad_norm": 0.4928759038448334,
       "learning_rate": 7.0788958736126e-08,
-      "loss": 0.6887,
+      "loss": 0.6886,
       "num_input_tokens_seen": 166707200,
       "step": 20350
     },
     {
       "epoch": 0.977671068427371,
-      "grad_norm": 0.5118972659111023,
+      "grad_norm": 0.5131922960281372,
       "learning_rate": 6.784108019988189e-08,
-      "loss": 0.8574,
+      "loss": 0.8575,
       "num_input_tokens_seen": 166789120,
       "step": 20360
     },
     {
       "epoch": 0.9781512605042016,
-      "grad_norm": 0.387319952249527,
+      "grad_norm": 0.3878309726715088,
       "learning_rate": 6.495581014202556e-08,
-      "loss": 0.7726,
+      "loss": 0.7712,
       "num_input_tokens_seen": 166871040,
       "step": 20370
     },
     {
       "epoch": 0.9786314525810325,
-      "grad_norm": 0.494081050157547,
+      "grad_norm": 0.4928576350212097,
       "learning_rate": 6.213315580791135e-08,
-      "loss": 0.9247,
+      "loss": 0.9243,
       "num_input_tokens_seen": 166952960,
       "step": 20380
     },
     {
       "epoch": 0.9791116446578632,
-      "grad_norm": 0.5668483376502991,
+      "grad_norm": 0.5688450336456299,
       "learning_rate": 5.9373124285661e-08,
-      "loss": 0.8177,
+      "loss": 0.8171,
       "num_input_tokens_seen": 167034880,
       "step": 20390
     },
     {
       "epoch": 0.9795918367346939,
-      "grad_norm": 0.5243094563484192,
+      "grad_norm": 0.5241358876228333,
       "learning_rate": 5.6675722506135956e-08,
-      "loss": 0.8603,
+      "loss": 0.86,
       "num_input_tokens_seen": 167116800,
       "step": 20400
     },
     {
       "epoch": 0.9800720288115247,
-      "grad_norm": 0.5012032389640808,
+      "grad_norm": 0.5009602308273315,
       "learning_rate": 5.404095724292346e-08,
-      "loss": 0.9751,
+      "loss": 0.9734,
       "num_input_tokens_seen": 167198720,
       "step": 20410
     },
     {
       "epoch": 0.9805522208883554,
-      "grad_norm": 0.48027274012565613,
+      "grad_norm": 0.4785056710243225,
       "learning_rate": 5.146883511232825e-08,
-      "loss": 0.9047,
+      "loss": 0.9059,
       "num_input_tokens_seen": 167280640,
       "step": 20420
     },
     {
       "epoch": 0.9810324129651861,
-      "grad_norm": 0.9182277321815491,
+      "grad_norm": 0.9318637251853943,
       "learning_rate": 4.8959362573341995e-08,
-      "loss": 0.8816,
+      "loss": 0.8805,
       "num_input_tokens_seen": 167362560,
       "step": 20430
     },
     {
       "epoch": 0.9815126050420168,
-      "grad_norm": 0.47748062014579773,
+      "grad_norm": 0.4752718508243561,
       "learning_rate": 4.6512545927632213e-08,
-      "loss": 0.8182,
+      "loss": 0.8186,
       "num_input_tokens_seen": 167444480,
       "step": 20440
     },
     {
       "epoch": 0.9819927971188476,
-      "grad_norm": 0.5384604930877686,
+      "grad_norm": 0.5406351685523987,
       "learning_rate": 4.412839131953395e-08,
-      "loss": 0.8999,
+      "loss": 0.9007,
       "num_input_tokens_seen": 167526400,
       "step": 20450
     },
     {
       "epoch": 0.9824729891956783,
-      "grad_norm": 0.49782511591911316,
+      "grad_norm": 0.4987322688102722,
       "learning_rate": 4.180690473602755e-08,
-      "loss": 0.9972,
+      "loss": 1.0011,
       "num_input_tokens_seen": 167608320,
       "step": 20460
     },
     {
       "epoch": 0.982953181272509,
-      "grad_norm": 0.46127596497535706,
+      "grad_norm": 0.4613893926143646,
       "learning_rate": 3.9548092006719275e-08,
-      "loss": 0.7583,
+      "loss": 0.7579,
       "num_input_tokens_seen": 167690240,
       "step": 20470
     },
     {
       "epoch": 0.9834333733493398,
-      "grad_norm": 0.8296525478363037,
+      "grad_norm": 0.8392557501792908,
       "learning_rate": 3.7351958803835685e-08,
-      "loss": 1.0416,
+      "loss": 1.0422,
       "num_input_tokens_seen": 167772160,
       "step": 20480
     },
     {
       "epoch": 0.9839135654261705,
-      "grad_norm": 0.48525142669677734,
+      "grad_norm": 0.4861963391304016,
       "learning_rate": 3.5218510642201496e-08,
       "loss": 1.1318,
       "num_input_tokens_seen": 167854080,
@@ -16402,23 +16402,23 @@
     },
     {
       "epoch": 0.9843937575030012,
-      "grad_norm": 0.656819224357605,
+      "grad_norm": 0.6637359857559204,
       "learning_rate": 3.314775287923677e-08,
-      "loss": 1.0568,
+      "loss": 1.0579,
       "num_input_tokens_seen": 167936000,
       "step": 20500
     },
     {
       "epoch": 0.984873949579832,
-      "grad_norm": 0.49924975633621216,
+      "grad_norm": 0.4990644156932831,
       "learning_rate": 3.1139690714931945e-08,
-      "loss": 0.9061,
+      "loss": 0.9065,
       "num_input_tokens_seen": 168017920,
       "step": 20510
     },
     {
       "epoch": 0.9853541416566627,
-      "grad_norm": 0.47021421790122986,
+      "grad_norm": 0.4722457528114319,
       "learning_rate": 2.919432919183396e-08,
       "loss": 0.7413,
       "num_input_tokens_seen": 168099840,
@@ -16426,151 +16426,151 @@
     },
     {
       "epoch": 0.9858343337334934,
-      "grad_norm": 0.4875717759132385,
+      "grad_norm": 0.48728999495506287,
       "learning_rate": 2.731167319505179e-08,
-      "loss": 0.7303,
+      "loss": 0.7307,
       "num_input_tokens_seen": 168181760,
       "step": 20530
     },
     {
       "epoch": 0.9863145258103241,
-      "grad_norm": 0.47753021121025085,
+      "grad_norm": 0.477203905582428,
       "learning_rate": 2.5491727452217616e-08,
-      "loss": 0.763,
+      "loss": 0.7627,
       "num_input_tokens_seen": 168263680,
       "step": 20540
     },
     {
       "epoch": 0.9867947178871549,
-      "grad_norm": 1.1384245157241821,
+      "grad_norm": 1.1573189496994019,
       "learning_rate": 2.3734496533497907e-08,
-      "loss": 0.9044,
+      "loss": 0.9042,
       "num_input_tokens_seen": 168345600,
       "step": 20550
     },
     {
       "epoch": 0.9872749099639856,
-      "grad_norm": 0.47026336193084717,
+      "grad_norm": 0.44929322600364685,
       "learning_rate": 2.203998485156844e-08,
-      "loss": 0.9609,
+      "loss": 0.9611,
       "num_input_tokens_seen": 168427520,
       "step": 20560
     },
     {
       "epoch": 0.9877551020408163,
-      "grad_norm": 0.48212888836860657,
+      "grad_norm": 0.48382875323295593,
       "learning_rate": 2.040819666160876e-08,
-      "loss": 1.0561,
+      "loss": 1.0542,
       "num_input_tokens_seen": 168509440,
       "step": 20570
     },
     {
       "epoch": 0.9882352941176471,
-      "grad_norm": 0.5447530150413513,
+      "grad_norm": 0.5462098717689514,
       "learning_rate": 1.8839136061288288e-08,
-      "loss": 1.0402,
+      "loss": 1.0403,
       "num_input_tokens_seen": 168591360,
       "step": 20580
     },
     {
       "epoch": 0.9887154861944778,
-      "grad_norm": 0.5337615609169006,
+      "grad_norm": 0.5324848890304565,
       "learning_rate": 1.7332806990758012e-08,
-      "loss": 1.1063,
+      "loss": 1.1069,
       "num_input_tokens_seen": 168673280,
       "step": 20590
     },
     {
       "epoch": 0.9891956782713085,
-      "grad_norm": 0.4831818640232086,
+      "grad_norm": 0.4525868892669678,
       "learning_rate": 1.5889213232644917e-08,
-      "loss": 0.8762,
+      "loss": 0.8757,
       "num_input_tokens_seen": 168755200,
       "step": 20600
     },
     {
       "epoch": 0.9896758703481393,
-      "grad_norm": 0.5846928954124451,
+      "grad_norm": 0.5697076320648193,
       "learning_rate": 1.4508358412032575e-08,
-      "loss": 1.2618,
+      "loss": 1.2612,
       "num_input_tokens_seen": 168837120,
       "step": 20610
     },
     {
       "epoch": 0.99015606242497,
-      "grad_norm": 0.4968603253364563,
+      "grad_norm": 0.49757134914398193,
       "learning_rate": 1.319024599645835e-08,
-      "loss": 0.959,
+      "loss": 0.9593,
       "num_input_tokens_seen": 168919040,
       "step": 20620
     },
     {
       "epoch": 0.9906362545018007,
-      "grad_norm": 0.5029283165931702,
+      "grad_norm": 0.5038264989852905,
       "learning_rate": 1.1934879295905089e-08,
-      "loss": 0.7467,
+      "loss": 0.7466,
       "num_input_tokens_seen": 169000960,
       "step": 20630
     },
     {
       "epoch": 0.9911164465786314,
-      "grad_norm": 0.5078738331794739,
+      "grad_norm": 0.5087673664093018,
       "learning_rate": 1.074226146279278e-08,
-      "loss": 1.0644,
+      "loss": 1.0652,
       "num_input_tokens_seen": 169082880,
       "step": 20640
     },
     {
       "epoch": 0.9915966386554622,
-      "grad_norm": 0.44298240542411804,
+      "grad_norm": 0.44474107027053833,
       "learning_rate": 9.612395491970239e-09,
-      "loss": 1.036,
+      "loss": 1.0358,
       "num_input_tokens_seen": 169164800,
       "step": 20650
     },
     {
       "epoch": 0.9920768307322929,
-      "grad_norm": 0.480905681848526,
+      "grad_norm": 0.482793927192688,
       "learning_rate": 8.545284220698446e-09,
-      "loss": 0.8798,
+      "loss": 0.8794,
       "num_input_tokens_seen": 169246720,
       "step": 20660
     },
     {
       "epoch": 0.9925570228091236,
-      "grad_norm": 0.4742453396320343,
+      "grad_norm": 0.4741480350494385,
       "learning_rate": 7.540930328658879e-09,
-      "loss": 0.848,
+      "loss": 0.8482,
       "num_input_tokens_seen": 169328640,
       "step": 20670
     },
     {
       "epoch": 0.9930372148859544,
-      "grad_norm": 0.4936705529689789,
+      "grad_norm": 0.4936019480228424,
       "learning_rate": 6.599336337942408e-09,
-      "loss": 1.0114,
+      "loss": 1.0116,
       "num_input_tokens_seen": 169410560,
       "step": 20680
     },
     {
       "epoch": 0.9935174069627851,
-      "grad_norm": 0.3984171152114868,
+      "grad_norm": 0.4134100377559662,
       "learning_rate": 5.720504613035416e-09,
-      "loss": 0.7468,
+      "loss": 0.7456,
       "num_input_tokens_seen": 169492480,
       "step": 20690
     },
     {
       "epoch": 0.9939975990396158,
-      "grad_norm": 0.5437182188034058,
+      "grad_norm": 0.5405466556549072,
       "learning_rate": 4.904437360814252e-09,
-      "loss": 0.8409,
+      "loss": 0.8412,
       "num_input_tokens_seen": 169574400,
       "step": 20700
     },
     {
       "epoch": 0.9944777911164466,
-      "grad_norm": 0.4836597144603729,
+      "grad_norm": 0.48422014713287354,
       "learning_rate": 4.151136630553554e-09,
       "loss": 1.024,
       "num_input_tokens_seen": 169656320,
@@ -16578,89 +16578,89 @@
     },
     {
       "epoch": 0.9949579831932773,
-      "grad_norm": 0.4973140358924866,
+      "grad_norm": 0.49525994062423706,
       "learning_rate": 3.4606043139068234e-09,
-      "loss": 1.1108,
+      "loss": 1.1113,
       "num_input_tokens_seen": 169738240,
       "step": 20720
     },
     {
       "epoch": 0.995438175270108,
-      "grad_norm": 0.4446418583393097,
+      "grad_norm": 0.44537219405174255,
       "learning_rate": 2.832842144903647e-09,
-      "loss": 0.9724,
+      "loss": 0.9721,
       "num_input_tokens_seen": 169820160,
       "step": 20730
     },
     {
       "epoch": 0.9959183673469387,
-      "grad_norm": 0.9471971392631531,
+      "grad_norm": 1.051952838897705,
       "learning_rate": 2.2678516999552478e-09,
-      "loss": 0.7679,
+      "loss": 0.7684,
       "num_input_tokens_seen": 169902080,
       "step": 20740
     },
     {
       "epoch": 0.9963985594237695,
-      "grad_norm": 0.5114956498146057,
+      "grad_norm": 0.5115790963172913,
       "learning_rate": 1.7656343978378342e-09,
-      "loss": 0.9723,
+      "loss": 0.9727,
       "num_input_tokens_seen": 169984000,
       "step": 20750
     },
     {
       "epoch": 0.9968787515006002,
-      "grad_norm": 0.4692475497722626,
+      "grad_norm": 0.46845543384552,
       "learning_rate": 1.3261914996953728e-09,
-      "loss": 1.1212,
+      "loss": 1.121,
       "num_input_tokens_seen": 170065920,
       "step": 20760
     },
     {
       "epoch": 0.9973589435774309,
-      "grad_norm": 0.49446624517440796,
+      "grad_norm": 0.4868726432323456,
       "learning_rate": 9.49524109034039e-10,
-      "loss": 0.9877,
+      "loss": 0.9879,
       "num_input_tokens_seen": 170147840,
       "step": 20770
     },
     {
       "epoch": 0.9978391356542617,
-      "grad_norm": 0.47214624285697937,
+      "grad_norm": 0.47099846601486206,
       "learning_rate": 6.356331717305431e-10,
-      "loss": 0.8818,
+      "loss": 0.8828,
       "num_input_tokens_seen": 170229760,
       "step": 20780
     },
     {
       "epoch": 0.9983193277310924,
-      "grad_norm": 0.5106399655342102,
+      "grad_norm": 0.5169798135757446,
       "learning_rate": 3.8451947600437466e-10,
-      "loss": 0.9168,
+      "loss": 0.9171,
       "num_input_tokens_seen": 170311680,
       "step": 20790
     },
     {
       "epoch": 0.9987995198079231,
-      "grad_norm": 0.4835795760154724,
+      "grad_norm": 0.48473504185676575,
       "learning_rate": 1.9618365244833404e-10,
-      "loss": 0.8026,
+      "loss": 0.803,
       "num_input_tokens_seen": 170393600,
       "step": 20800
     },
     {
       "epoch": 0.999279711884754,
-      "grad_norm": 0.5011347532272339,
+      "grad_norm": 0.5023286938667297,
       "learning_rate": 7.062617399800075e-11,
-      "loss": 0.9025,
+      "loss": 0.9032,
       "num_input_tokens_seen": 170475520,
       "step": 20810
     },
     {
       "epoch": 0.9997599039615847,
-      "grad_norm": 0.48915189504623413,
+      "grad_norm": 0.4894420802593231,
       "learning_rate": 7.847355951162705e-12,
-      "loss": 0.9373,
+      "loss": 0.9376,
       "num_input_tokens_seen": 170557440,
       "step": 20820
     },
@@ -16669,10 +16669,10 @@
       "num_input_tokens_seen": 170598400,
       "step": 20825,
       "total_flos": 7.767833833163981e+18,
-      "train_loss": 0.9246458945440359,
-      "train_runtime": 106837.984,
-      "train_samples_per_second": 0.195,
-      "train_steps_per_second": 0.195
+      "train_loss": 0.2516314516834566,
+      "train_runtime": 26805.2364,
+      "train_samples_per_second": 0.777,
+      "train_steps_per_second": 0.777
     }
   ],
   "logging_steps": 10,