diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.2499897072748981,
+  "epoch": 0.4999794145497962,
   "eval_steps": 759,
-  "global_step": 759,
+  "global_step": 1518,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -5336,6 +5336,5327 @@
       "eval_samples_per_second": 3.215,
       "eval_steps_per_second": 1.608,
       "step": 759
+    },
+    {
+      "epoch": 0.2503190744781588,
+      "grad_norm": 2.326564073562622,
+      "learning_rate": 8.668665207923482e-05,
+      "loss": 2.0632,
+      "step": 760
+    },
+    {
+      "epoch": 0.2506484416814196,
+      "grad_norm": 2.884378671646118,
+      "learning_rate": 8.665088970082331e-05,
+      "loss": 2.5123,
+      "step": 761
+    },
+    {
+      "epoch": 0.2509778088846803,
+      "grad_norm": 2.3617143630981445,
+      "learning_rate": 8.661508675244801e-05,
+      "loss": 2.1108,
+      "step": 762
+    },
+    {
+      "epoch": 0.25130717608794106,
+      "grad_norm": 2.4837584495544434,
+      "learning_rate": 8.657924327374025e-05,
+      "loss": 2.3088,
+      "step": 763
+    },
+    {
+      "epoch": 0.2516365432912018,
+      "grad_norm": 2.4218530654907227,
+      "learning_rate": 8.654335930437627e-05,
+      "loss": 1.8391,
+      "step": 764
+    },
+    {
+      "epoch": 0.25196591049446254,
+      "grad_norm": 2.5416500568389893,
+      "learning_rate": 8.650743488407709e-05,
+      "loss": 2.1901,
+      "step": 765
+    },
+    {
+      "epoch": 0.25229527769772325,
+      "grad_norm": 2.6304333209991455,
+      "learning_rate": 8.647147005260854e-05,
+      "loss": 2.4257,
+      "step": 766
+    },
+    {
+      "epoch": 0.25262464490098396,
+      "grad_norm": 3.04967999458313,
+      "learning_rate": 8.643546484978115e-05,
+      "loss": 2.1485,
+      "step": 767
+    },
+    {
+      "epoch": 0.25295401210424473,
+      "grad_norm": 2.736220359802246,
+      "learning_rate": 8.639941931545017e-05,
+      "loss": 2.0484,
+      "step": 768
+    },
+    {
+      "epoch": 0.25328337930750544,
+      "grad_norm": 2.4498915672302246,
+      "learning_rate": 8.636333348951546e-05,
+      "loss": 1.9845,
+      "step": 769
+    },
+    {
+      "epoch": 0.2536127465107662,
+      "grad_norm": 3.074000597000122,
+      "learning_rate": 8.63272074119215e-05,
+      "loss": 2.0217,
+      "step": 770
+    },
+    {
+      "epoch": 0.2539421137140269,
+      "grad_norm": 2.7875514030456543,
+      "learning_rate": 8.629104112265735e-05,
+      "loss": 2.1935,
+      "step": 771
+    },
+    {
+      "epoch": 0.2542714809172877,
+      "grad_norm": 2.7638540267944336,
+      "learning_rate": 8.625483466175652e-05,
+      "loss": 2.1215,
+      "step": 772
+    },
+    {
+      "epoch": 0.2546008481205484,
+      "grad_norm": 2.99078106880188,
+      "learning_rate": 8.621858806929705e-05,
+      "loss": 1.8759,
+      "step": 773
+    },
+    {
+      "epoch": 0.2549302153238091,
+      "grad_norm": 3.0909841060638428,
+      "learning_rate": 8.618230138540136e-05,
+      "loss": 1.8021,
+      "step": 774
+    },
+    {
+      "epoch": 0.2552595825270699,
+      "grad_norm": 3.437960147857666,
+      "learning_rate": 8.614597465023626e-05,
+      "loss": 1.9589,
+      "step": 775
+    },
+    {
+      "epoch": 0.2555889497303306,
+      "grad_norm": 2.2460529804229736,
+      "learning_rate": 8.61096079040129e-05,
+      "loss": 2.2812,
+      "step": 776
+    },
+    {
+      "epoch": 0.25591831693359135,
+      "grad_norm": 2.4062023162841797,
+      "learning_rate": 8.607320118698674e-05,
+      "loss": 2.1588,
+      "step": 777
+    },
+    {
+      "epoch": 0.25624768413685206,
+      "grad_norm": 3.011057138442993,
+      "learning_rate": 8.603675453945742e-05,
+      "loss": 2.5091,
+      "step": 778
+    },
+    {
+      "epoch": 0.2565770513401128,
+      "grad_norm": 2.4397964477539062,
+      "learning_rate": 8.600026800176885e-05,
+      "loss": 2.4397,
+      "step": 779
+    },
+    {
+      "epoch": 0.25690641854337354,
+      "grad_norm": 2.548110246658325,
+      "learning_rate": 8.596374161430907e-05,
+      "loss": 2.4302,
+      "step": 780
+    },
+    {
+      "epoch": 0.2572357857466343,
+      "grad_norm": 2.477574586868286,
+      "learning_rate": 8.592717541751021e-05,
+      "loss": 2.2464,
+      "step": 781
+    },
+    {
+      "epoch": 0.257565152949895,
+      "grad_norm": 2.5044217109680176,
+      "learning_rate": 8.58905694518485e-05,
+      "loss": 2.13,
+      "step": 782
+    },
+    {
+      "epoch": 0.2578945201531557,
+      "grad_norm": 2.3912570476531982,
+      "learning_rate": 8.585392375784418e-05,
+      "loss": 2.3024,
+      "step": 783
+    },
+    {
+      "epoch": 0.2582238873564165,
+      "grad_norm": 2.51187801361084,
+      "learning_rate": 8.581723837606144e-05,
+      "loss": 2.3613,
+      "step": 784
+    },
+    {
+      "epoch": 0.2585532545596772,
+      "grad_norm": 2.4491524696350098,
+      "learning_rate": 8.578051334710844e-05,
+      "loss": 2.0575,
+      "step": 785
+    },
+    {
+      "epoch": 0.25888262176293797,
+      "grad_norm": 2.263449192047119,
+      "learning_rate": 8.574374871163721e-05,
+      "loss": 2.0833,
+      "step": 786
+    },
+    {
+      "epoch": 0.2592119889661987,
+      "grad_norm": 2.714989423751831,
+      "learning_rate": 8.570694451034362e-05,
+      "loss": 2.2463,
+      "step": 787
+    },
+    {
+      "epoch": 0.25954135616945945,
+      "grad_norm": 2.5804946422576904,
+      "learning_rate": 8.567010078396735e-05,
+      "loss": 2.2576,
+      "step": 788
+    },
+    {
+      "epoch": 0.25987072337272016,
+      "grad_norm": 2.759988784790039,
+      "learning_rate": 8.56332175732918e-05,
+      "loss": 2.2353,
+      "step": 789
+    },
+    {
+      "epoch": 0.26020009057598087,
+      "grad_norm": 2.70926570892334,
+      "learning_rate": 8.559629491914412e-05,
+      "loss": 2.0866,
+      "step": 790
+    },
+    {
+      "epoch": 0.26052945777924164,
+      "grad_norm": 3.139805555343628,
+      "learning_rate": 8.555933286239508e-05,
+      "loss": 2.4547,
+      "step": 791
+    },
+    {
+      "epoch": 0.26085882498250235,
+      "grad_norm": 2.390231132507324,
+      "learning_rate": 8.552233144395907e-05,
+      "loss": 1.9382,
+      "step": 792
+    },
+    {
+      "epoch": 0.2611881921857631,
+      "grad_norm": 2.88167142868042,
+      "learning_rate": 8.54852907047941e-05,
+      "loss": 2.1961,
+      "step": 793
+    },
+    {
+      "epoch": 0.2615175593890238,
+      "grad_norm": 2.3817718029022217,
+      "learning_rate": 8.544821068590165e-05,
+      "loss": 1.7874,
+      "step": 794
+    },
+    {
+      "epoch": 0.2618469265922846,
+      "grad_norm": 2.8125686645507812,
+      "learning_rate": 8.54110914283267e-05,
+      "loss": 2.1755,
+      "step": 795
+    },
+    {
+      "epoch": 0.2621762937955453,
+      "grad_norm": 2.719560384750366,
+      "learning_rate": 8.537393297315767e-05,
+      "loss": 2.0847,
+      "step": 796
+    },
+    {
+      "epoch": 0.26250566099880607,
+      "grad_norm": 2.7798001766204834,
+      "learning_rate": 8.533673536152638e-05,
+      "loss": 1.8334,
+      "step": 797
+    },
+    {
+      "epoch": 0.2628350282020668,
+      "grad_norm": 4.041043758392334,
+      "learning_rate": 8.529949863460793e-05,
+      "loss": 2.3578,
+      "step": 798
+    },
+    {
+      "epoch": 0.2631643954053275,
+      "grad_norm": 3.4026458263397217,
+      "learning_rate": 8.52622228336208e-05,
+      "loss": 2.0998,
+      "step": 799
+    },
+    {
+      "epoch": 0.26349376260858826,
+      "grad_norm": 4.440029144287109,
+      "learning_rate": 8.522490799982669e-05,
+      "loss": 1.7791,
+      "step": 800
+    },
+    {
+      "epoch": 0.26382312981184897,
+      "grad_norm": 2.11838436126709,
+      "learning_rate": 8.518755417453048e-05,
+      "loss": 2.4855,
+      "step": 801
+    },
+    {
+      "epoch": 0.26415249701510973,
+      "grad_norm": 2.5172719955444336,
+      "learning_rate": 8.515016139908024e-05,
+      "loss": 2.2894,
+      "step": 802
+    },
+    {
+      "epoch": 0.26448186421837044,
+      "grad_norm": 2.1991279125213623,
+      "learning_rate": 8.511272971486717e-05,
+      "loss": 2.2683,
+      "step": 803
+    },
+    {
+      "epoch": 0.2648112314216312,
+      "grad_norm": 2.425428867340088,
+      "learning_rate": 8.507525916332549e-05,
+      "loss": 2.2494,
+      "step": 804
+    },
+    {
+      "epoch": 0.2651405986248919,
+      "grad_norm": 2.5921618938446045,
+      "learning_rate": 8.503774978593248e-05,
+      "loss": 2.2534,
+      "step": 805
+    },
+    {
+      "epoch": 0.2654699658281527,
+      "grad_norm": 2.6209821701049805,
+      "learning_rate": 8.500020162420841e-05,
+      "loss": 2.3588,
+      "step": 806
+    },
+    {
+      "epoch": 0.2657993330314134,
+      "grad_norm": 2.933804750442505,
+      "learning_rate": 8.496261471971645e-05,
+      "loss": 2.5508,
+      "step": 807
+    },
+    {
+      "epoch": 0.2661287002346741,
+      "grad_norm": 2.4999735355377197,
+      "learning_rate": 8.492498911406266e-05,
+      "loss": 2.0871,
+      "step": 808
+    },
+    {
+      "epoch": 0.2664580674379349,
+      "grad_norm": 2.6054482460021973,
+      "learning_rate": 8.488732484889594e-05,
+      "loss": 2.1609,
+      "step": 809
+    },
+    {
+      "epoch": 0.2667874346411956,
+      "grad_norm": 2.588648557662964,
+      "learning_rate": 8.4849621965908e-05,
+      "loss": 2.2175,
+      "step": 810
+    },
+    {
+      "epoch": 0.26711680184445635,
+      "grad_norm": 2.7105255126953125,
+      "learning_rate": 8.481188050683328e-05,
+      "loss": 2.1727,
+      "step": 811
+    },
+    {
+      "epoch": 0.26744616904771706,
+      "grad_norm": 2.858750581741333,
+      "learning_rate": 8.477410051344895e-05,
+      "loss": 2.6405,
+      "step": 812
+    },
+    {
+      "epoch": 0.26777553625097783,
+      "grad_norm": 2.562743663787842,
+      "learning_rate": 8.473628202757478e-05,
+      "loss": 2.2222,
+      "step": 813
+    },
+    {
+      "epoch": 0.26810490345423854,
+      "grad_norm": 2.7561135292053223,
+      "learning_rate": 8.46984250910732e-05,
+      "loss": 2.3755,
+      "step": 814
+    },
+    {
+      "epoch": 0.26843427065749925,
+      "grad_norm": 2.5868582725524902,
+      "learning_rate": 8.466052974584918e-05,
+      "loss": 2.3861,
+      "step": 815
+    },
+    {
+      "epoch": 0.26876363786076,
+      "grad_norm": 2.836190938949585,
+      "learning_rate": 8.46225960338502e-05,
+      "loss": 2.1303,
+      "step": 816
+    },
+    {
+      "epoch": 0.26909300506402073,
+      "grad_norm": 3.028157949447632,
+      "learning_rate": 8.458462399706623e-05,
+      "loss": 2.2734,
+      "step": 817
+    },
+    {
+      "epoch": 0.2694223722672815,
+      "grad_norm": 2.804187536239624,
+      "learning_rate": 8.454661367752965e-05,
+      "loss": 2.1357,
+      "step": 818
+    },
+    {
+      "epoch": 0.2697517394705422,
+      "grad_norm": 2.80621075630188,
+      "learning_rate": 8.450856511731519e-05,
+      "loss": 1.732,
+      "step": 819
+    },
+    {
+      "epoch": 0.270081106673803,
+      "grad_norm": 2.5028510093688965,
+      "learning_rate": 8.447047835853999e-05,
+      "loss": 1.8806,
+      "step": 820
+    },
+    {
+      "epoch": 0.2704104738770637,
+      "grad_norm": 2.597686529159546,
+      "learning_rate": 8.443235344336337e-05,
+      "loss": 2.1617,
+      "step": 821
+    },
+    {
+      "epoch": 0.27073984108032445,
+      "grad_norm": 2.7407145500183105,
+      "learning_rate": 8.439419041398698e-05,
+      "loss": 2.157,
+      "step": 822
+    },
+    {
+      "epoch": 0.27106920828358516,
+      "grad_norm": 3.096123695373535,
+      "learning_rate": 8.435598931265459e-05,
+      "loss": 1.9549,
+      "step": 823
+    },
+    {
+      "epoch": 0.27139857548684587,
+      "grad_norm": 4.0686516761779785,
+      "learning_rate": 8.431775018165211e-05,
+      "loss": 2.139,
+      "step": 824
+    },
+    {
+      "epoch": 0.27172794269010664,
+      "grad_norm": 2.7538039684295654,
+      "learning_rate": 8.427947306330764e-05,
+      "loss": 1.8598,
+      "step": 825
+    },
+    {
+      "epoch": 0.27205730989336735,
+      "grad_norm": 2.3142900466918945,
+      "learning_rate": 8.424115799999122e-05,
+      "loss": 2.3159,
+      "step": 826
+    },
+    {
+      "epoch": 0.2723866770966281,
+      "grad_norm": 2.2066521644592285,
+      "learning_rate": 8.420280503411495e-05,
+      "loss": 2.2287,
+      "step": 827
+    },
+    {
+      "epoch": 0.2727160442998888,
+      "grad_norm": 2.188645839691162,
+      "learning_rate": 8.416441420813288e-05,
+      "loss": 2.2174,
+      "step": 828
+    },
+    {
+      "epoch": 0.2730454115031496,
+      "grad_norm": 2.808419704437256,
+      "learning_rate": 8.412598556454096e-05,
+      "loss": 2.6173,
+      "step": 829
+    },
+    {
+      "epoch": 0.2733747787064103,
+      "grad_norm": 2.461317300796509,
+      "learning_rate": 8.408751914587699e-05,
+      "loss": 2.5954,
+      "step": 830
+    },
+    {
+      "epoch": 0.27370414590967107,
+      "grad_norm": 2.9252431392669678,
+      "learning_rate": 8.40490149947206e-05,
+      "loss": 2.4617,
+      "step": 831
+    },
+    {
+      "epoch": 0.2740335131129318,
+      "grad_norm": 2.5286619663238525,
+      "learning_rate": 8.401047315369319e-05,
+      "loss": 2.5215,
+      "step": 832
+    },
+    {
+      "epoch": 0.2743628803161925,
+      "grad_norm": 2.650164842605591,
+      "learning_rate": 8.397189366545786e-05,
+      "loss": 2.1433,
+      "step": 833
+    },
+    {
+      "epoch": 0.27469224751945326,
+      "grad_norm": 2.2741010189056396,
+      "learning_rate": 8.393327657271944e-05,
+      "loss": 2.0067,
+      "step": 834
+    },
+    {
+      "epoch": 0.27502161472271397,
+      "grad_norm": 2.3971540927886963,
+      "learning_rate": 8.389462191822428e-05,
+      "loss": 2.2778,
+      "step": 835
+    },
+    {
+      "epoch": 0.27535098192597474,
+      "grad_norm": 2.9199283123016357,
+      "learning_rate": 8.385592974476042e-05,
+      "loss": 2.375,
+      "step": 836
+    },
+    {
+      "epoch": 0.27568034912923545,
+      "grad_norm": 2.3536462783813477,
+      "learning_rate": 8.381720009515736e-05,
+      "loss": 2.2465,
+      "step": 837
+    },
+    {
+      "epoch": 0.2760097163324962,
+      "grad_norm": 2.7381389141082764,
+      "learning_rate": 8.377843301228611e-05,
+      "loss": 2.201,
+      "step": 838
+    },
+    {
+      "epoch": 0.2763390835357569,
+      "grad_norm": 2.795403003692627,
+      "learning_rate": 8.373962853905912e-05,
+      "loss": 2.1858,
+      "step": 839
+    },
+    {
+      "epoch": 0.27666845073901764,
+      "grad_norm": 2.549487829208374,
+      "learning_rate": 8.37007867184302e-05,
+      "loss": 2.0456,
+      "step": 840
+    },
+    {
+      "epoch": 0.2769978179422784,
+      "grad_norm": 2.419311046600342,
+      "learning_rate": 8.366190759339453e-05,
+      "loss": 1.8163,
+      "step": 841
+    },
+    {
+      "epoch": 0.2773271851455391,
+      "grad_norm": 2.8960893154144287,
+      "learning_rate": 8.362299120698858e-05,
+      "loss": 2.0316,
+      "step": 842
+    },
+    {
+      "epoch": 0.2776565523487999,
+      "grad_norm": 2.440534830093384,
+      "learning_rate": 8.358403760229006e-05,
+      "loss": 1.8316,
+      "step": 843
+    },
+    {
+      "epoch": 0.2779859195520606,
+      "grad_norm": 2.9307198524475098,
+      "learning_rate": 8.354504682241786e-05,
+      "loss": 2.5626,
+      "step": 844
+    },
+    {
+      "epoch": 0.27831528675532136,
+      "grad_norm": 2.7770323753356934,
+      "learning_rate": 8.350601891053207e-05,
+      "loss": 2.2409,
+      "step": 845
+    },
+    {
+      "epoch": 0.27864465395858207,
+      "grad_norm": 2.7087655067443848,
+      "learning_rate": 8.346695390983382e-05,
+      "loss": 1.9896,
+      "step": 846
+    },
+    {
+      "epoch": 0.27897402116184283,
+      "grad_norm": 2.673274517059326,
+      "learning_rate": 8.342785186356534e-05,
+      "loss": 2.0128,
+      "step": 847
+    },
+    {
+      "epoch": 0.27930338836510354,
+      "grad_norm": 3.0754430294036865,
+      "learning_rate": 8.338871281500984e-05,
+      "loss": 2.008,
+      "step": 848
+    },
+    {
+      "epoch": 0.27963275556836426,
+      "grad_norm": 2.8769547939300537,
+      "learning_rate": 8.334953680749152e-05,
+      "loss": 1.8857,
+      "step": 849
+    },
+    {
+      "epoch": 0.279962122771625,
+      "grad_norm": 3.0530593395233154,
+      "learning_rate": 8.331032388437546e-05,
+      "loss": 1.967,
+      "step": 850
+    },
+    {
+      "epoch": 0.28029148997488573,
+      "grad_norm": 1.8923742771148682,
+      "learning_rate": 8.32710740890676e-05,
+      "loss": 2.4918,
+      "step": 851
+    },
+    {
+      "epoch": 0.2806208571781465,
+      "grad_norm": 2.0591697692871094,
+      "learning_rate": 8.323178746501475e-05,
+      "loss": 2.2662,
+      "step": 852
+    },
+    {
+      "epoch": 0.2809502243814072,
+      "grad_norm": 2.4539167881011963,
+      "learning_rate": 8.319246405570441e-05,
+      "loss": 2.4903,
+      "step": 853
+    },
+    {
+      "epoch": 0.281279591584668,
+      "grad_norm": 2.1013600826263428,
+      "learning_rate": 8.315310390466487e-05,
+      "loss": 2.1026,
+      "step": 854
+    },
+    {
+      "epoch": 0.2816089587879287,
+      "grad_norm": 2.0880558490753174,
+      "learning_rate": 8.311370705546501e-05,
+      "loss": 2.3886,
+      "step": 855
+    },
+    {
+      "epoch": 0.28193832599118945,
+      "grad_norm": 2.4675230979919434,
+      "learning_rate": 8.307427355171443e-05,
+      "loss": 2.2145,
+      "step": 856
+    },
+    {
+      "epoch": 0.28226769319445016,
+      "grad_norm": 2.3811771869659424,
+      "learning_rate": 8.303480343706321e-05,
+      "loss": 2.6376,
+      "step": 857
+    },
+    {
+      "epoch": 0.2825970603977109,
+      "grad_norm": 2.4288272857666016,
+      "learning_rate": 8.299529675520201e-05,
+      "loss": 2.2381,
+      "step": 858
+    },
+    {
+      "epoch": 0.28292642760097164,
+      "grad_norm": 2.4424099922180176,
+      "learning_rate": 8.295575354986196e-05,
+      "loss": 2.4726,
+      "step": 859
+    },
+    {
+      "epoch": 0.28325579480423235,
+      "grad_norm": 2.6503515243530273,
+      "learning_rate": 8.291617386481463e-05,
+      "loss": 2.4256,
+      "step": 860
+    },
+    {
+      "epoch": 0.2835851620074931,
+      "grad_norm": 2.2516067028045654,
+      "learning_rate": 8.287655774387193e-05,
+      "loss": 1.9415,
+      "step": 861
+    },
+    {
+      "epoch": 0.28391452921075383,
+      "grad_norm": 2.435704231262207,
+      "learning_rate": 8.283690523088613e-05,
+      "loss": 2.0356,
+      "step": 862
+    },
+    {
+      "epoch": 0.2842438964140146,
+      "grad_norm": 2.6750681400299072,
+      "learning_rate": 8.279721636974978e-05,
+      "loss": 2.2815,
+      "step": 863
+    },
+    {
+      "epoch": 0.2845732636172753,
+      "grad_norm": 3.6107234954833984,
+      "learning_rate": 8.275749120439569e-05,
+      "loss": 2.2028,
+      "step": 864
+    },
+    {
+      "epoch": 0.284902630820536,
+      "grad_norm": 2.5893361568450928,
+      "learning_rate": 8.27177297787968e-05,
+      "loss": 2.1027,
+      "step": 865
+    },
+    {
+      "epoch": 0.2852319980237968,
+      "grad_norm": 2.4202561378479004,
+      "learning_rate": 8.267793213696624e-05,
+      "loss": 2.0077,
+      "step": 866
+    },
+    {
+      "epoch": 0.2855613652270575,
+      "grad_norm": 2.8321430683135986,
+      "learning_rate": 8.263809832295719e-05,
+      "loss": 1.9374,
+      "step": 867
+    },
+    {
+      "epoch": 0.28589073243031826,
+      "grad_norm": 2.830946683883667,
+      "learning_rate": 8.25982283808629e-05,
+      "loss": 2.3701,
+      "step": 868
+    },
+    {
+      "epoch": 0.286220099633579,
+      "grad_norm": 2.626882791519165,
+      "learning_rate": 8.255832235481659e-05,
+      "loss": 2.2236,
+      "step": 869
+    },
+    {
+      "epoch": 0.28654946683683974,
+      "grad_norm": 2.398036003112793,
+      "learning_rate": 8.251838028899143e-05,
+      "loss": 2.2344,
+      "step": 870
+    },
+    {
+      "epoch": 0.28687883404010045,
+      "grad_norm": 3.660982608795166,
+      "learning_rate": 8.247840222760052e-05,
+      "loss": 2.2156,
+      "step": 871
+    },
+    {
+      "epoch": 0.2872082012433612,
+      "grad_norm": 3.3050460815429688,
+      "learning_rate": 8.243838821489671e-05,
+      "loss": 2.0017,
+      "step": 872
+    },
+    {
+      "epoch": 0.2875375684466219,
+      "grad_norm": 3.096376895904541,
+      "learning_rate": 8.239833829517276e-05,
+      "loss": 1.9194,
+      "step": 873
+    },
+    {
+      "epoch": 0.28786693564988264,
+      "grad_norm": 2.761716365814209,
+      "learning_rate": 8.235825251276108e-05,
+      "loss": 1.9807,
+      "step": 874
+    },
+    {
+      "epoch": 0.2881963028531434,
+      "grad_norm": 3.867349624633789,
+      "learning_rate": 8.231813091203385e-05,
+      "loss": 1.9222,
+      "step": 875
+    },
+    {
+      "epoch": 0.2885256700564041,
+      "grad_norm": 2.335017442703247,
+      "learning_rate": 8.227797353740286e-05,
+      "loss": 2.4247,
+      "step": 876
+    },
+    {
+      "epoch": 0.2888550372596649,
+      "grad_norm": 2.4474287033081055,
+      "learning_rate": 8.223778043331948e-05,
+      "loss": 2.6098,
+      "step": 877
+    },
+    {
+      "epoch": 0.2891844044629256,
+      "grad_norm": 2.22808575630188,
+      "learning_rate": 8.219755164427469e-05,
+      "loss": 2.2784,
+      "step": 878
+    },
+    {
+      "epoch": 0.28951377166618636,
+      "grad_norm": 2.265848398208618,
+      "learning_rate": 8.215728721479892e-05,
+      "loss": 2.2767,
+      "step": 879
+    },
+    {
+      "epoch": 0.28984313886944707,
+      "grad_norm": 2.5640437602996826,
+      "learning_rate": 8.211698718946208e-05,
+      "loss": 1.9147,
+      "step": 880
+    },
+    {
+      "epoch": 0.29017250607270784,
+      "grad_norm": 2.4126553535461426,
+      "learning_rate": 8.207665161287345e-05,
+      "loss": 2.2837,
+      "step": 881
+    },
+    {
+      "epoch": 0.29050187327596855,
+      "grad_norm": 2.4454050064086914,
+      "learning_rate": 8.20362805296817e-05,
+      "loss": 2.3148,
+      "step": 882
+    },
+    {
+      "epoch": 0.29083124047922926,
+      "grad_norm": 2.776961088180542,
+      "learning_rate": 8.19958739845748e-05,
+      "loss": 2.7321,
+      "step": 883
+    },
+    {
+      "epoch": 0.29116060768249,
+      "grad_norm": 2.721742630004883,
+      "learning_rate": 8.195543202227993e-05,
+      "loss": 2.3516,
+      "step": 884
+    },
+    {
+      "epoch": 0.29148997488575074,
+      "grad_norm": 2.417945384979248,
+      "learning_rate": 8.191495468756354e-05,
+      "loss": 2.0815,
+      "step": 885
+    },
+    {
+      "epoch": 0.2918193420890115,
+      "grad_norm": 2.7946279048919678,
+      "learning_rate": 8.187444202523116e-05,
+      "loss": 2.5238,
+      "step": 886
+    },
+    {
+      "epoch": 0.2921487092922722,
+      "grad_norm": 2.2200636863708496,
+      "learning_rate": 8.183389408012752e-05,
+      "loss": 2.2116,
+      "step": 887
+    },
+    {
+      "epoch": 0.292478076495533,
+      "grad_norm": 2.7158875465393066,
+      "learning_rate": 8.179331089713629e-05,
+      "loss": 2.2724,
+      "step": 888
+    },
+    {
+      "epoch": 0.2928074436987937,
+      "grad_norm": 3.0811691284179688,
+      "learning_rate": 8.175269252118023e-05,
+      "loss": 2.481,
+      "step": 889
+    },
+    {
+      "epoch": 0.2931368109020544,
+      "grad_norm": 2.898715019226074,
+      "learning_rate": 8.171203899722105e-05,
+      "loss": 1.9759,
+      "step": 890
+    },
+    {
+      "epoch": 0.29346617810531517,
+      "grad_norm": 3.110511541366577,
+      "learning_rate": 8.167135037025933e-05,
+      "loss": 2.3635,
+      "step": 891
+    },
+    {
+      "epoch": 0.2937955453085759,
+      "grad_norm": 2.4876153469085693,
+      "learning_rate": 8.163062668533454e-05,
+      "loss": 2.2224,
+      "step": 892
+    },
+    {
+      "epoch": 0.29412491251183664,
+      "grad_norm": 2.962139844894409,
+      "learning_rate": 8.158986798752492e-05,
+      "loss": 2.1494,
+      "step": 893
+    },
+    {
+      "epoch": 0.29445427971509736,
+      "grad_norm": 2.6857833862304688,
+      "learning_rate": 8.154907432194751e-05,
+      "loss": 1.9066,
+      "step": 894
+    },
+    {
+      "epoch": 0.2947836469183581,
+      "grad_norm": 3.1264843940734863,
+      "learning_rate": 8.150824573375804e-05,
+      "loss": 2.1503,
+      "step": 895
+    },
+    {
+      "epoch": 0.29511301412161883,
+      "grad_norm": 3.0480618476867676,
+      "learning_rate": 8.146738226815087e-05,
+      "loss": 2.2263,
+      "step": 896
+    },
+    {
+      "epoch": 0.2954423813248796,
+      "grad_norm": 2.6517231464385986,
+      "learning_rate": 8.142648397035899e-05,
+      "loss": 1.98,
+      "step": 897
+    },
+    {
+      "epoch": 0.2957717485281403,
+      "grad_norm": 2.765357255935669,
+      "learning_rate": 8.138555088565398e-05,
+      "loss": 1.8292,
+      "step": 898
+    },
+    {
+      "epoch": 0.296101115731401,
+      "grad_norm": 2.788534641265869,
+      "learning_rate": 8.134458305934587e-05,
+      "loss": 1.7272,
+      "step": 899
+    },
+    {
+      "epoch": 0.2964304829346618,
+      "grad_norm": 3.0059924125671387,
+      "learning_rate": 8.130358053678315e-05,
+      "loss": 2.0136,
+      "step": 900
+    },
+    {
+      "epoch": 0.2967598501379225,
+      "grad_norm": 2.0941505432128906,
+      "learning_rate": 8.126254336335279e-05,
+      "loss": 2.5732,
+      "step": 901
+    },
+    {
+      "epoch": 0.29708921734118326,
+      "grad_norm": 2.0665740966796875,
+      "learning_rate": 8.122147158448002e-05,
+      "loss": 2.4054,
+      "step": 902
+    },
+    {
+      "epoch": 0.297418584544444,
+      "grad_norm": 2.2430355548858643,
+      "learning_rate": 8.118036524562841e-05,
+      "loss": 2.4054,
+      "step": 903
+    },
+    {
+      "epoch": 0.29774795174770474,
+      "grad_norm": 2.2699286937713623,
+      "learning_rate": 8.113922439229982e-05,
+      "loss": 2.1162,
+      "step": 904
+    },
+    {
+      "epoch": 0.29807731895096545,
+      "grad_norm": 2.1671202182769775,
+      "learning_rate": 8.109804907003429e-05,
+      "loss": 2.3838,
+      "step": 905
+    },
+    {
+      "epoch": 0.2984066861542262,
+      "grad_norm": 2.175811529159546,
+      "learning_rate": 8.105683932441e-05,
+      "loss": 2.0848,
+      "step": 906
+    },
+    {
+      "epoch": 0.29873605335748693,
+      "grad_norm": 3.168586492538452,
+      "learning_rate": 8.101559520104323e-05,
+      "loss": 2.4891,
+      "step": 907
+    },
+    {
+      "epoch": 0.29906542056074764,
+      "grad_norm": 2.39568829536438,
+      "learning_rate": 8.097431674558838e-05,
+      "loss": 2.3179,
+      "step": 908
+    },
+    {
+      "epoch": 0.2993947877640084,
+      "grad_norm": 2.7230722904205322,
+      "learning_rate": 8.093300400373775e-05,
+      "loss": 2.2789,
+      "step": 909
+    },
+    {
+      "epoch": 0.2997241549672691,
+      "grad_norm": 2.686748504638672,
+      "learning_rate": 8.08916570212217e-05,
+      "loss": 2.1697,
+      "step": 910
+    },
+    {
+      "epoch": 0.3000535221705299,
+      "grad_norm": 2.3742809295654297,
+      "learning_rate": 8.08502758438084e-05,
+      "loss": 1.9485,
+      "step": 911
+    },
+    {
+      "epoch": 0.3003828893737906,
+      "grad_norm": 2.234506607055664,
+      "learning_rate": 8.080886051730391e-05,
+      "loss": 2.2139,
+      "step": 912
+    },
+    {
+      "epoch": 0.30071225657705136,
+      "grad_norm": 2.438096284866333,
+      "learning_rate": 8.076741108755212e-05,
+      "loss": 2.338,
+      "step": 913
+    },
+    {
+      "epoch": 0.3010416237803121,
+      "grad_norm": 2.3559389114379883,
+      "learning_rate": 8.072592760043463e-05,
+      "loss": 2.3608,
+      "step": 914
+    },
+    {
+      "epoch": 0.3013709909835728,
+      "grad_norm": 2.8507134914398193,
+      "learning_rate": 8.068441010187073e-05,
+      "loss": 2.3359,
+      "step": 915
+    },
+    {
+      "epoch": 0.30170035818683355,
+      "grad_norm": 3.0053601264953613,
+      "learning_rate": 8.06428586378174e-05,
+      "loss": 2.2598,
+      "step": 916
+    },
+    {
+      "epoch": 0.30202972539009426,
+      "grad_norm": 2.6545512676239014,
+      "learning_rate": 8.06012732542692e-05,
+      "loss": 2.2021,
+      "step": 917
+    },
+    {
+      "epoch": 0.302359092593355,
+      "grad_norm": 2.2311315536499023,
+      "learning_rate": 8.05596539972582e-05,
+      "loss": 2.0849,
+      "step": 918
+    },
+    {
+      "epoch": 0.30268845979661574,
+      "grad_norm": 3.3348686695098877,
+      "learning_rate": 8.051800091285404e-05,
+      "loss": 2.2344,
+      "step": 919
+    },
+    {
+      "epoch": 0.3030178269998765,
+      "grad_norm": 2.89279842376709,
+      "learning_rate": 8.047631404716374e-05,
+      "loss": 2.5125,
+      "step": 920
+    },
+    {
+      "epoch": 0.3033471942031372,
+      "grad_norm": 2.6193151473999023,
+      "learning_rate": 8.043459344633173e-05,
+      "loss": 1.7652,
+      "step": 921
+    },
+    {
+      "epoch": 0.303676561406398,
+      "grad_norm": 2.6484382152557373,
+      "learning_rate": 8.039283915653979e-05,
+      "loss": 1.9555,
+      "step": 922
+    },
+    {
+      "epoch": 0.3040059286096587,
+      "grad_norm": 2.553636312484741,
+      "learning_rate": 8.035105122400701e-05,
+      "loss": 1.8701,
+      "step": 923
+    },
+    {
+      "epoch": 0.3043352958129194,
+      "grad_norm": 3.712156057357788,
+      "learning_rate": 8.030922969498968e-05,
+      "loss": 2.2537,
+      "step": 924
+    },
+    {
+      "epoch": 0.30466466301618017,
+      "grad_norm": 3.272002696990967,
+      "learning_rate": 8.026737461578132e-05,
+      "loss": 2.1644,
+      "step": 925
+    },
+    {
+      "epoch": 0.3049940302194409,
+      "grad_norm": 2.4612069129943848,
+      "learning_rate": 8.022548603271252e-05,
+      "loss": 2.4379,
+      "step": 926
+    },
+    {
+      "epoch": 0.30532339742270165,
+      "grad_norm": 3.1458165645599365,
+      "learning_rate": 8.018356399215104e-05,
+      "loss": 2.6069,
+      "step": 927
+    },
+    {
+      "epoch": 0.30565276462596236,
+      "grad_norm": 2.0101094245910645,
+      "learning_rate": 8.014160854050164e-05,
+      "loss": 2.185,
+      "step": 928
+    },
+    {
+      "epoch": 0.3059821318292231,
+      "grad_norm": 2.1659107208251953,
+      "learning_rate": 8.009961972420607e-05,
+      "loss": 2.1568,
+      "step": 929
+    },
+    {
+      "epoch": 0.30631149903248384,
+      "grad_norm": 2.243806838989258,
+      "learning_rate": 8.005759758974296e-05,
+      "loss": 2.4728,
+      "step": 930
+    },
+    {
+      "epoch": 0.3066408662357446,
+      "grad_norm": 2.525348424911499,
+      "learning_rate": 8.001554218362791e-05,
+      "loss": 2.6265,
+      "step": 931
+    },
+    {
+      "epoch": 0.3069702334390053,
+      "grad_norm": 2.3284335136413574,
+      "learning_rate": 7.997345355241328e-05,
+      "loss": 2.3675,
+      "step": 932
+    },
+    {
+      "epoch": 0.307299600642266,
+      "grad_norm": 2.4215810298919678,
+      "learning_rate": 7.993133174268826e-05,
+      "loss": 2.1221,
+      "step": 933
+    },
+    {
+      "epoch": 0.3076289678455268,
+      "grad_norm": 2.8312551975250244,
+      "learning_rate": 7.988917680107871e-05,
+      "loss": 2.5939,
+      "step": 934
+    },
+    {
+      "epoch": 0.3079583350487875,
+      "grad_norm": 2.665449857711792,
+      "learning_rate": 7.984698877424718e-05,
+      "loss": 2.24,
+      "step": 935
+    },
+    {
+      "epoch": 0.30828770225204827,
+      "grad_norm": 2.5410232543945312,
+      "learning_rate": 7.980476770889289e-05,
+      "loss": 2.2135,
+      "step": 936
+    },
+    {
+      "epoch": 0.308617069455309,
+      "grad_norm": 2.556459426879883,
+      "learning_rate": 7.976251365175158e-05,
+      "loss": 2.2969,
+      "step": 937
+    },
+    {
+      "epoch": 0.30894643665856975,
+      "grad_norm": 3.4073662757873535,
+      "learning_rate": 7.972022664959554e-05,
+      "loss": 2.3814,
+      "step": 938
+    },
+    {
+      "epoch": 0.30927580386183046,
+      "grad_norm": 2.888021230697632,
+      "learning_rate": 7.96779067492335e-05,
+      "loss": 2.4569,
+      "step": 939
+    },
+    {
+      "epoch": 0.30960517106509117,
+      "grad_norm": 2.819286584854126,
+      "learning_rate": 7.963555399751063e-05,
+      "loss": 2.1128,
+      "step": 940
+    },
+    {
+      "epoch": 0.30993453826835193,
+      "grad_norm": 2.760483741760254,
+      "learning_rate": 7.959316844130846e-05,
+      "loss": 2.4026,
+      "step": 941
+    },
+    {
+      "epoch": 0.31026390547161264,
+      "grad_norm": 2.6997532844543457,
+      "learning_rate": 7.95507501275448e-05,
+      "loss": 2.0134,
+      "step": 942
+    },
+    {
+      "epoch": 0.3105932726748734,
+      "grad_norm": 2.7757112979888916,
+      "learning_rate": 7.950829910317379e-05,
+      "loss": 2.207,
+      "step": 943
+    },
+    {
+      "epoch": 0.3109226398781341,
+      "grad_norm": 2.5524275302886963,
+      "learning_rate": 7.946581541518569e-05,
+      "loss": 1.7721,
+      "step": 944
+    },
+    {
+      "epoch": 0.3112520070813949,
+      "grad_norm": 2.4421706199645996,
+      "learning_rate": 7.942329911060703e-05,
+      "loss": 2.1257,
+      "step": 945
+    },
+    {
+      "epoch": 0.3115813742846556,
+      "grad_norm": 2.7249844074249268,
+      "learning_rate": 7.938075023650029e-05,
+      "loss": 1.9599,
+      "step": 946
+    },
+    {
+      "epoch": 0.31191074148791637,
+      "grad_norm": 2.9589931964874268,
+      "learning_rate": 7.933816883996415e-05,
+      "loss": 2.3046,
+      "step": 947
+    },
+    {
+      "epoch": 0.3122401086911771,
+      "grad_norm": 2.614107608795166,
+      "learning_rate": 7.92955549681332e-05,
+      "loss": 1.8874,
+      "step": 948
+    },
+    {
+      "epoch": 0.3125694758944378,
+      "grad_norm": 3.2241110801696777,
+      "learning_rate": 7.925290866817802e-05,
+      "loss": 2.3943,
+      "step": 949
+    },
+    {
+      "epoch": 0.31289884309769855,
+      "grad_norm": 3.045088291168213,
+      "learning_rate": 7.921022998730507e-05,
+      "loss": 1.961,
+      "step": 950
+    },
+    {
+      "epoch": 0.31322821030095926,
+      "grad_norm": 2.1225154399871826,
+      "learning_rate": 7.916751897275665e-05,
+      "loss": 2.749,
+      "step": 951
+    },
+    {
+      "epoch": 0.31355757750422003,
+      "grad_norm": 2.060499668121338,
+      "learning_rate": 7.912477567181086e-05,
+      "loss": 2.2708,
+      "step": 952
+    },
+    {
+      "epoch": 0.31388694470748074,
+      "grad_norm": 2.4853827953338623,
+      "learning_rate": 7.908200013178156e-05,
+      "loss": 2.2059,
+      "step": 953
+    },
+    {
+      "epoch": 0.3142163119107415,
+      "grad_norm": 2.317662477493286,
+      "learning_rate": 7.903919240001824e-05,
+      "loss": 2.4268,
+      "step": 954
+    },
+    {
+      "epoch": 0.3145456791140022,
+      "grad_norm": 2.5555713176727295,
+      "learning_rate": 7.899635252390606e-05,
+      "loss": 2.4417,
+      "step": 955
+    },
+    {
+      "epoch": 0.314875046317263,
+      "grad_norm": 2.6387650966644287,
+      "learning_rate": 7.895348055086577e-05,
+      "loss": 2.3655,
+      "step": 956
+    },
+    {
+      "epoch": 0.3152044135205237,
+      "grad_norm": 2.4570939540863037,
+      "learning_rate": 7.891057652835361e-05,
+      "loss": 2.4014,
+      "step": 957
+    },
+    {
+      "epoch": 0.3155337807237844,
+      "grad_norm": 2.4972894191741943,
+      "learning_rate": 7.886764050386135e-05,
+      "loss": 2.3593,
+      "step": 958
+    },
+    {
+      "epoch": 0.3158631479270452,
+      "grad_norm": 2.601262092590332,
+      "learning_rate": 7.882467252491617e-05,
+      "loss": 2.4721,
+      "step": 959
+    },
+    {
+      "epoch": 0.3161925151303059,
+      "grad_norm": 2.5913026332855225,
+      "learning_rate": 7.878167263908056e-05,
+      "loss": 2.2818,
+      "step": 960
+    },
+    {
+      "epoch": 0.31652188233356665,
+      "grad_norm": 2.6688802242279053,
+      "learning_rate": 7.873864089395243e-05,
+      "loss": 2.1604,
+      "step": 961
+    },
+    {
+      "epoch": 0.31685124953682736,
+      "grad_norm": 2.4998068809509277,
+      "learning_rate": 7.869557733716488e-05,
+      "loss": 1.95,
+      "step": 962
+    },
+    {
+      "epoch": 0.31718061674008813,
+      "grad_norm": 2.5691847801208496,
+      "learning_rate": 7.865248201638623e-05,
+      "loss": 2.1274,
+      "step": 963
+    },
+    {
+      "epoch": 0.31750998394334884,
+      "grad_norm": 2.2748234272003174,
+      "learning_rate": 7.860935497932e-05,
+      "loss": 1.9988,
+      "step": 964
+    },
+    {
+      "epoch": 0.31783935114660955,
+      "grad_norm": 2.8164100646972656,
+      "learning_rate": 7.856619627370479e-05,
+      "loss": 2.3755,
+      "step": 965
+    },
+    {
+      "epoch": 0.3181687183498703,
+      "grad_norm": 2.8128836154937744,
+      "learning_rate": 7.852300594731425e-05,
+      "loss": 2.1516,
+      "step": 966
+    },
+    {
+      "epoch": 0.318498085553131,
+      "grad_norm": 2.7594070434570312,
+      "learning_rate": 7.847978404795704e-05,
+      "loss": 2.177,
+      "step": 967
+    },
+    {
+      "epoch": 0.3188274527563918,
+      "grad_norm": 2.687586784362793,
+      "learning_rate": 7.843653062347679e-05,
+      "loss": 2.4126,
+      "step": 968
+    },
+    {
+      "epoch": 0.3191568199596525,
+      "grad_norm": 2.610460042953491,
+      "learning_rate": 7.8393245721752e-05,
+      "loss": 2.3253,
+      "step": 969
+    },
+    {
+      "epoch": 0.31948618716291327,
+      "grad_norm": 2.525682210922241,
+      "learning_rate": 7.8349929390696e-05,
+      "loss": 1.9292,
+      "step": 970
+    },
+    {
+      "epoch": 0.319815554366174,
+      "grad_norm": 3.1786725521087646,
+      "learning_rate": 7.830658167825696e-05,
+      "loss": 1.956,
+      "step": 971
+    },
+    {
+      "epoch": 0.32014492156943475,
+      "grad_norm": 2.809406280517578,
+      "learning_rate": 7.826320263241771e-05,
+      "loss": 2.1892,
+      "step": 972
+    },
+    {
+      "epoch": 0.32047428877269546,
+      "grad_norm": 3.148411750793457,
+      "learning_rate": 7.821979230119587e-05,
+      "loss": 2.0155,
+      "step": 973
+    },
+    {
+      "epoch": 0.32080365597595617,
+      "grad_norm": 3.3248815536499023,
+      "learning_rate": 7.81763507326436e-05,
+      "loss": 1.7539,
+      "step": 974
+    },
+    {
+      "epoch": 0.32113302317921694,
+      "grad_norm": 3.2047150135040283,
+      "learning_rate": 7.813287797484768e-05,
+      "loss": 2.028,
+      "step": 975
+    },
+    {
+      "epoch": 0.32146239038247765,
+      "grad_norm": 2.169050455093384,
+      "learning_rate": 7.808937407592938e-05,
+      "loss": 2.4696,
+      "step": 976
+    },
+    {
+      "epoch": 0.3217917575857384,
+      "grad_norm": 2.0645432472229004,
+      "learning_rate": 7.804583908404448e-05,
+      "loss": 2.1029,
+      "step": 977
+    },
+    {
+      "epoch": 0.3221211247889991,
+      "grad_norm": 2.496066093444824,
+      "learning_rate": 7.800227304738317e-05,
+      "loss": 2.3193,
+      "step": 978
+    },
+    {
+      "epoch": 0.3224504919922599,
+      "grad_norm": 2.2872400283813477,
+      "learning_rate": 7.795867601416998e-05,
+      "loss": 2.4506,
+      "step": 979
+    },
+    {
+      "epoch": 0.3227798591955206,
+      "grad_norm": 2.6444544792175293,
+      "learning_rate": 7.791504803266377e-05,
+      "loss": 2.818,
+      "step": 980
+    },
+    {
+      "epoch": 0.32310922639878137,
+      "grad_norm": 2.2178843021392822,
+      "learning_rate": 7.787138915115768e-05,
+      "loss": 1.9535,
+      "step": 981
+    },
+    {
+      "epoch": 0.3234385936020421,
+      "grad_norm": 2.528449535369873,
+      "learning_rate": 7.782769941797899e-05,
+      "loss": 2.1525,
+      "step": 982
+    },
+    {
+      "epoch": 0.3237679608053028,
+      "grad_norm": 2.597266435623169,
+      "learning_rate": 7.778397888148921e-05,
+      "loss": 2.4295,
+      "step": 983
+    },
+    {
+      "epoch": 0.32409732800856356,
+      "grad_norm": 2.5413691997528076,
+      "learning_rate": 7.774022759008386e-05,
+      "loss": 2.3796,
+      "step": 984
+    },
+    {
+      "epoch": 0.32442669521182427,
+      "grad_norm": 2.5218312740325928,
+      "learning_rate": 7.76964455921926e-05,
+      "loss": 2.4834,
+      "step": 985
+    },
+    {
+      "epoch": 0.32475606241508503,
+      "grad_norm": 2.420651435852051,
+      "learning_rate": 7.7652632936279e-05,
+      "loss": 2.1914,
+      "step": 986
+    },
+    {
+      "epoch": 0.32508542961834574,
+      "grad_norm": 2.723442792892456,
+      "learning_rate": 7.760878967084059e-05,
+      "loss": 2.5173,
+      "step": 987
+    },
+    {
+      "epoch": 0.3254147968216065,
+      "grad_norm": 2.400263786315918,
+      "learning_rate": 7.756491584440882e-05,
+      "loss": 2.1829,
+      "step": 988
+    },
+    {
+      "epoch": 0.3257441640248672,
+      "grad_norm": 2.4888710975646973,
+      "learning_rate": 7.75210115055489e-05,
+      "loss": 2.1607,
+      "step": 989
+    },
+    {
+      "epoch": 0.32607353122812793,
+      "grad_norm": 2.6729047298431396,
+      "learning_rate": 7.747707670285989e-05,
+      "loss": 2.3433,
+      "step": 990
+    },
+    {
+      "epoch": 0.3264028984313887,
+      "grad_norm": 2.9809768199920654,
+      "learning_rate": 7.743311148497452e-05,
+      "loss": 2.3124,
+      "step": 991
+    },
+    {
+      "epoch": 0.3267322656346494,
+      "grad_norm": 2.287365198135376,
+      "learning_rate": 7.73891159005592e-05,
+      "loss": 2.0576,
+      "step": 992
+    },
+    {
+      "epoch": 0.3270616328379102,
+      "grad_norm": 2.4659500122070312,
+      "learning_rate": 7.734508999831394e-05,
+      "loss": 2.0228,
+      "step": 993
+    },
+    {
+      "epoch": 0.3273910000411709,
+      "grad_norm": 2.66711163520813,
+      "learning_rate": 7.730103382697236e-05,
+      "loss": 2.1277,
+      "step": 994
+    },
+    {
+      "epoch": 0.32772036724443165,
+      "grad_norm": 2.902860641479492,
+      "learning_rate": 7.725694743530153e-05,
+      "loss": 2.3815,
+      "step": 995
+    },
+    {
+      "epoch": 0.32804973444769236,
+      "grad_norm": 2.8413193225860596,
+      "learning_rate": 7.721283087210199e-05,
+      "loss": 2.2505,
+      "step": 996
+    },
+    {
+      "epoch": 0.32837910165095313,
+      "grad_norm": 3.3534953594207764,
+      "learning_rate": 7.716868418620768e-05,
+      "loss": 2.2123,
+      "step": 997
+    },
+    {
+      "epoch": 0.32870846885421384,
+      "grad_norm": 2.6357085704803467,
+      "learning_rate": 7.71245074264859e-05,
+      "loss": 1.8754,
+      "step": 998
+    },
+    {
+      "epoch": 0.32903783605747455,
+      "grad_norm": 3.168962001800537,
+      "learning_rate": 7.70803006418372e-05,
+      "loss": 1.9192,
+      "step": 999
+    },
+    {
+      "epoch": 0.3293672032607353,
+      "grad_norm": 3.3432531356811523,
+      "learning_rate": 7.703606388119542e-05,
+      "loss": 1.7519,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32969657046399603,
+      "grad_norm": 1.857453465461731,
+      "learning_rate": 7.699179719352752e-05,
+      "loss": 2.5652,
+      "step": 1001
+    },
+    {
+      "epoch": 0.3300259376672568,
+      "grad_norm": 1.9321167469024658,
+      "learning_rate": 7.694750062783363e-05,
+      "loss": 2.1067,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3303553048705175,
+      "grad_norm": 2.123394250869751,
+      "learning_rate": 7.690317423314696e-05,
+      "loss": 2.311,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3306846720737783,
+      "grad_norm": 2.345137357711792,
+      "learning_rate": 7.685881805853369e-05,
+      "loss": 2.1725,
+      "step": 1004
+    },
+    {
+      "epoch": 0.331014039277039,
+      "grad_norm": 2.3156449794769287,
+      "learning_rate": 7.6814432153093e-05,
+      "loss": 2.3528,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33134340648029975,
+      "grad_norm": 2.4214887619018555,
+      "learning_rate": 7.6770016565957e-05,
+      "loss": 2.0471,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33167277368356046,
+      "grad_norm": 2.3343491554260254,
+      "learning_rate": 7.672557134629059e-05,
+      "loss": 2.1914,
+      "step": 1007
+    },
+    {
+      "epoch": 0.3320021408868212,
+      "grad_norm": 2.713235378265381,
+      "learning_rate": 7.668109654329154e-05,
+      "loss": 2.4468,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33233150809008194,
+      "grad_norm": 3.0459039211273193,
+      "learning_rate": 7.663659220619033e-05,
+      "loss": 2.4859,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33266087529334265,
+      "grad_norm": 2.7493438720703125,
+      "learning_rate": 7.659205838425013e-05,
+      "loss": 2.4392,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3329902424966034,
+      "grad_norm": 2.3039515018463135,
+      "learning_rate": 7.654749512676676e-05,
+      "loss": 2.0348,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33331960969986413,
+      "grad_norm": 2.667750597000122,
+      "learning_rate": 7.650290248306863e-05,
+      "loss": 1.949,
+      "step": 1012
+    },
+    {
+      "epoch": 0.3336489769031249,
+      "grad_norm": 2.9373788833618164,
+      "learning_rate": 7.645828050251665e-05,
+      "loss": 2.591,
+      "step": 1013
+    },
+    {
+      "epoch": 0.3339783441063856,
+      "grad_norm": 3.202880620956421,
+      "learning_rate": 7.641362923450424e-05,
+      "loss": 2.5421,
+      "step": 1014
+    },
+    {
+      "epoch": 0.3343077113096463,
+      "grad_norm": 2.4246268272399902,
+      "learning_rate": 7.636894872845722e-05,
+      "loss": 2.1465,
+      "step": 1015
+    },
+    {
+      "epoch": 0.3346370785129071,
+      "grad_norm": 2.732734203338623,
+      "learning_rate": 7.632423903383374e-05,
+      "loss": 2.1029,
+      "step": 1016
+    },
+    {
+      "epoch": 0.3349664457161678,
+      "grad_norm": 3.3760523796081543,
+      "learning_rate": 7.627950020012434e-05,
+      "loss": 2.0353,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33529581291942856,
+      "grad_norm": 3.0070247650146484,
+      "learning_rate": 7.623473227685176e-05,
+      "loss": 2.365,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33562518012268927,
+      "grad_norm": 2.8173720836639404,
+      "learning_rate": 7.618993531357094e-05,
+      "loss": 2.0214,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33595454732595004,
+      "grad_norm": 2.5577118396759033,
+      "learning_rate": 7.614510935986898e-05,
+      "loss": 1.8253,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33628391452921075,
+      "grad_norm": 2.8958218097686768,
+      "learning_rate": 7.610025446536509e-05,
+      "loss": 2.2123,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3366132817324715,
+      "grad_norm": 3.0846564769744873,
+      "learning_rate": 7.605537067971045e-05,
+      "loss": 2.0833,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3369426489357322,
+      "grad_norm": 3.0642452239990234,
+      "learning_rate": 7.601045805258828e-05,
+      "loss": 1.8784,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33727201613899294,
+      "grad_norm": 4.532246112823486,
+      "learning_rate": 7.596551663371372e-05,
+      "loss": 2.4327,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3376013833422537,
+      "grad_norm": 3.3755955696105957,
+      "learning_rate": 7.592054647283375e-05,
+      "loss": 1.7789,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3379307505455144,
+      "grad_norm": 1.943467140197754,
+      "learning_rate": 7.587554761972718e-05,
+      "loss": 2.3184,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3382601177487752,
+      "grad_norm": 2.520038604736328,
+      "learning_rate": 7.583052012420461e-05,
+      "loss": 2.2305,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3385894849520359,
+      "grad_norm": 2.3161160945892334,
+      "learning_rate": 7.57854640361083e-05,
+      "loss": 2.2228,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33891885215529666,
+      "grad_norm": 2.4692294597625732,
+      "learning_rate": 7.574037940531218e-05,
+      "loss": 2.251,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33924821935855737,
+      "grad_norm": 2.4752914905548096,
+      "learning_rate": 7.569526628172177e-05,
+      "loss": 2.2454,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33957758656181813,
+      "grad_norm": 2.6334524154663086,
+      "learning_rate": 7.565012471527416e-05,
+      "loss": 2.5675,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33990695376507885,
+      "grad_norm": 2.3727474212646484,
+      "learning_rate": 7.560495475593785e-05,
+      "loss": 2.2345,
+      "step": 1032
+    },
+    {
+      "epoch": 0.34023632096833956,
+      "grad_norm": 2.6379683017730713,
+      "learning_rate": 7.555975645371285e-05,
+      "loss": 2.4907,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3405656881716003,
+      "grad_norm": 2.5622193813323975,
+      "learning_rate": 7.55145298586305e-05,
+      "loss": 2.3077,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34089505537486103,
+      "grad_norm": 2.888123035430908,
+      "learning_rate": 7.546927502075348e-05,
+      "loss": 2.2664,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3412244225781218,
+      "grad_norm": 2.4946439266204834,
+      "learning_rate": 7.542399199017568e-05,
+      "loss": 2.3239,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3415537897813825,
+      "grad_norm": 2.6942648887634277,
+      "learning_rate": 7.53786808170223e-05,
+      "loss": 2.501,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3418831569846433,
+      "grad_norm": 2.577432632446289,
+      "learning_rate": 7.53333415514496e-05,
+      "loss": 2.3515,
+      "step": 1038
+    },
+    {
+      "epoch": 0.342212524187904,
+      "grad_norm": 2.2709529399871826,
+      "learning_rate": 7.528797424364496e-05,
+      "loss": 1.9878,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3425418913911647,
+      "grad_norm": 2.2214181423187256,
+      "learning_rate": 7.524257894382681e-05,
+      "loss": 1.8414,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34287125859442547,
+      "grad_norm": 2.8622872829437256,
+      "learning_rate": 7.519715570224457e-05,
+      "loss": 1.9751,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3432006257976862,
+      "grad_norm": 2.7308106422424316,
+      "learning_rate": 7.515170456917857e-05,
+      "loss": 2.1777,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34352999300094694,
+      "grad_norm": 3.008892297744751,
+      "learning_rate": 7.510622559494002e-05,
+      "loss": 2.1712,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34385936020420765,
+      "grad_norm": 2.538496732711792,
+      "learning_rate": 7.5060718829871e-05,
+      "loss": 1.5176,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3441887274074684,
+      "grad_norm": 2.9815316200256348,
+      "learning_rate": 7.501518432434424e-05,
+      "loss": 1.9269,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34451809461072913,
+      "grad_norm": 2.786571741104126,
+      "learning_rate": 7.49696221287633e-05,
+      "loss": 2.0934,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3448474618139899,
+      "grad_norm": 2.8320257663726807,
+      "learning_rate": 7.49240322935623e-05,
+      "loss": 1.7772,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3451768290172506,
+      "grad_norm": 3.3367862701416016,
+      "learning_rate": 7.487841486920599e-05,
+      "loss": 2.2513,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3455061962205113,
+      "grad_norm": 3.6575772762298584,
+      "learning_rate": 7.48327699061897e-05,
+      "loss": 2.2954,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3458355634237721,
+      "grad_norm": 3.485363721847534,
+      "learning_rate": 7.478709745503913e-05,
+      "loss": 1.9509,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3461649306270328,
+      "grad_norm": 2.551144599914551,
+      "learning_rate": 7.474139756631056e-05,
+      "loss": 2.5046,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34649429783029356,
+      "grad_norm": 2.3824918270111084,
+      "learning_rate": 7.46956702905905e-05,
+      "loss": 2.3288,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3468236650335543,
+      "grad_norm": 2.3063457012176514,
+      "learning_rate": 7.464991567849586e-05,
+      "loss": 2.2688,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34715303223681504,
+      "grad_norm": 2.4683382511138916,
+      "learning_rate": 7.460413378067379e-05,
+      "loss": 2.4154,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34748239944007575,
+      "grad_norm": 2.572463035583496,
+      "learning_rate": 7.455832464780162e-05,
+      "loss": 2.4033,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3478117666433365,
+      "grad_norm": 2.575000762939453,
+      "learning_rate": 7.451248833058687e-05,
+      "loss": 2.2997,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34814113384659723,
+      "grad_norm": 2.226249933242798,
+      "learning_rate": 7.446662487976713e-05,
+      "loss": 2.1283,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34847050104985794,
+      "grad_norm": 2.5053114891052246,
+      "learning_rate": 7.442073434610997e-05,
+      "loss": 2.1099,
+      "step": 1058
+    },
+    {
+      "epoch": 0.3487998682531187,
+      "grad_norm": 2.71102237701416,
+      "learning_rate": 7.437481678041307e-05,
+      "loss": 2.3652,
+      "step": 1059
+    },
+    {
+      "epoch": 0.3491292354563794,
+      "grad_norm": 3.036839723587036,
+      "learning_rate": 7.43288722335039e-05,
+      "loss": 1.9926,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3494586026596402,
+      "grad_norm": 2.7319066524505615,
+      "learning_rate": 7.428290075623987e-05,
+      "loss": 2.5434,
+      "step": 1061
+    },
+    {
+      "epoch": 0.3497879698629009,
+      "grad_norm": 2.671752691268921,
+      "learning_rate": 7.423690239950818e-05,
+      "loss": 2.1477,
+      "step": 1062
+    },
+    {
+      "epoch": 0.35011733706616166,
+      "grad_norm": 3.032055616378784,
+      "learning_rate": 7.419087721422576e-05,
+      "loss": 2.3655,
+      "step": 1063
+    },
+    {
+      "epoch": 0.35044670426942237,
+      "grad_norm": 2.8573625087738037,
+      "learning_rate": 7.414482525133928e-05,
+      "loss": 2.6337,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3507760714726831,
+      "grad_norm": 2.5970046520233154,
+      "learning_rate": 7.409874656182506e-05,
+      "loss": 1.9232,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35110543867594385,
+      "grad_norm": 2.5858652591705322,
+      "learning_rate": 7.405264119668894e-05,
+      "loss": 2.4204,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35143480587920456,
+      "grad_norm": 3.2637603282928467,
+      "learning_rate": 7.400650920696633e-05,
+      "loss": 2.6369,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3517641730824653,
+      "grad_norm": 2.568941831588745,
+      "learning_rate": 7.396035064372214e-05,
+      "loss": 2.2192,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35209354028572604,
+      "grad_norm": 3.0660107135772705,
+      "learning_rate": 7.39141655580506e-05,
+      "loss": 2.1387,
+      "step": 1069
+    },
+    {
+      "epoch": 0.3524229074889868,
+      "grad_norm": 3.537260055541992,
+      "learning_rate": 7.386795400107539e-05,
+      "loss": 2.1265,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3527522746922475,
+      "grad_norm": 2.7157890796661377,
+      "learning_rate": 7.382171602394948e-05,
+      "loss": 1.7636,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3530816418955083,
+      "grad_norm": 3.227700710296631,
+      "learning_rate": 7.377545167785506e-05,
+      "loss": 2.1963,
+      "step": 1072
+    },
+    {
+      "epoch": 0.353411009098769,
+      "grad_norm": 2.9469199180603027,
+      "learning_rate": 7.372916101400349e-05,
+      "loss": 1.8702,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3537403763020297,
+      "grad_norm": 2.9933621883392334,
+      "learning_rate": 7.368284408363531e-05,
+      "loss": 1.6875,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35406974350529047,
+      "grad_norm": 3.143301248550415,
+      "learning_rate": 7.363650093802012e-05,
+      "loss": 1.8475,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3543991107085512,
+      "grad_norm": 1.9010746479034424,
+      "learning_rate": 7.35901316284565e-05,
+      "loss": 2.4547,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35472847791181195,
+      "grad_norm": 1.9302737712860107,
+      "learning_rate": 7.354373620627205e-05,
+      "loss": 2.2665,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35505784511507266,
+      "grad_norm": 2.267845392227173,
+      "learning_rate": 7.349731472282325e-05,
+      "loss": 2.3649,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3553872123183334,
+      "grad_norm": 2.5052695274353027,
+      "learning_rate": 7.345086722949539e-05,
+      "loss": 2.3261,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35571657952159413,
+      "grad_norm": 2.4212067127227783,
+      "learning_rate": 7.340439377770263e-05,
+      "loss": 2.1668,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35604594672485484,
+      "grad_norm": 2.1286873817443848,
+      "learning_rate": 7.335789441888781e-05,
+      "loss": 2.1269,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3563753139281156,
+      "grad_norm": 2.6201021671295166,
+      "learning_rate": 7.331136920452244e-05,
+      "loss": 2.498,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3567046811313763,
+      "grad_norm": 2.8522377014160156,
+      "learning_rate": 7.326481818610668e-05,
+      "loss": 2.2456,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3570340483346371,
+      "grad_norm": 2.749220609664917,
+      "learning_rate": 7.321824141516926e-05,
+      "loss": 2.2296,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3573634155378978,
+      "grad_norm": 2.6423492431640625,
+      "learning_rate": 7.317163894326735e-05,
+      "loss": 2.0497,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35769278274115857,
+      "grad_norm": 2.9978692531585693,
+      "learning_rate": 7.312501082198666e-05,
+      "loss": 2.4605,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3580221499444193,
+      "grad_norm": 2.5553252696990967,
+      "learning_rate": 7.307835710294125e-05,
+      "loss": 1.9969,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35835151714768004,
+      "grad_norm": 3.3244998455047607,
+      "learning_rate": 7.303167783777349e-05,
+      "loss": 2.2522,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35868088435094075,
+      "grad_norm": 2.7964744567871094,
+      "learning_rate": 7.298497307815406e-05,
+      "loss": 2.2721,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35901025155420146,
+      "grad_norm": 2.972792625427246,
+      "learning_rate": 7.293824287578185e-05,
+      "loss": 2.2599,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35933961875746223,
+      "grad_norm": 2.953064441680908,
+      "learning_rate": 7.289148728238392e-05,
+      "loss": 2.4102,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35966898596072294,
+      "grad_norm": 3.144014835357666,
+      "learning_rate": 7.284470634971544e-05,
+      "loss": 2.4154,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3599983531639837,
+      "grad_norm": 2.5227506160736084,
+      "learning_rate": 7.279790012955961e-05,
+      "loss": 1.9305,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3603277203672444,
+      "grad_norm": 3.1171419620513916,
+      "learning_rate": 7.275106867372762e-05,
+      "loss": 2.2716,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3606570875705052,
+      "grad_norm": 2.6330418586730957,
+      "learning_rate": 7.270421203405863e-05,
+      "loss": 1.8731,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3609864547737659,
+      "grad_norm": 2.5206286907196045,
+      "learning_rate": 7.265733026241966e-05,
+      "loss": 1.8349,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36131582197702666,
+      "grad_norm": 2.8334505558013916,
+      "learning_rate": 7.261042341070552e-05,
+      "loss": 1.4885,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3616451891802874,
+      "grad_norm": 3.205961227416992,
+      "learning_rate": 7.256349153083881e-05,
+      "loss": 1.9328,
+      "step": 1098
+    },
+    {
+      "epoch": 0.3619745563835481,
+      "grad_norm": 3.352240562438965,
+      "learning_rate": 7.251653467476983e-05,
+      "loss": 2.0997,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36230392358680885,
+      "grad_norm": 3.4709959030151367,
+      "learning_rate": 7.246955289447653e-05,
+      "loss": 1.7248,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36263329079006956,
+      "grad_norm": 2.071342945098877,
+      "learning_rate": 7.242254624196443e-05,
+      "loss": 2.2652,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36296265799333033,
+      "grad_norm": 2.2652060985565186,
+      "learning_rate": 7.237551476926661e-05,
+      "loss": 2.3592,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36329202519659104,
+      "grad_norm": 2.8345634937286377,
+      "learning_rate": 7.232845852844361e-05,
+      "loss": 2.6346,
+      "step": 1103
+    },
+    {
+      "epoch": 0.3636213923998518,
+      "grad_norm": 2.3296942710876465,
+      "learning_rate": 7.228137757158338e-05,
+      "loss": 2.0973,
+      "step": 1104
+    },
+    {
+      "epoch": 0.3639507596031125,
+      "grad_norm": 2.4769644737243652,
+      "learning_rate": 7.223427195080126e-05,
+      "loss": 2.2648,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36428012680637323,
+      "grad_norm": 2.285839080810547,
+      "learning_rate": 7.218714171823984e-05,
+      "loss": 2.1648,
+      "step": 1106
+    },
+    {
+      "epoch": 0.364609494009634,
+      "grad_norm": 2.404283046722412,
+      "learning_rate": 7.2139986926069e-05,
+      "loss": 2.3887,
+      "step": 1107
+    },
+    {
+      "epoch": 0.3649388612128947,
+      "grad_norm": 2.4443814754486084,
+      "learning_rate": 7.209280762648576e-05,
+      "loss": 2.227,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36526822841615547,
+      "grad_norm": 2.4258129596710205,
+      "learning_rate": 7.204560387171432e-05,
+      "loss": 2.2972,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3655975956194162,
+      "grad_norm": 2.630659341812134,
+      "learning_rate": 7.199837571400591e-05,
+      "loss": 2.3323,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36592696282267695,
+      "grad_norm": 2.332148313522339,
+      "learning_rate": 7.195112320563881e-05,
+      "loss": 2.0955,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36625633002593766,
+      "grad_norm": 2.656656503677368,
+      "learning_rate": 7.190384639891822e-05,
+      "loss": 2.016,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3665856972291984,
+      "grad_norm": 2.663341760635376,
+      "learning_rate": 7.185654534617623e-05,
+      "loss": 2.3012,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36691506443245914,
+      "grad_norm": 2.5622754096984863,
+      "learning_rate": 7.180922009977181e-05,
+      "loss": 1.9749,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36724443163571985,
+      "grad_norm": 2.865852117538452,
+      "learning_rate": 7.176187071209069e-05,
+      "loss": 1.648,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3675737988389806,
+      "grad_norm": 2.8726089000701904,
+      "learning_rate": 7.171449723554531e-05,
+      "loss": 2.0493,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3679031660422413,
+      "grad_norm": 2.720703125,
+      "learning_rate": 7.166709972257478e-05,
+      "loss": 2.3402,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3682325332455021,
+      "grad_norm": 3.307445764541626,
+      "learning_rate": 7.161967822564483e-05,
+      "loss": 1.8724,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3685619004487628,
+      "grad_norm": 3.013970375061035,
+      "learning_rate": 7.157223279724775e-05,
+      "loss": 2.1087,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36889126765202357,
+      "grad_norm": 2.9190618991851807,
+      "learning_rate": 7.152476348990224e-05,
+      "loss": 2.1204,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3692206348552843,
+      "grad_norm": 2.860560178756714,
+      "learning_rate": 7.147727035615355e-05,
+      "loss": 2.0234,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36955000205854505,
+      "grad_norm": 3.1333353519439697,
+      "learning_rate": 7.142975344857325e-05,
+      "loss": 2.1354,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36987936926180576,
+      "grad_norm": 3.1250181198120117,
+      "learning_rate": 7.138221281975919e-05,
+      "loss": 2.2142,
+      "step": 1123
+    },
+    {
+      "epoch": 0.37020873646506647,
+      "grad_norm": 3.049356460571289,
+      "learning_rate": 7.133464852233553e-05,
+      "loss": 1.9174,
+      "step": 1124
+    },
+    {
+      "epoch": 0.37053810366832723,
+      "grad_norm": 3.432244062423706,
+      "learning_rate": 7.12870606089526e-05,
+      "loss": 1.4808,
+      "step": 1125
+    },
+    {
+      "epoch": 0.37086747087158795,
+      "grad_norm": 1.7517294883728027,
+      "learning_rate": 7.123944913228688e-05,
+      "loss": 2.5009,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3711968380748487,
+      "grad_norm": 2.627819299697876,
+      "learning_rate": 7.119181414504095e-05,
+      "loss": 2.4364,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3715262052781094,
+      "grad_norm": 2.8850955963134766,
+      "learning_rate": 7.11441556999434e-05,
+      "loss": 2.4496,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3718555724813702,
+      "grad_norm": 2.320338726043701,
+      "learning_rate": 7.109647384974876e-05,
+      "loss": 1.9873,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3721849396846309,
+      "grad_norm": 2.7479662895202637,
+      "learning_rate": 7.104876864723751e-05,
+      "loss": 2.4404,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3725143068878916,
+      "grad_norm": 2.618581533432007,
+      "learning_rate": 7.100104014521598e-05,
+      "loss": 2.0846,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3728436740911524,
+      "grad_norm": 3.1042213439941406,
+      "learning_rate": 7.095328839651625e-05,
+      "loss": 2.6769,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3731730412944131,
+      "grad_norm": 2.402482271194458,
+      "learning_rate": 7.090551345399616e-05,
+      "loss": 2.1385,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37350240849767385,
+      "grad_norm": 2.716562271118164,
+      "learning_rate": 7.085771537053923e-05,
+      "loss": 2.2691,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 2.5292835235595703,
+      "learning_rate": 7.080989419905456e-05,
+      "loss": 2.0973,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37416114290419533,
+      "grad_norm": 2.4309535026550293,
+      "learning_rate": 7.076204999247686e-05,
+      "loss": 2.2923,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37449051010745604,
+      "grad_norm": 2.3152859210968018,
+      "learning_rate": 7.071418280376629e-05,
+      "loss": 1.898,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3748198773107168,
+      "grad_norm": 2.7684056758880615,
+      "learning_rate": 7.06662926859085e-05,
+      "loss": 2.192,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3751492445139775,
+      "grad_norm": 2.886624813079834,
+      "learning_rate": 7.061837969191445e-05,
+      "loss": 2.2443,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37547861171723823,
+      "grad_norm": 2.6988983154296875,
+      "learning_rate": 7.05704438748205e-05,
+      "loss": 2.0046,
+      "step": 1140
+    },
+    {
+      "epoch": 0.375807978920499,
+      "grad_norm": 2.849220037460327,
+      "learning_rate": 7.05224852876882e-05,
+      "loss": 2.2765,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3761373461237597,
+      "grad_norm": 2.494020462036133,
+      "learning_rate": 7.047450398360438e-05,
+      "loss": 2.1473,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3764667133270205,
+      "grad_norm": 2.90225887298584,
+      "learning_rate": 7.042650001568097e-05,
+      "loss": 2.3484,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3767960805302812,
+      "grad_norm": 2.873544692993164,
+      "learning_rate": 7.037847343705496e-05,
+      "loss": 2.0771,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37712544773354195,
+      "grad_norm": 2.2210004329681396,
+      "learning_rate": 7.033042430088844e-05,
+      "loss": 1.6013,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37745481493680266,
+      "grad_norm": 2.627661943435669,
+      "learning_rate": 7.028235266036841e-05,
+      "loss": 2.0004,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37778418214006343,
+      "grad_norm": 2.7000410556793213,
+      "learning_rate": 7.023425856870683e-05,
+      "loss": 1.8592,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37811354934332414,
+      "grad_norm": 3.242762327194214,
+      "learning_rate": 7.018614207914047e-05,
+      "loss": 2.1389,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37844291654658485,
+      "grad_norm": 3.2626426219940186,
+      "learning_rate": 7.013800324493089e-05,
+      "loss": 2.0661,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3787722837498456,
+      "grad_norm": 4.664367198944092,
+      "learning_rate": 7.008984211936446e-05,
+      "loss": 2.0999,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37910165095310633,
+      "grad_norm": 1.8425556421279907,
+      "learning_rate": 7.00416587557521e-05,
+      "loss": 2.4092,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3794310181563671,
+      "grad_norm": 2.1894047260284424,
+      "learning_rate": 6.999345320742945e-05,
+      "loss": 2.4542,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3797603853596278,
+      "grad_norm": 2.095588207244873,
+      "learning_rate": 6.994522552775666e-05,
+      "loss": 2.1205,
+      "step": 1153
+    },
+    {
+      "epoch": 0.38008975256288857,
+      "grad_norm": 2.427074909210205,
+      "learning_rate": 6.98969757701184e-05,
+      "loss": 2.2412,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3804191197661493,
+      "grad_norm": 2.322599411010742,
+      "learning_rate": 6.984870398792374e-05,
+      "loss": 2.0839,
+      "step": 1155
+    },
+    {
+      "epoch": 0.38074848696941,
+      "grad_norm": 2.316253900527954,
+      "learning_rate": 6.980041023460619e-05,
+      "loss": 2.2265,
+      "step": 1156
+    },
+    {
+      "epoch": 0.38107785417267076,
+      "grad_norm": 2.33427095413208,
+      "learning_rate": 6.975209456362353e-05,
+      "loss": 2.3765,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38140722137593147,
+      "grad_norm": 2.23207426071167,
+      "learning_rate": 6.97037570284578e-05,
+      "loss": 1.9018,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38173658857919224,
+      "grad_norm": 2.3566267490386963,
+      "learning_rate": 6.965539768261531e-05,
+      "loss": 2.1916,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38206595578245295,
+      "grad_norm": 2.377549409866333,
+      "learning_rate": 6.960701657962641e-05,
+      "loss": 2.4266,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3823953229857137,
+      "grad_norm": 2.494011163711548,
+      "learning_rate": 6.955861377304564e-05,
+      "loss": 2.1707,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3827246901889744,
+      "grad_norm": 2.6809515953063965,
+      "learning_rate": 6.951018931645146e-05,
+      "loss": 2.3089,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3830540573922352,
+      "grad_norm": 2.6300032138824463,
+      "learning_rate": 6.946174326344637e-05,
+      "loss": 2.1554,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3833834245954959,
+      "grad_norm": 2.4809701442718506,
+      "learning_rate": 6.941327566765675e-05,
+      "loss": 2.1591,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3837127917987566,
+      "grad_norm": 2.6750636100769043,
+      "learning_rate": 6.936478658273285e-05,
+      "loss": 2.3151,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3840421590020174,
+      "grad_norm": 2.421334981918335,
+      "learning_rate": 6.931627606234865e-05,
+      "loss": 1.9933,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3843715262052781,
+      "grad_norm": 2.758087158203125,
+      "learning_rate": 6.92677441602019e-05,
+      "loss": 2.2134,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38470089340853886,
+      "grad_norm": 3.3581409454345703,
+      "learning_rate": 6.921919093001402e-05,
+      "loss": 2.3097,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38503026061179957,
+      "grad_norm": 3.3593056201934814,
+      "learning_rate": 6.917061642553005e-05,
+      "loss": 1.9195,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38535962781506033,
+      "grad_norm": 2.783876895904541,
+      "learning_rate": 6.91220207005185e-05,
+      "loss": 1.8634,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38568899501832105,
+      "grad_norm": 2.8132247924804688,
+      "learning_rate": 6.907340380877149e-05,
+      "loss": 2.1477,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3860183622215818,
+      "grad_norm": 2.878632068634033,
+      "learning_rate": 6.902476580410449e-05,
+      "loss": 2.254,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3863477294248425,
+      "grad_norm": 3.4279277324676514,
+      "learning_rate": 6.897610674035634e-05,
+      "loss": 1.95,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38667709662810323,
+      "grad_norm": 3.4778835773468018,
+      "learning_rate": 6.892742667138923e-05,
+      "loss": 1.8198,
+      "step": 1174
+    },
+    {
+      "epoch": 0.387006463831364,
+      "grad_norm": 3.2819323539733887,
+      "learning_rate": 6.887872565108859e-05,
+      "loss": 1.8538,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3873358310346247,
+      "grad_norm": 1.8319649696350098,
+      "learning_rate": 6.883000373336299e-05,
+      "loss": 2.4531,
+      "step": 1176
+    },
+    {
+      "epoch": 0.3876651982378855,
+      "grad_norm": 2.054316520690918,
+      "learning_rate": 6.878126097214421e-05,
+      "loss": 1.9968,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3879945654411462,
+      "grad_norm": 2.3420443534851074,
+      "learning_rate": 6.873249742138709e-05,
+      "loss": 2.2451,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38832393264440696,
+      "grad_norm": 1.8880903720855713,
+      "learning_rate": 6.868371313506941e-05,
+      "loss": 2.3786,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38865329984766767,
+      "grad_norm": 2.393928289413452,
+      "learning_rate": 6.863490816719196e-05,
+      "loss": 2.1902,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3889826670509284,
+      "grad_norm": 2.3493289947509766,
+      "learning_rate": 6.858608257177846e-05,
+      "loss": 2.053,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38931203425418914,
+      "grad_norm": 2.4995625019073486,
+      "learning_rate": 6.853723640287535e-05,
+      "loss": 2.3704,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38964140145744985,
+      "grad_norm": 2.5445573329925537,
+      "learning_rate": 6.848836971455197e-05,
+      "loss": 1.9575,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3899707686607106,
+      "grad_norm": 2.5418543815612793,
+      "learning_rate": 6.84394825609003e-05,
+      "loss": 2.4032,
+      "step": 1184
+    },
+    {
+      "epoch": 0.39030013586397133,
+      "grad_norm": 3.205739736557007,
+      "learning_rate": 6.839057499603497e-05,
+      "loss": 2.6626,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3906295030672321,
+      "grad_norm": 2.529710531234741,
+      "learning_rate": 6.834164707409326e-05,
+      "loss": 2.1196,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3909588702704928,
+      "grad_norm": 2.9174511432647705,
+      "learning_rate": 6.829269884923491e-05,
+      "loss": 2.4384,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3912882374737536,
+      "grad_norm": 2.5731706619262695,
+      "learning_rate": 6.82437303756422e-05,
+      "loss": 2.21,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3916176046770143,
+      "grad_norm": 3.202188014984131,
+      "learning_rate": 6.819474170751978e-05,
+      "loss": 2.5139,
+      "step": 1189
+    },
+    {
+      "epoch": 0.391946971880275,
+      "grad_norm": 2.6559245586395264,
+      "learning_rate": 6.814573289909466e-05,
+      "loss": 2.3044,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39227633908353576,
+      "grad_norm": 3.1062824726104736,
+      "learning_rate": 6.809670400461618e-05,
+      "loss": 2.0021,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3926057062867965,
+      "grad_norm": 2.513209342956543,
+      "learning_rate": 6.804765507835587e-05,
+      "loss": 1.8991,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39293507349005724,
+      "grad_norm": 2.8009233474731445,
+      "learning_rate": 6.799858617460744e-05,
+      "loss": 2.3419,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39326444069331795,
+      "grad_norm": 2.8235421180725098,
+      "learning_rate": 6.794949734768674e-05,
+      "loss": 2.2565,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3935938078965787,
+      "grad_norm": 2.455665111541748,
+      "learning_rate": 6.790038865193167e-05,
+      "loss": 1.9415,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39392317509983943,
+      "grad_norm": 2.5987555980682373,
+      "learning_rate": 6.785126014170207e-05,
+      "loss": 1.8636,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3942525423031002,
+      "grad_norm": 2.7285478115081787,
+      "learning_rate": 6.780211187137981e-05,
+      "loss": 2.0818,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3945819095063609,
+      "grad_norm": 3.124101161956787,
+      "learning_rate": 6.775294389536852e-05,
+      "loss": 2.1855,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3949112767096216,
+      "grad_norm": 2.466757297515869,
+      "learning_rate": 6.770375626809373e-05,
+      "loss": 1.6696,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3952406439128824,
+      "grad_norm": 3.0805959701538086,
+      "learning_rate": 6.76545490440027e-05,
+      "loss": 2.1575,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3955700111161431,
+      "grad_norm": 2.310868978500366,
+      "learning_rate": 6.760532227756435e-05,
+      "loss": 2.3745,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39589937831940386,
+      "grad_norm": 2.0340662002563477,
+      "learning_rate": 6.755607602326928e-05,
+      "loss": 2.2478,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39622874552266457,
+      "grad_norm": 2.500797748565674,
+      "learning_rate": 6.750681033562964e-05,
+      "loss": 2.593,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39655811272592534,
+      "grad_norm": 2.07173228263855,
+      "learning_rate": 6.745752526917907e-05,
+      "loss": 2.3154,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39688747992918605,
+      "grad_norm": 2.0684971809387207,
+      "learning_rate": 6.74082208784727e-05,
+      "loss": 2.0726,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39721684713244676,
+      "grad_norm": 2.2718148231506348,
+      "learning_rate": 6.735889721808703e-05,
+      "loss": 2.2801,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3975462143357075,
+      "grad_norm": 2.2261784076690674,
+      "learning_rate": 6.730955434261986e-05,
+      "loss": 2.1202,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39787558153896824,
+      "grad_norm": 2.9758946895599365,
+      "learning_rate": 6.726019230669034e-05,
+      "loss": 2.4945,
+      "step": 1208
+    },
+    {
+      "epoch": 0.398204948742229,
+      "grad_norm": 2.5161709785461426,
+      "learning_rate": 6.721081116493874e-05,
+      "loss": 2.1904,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3985343159454897,
+      "grad_norm": 2.922379493713379,
+      "learning_rate": 6.716141097202657e-05,
+      "loss": 2.5583,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3988636831487505,
+      "grad_norm": 2.4563610553741455,
+      "learning_rate": 6.711199178263632e-05,
+      "loss": 2.355,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3991930503520112,
+      "grad_norm": 3.078686237335205,
+      "learning_rate": 6.706255365147161e-05,
+      "loss": 2.604,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39952241755527196,
+      "grad_norm": 2.5945472717285156,
+      "learning_rate": 6.701309663325696e-05,
+      "loss": 2.3826,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39985178475853267,
+      "grad_norm": 2.8002431392669678,
+      "learning_rate": 6.696362078273781e-05,
+      "loss": 2.1889,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4001811519617934,
+      "grad_norm": 2.4396965503692627,
+      "learning_rate": 6.69141261546805e-05,
+      "loss": 1.9122,
+      "step": 1215
+    },
+    {
+      "epoch": 0.40051051916505415,
+      "grad_norm": 2.8927268981933594,
+      "learning_rate": 6.686461280387203e-05,
+      "loss": 2.2138,
+      "step": 1216
+    },
+    {
+      "epoch": 0.40083988636831486,
+      "grad_norm": 2.633314847946167,
+      "learning_rate": 6.681508078512031e-05,
+      "loss": 1.7317,
+      "step": 1217
+    },
+    {
+      "epoch": 0.4011692535715756,
+      "grad_norm": 3.017258882522583,
+      "learning_rate": 6.676553015325372e-05,
+      "loss": 2.1284,
+      "step": 1218
+    },
+    {
+      "epoch": 0.40149862077483633,
+      "grad_norm": 2.7072062492370605,
+      "learning_rate": 6.67159609631214e-05,
+      "loss": 1.8943,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4018279879780971,
+      "grad_norm": 2.961946487426758,
+      "learning_rate": 6.666637326959293e-05,
+      "loss": 2.2615,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4021573551813578,
+      "grad_norm": 2.648956775665283,
+      "learning_rate": 6.661676712755842e-05,
+      "loss": 1.9888,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4024867223846186,
+      "grad_norm": 2.841571092605591,
+      "learning_rate": 6.656714259192839e-05,
+      "loss": 2.2434,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4028160895878793,
+      "grad_norm": 2.6502773761749268,
+      "learning_rate": 6.65174997176337e-05,
+      "loss": 1.9564,
+      "step": 1223
+    },
+    {
+      "epoch": 0.40314545679114,
+      "grad_norm": 2.589456796646118,
+      "learning_rate": 6.646783855962555e-05,
+      "loss": 1.664,
+      "step": 1224
+    },
+    {
+      "epoch": 0.40347482399440077,
+      "grad_norm": 2.849358320236206,
+      "learning_rate": 6.641815917287535e-05,
+      "loss": 1.6624,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4038041911976615,
+      "grad_norm": 2.1147572994232178,
+      "learning_rate": 6.63684616123747e-05,
+      "loss": 2.2939,
+      "step": 1226
+    },
+    {
+      "epoch": 0.40413355840092224,
+      "grad_norm": 2.6644229888916016,
+      "learning_rate": 6.631874593313531e-05,
+      "loss": 2.2582,
+      "step": 1227
+    },
+    {
+      "epoch": 0.40446292560418295,
+      "grad_norm": 2.094411849975586,
+      "learning_rate": 6.626901219018895e-05,
+      "loss": 2.3132,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4047922928074437,
+      "grad_norm": 2.631877899169922,
+      "learning_rate": 6.621926043858739e-05,
+      "loss": 2.1575,
+      "step": 1229
+    },
+    {
+      "epoch": 0.40512166001070443,
+      "grad_norm": 2.610651731491089,
+      "learning_rate": 6.616949073340232e-05,
+      "loss": 2.3285,
+      "step": 1230
+    },
+    {
+      "epoch": 0.40545102721396514,
+      "grad_norm": 2.4871439933776855,
+      "learning_rate": 6.611970312972531e-05,
+      "loss": 2.0776,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4057803944172259,
+      "grad_norm": 2.7471463680267334,
+      "learning_rate": 6.606989768266776e-05,
+      "loss": 2.5548,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4061097616204866,
+      "grad_norm": 2.746669292449951,
+      "learning_rate": 6.602007444736077e-05,
+      "loss": 2.2351,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4064391288237474,
+      "grad_norm": 2.907914161682129,
+      "learning_rate": 6.597023347895524e-05,
+      "loss": 2.7394,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4067684960270081,
+      "grad_norm": 2.477755308151245,
+      "learning_rate": 6.592037483262156e-05,
+      "loss": 2.0522,
+      "step": 1235
+    },
+    {
+      "epoch": 0.40709786323026886,
+      "grad_norm": 2.537019968032837,
+      "learning_rate": 6.587049856354977e-05,
+      "loss": 2.2179,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4074272304335296,
+      "grad_norm": 2.399907350540161,
+      "learning_rate": 6.582060472694939e-05,
+      "loss": 2.2994,
+      "step": 1237
+    },
+    {
+      "epoch": 0.40775659763679034,
+      "grad_norm": 2.9592201709747314,
+      "learning_rate": 6.577069337804944e-05,
+      "loss": 2.246,
+      "step": 1238
+    },
+    {
+      "epoch": 0.40808596484005105,
+      "grad_norm": 2.831124782562256,
+      "learning_rate": 6.572076457209822e-05,
+      "loss": 2.0093,
+      "step": 1239
+    },
+    {
+      "epoch": 0.40841533204331176,
+      "grad_norm": 2.620274066925049,
+      "learning_rate": 6.567081836436346e-05,
+      "loss": 1.8906,
+      "step": 1240
+    },
+    {
+      "epoch": 0.40874469924657253,
+      "grad_norm": 2.5152428150177,
+      "learning_rate": 6.562085481013211e-05,
+      "loss": 2.2982,
+      "step": 1241
+    },
+    {
+      "epoch": 0.40907406644983324,
+      "grad_norm": 3.2875349521636963,
+      "learning_rate": 6.55708739647103e-05,
+      "loss": 2.4582,
+      "step": 1242
+    },
+    {
+      "epoch": 0.409403433653094,
+      "grad_norm": 2.650205612182617,
+      "learning_rate": 6.552087588342332e-05,
+      "loss": 2.1196,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4097328008563547,
+      "grad_norm": 2.7873198986053467,
+      "learning_rate": 6.547086062161555e-05,
+      "loss": 1.9238,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4100621680596155,
+      "grad_norm": 2.823976993560791,
+      "learning_rate": 6.542082823465037e-05,
+      "loss": 1.9929,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4103915352628762,
+      "grad_norm": 3.1395912170410156,
+      "learning_rate": 6.537077877791011e-05,
+      "loss": 2.3549,
+      "step": 1246
+    },
+    {
+      "epoch": 0.41072090246613696,
+      "grad_norm": 2.8609724044799805,
+      "learning_rate": 6.532071230679604e-05,
+      "loss": 2.1222,
+      "step": 1247
+    },
+    {
+      "epoch": 0.41105026966939767,
+      "grad_norm": 2.8728225231170654,
+      "learning_rate": 6.527062887672819e-05,
+      "loss": 2.1451,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4113796368726584,
+      "grad_norm": 3.3821842670440674,
+      "learning_rate": 6.522052854314544e-05,
+      "loss": 2.3055,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41170900407591915,
+      "grad_norm": 3.682586431503296,
+      "learning_rate": 6.517041136150534e-05,
+      "loss": 1.9983,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41203837127917986,
+      "grad_norm": 2.3784186840057373,
+      "learning_rate": 6.512027738728407e-05,
+      "loss": 2.4458,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4123677384824406,
+      "grad_norm": 2.384232997894287,
+      "learning_rate": 6.507012667597643e-05,
+      "loss": 2.5749,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41269710568570134,
+      "grad_norm": 2.3909189701080322,
+      "learning_rate": 6.501995928309577e-05,
+      "loss": 2.3923,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4130264728889621,
+      "grad_norm": 2.478942394256592,
+      "learning_rate": 6.496977526417383e-05,
+      "loss": 2.5673,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4133558400922228,
+      "grad_norm": 2.6806704998016357,
+      "learning_rate": 6.491957467476081e-05,
+      "loss": 2.1562,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4136852072954835,
+      "grad_norm": 2.777129888534546,
+      "learning_rate": 6.486935757042529e-05,
+      "loss": 2.3145,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4140145744987443,
+      "grad_norm": 2.46478533744812,
+      "learning_rate": 6.481912400675402e-05,
+      "loss": 2.1051,
+      "step": 1257
+    },
+    {
+      "epoch": 0.414343941702005,
+      "grad_norm": 2.9172468185424805,
+      "learning_rate": 6.476887403935204e-05,
+      "loss": 2.32,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41467330890526577,
+      "grad_norm": 2.7249600887298584,
+      "learning_rate": 6.471860772384256e-05,
+      "loss": 2.2816,
+      "step": 1259
+    },
+    {
+      "epoch": 0.4150026761085265,
+      "grad_norm": 2.6038973331451416,
+      "learning_rate": 6.466832511586687e-05,
+      "loss": 2.1756,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41533204331178725,
+      "grad_norm": 2.6487905979156494,
+      "learning_rate": 6.461802627108426e-05,
+      "loss": 2.6021,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41566141051504796,
+      "grad_norm": 2.545063018798828,
+      "learning_rate": 6.456771124517205e-05,
+      "loss": 2.0434,
+      "step": 1262
+    },
+    {
+      "epoch": 0.4159907777183087,
+      "grad_norm": 2.7668240070343018,
+      "learning_rate": 6.451738009382542e-05,
+      "loss": 2.3299,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41632014492156943,
+      "grad_norm": 2.4981627464294434,
+      "learning_rate": 6.446703287275745e-05,
+      "loss": 1.9392,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41664951212483015,
+      "grad_norm": 2.7257003784179688,
+      "learning_rate": 6.441666963769897e-05,
+      "loss": 2.0421,
+      "step": 1265
+    },
+    {
+      "epoch": 0.4169788793280909,
+      "grad_norm": 2.8627986907958984,
+      "learning_rate": 6.436629044439854e-05,
+      "loss": 2.3554,
+      "step": 1266
+    },
+    {
+      "epoch": 0.4173082465313516,
+      "grad_norm": 2.81748628616333,
+      "learning_rate": 6.43158953486224e-05,
+      "loss": 1.9084,
+      "step": 1267
+    },
+    {
+      "epoch": 0.4176376137346124,
+      "grad_norm": 2.941399335861206,
+      "learning_rate": 6.426548440615438e-05,
+      "loss": 2.0619,
+      "step": 1268
+    },
+    {
+      "epoch": 0.4179669809378731,
+      "grad_norm": 3.05411434173584,
+      "learning_rate": 6.421505767279588e-05,
+      "loss": 2.1444,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41829634814113387,
+      "grad_norm": 2.318077564239502,
+      "learning_rate": 6.416461520436571e-05,
+      "loss": 1.7884,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4186257153443946,
+      "grad_norm": 2.866882562637329,
+      "learning_rate": 6.411415705670021e-05,
+      "loss": 2.0699,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41895508254765534,
+      "grad_norm": 2.7729244232177734,
+      "learning_rate": 6.406368328565295e-05,
+      "loss": 2.2929,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41928444975091606,
+      "grad_norm": 3.3135483264923096,
+      "learning_rate": 6.401319394709489e-05,
+      "loss": 2.4658,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41961381695417677,
+      "grad_norm": 3.1503915786743164,
+      "learning_rate": 6.396268909691414e-05,
+      "loss": 2.117,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41994318415743753,
+      "grad_norm": 2.9462268352508545,
+      "learning_rate": 6.391216879101608e-05,
+      "loss": 2.0571,
+      "step": 1275
+    },
+    {
+      "epoch": 0.42027255136069824,
+      "grad_norm": 2.186408042907715,
+      "learning_rate": 6.386163308532314e-05,
+      "loss": 2.5956,
+      "step": 1276
+    },
+    {
+      "epoch": 0.420601918563959,
+      "grad_norm": 2.4059898853302,
+      "learning_rate": 6.381108203577476e-05,
+      "loss": 2.4601,
+      "step": 1277
+    },
+    {
+      "epoch": 0.4209312857672197,
+      "grad_norm": 2.4408154487609863,
+      "learning_rate": 6.376051569832742e-05,
+      "loss": 2.537,
+      "step": 1278
+    },
+    {
+      "epoch": 0.4212606529704805,
+      "grad_norm": 2.3048956394195557,
+      "learning_rate": 6.370993412895454e-05,
+      "loss": 2.4092,
+      "step": 1279
+    },
+    {
+      "epoch": 0.4215900201737412,
+      "grad_norm": 2.4109179973602295,
+      "learning_rate": 6.365933738364634e-05,
+      "loss": 2.2523,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4219193873770019,
+      "grad_norm": 2.6136887073516846,
+      "learning_rate": 6.360872551840988e-05,
+      "loss": 2.0975,
+      "step": 1281
+    },
+    {
+      "epoch": 0.4222487545802627,
+      "grad_norm": 2.697741985321045,
+      "learning_rate": 6.355809858926893e-05,
+      "loss": 2.5818,
+      "step": 1282
+    },
+    {
+      "epoch": 0.4225781217835234,
+      "grad_norm": 2.3792057037353516,
+      "learning_rate": 6.350745665226396e-05,
+      "loss": 2.1916,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42290748898678415,
+      "grad_norm": 2.742786169052124,
+      "learning_rate": 6.345679976345205e-05,
+      "loss": 2.5287,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42323685619004486,
+      "grad_norm": 2.550384998321533,
+      "learning_rate": 6.34061279789068e-05,
+      "loss": 2.3782,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42356622339330563,
+      "grad_norm": 3.0929136276245117,
+      "learning_rate": 6.335544135471834e-05,
+      "loss": 2.3501,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42389559059656634,
+      "grad_norm": 2.920880079269409,
+      "learning_rate": 6.330473994699318e-05,
+      "loss": 2.3128,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4242249577998271,
+      "grad_norm": 2.258695125579834,
+      "learning_rate": 6.325402381185426e-05,
+      "loss": 2.0221,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4245543250030878,
+      "grad_norm": 2.6346912384033203,
+      "learning_rate": 6.320329300544076e-05,
+      "loss": 2.2641,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42488369220634853,
+      "grad_norm": 3.050056219100952,
+      "learning_rate": 6.315254758390814e-05,
+      "loss": 2.4125,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4252130594096093,
+      "grad_norm": 2.530167818069458,
+      "learning_rate": 6.3101787603428e-05,
+      "loss": 2.0573,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42554242661287,
+      "grad_norm": 3.0110995769500732,
+      "learning_rate": 6.305101312018809e-05,
+      "loss": 2.6722,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4258717938161308,
+      "grad_norm": 2.6352243423461914,
+      "learning_rate": 6.300022419039219e-05,
+      "loss": 2.0293,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4262011610193915,
+      "grad_norm": 2.864758253097534,
+      "learning_rate": 6.294942087026011e-05,
+      "loss": 1.9841,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42653052822265225,
+      "grad_norm": 3.216989517211914,
+      "learning_rate": 6.289860321602754e-05,
+      "loss": 2.005,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42685989542591296,
+      "grad_norm": 3.134946584701538,
+      "learning_rate": 6.284777128394603e-05,
+      "loss": 2.3214,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4271892626291737,
+      "grad_norm": 3.151312828063965,
+      "learning_rate": 6.279692513028304e-05,
+      "loss": 2.2082,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42751862983243444,
+      "grad_norm": 2.858823776245117,
+      "learning_rate": 6.274606481132163e-05,
+      "loss": 2.0439,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42784799703569515,
+      "grad_norm": 2.6305992603302,
+      "learning_rate": 6.269519038336062e-05,
+      "loss": 1.7326,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4281773642389559,
+      "grad_norm": 4.182598114013672,
+      "learning_rate": 6.264430190271444e-05,
+      "loss": 1.8915,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4285067314422166,
+      "grad_norm": 2.144503116607666,
+      "learning_rate": 6.259339942571307e-05,
+      "loss": 2.3162,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4288360986454774,
+      "grad_norm": 2.138836622238159,
+      "learning_rate": 6.254248300870198e-05,
+      "loss": 2.3545,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4291654658487381,
+      "grad_norm": 2.6747050285339355,
+      "learning_rate": 6.249155270804206e-05,
+      "loss": 2.6766,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42949483305199887,
+      "grad_norm": 2.2941064834594727,
+      "learning_rate": 6.24406085801096e-05,
+      "loss": 2.6815,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4298242002552596,
+      "grad_norm": 2.450798988342285,
+      "learning_rate": 6.238965068129616e-05,
+      "loss": 2.3887,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4301535674585203,
+      "grad_norm": 2.7236177921295166,
+      "learning_rate": 6.233867906800856e-05,
+      "loss": 2.6746,
+      "step": 1306
+    },
+    {
+      "epoch": 0.43048293466178106,
+      "grad_norm": 2.221660852432251,
+      "learning_rate": 6.22876937966688e-05,
+      "loss": 2.4207,
+      "step": 1307
+    },
+    {
+      "epoch": 0.43081230186504177,
+      "grad_norm": 2.9190986156463623,
+      "learning_rate": 6.2236694923714e-05,
+      "loss": 2.5129,
+      "step": 1308
+    },
+    {
+      "epoch": 0.43114166906830254,
+      "grad_norm": 2.183797597885132,
+      "learning_rate": 6.218568250559634e-05,
+      "loss": 2.209,
+      "step": 1309
+    },
+    {
+      "epoch": 0.43147103627156325,
+      "grad_norm": 2.7680306434631348,
+      "learning_rate": 6.2134656598783e-05,
+      "loss": 1.9334,
+      "step": 1310
+    },
+    {
+      "epoch": 0.431800403474824,
+      "grad_norm": 2.6052422523498535,
+      "learning_rate": 6.208361725975605e-05,
+      "loss": 2.2873,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4321297706780847,
+      "grad_norm": 2.873652696609497,
+      "learning_rate": 6.203256454501248e-05,
+      "loss": 2.5024,
+      "step": 1312
+    },
+    {
+      "epoch": 0.4324591378813455,
+      "grad_norm": 3.202440023422241,
+      "learning_rate": 6.198149851106407e-05,
+      "loss": 2.2169,
+      "step": 1313
+    },
+    {
+      "epoch": 0.4327885050846062,
+      "grad_norm": 2.7076804637908936,
+      "learning_rate": 6.19304192144373e-05,
+      "loss": 2.0037,
+      "step": 1314
+    },
+    {
+      "epoch": 0.4331178722878669,
+      "grad_norm": 3.2877442836761475,
+      "learning_rate": 6.187932671167342e-05,
+      "loss": 2.6063,
+      "step": 1315
+    },
+    {
+      "epoch": 0.4334472394911277,
+      "grad_norm": 2.6223089694976807,
+      "learning_rate": 6.18282210593282e-05,
+      "loss": 1.959,
+      "step": 1316
+    },
+    {
+      "epoch": 0.4337766066943884,
+      "grad_norm": 3.001115083694458,
+      "learning_rate": 6.177710231397203e-05,
+      "loss": 2.1598,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43410597389764916,
+      "grad_norm": 2.4432201385498047,
+      "learning_rate": 6.172597053218978e-05,
+      "loss": 1.7752,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43443534110090987,
+      "grad_norm": 2.4193572998046875,
+      "learning_rate": 6.167482577058075e-05,
+      "loss": 2.0442,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43476470830417063,
+      "grad_norm": 2.7738735675811768,
+      "learning_rate": 6.162366808575857e-05,
+      "loss": 2.1012,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43509407550743134,
+      "grad_norm": 2.398259401321411,
+      "learning_rate": 6.157249753435124e-05,
+      "loss": 1.641,
+      "step": 1321
+    },
+    {
+      "epoch": 0.4354234427106921,
+      "grad_norm": 2.803773880004883,
+      "learning_rate": 6.152131417300098e-05,
+      "loss": 2.1607,
+      "step": 1322
+    },
+    {
+      "epoch": 0.4357528099139528,
+      "grad_norm": 3.041407346725464,
+      "learning_rate": 6.147011805836414e-05,
+      "loss": 1.8865,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43608217711721353,
+      "grad_norm": 4.000000476837158,
+      "learning_rate": 6.141890924711126e-05,
+      "loss": 1.8523,
+      "step": 1324
+    },
+    {
+      "epoch": 0.4364115443204743,
+      "grad_norm": 3.650296688079834,
+      "learning_rate": 6.136768779592691e-05,
+      "loss": 2.2463,
+      "step": 1325
+    },
+    {
+      "epoch": 0.436740911523735,
+      "grad_norm": 2.288025140762329,
+      "learning_rate": 6.13164537615096e-05,
+      "loss": 2.5091,
+      "step": 1326
+    },
+    {
+      "epoch": 0.4370702787269958,
+      "grad_norm": 2.0283560752868652,
+      "learning_rate": 6.126520720057186e-05,
+      "loss": 2.2552,
+      "step": 1327
+    },
+    {
+      "epoch": 0.4373996459302565,
+      "grad_norm": 2.147888422012329,
+      "learning_rate": 6.121394816984e-05,
+      "loss": 2.3754,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43772901313351725,
+      "grad_norm": 2.3866569995880127,
+      "learning_rate": 6.11626767260542e-05,
+      "loss": 2.2485,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43805838033677796,
+      "grad_norm": 2.4724600315093994,
+      "learning_rate": 6.111139292596834e-05,
+      "loss": 2.3486,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4383877475400387,
+      "grad_norm": 2.1698498725891113,
+      "learning_rate": 6.106009682634997e-05,
+      "loss": 2.2673,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43871711474329944,
+      "grad_norm": 2.1860690116882324,
+      "learning_rate": 6.100878848398032e-05,
+      "loss": 2.1531,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43904648194656015,
+      "grad_norm": 2.3432462215423584,
+      "learning_rate": 6.095746795565408e-05,
+      "loss": 2.3907,
+      "step": 1333
+    },
+    {
+      "epoch": 0.4393758491498209,
+      "grad_norm": 3.00425386428833,
+      "learning_rate": 6.090613529817949e-05,
+      "loss": 2.5884,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43970521635308163,
+      "grad_norm": 3.2894668579101562,
+      "learning_rate": 6.085479056837821e-05,
+      "loss": 2.2589,
+      "step": 1335
+    },
+    {
+      "epoch": 0.4400345835563424,
+      "grad_norm": 2.9489943981170654,
+      "learning_rate": 6.0803433823085244e-05,
+      "loss": 2.2962,
+      "step": 1336
+    },
+    {
+      "epoch": 0.4403639507596031,
+      "grad_norm": 3.29754638671875,
+      "learning_rate": 6.075206511914891e-05,
+      "loss": 2.4655,
+      "step": 1337
+    },
+    {
+      "epoch": 0.4406933179628639,
+      "grad_norm": 2.559438705444336,
+      "learning_rate": 6.070068451343074e-05,
+      "loss": 2.2637,
+      "step": 1338
+    },
+    {
+      "epoch": 0.4410226851661246,
+      "grad_norm": 3.065319061279297,
+      "learning_rate": 6.0649292062805494e-05,
+      "loss": 2.1736,
+      "step": 1339
+    },
+    {
+      "epoch": 0.4413520523693853,
+      "grad_norm": 2.943509101867676,
+      "learning_rate": 6.059788782416099e-05,
+      "loss": 2.0683,
+      "step": 1340
+    },
+    {
+      "epoch": 0.44168141957264606,
+      "grad_norm": 2.7584707736968994,
+      "learning_rate": 6.054647185439814e-05,
+      "loss": 2.3206,
+      "step": 1341
+    },
+    {
+      "epoch": 0.44201078677590677,
+      "grad_norm": 2.789400577545166,
+      "learning_rate": 6.049504421043078e-05,
+      "loss": 1.8737,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44234015397916754,
+      "grad_norm": 2.9476754665374756,
+      "learning_rate": 6.0443604949185706e-05,
+      "loss": 2.2538,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44266952118242825,
+      "grad_norm": 2.6708343029022217,
+      "learning_rate": 6.0392154127602595e-05,
+      "loss": 2.156,
+      "step": 1344
+    },
+    {
+      "epoch": 0.442998888385689,
+      "grad_norm": 2.7900948524475098,
+      "learning_rate": 6.0340691802633884e-05,
+      "loss": 1.8873,
+      "step": 1345
+    },
+    {
+      "epoch": 0.4433282555889497,
+      "grad_norm": 2.6233713626861572,
+      "learning_rate": 6.028921803124476e-05,
+      "loss": 1.7659,
+      "step": 1346
+    },
+    {
+      "epoch": 0.4436576227922105,
+      "grad_norm": 2.7721614837646484,
+      "learning_rate": 6.023773287041308e-05,
+      "loss": 2.0038,
+      "step": 1347
+    },
+    {
+      "epoch": 0.4439869899954712,
+      "grad_norm": 2.7499945163726807,
+      "learning_rate": 6.01862363771293e-05,
+      "loss": 2.1821,
+      "step": 1348
+    },
+    {
+      "epoch": 0.4443163571987319,
+      "grad_norm": 2.655479907989502,
+      "learning_rate": 6.013472860839642e-05,
+      "loss": 1.9229,
+      "step": 1349
+    },
+    {
+      "epoch": 0.4446457244019927,
+      "grad_norm": 2.927269458770752,
+      "learning_rate": 6.008320962122994e-05,
+      "loss": 1.8664,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4449750916052534,
+      "grad_norm": 1.9483531713485718,
+      "learning_rate": 6.003167947265777e-05,
+      "loss": 2.5558,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44530445880851416,
+      "grad_norm": 2.5765058994293213,
+      "learning_rate": 5.9980138219720125e-05,
+      "loss": 2.6047,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44563382601177487,
+      "grad_norm": 2.483147382736206,
+      "learning_rate": 5.992858591946961e-05,
+      "loss": 2.324,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44596319321503564,
+      "grad_norm": 2.8535609245300293,
+      "learning_rate": 5.987702262897098e-05,
+      "loss": 2.395,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44629256041829635,
+      "grad_norm": 2.455509662628174,
+      "learning_rate": 5.9825448405301175e-05,
+      "loss": 2.3842,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44662192762155706,
+      "grad_norm": 2.347088575363159,
+      "learning_rate": 5.977386330554926e-05,
+      "loss": 2.2313,
+      "step": 1356
+    },
+    {
+      "epoch": 0.4469512948248178,
+      "grad_norm": 2.4735326766967773,
+      "learning_rate": 5.9722267386816324e-05,
+      "loss": 2.1965,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44728066202807854,
+      "grad_norm": 2.5871574878692627,
+      "learning_rate": 5.967066070621541e-05,
+      "loss": 2.2823,
+      "step": 1358
+    },
+    {
+      "epoch": 0.4476100292313393,
+      "grad_norm": 2.4709925651550293,
+      "learning_rate": 5.9619043320871494e-05,
+      "loss": 2.2467,
+      "step": 1359
+    },
+    {
+      "epoch": 0.4479393964346,
+      "grad_norm": 2.6086020469665527,
+      "learning_rate": 5.956741528792142e-05,
+      "loss": 2.2654,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4482687636378608,
+      "grad_norm": 2.7858760356903076,
+      "learning_rate": 5.951577666451379e-05,
+      "loss": 2.3658,
+      "step": 1361
+    },
+    {
+      "epoch": 0.4485981308411215,
+      "grad_norm": 2.6391570568084717,
+      "learning_rate": 5.946412750780892e-05,
+      "loss": 2.1763,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44892749804438226,
+      "grad_norm": 2.597666025161743,
+      "learning_rate": 5.941246787497884e-05,
+      "loss": 2.1193,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44925686524764297,
+      "grad_norm": 2.591607093811035,
+      "learning_rate": 5.9360797823207104e-05,
+      "loss": 1.9894,
+      "step": 1364
+    },
+    {
+      "epoch": 0.4495862324509037,
+      "grad_norm": 2.4428396224975586,
+      "learning_rate": 5.930911740968884e-05,
+      "loss": 2.1478,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44991559965416444,
+      "grad_norm": 2.8427088260650635,
+      "learning_rate": 5.9257426691630656e-05,
+      "loss": 2.0804,
+      "step": 1366
+    },
+    {
+      "epoch": 0.45024496685742516,
+      "grad_norm": 2.4200117588043213,
+      "learning_rate": 5.920572572625056e-05,
+      "loss": 2.0263,
+      "step": 1367
+    },
+    {
+      "epoch": 0.4505743340606859,
+      "grad_norm": 2.4792933464050293,
+      "learning_rate": 5.915401457077785e-05,
+      "loss": 2.1457,
+      "step": 1368
+    },
+    {
+      "epoch": 0.45090370126394663,
+      "grad_norm": 3.0544183254241943,
+      "learning_rate": 5.910229328245319e-05,
+      "loss": 2.476,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4512330684672074,
+      "grad_norm": 3.122922897338867,
+      "learning_rate": 5.90505619185284e-05,
+      "loss": 1.8849,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4515624356704681,
+      "grad_norm": 3.2922286987304688,
+      "learning_rate": 5.899882053626646e-05,
+      "loss": 2.1517,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4518918028737289,
+      "grad_norm": 3.095803737640381,
+      "learning_rate": 5.8947069192941493e-05,
+      "loss": 2.1022,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4522211700769896,
+      "grad_norm": 3.426305055618286,
+      "learning_rate": 5.889530794583855e-05,
+      "loss": 2.1558,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4525505372802503,
+      "grad_norm": 3.2254509925842285,
+      "learning_rate": 5.8843536852253745e-05,
+      "loss": 2.2135,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45287990448351106,
+      "grad_norm": 3.3690662384033203,
+      "learning_rate": 5.879175596949401e-05,
+      "loss": 2.0148,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4532092716867718,
+      "grad_norm": 2.666074275970459,
+      "learning_rate": 5.8739965354877194e-05,
+      "loss": 2.445,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45353863889003254,
+      "grad_norm": 2.3343346118927,
+      "learning_rate": 5.8688165065731826e-05,
+      "loss": 2.332,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45386800609329325,
+      "grad_norm": 2.3201119899749756,
+      "learning_rate": 5.8636355159397225e-05,
+      "loss": 2.5834,
+      "step": 1378
+    },
+    {
+      "epoch": 0.454197373296554,
+      "grad_norm": 2.493422269821167,
+      "learning_rate": 5.858453569322332e-05,
+      "loss": 2.5553,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45452674049981473,
+      "grad_norm": 2.374323606491089,
+      "learning_rate": 5.853270672457061e-05,
+      "loss": 2.1962,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45485610770307544,
+      "grad_norm": 2.812485694885254,
+      "learning_rate": 5.8480868310810124e-05,
+      "loss": 2.3164,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4551854749063362,
+      "grad_norm": 2.3375439643859863,
+      "learning_rate": 5.8429020509323385e-05,
+      "loss": 2.4357,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4555148421095969,
+      "grad_norm": 2.376716136932373,
+      "learning_rate": 5.837716337750223e-05,
+      "loss": 2.5416,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4558442093128577,
+      "grad_norm": 2.5697383880615234,
+      "learning_rate": 5.8325296972748864e-05,
+      "loss": 2.3119,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4561735765161184,
+      "grad_norm": 2.5520308017730713,
+      "learning_rate": 5.827342135247581e-05,
+      "loss": 2.3784,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45650294371937916,
+      "grad_norm": 2.750321865081787,
+      "learning_rate": 5.8221536574105694e-05,
+      "loss": 2.5948,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4568323109226399,
+      "grad_norm": 2.980220317840576,
+      "learning_rate": 5.816964269507135e-05,
+      "loss": 2.5563,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45716167812590064,
+      "grad_norm": 2.898198366165161,
+      "learning_rate": 5.811773977281565e-05,
+      "loss": 2.2648,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45749104532916135,
+      "grad_norm": 3.0173451900482178,
+      "learning_rate": 5.806582786479149e-05,
+      "loss": 2.2571,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45782041253242206,
+      "grad_norm": 2.6375420093536377,
+      "learning_rate": 5.801390702846171e-05,
+      "loss": 2.0654,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4581497797356828,
+      "grad_norm": 2.8282692432403564,
+      "learning_rate": 5.796197732129905e-05,
+      "loss": 2.3666,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45847914693894354,
+      "grad_norm": 2.815434694290161,
+      "learning_rate": 5.7910038800786e-05,
+      "loss": 2.2048,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4588085141422043,
+      "grad_norm": 2.911081075668335,
+      "learning_rate": 5.7858091524414926e-05,
+      "loss": 2.3954,
+      "step": 1393
+    },
+    {
+      "epoch": 0.459137881345465,
+      "grad_norm": 2.4357879161834717,
+      "learning_rate": 5.780613554968777e-05,
+      "loss": 1.9379,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4594672485487258,
+      "grad_norm": 3.2798242568969727,
+      "learning_rate": 5.775417093411619e-05,
+      "loss": 2.3849,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4597966157519865,
+      "grad_norm": 2.6205365657806396,
+      "learning_rate": 5.770219773522133e-05,
+      "loss": 2.2528,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4601259829552472,
+      "grad_norm": 2.84399676322937,
+      "learning_rate": 5.765021601053391e-05,
+      "loss": 1.8288,
+      "step": 1397
+    },
+    {
+      "epoch": 0.46045535015850797,
+      "grad_norm": 2.869101047515869,
+      "learning_rate": 5.7598225817594035e-05,
+      "loss": 2.3284,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4607847173617687,
+      "grad_norm": 2.6911323070526123,
+      "learning_rate": 5.754622721395119e-05,
+      "loss": 2.0477,
+      "step": 1399
+    },
+    {
+      "epoch": 0.46111408456502945,
+      "grad_norm": 3.472745656967163,
+      "learning_rate": 5.74942202571642e-05,
+      "loss": 1.8559,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46144345176829016,
+      "grad_norm": 2.722027063369751,
+      "learning_rate": 5.744220500480113e-05,
+      "loss": 2.5318,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4617728189715509,
+      "grad_norm": 2.0918314456939697,
+      "learning_rate": 5.739018151443918e-05,
+      "loss": 2.1847,
+      "step": 1402
+    },
+    {
+      "epoch": 0.46210218617481164,
+      "grad_norm": 2.0019867420196533,
+      "learning_rate": 5.733814984366474e-05,
+      "loss": 2.2064,
+      "step": 1403
+    },
+    {
+      "epoch": 0.4624315533780724,
+      "grad_norm": 2.3096742630004883,
+      "learning_rate": 5.7286110050073194e-05,
+      "loss": 2.0878,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4627609205813331,
+      "grad_norm": 2.7274067401885986,
+      "learning_rate": 5.723406219126895e-05,
+      "loss": 2.2556,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4630902877845938,
+      "grad_norm": 2.2362143993377686,
+      "learning_rate": 5.718200632486534e-05,
+      "loss": 2.0013,
+      "step": 1406
+    },
+    {
+      "epoch": 0.4634196549878546,
+      "grad_norm": 2.458976984024048,
+      "learning_rate": 5.7129942508484556e-05,
+      "loss": 2.1049,
+      "step": 1407
+    },
+    {
+      "epoch": 0.4637490221911153,
+      "grad_norm": 2.278069496154785,
+      "learning_rate": 5.707787079975758e-05,
+      "loss": 2.0951,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46407838939437607,
+      "grad_norm": 2.235696315765381,
+      "learning_rate": 5.702579125632416e-05,
+      "loss": 2.4472,
+      "step": 1409
+    },
+    {
+      "epoch": 0.4644077565976368,
+      "grad_norm": 2.662332773208618,
+      "learning_rate": 5.697370393583269e-05,
+      "loss": 2.1896,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46473712380089754,
+      "grad_norm": 2.677049398422241,
+      "learning_rate": 5.692160889594017e-05,
+      "loss": 2.0806,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46506649100415826,
+      "grad_norm": 2.7263095378875732,
+      "learning_rate": 5.686950619431215e-05,
+      "loss": 2.1826,
+      "step": 1412
+    },
+    {
+      "epoch": 0.465395858207419,
+      "grad_norm": 2.625284194946289,
+      "learning_rate": 5.6817395888622694e-05,
+      "loss": 2.4464,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46572522541067973,
+      "grad_norm": 2.6425976753234863,
+      "learning_rate": 5.6765278036554225e-05,
+      "loss": 2.0412,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46605459261394044,
+      "grad_norm": 2.918874502182007,
+      "learning_rate": 5.671315269579756e-05,
+      "loss": 2.3,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4663839598172012,
+      "grad_norm": 2.5336811542510986,
+      "learning_rate": 5.6661019924051814e-05,
+      "loss": 2.4006,
+      "step": 1416
+    },
+    {
+      "epoch": 0.4667133270204619,
+      "grad_norm": 2.6169357299804688,
+      "learning_rate": 5.6608879779024274e-05,
+      "loss": 2.1464,
+      "step": 1417
+    },
+    {
+      "epoch": 0.4670426942237227,
+      "grad_norm": 2.734600305557251,
+      "learning_rate": 5.6556732318430437e-05,
+      "loss": 2.0906,
+      "step": 1418
+    },
+    {
+      "epoch": 0.4673720614269834,
+      "grad_norm": 2.7634074687957764,
+      "learning_rate": 5.650457759999389e-05,
+      "loss": 2.1538,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46770142863024416,
+      "grad_norm": 2.391335964202881,
+      "learning_rate": 5.6452415681446256e-05,
+      "loss": 1.7893,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4680307958335049,
+      "grad_norm": 2.9604604244232178,
+      "learning_rate": 5.6400246620527096e-05,
+      "loss": 2.1226,
+      "step": 1421
+    },
+    {
+      "epoch": 0.4683601630367656,
+      "grad_norm": 2.650703191757202,
+      "learning_rate": 5.6348070474983905e-05,
+      "loss": 1.8164,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46868953024002635,
+      "grad_norm": 2.603179454803467,
+      "learning_rate": 5.629588730257205e-05,
+      "loss": 1.6553,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46901889744328706,
+      "grad_norm": 2.903170108795166,
+      "learning_rate": 5.6243697161054584e-05,
+      "loss": 1.8419,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46934826464654783,
+      "grad_norm": 3.1051528453826904,
+      "learning_rate": 5.619150010820238e-05,
+      "loss": 1.6226,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46967763184980854,
+      "grad_norm": 1.8390017747879028,
+      "learning_rate": 5.613929620179389e-05,
+      "loss": 2.2257,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4700069990530693,
+      "grad_norm": 2.490265130996704,
+      "learning_rate": 5.608708549961519e-05,
+      "loss": 2.3091,
+      "step": 1427
+    },
+    {
+      "epoch": 0.47033636625633,
+      "grad_norm": 2.267498254776001,
+      "learning_rate": 5.603486805945984e-05,
+      "loss": 2.4721,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4706657334595908,
+      "grad_norm": 2.788170337677002,
+      "learning_rate": 5.598264393912891e-05,
+      "loss": 2.453,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4709951006628515,
+      "grad_norm": 2.2585389614105225,
+      "learning_rate": 5.593041319643083e-05,
+      "loss": 2.2228,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4713244678661122,
+      "grad_norm": 2.556220531463623,
+      "learning_rate": 5.587817588918137e-05,
+      "loss": 2.3444,
+      "step": 1431
+    },
+    {
+      "epoch": 0.471653835069373,
+      "grad_norm": 2.604313611984253,
+      "learning_rate": 5.582593207520357e-05,
+      "loss": 2.1515,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4719832022726337,
+      "grad_norm": 2.6779942512512207,
+      "learning_rate": 5.577368181232764e-05,
+      "loss": 2.1363,
+      "step": 1433
+    },
+    {
+      "epoch": 0.47231256947589445,
+      "grad_norm": 2.4980621337890625,
+      "learning_rate": 5.572142515839098e-05,
+      "loss": 2.0295,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47264193667915516,
+      "grad_norm": 2.331618309020996,
+      "learning_rate": 5.5669162171238046e-05,
+      "loss": 2.1247,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4729713038824159,
+      "grad_norm": 3.248292922973633,
+      "learning_rate": 5.5616892908720274e-05,
+      "loss": 2.2727,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47330067108567664,
+      "grad_norm": 2.3975675106048584,
+      "learning_rate": 5.556461742869609e-05,
+      "loss": 2.3504,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4736300382889374,
+      "grad_norm": 2.7226855754852295,
+      "learning_rate": 5.551233578903078e-05,
+      "loss": 2.0893,
+      "step": 1438
+    },
+    {
+      "epoch": 0.4739594054921981,
+      "grad_norm": 2.8404018878936768,
+      "learning_rate": 5.5460048047596434e-05,
+      "loss": 2.4199,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4742887726954588,
+      "grad_norm": 2.4289791584014893,
+      "learning_rate": 5.540775426227194e-05,
+      "loss": 2.4747,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4746181398987196,
+      "grad_norm": 2.772507905960083,
+      "learning_rate": 5.535545449094283e-05,
+      "loss": 2.054,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4749475071019803,
+      "grad_norm": 2.7915074825286865,
+      "learning_rate": 5.5303148791501305e-05,
+      "loss": 2.3271,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47527687430524107,
+      "grad_norm": 2.894771099090576,
+      "learning_rate": 5.525083722184607e-05,
+      "loss": 2.2738,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4756062415085018,
+      "grad_norm": 2.797031879425049,
+      "learning_rate": 5.519851983988239e-05,
+      "loss": 2.3225,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47593560871176255,
+      "grad_norm": 3.053269386291504,
+      "learning_rate": 5.514619670352192e-05,
+      "loss": 2.1418,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47626497591502326,
+      "grad_norm": 2.743236780166626,
+      "learning_rate": 5.5093867870682725e-05,
+      "loss": 1.966,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47659434311828397,
+      "grad_norm": 2.998206615447998,
+      "learning_rate": 5.504153339928914e-05,
+      "loss": 2.2283,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47692371032154474,
+      "grad_norm": 3.0275509357452393,
+      "learning_rate": 5.498919334727175e-05,
+      "loss": 2.0081,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47725307752480545,
+      "grad_norm": 3.7755558490753174,
+      "learning_rate": 5.4936847772567314e-05,
+      "loss": 2.3072,
+      "step": 1449
+    },
+    {
+      "epoch": 0.4775824447280662,
+      "grad_norm": 3.9976677894592285,
+      "learning_rate": 5.488449673311872e-05,
+      "loss": 2.2594,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4779118119313269,
+      "grad_norm": 2.3137824535369873,
+      "learning_rate": 5.48321402868749e-05,
+      "loss": 2.6206,
+      "step": 1451
+    },
+    {
+      "epoch": 0.4782411791345877,
+      "grad_norm": 2.003905773162842,
+      "learning_rate": 5.477977849179076e-05,
+      "loss": 2.221,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4785705463378484,
+      "grad_norm": 2.37841534614563,
+      "learning_rate": 5.4727411405827136e-05,
+      "loss": 2.5322,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47889991354110917,
+      "grad_norm": 2.259744644165039,
+      "learning_rate": 5.467503908695073e-05,
+      "loss": 1.9622,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4792292807443699,
+      "grad_norm": 2.1672475337982178,
+      "learning_rate": 5.4622661593133996e-05,
+      "loss": 2.0246,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4795586479476306,
+      "grad_norm": 2.359017848968506,
+      "learning_rate": 5.457027898235517e-05,
+      "loss": 2.0392,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47988801515089136,
+      "grad_norm": 2.5766847133636475,
+      "learning_rate": 5.451789131259814e-05,
+      "loss": 2.2329,
+      "step": 1457
+    },
+    {
+      "epoch": 0.48021738235415207,
+      "grad_norm": 2.3741567134857178,
+      "learning_rate": 5.446549864185233e-05,
+      "loss": 2.2631,
+      "step": 1458
+    },
+    {
+      "epoch": 0.48054674955741283,
+      "grad_norm": 2.526066541671753,
+      "learning_rate": 5.44131010281128e-05,
+      "loss": 2.1493,
+      "step": 1459
+    },
+    {
+      "epoch": 0.48087611676067354,
+      "grad_norm": 2.551849126815796,
+      "learning_rate": 5.4360698529380004e-05,
+      "loss": 2.3144,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4812054839639343,
+      "grad_norm": 2.2339227199554443,
+      "learning_rate": 5.4308291203659855e-05,
+      "loss": 2.2004,
+      "step": 1461
+    },
+    {
+      "epoch": 0.481534851167195,
+      "grad_norm": 2.5119829177856445,
+      "learning_rate": 5.425587910896357e-05,
+      "loss": 2.0944,
+      "step": 1462
+    },
+    {
+      "epoch": 0.4818642183704558,
+      "grad_norm": 2.8552167415618896,
+      "learning_rate": 5.4203462303307685e-05,
+      "loss": 2.1718,
+      "step": 1463
+    },
+    {
+      "epoch": 0.4821935855737165,
+      "grad_norm": 2.6687982082366943,
+      "learning_rate": 5.4151040844713886e-05,
+      "loss": 2.2164,
+      "step": 1464
+    },
+    {
+      "epoch": 0.4825229527769772,
+      "grad_norm": 2.963787317276001,
+      "learning_rate": 5.409861479120908e-05,
+      "loss": 2.22,
+      "step": 1465
+    },
+    {
+      "epoch": 0.482852319980238,
+      "grad_norm": 2.990370750427246,
+      "learning_rate": 5.404618420082521e-05,
+      "loss": 2.2234,
+      "step": 1466
+    },
+    {
+      "epoch": 0.4831816871834987,
+      "grad_norm": 2.695317506790161,
+      "learning_rate": 5.39937491315993e-05,
+      "loss": 2.2448,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48351105438675945,
+      "grad_norm": 2.775763988494873,
+      "learning_rate": 5.394130964157324e-05,
+      "loss": 2.1806,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48384042159002016,
+      "grad_norm": 2.3641860485076904,
+      "learning_rate": 5.388886578879392e-05,
+      "loss": 1.927,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48416978879328093,
+      "grad_norm": 2.720329999923706,
+      "learning_rate": 5.383641763131297e-05,
+      "loss": 1.9248,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48449915599654164,
+      "grad_norm": 2.7788572311401367,
+      "learning_rate": 5.378396522718683e-05,
+      "loss": 1.9824,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48482852319980235,
+      "grad_norm": 3.20474910736084,
+      "learning_rate": 5.373150863447662e-05,
+      "loss": 1.5423,
+      "step": 1472
+    },
+    {
+      "epoch": 0.4851578904030631,
+      "grad_norm": 2.918916940689087,
+      "learning_rate": 5.367904791124815e-05,
+      "loss": 1.9387,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48548725760632383,
+      "grad_norm": 3.4597556591033936,
+      "learning_rate": 5.3626583115571716e-05,
+      "loss": 2.226,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4858166248095846,
+      "grad_norm": 3.169464588165283,
+      "learning_rate": 5.357411430552216e-05,
+      "loss": 2.0575,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4861459920128453,
+      "grad_norm": 2.0502803325653076,
+      "learning_rate": 5.352164153917882e-05,
+      "loss": 2.3078,
+      "step": 1476
+    },
+    {
+      "epoch": 0.4864753592161061,
+      "grad_norm": 2.205932855606079,
+      "learning_rate": 5.3469164874625345e-05,
+      "loss": 2.3107,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4868047264193668,
+      "grad_norm": 1.9236304759979248,
+      "learning_rate": 5.341668436994971e-05,
+      "loss": 2.22,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48713409362262755,
+      "grad_norm": 2.331031322479248,
+      "learning_rate": 5.3364200083244175e-05,
+      "loss": 2.1931,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48746346082588826,
+      "grad_norm": 2.279069185256958,
+      "learning_rate": 5.3311712072605136e-05,
+      "loss": 2.2918,
+      "step": 1480
+    },
+    {
+      "epoch": 0.487792828029149,
+      "grad_norm": 2.3501081466674805,
+      "learning_rate": 5.325922039613316e-05,
+      "loss": 2.28,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48812219523240974,
+      "grad_norm": 2.4179487228393555,
+      "learning_rate": 5.320672511193285e-05,
+      "loss": 2.4011,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48845156243567045,
+      "grad_norm": 2.3031511306762695,
+      "learning_rate": 5.315422627811278e-05,
+      "loss": 2.0346,
+      "step": 1483
+    },
+    {
+      "epoch": 0.4887809296389312,
+      "grad_norm": 3.2005553245544434,
+      "learning_rate": 5.310172395278551e-05,
+      "loss": 2.5121,
+      "step": 1484
+    },
+    {
+      "epoch": 0.4891102968421919,
+      "grad_norm": 2.487902879714966,
+      "learning_rate": 5.3049218194067394e-05,
+      "loss": 2.0688,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4894396640454527,
+      "grad_norm": 2.2701635360717773,
+      "learning_rate": 5.299670906007866e-05,
+      "loss": 2.0644,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4897690312487134,
+      "grad_norm": 3.022916078567505,
+      "learning_rate": 5.294419660894322e-05,
+      "loss": 2.3781,
+      "step": 1487
+    },
+    {
+      "epoch": 0.49009839845197417,
+      "grad_norm": 2.469449758529663,
+      "learning_rate": 5.2891680898788665e-05,
+      "loss": 1.7029,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4904277656552349,
+      "grad_norm": 2.625373363494873,
+      "learning_rate": 5.283916198774621e-05,
+      "loss": 2.0506,
+      "step": 1489
+    },
+    {
+      "epoch": 0.4907571328584956,
+      "grad_norm": 2.7571139335632324,
+      "learning_rate": 5.2786639933950597e-05,
+      "loss": 2.1408,
+      "step": 1490
+    },
+    {
+      "epoch": 0.49108650006175636,
+      "grad_norm": 2.4763495922088623,
+      "learning_rate": 5.273411479554008e-05,
+      "loss": 1.9425,
+      "step": 1491
+    },
+    {
+      "epoch": 0.49141586726501707,
+      "grad_norm": 3.169473171234131,
+      "learning_rate": 5.2681586630656276e-05,
+      "loss": 2.4006,
+      "step": 1492
+    },
+    {
+      "epoch": 0.49174523446827784,
+      "grad_norm": 2.6018457412719727,
+      "learning_rate": 5.262905549744419e-05,
+      "loss": 2.0221,
+      "step": 1493
+    },
+    {
+      "epoch": 0.49207460167153855,
+      "grad_norm": 2.7356503009796143,
+      "learning_rate": 5.25765214540521e-05,
+      "loss": 2.1555,
+      "step": 1494
+    },
+    {
+      "epoch": 0.4924039688747993,
+      "grad_norm": 2.443387269973755,
+      "learning_rate": 5.2523984558631514e-05,
+      "loss": 2.0985,
+      "step": 1495
+    },
+    {
+      "epoch": 0.49273333607806,
+      "grad_norm": 2.6359920501708984,
+      "learning_rate": 5.247144486933706e-05,
+      "loss": 2.217,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49306270328132074,
+      "grad_norm": 2.7415084838867188,
+      "learning_rate": 5.241890244432652e-05,
+      "loss": 1.8597,
+      "step": 1497
+    },
+    {
+      "epoch": 0.4933920704845815,
+      "grad_norm": 2.734299421310425,
+      "learning_rate": 5.236635734176069e-05,
+      "loss": 2.2436,
+      "step": 1498
+    },
+    {
+      "epoch": 0.4937214376878422,
+      "grad_norm": 2.9822301864624023,
+      "learning_rate": 5.231380961980326e-05,
+      "loss": 2.0025,
+      "step": 1499
+    },
+    {
+      "epoch": 0.494050804891103,
+      "grad_norm": 3.2635011672973633,
+      "learning_rate": 5.226125933662088e-05,
+      "loss": 2.073,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4943801720943637,
+      "grad_norm": 2.0584778785705566,
+      "learning_rate": 5.220870655038308e-05,
+      "loss": 2.2952,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49470953929762446,
+      "grad_norm": 2.4790139198303223,
+      "learning_rate": 5.2156151319262045e-05,
+      "loss": 2.5564,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49503890650088517,
+      "grad_norm": 2.687093496322632,
+      "learning_rate": 5.2103593701432776e-05,
+      "loss": 2.0692,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49536827370414593,
+      "grad_norm": 2.1168501377105713,
+      "learning_rate": 5.2051033755072834e-05,
+      "loss": 2.065,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49569764090740664,
+      "grad_norm": 2.593621253967285,
+      "learning_rate": 5.199847153836241e-05,
+      "loss": 2.2862,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49602700811066736,
+      "grad_norm": 2.273329257965088,
+      "learning_rate": 5.194590710948419e-05,
+      "loss": 1.9901,
+      "step": 1506
+    },
+    {
+      "epoch": 0.4963563753139281,
+      "grad_norm": 2.755690336227417,
+      "learning_rate": 5.189334052662331e-05,
+      "loss": 2.1752,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49668574251718883,
+      "grad_norm": 2.4404118061065674,
+      "learning_rate": 5.1840771847967286e-05,
+      "loss": 2.1709,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4970151097204496,
+      "grad_norm": 2.30769681930542,
+      "learning_rate": 5.178820113170596e-05,
+      "loss": 2.2124,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4973444769237103,
+      "grad_norm": 2.8207881450653076,
+      "learning_rate": 5.1735628436031436e-05,
+      "loss": 2.2862,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4976738441269711,
+      "grad_norm": 2.278991460800171,
+      "learning_rate": 5.1683053819137975e-05,
+      "loss": 2.0548,
+      "step": 1511
+    },
+    {
+      "epoch": 0.4980032113302318,
+      "grad_norm": 2.4886574745178223,
+      "learning_rate": 5.1630477339221994e-05,
+      "loss": 2.3737,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49833257853349255,
+      "grad_norm": 2.911540985107422,
+      "learning_rate": 5.1577899054481996e-05,
+      "loss": 2.0253,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49866194573675326,
+      "grad_norm": 2.7352163791656494,
+      "learning_rate": 5.1525319023118434e-05,
+      "loss": 2.3057,
+      "step": 1514
+    },
+    {
+      "epoch": 0.498991312940014,
+      "grad_norm": 2.4180266857147217,
+      "learning_rate": 5.147273730333372e-05,
+      "loss": 2.3452,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49932068014327474,
+      "grad_norm": 2.5061450004577637,
+      "learning_rate": 5.14201539533321e-05,
+      "loss": 1.8195,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49965004734653545,
+      "grad_norm": 2.4668986797332764,
+      "learning_rate": 5.1367569031319715e-05,
+      "loss": 1.9367,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4999794145497962,
+      "grad_norm": 2.661878824234009,
+      "learning_rate": 5.1314982595504335e-05,
+      "loss": 1.9323,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4999794145497962,
+      "eval_loss": 2.212698221206665,
+      "eval_runtime": 764.8335,
+      "eval_samples_per_second": 3.343,
+      "eval_steps_per_second": 1.672,
+      "step": 1518
     }
   ],
   "logging_steps": 1,
@@ -5355,7 +10676,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.164145589886124e+18,
+  "total_flos": 2.328291179772248e+18,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null