End of training

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 3.0,
-    "eval_loss": 0.28307044506073,
-    "eval_runtime": 73.6303,
-    "eval_samples_per_second": 16.379,
-    "eval_steps_per_second": 2.051,
-    "total_flos": 5859923904498510.0,
-    "train_loss": 0.014361034947702911,
-    "train_runtime": 3221.5046,
-    "train_samples_per_second": 10.11,
-    "train_steps_per_second": 5.056
+    "eval_loss": 0.5312151312828064,
+    "eval_runtime": 75.4102,
+    "eval_samples_per_second": 16.006,
+    "eval_steps_per_second": 2.002,
+    "total_flos": 5859024508620360.0,
+    "train_loss": 0.093766236650771,
+    "train_runtime": 12421.8339,
+    "train_samples_per_second": 2.622,
+    "train_steps_per_second": 1.311
 }

eval_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.0,
-    "eval_loss": 0.28307044506073,
-    "eval_runtime": 73.6303,
-    "eval_samples_per_second": 16.379,
-    "eval_steps_per_second": 2.051
+    "eval_loss": 0.5312151312828064,
+    "eval_runtime": 75.4102,
+    "eval_samples_per_second": 16.006,
+    "eval_steps_per_second": 2.002
 }

runs/Aug22_08-32-24_winter-park/events.out.tfevents.1724349760.winter-park.2640878.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fad42edcc5823a77599fad269ce58894502083c767430f8f27b0308d832d6753
+size 311

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.0,
-    "total_flos": 5859923904498510.0,
-    "train_loss": 0.014361034947702911,
-    "train_runtime": 3221.5046,
-    "train_samples_per_second": 10.11,
-    "train_steps_per_second": 5.056
+    "total_flos": 5859024508620360.0,
+    "train_loss": 0.093766236650771,
+    "train_runtime": 12421.8339,
+    "train_samples_per_second": 2.622,
+    "train_steps_per_second": 1.311
 }

trainer_state.json

@@ -1,505 +1,505 @@
 {
-  "best_metric": 0.28307044506073,
+  "best_metric": 0.5312151312828064,
   "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e3l57-l/checkpoint-16000",
   "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 16287,
+  "global_step": 16284,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.09211495946941783,
-      "grad_norm": 3.071866512298584,
+      "grad_norm": 3.083087682723999,
       "learning_rate": 4.84647506755097e-07,
-      "loss": 0.3448,
+      "loss": 0.2708,
       "step": 500
     },
     {
       "epoch": 0.09211495946941783,
-      "eval_loss": 1.0838947296142578,
-      "eval_runtime": 76.5304,
-      "eval_samples_per_second": 15.772,
-      "eval_steps_per_second": 1.973,
+      "eval_loss": 1.0631108283996582,
+      "eval_runtime": 74.1964,
+      "eval_samples_per_second": 16.268,
+      "eval_steps_per_second": 2.035,
       "step": 500
     },
     {
       "epoch": 0.18422991893883567,
-      "grad_norm": 2.2681832313537598,
+      "grad_norm": 7.726487159729004,
       "learning_rate": 4.69295013510194e-07,
-      "loss": 0.2708,
+      "loss": 0.251,
       "step": 1000
     },
     {
       "epoch": 0.18422991893883567,
-      "eval_loss": 0.8948286175727844,
-      "eval_runtime": 76.7259,
-      "eval_samples_per_second": 15.731,
-      "eval_steps_per_second": 1.968,
+      "eval_loss": 0.942773699760437,
+      "eval_runtime": 74.6047,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.024,
       "step": 1000
     },
     {
       "epoch": 0.2763448784082535,
-      "grad_norm": 0.00790461990982294,
+      "grad_norm": 0.02718021161854267,
       "learning_rate": 4.5394252026529107e-07,
-      "loss": 0.177,
+      "loss": 0.235,
       "step": 1500
     },
     {
       "epoch": 0.2763448784082535,
-      "eval_loss": 0.8396429419517517,
-      "eval_runtime": 78.5505,
-      "eval_samples_per_second": 15.366,
-      "eval_steps_per_second": 1.922,
+      "eval_loss": 0.8716810345649719,
+      "eval_runtime": 75.03,
+      "eval_samples_per_second": 16.087,
+      "eval_steps_per_second": 2.013,
       "step": 1500
     },
     {
       "epoch": 0.36845983787767134,
-      "grad_norm": 0.29154184460639954,
+      "grad_norm": 0.0001215290030813776,
       "learning_rate": 4.385900270203881e-07,
-      "loss": 0.1831,
+      "loss": 0.1529,
       "step": 2000
     },
     {
       "epoch": 0.36845983787767134,
-      "eval_loss": 0.7720839977264404,
-      "eval_runtime": 78.1799,
-      "eval_samples_per_second": 15.439,
-      "eval_steps_per_second": 1.931,
+      "eval_loss": 0.8318113088607788,
+      "eval_runtime": 76.6917,
+      "eval_samples_per_second": 15.738,
+      "eval_steps_per_second": 1.969,
       "step": 2000
     },
     {
       "epoch": 0.46057479734708917,
-      "grad_norm": 186.6424102783203,
+      "grad_norm": 419.545166015625,
       "learning_rate": 4.232375337754851e-07,
-      "loss": 0.2038,
+      "loss": 0.1781,
       "step": 2500
     },
     {
       "epoch": 0.46057479734708917,
-      "eval_loss": 0.744644284248352,
-      "eval_runtime": 76.5359,
-      "eval_samples_per_second": 15.77,
-      "eval_steps_per_second": 1.973,
+      "eval_loss": 0.7549223303794861,
+      "eval_runtime": 75.2036,
+      "eval_samples_per_second": 16.05,
+      "eval_steps_per_second": 2.008,
       "step": 2500
     },
     {
       "epoch": 0.552689756816507,
-      "grad_norm": 0.009510820731520653,
+      "grad_norm": 368.7484436035156,
       "learning_rate": 4.0788504053058217e-07,
-      "loss": 0.1309,
+      "loss": 0.1681,
       "step": 3000
     },
     {
       "epoch": 0.552689756816507,
-      "eval_loss": 0.7234784960746765,
-      "eval_runtime": 76.6741,
-      "eval_samples_per_second": 15.742,
-      "eval_steps_per_second": 1.969,
+      "eval_loss": 0.7217888236045837,
+      "eval_runtime": 76.0606,
+      "eval_samples_per_second": 15.869,
+      "eval_steps_per_second": 1.985,
       "step": 3000
     },
     {
       "epoch": 0.6448047162859248,
-      "grad_norm": 542.5567626953125,
+      "grad_norm": 20.03949737548828,
       "learning_rate": 3.925325472856792e-07,
-      "loss": 0.1431,
+      "loss": 0.1064,
       "step": 3500
     },
     {
       "epoch": 0.6448047162859248,
-      "eval_loss": 0.6690872311592102,
-      "eval_runtime": 76.4873,
-      "eval_samples_per_second": 15.78,
-      "eval_steps_per_second": 1.974,
+      "eval_loss": 0.7048410177230835,
+      "eval_runtime": 76.0578,
+      "eval_samples_per_second": 15.87,
+      "eval_steps_per_second": 1.985,
       "step": 3500
     },
     {
       "epoch": 0.7369196757553427,
-      "grad_norm": 3.6845791339874268,
+      "grad_norm": 363.6124267578125,
       "learning_rate": 3.771800540407762e-07,
-      "loss": 0.1411,
+      "loss": 0.1357,
       "step": 4000
     },
     {
       "epoch": 0.7369196757553427,
-      "eval_loss": 0.643679141998291,
-      "eval_runtime": 76.4578,
-      "eval_samples_per_second": 15.786,
-      "eval_steps_per_second": 1.975,
+      "eval_loss": 0.6961681246757507,
+      "eval_runtime": 76.3574,
+      "eval_samples_per_second": 15.807,
+      "eval_steps_per_second": 1.978,
       "step": 4000
     },
     {
       "epoch": 0.8290346352247605,
-      "grad_norm": 0.005153482314199209,
+      "grad_norm": 494.03570556640625,
       "learning_rate": 3.618275607958732e-07,
-      "loss": 0.0849,
+      "loss": 0.1098,
       "step": 4500
     },
     {
       "epoch": 0.8290346352247605,
-      "eval_loss": 0.48050031065940857,
-      "eval_runtime": 75.8694,
-      "eval_samples_per_second": 15.909,
-      "eval_steps_per_second": 1.99,
+      "eval_loss": 0.6778352856636047,
+      "eval_runtime": 76.4844,
+      "eval_samples_per_second": 15.781,
+      "eval_steps_per_second": 1.974,
       "step": 4500
     },
     {
       "epoch": 0.9211495946941783,
-      "grad_norm": 2.631653070449829,
+      "grad_norm": 4.2969353671651334e-05,
       "learning_rate": 3.464750675509703e-07,
-      "loss": 0.1026,
+      "loss": 0.1142,
       "step": 5000
     },
     {
       "epoch": 0.9211495946941783,
-      "eval_loss": 0.4800306260585785,
-      "eval_runtime": 75.365,
-      "eval_samples_per_second": 16.015,
-      "eval_steps_per_second": 2.004,
+      "eval_loss": 0.6656659245491028,
+      "eval_runtime": 76.6475,
+      "eval_samples_per_second": 15.747,
+      "eval_steps_per_second": 1.97,
       "step": 5000
     },
     {
       "epoch": 1.013264554163596,
-      "grad_norm": 1.8651682694326155e-05,
+      "grad_norm": 0.0010161151876673102,
       "learning_rate": 3.311225743060673e-07,
-      "loss": 0.1201,
+      "loss": 0.1113,
       "step": 5500
     },
     {
       "epoch": 1.013264554163596,
-      "eval_loss": 0.4810214042663574,
-      "eval_runtime": 74.984,
-      "eval_samples_per_second": 16.097,
-      "eval_steps_per_second": 2.014,
+      "eval_loss": 0.6430702209472656,
+      "eval_runtime": 76.4915,
+      "eval_samples_per_second": 15.78,
+      "eval_steps_per_second": 1.974,
       "step": 5500
     },
     {
       "epoch": 1.105379513633014,
-      "grad_norm": 0.5627817511558533,
+      "grad_norm": 0.05515381693840027,
       "learning_rate": 3.157700810611643e-07,
-      "loss": 0.0757,
+      "loss": 0.0572,
       "step": 6000
     },
     {
       "epoch": 1.105379513633014,
-      "eval_loss": 0.46917983889579773,
-      "eval_runtime": 74.883,
-      "eval_samples_per_second": 16.118,
-      "eval_steps_per_second": 2.016,
+      "eval_loss": 0.636746883392334,
+      "eval_runtime": 76.0169,
+      "eval_samples_per_second": 15.878,
+      "eval_steps_per_second": 1.986,
       "step": 6000
     },
     {
       "epoch": 1.1974944731024317,
-      "grad_norm": 0.0001344973425148055,
+      "grad_norm": 0.0038496837951242924,
       "learning_rate": 3.0041758781626137e-07,
-      "loss": 0.0696,
+      "loss": 0.0746,
       "step": 6500
     },
     {
       "epoch": 1.1974944731024317,
-      "eval_loss": 0.47636929154396057,
-      "eval_runtime": 76.4445,
-      "eval_samples_per_second": 15.789,
-      "eval_steps_per_second": 1.975,
+      "eval_loss": 0.6261336207389832,
+      "eval_runtime": 76.1916,
+      "eval_samples_per_second": 15.842,
+      "eval_steps_per_second": 1.982,
       "step": 6500
     },
     {
       "epoch": 1.2896094325718497,
-      "grad_norm": 0.007305172737687826,
+      "grad_norm": 2.594869386030041e-07,
       "learning_rate": 2.8506509457135833e-07,
-      "loss": 0.0911,
+      "loss": 0.0494,
       "step": 7000
     },
     {
       "epoch": 1.2896094325718497,
-      "eval_loss": 0.4601423740386963,
-      "eval_runtime": 76.451,
-      "eval_samples_per_second": 15.788,
-      "eval_steps_per_second": 1.975,
+      "eval_loss": 0.6245251893997192,
+      "eval_runtime": 75.9712,
+      "eval_samples_per_second": 15.888,
+      "eval_steps_per_second": 1.988,
       "step": 7000
     },
     {
       "epoch": 1.3817243920412676,
-      "grad_norm": 9.710428905407298e-09,
+      "grad_norm": 25.23366928100586,
       "learning_rate": 2.697126013264554e-07,
-      "loss": 0.0806,
+      "loss": 0.0788,
       "step": 7500
     },
     {
       "epoch": 1.3817243920412676,
-      "eval_loss": 0.45901384949684143,
-      "eval_runtime": 74.7174,
-      "eval_samples_per_second": 16.154,
-      "eval_steps_per_second": 2.021,
+      "eval_loss": 0.611993670463562,
+      "eval_runtime": 75.1885,
+      "eval_samples_per_second": 16.053,
+      "eval_steps_per_second": 2.008,
       "step": 7500
     },
     {
       "epoch": 1.4738393515106853,
-      "grad_norm": 0.008983594365417957,
+      "grad_norm": 0.0002172550739487633,
       "learning_rate": 2.5436010808155247e-07,
-      "loss": 0.088,
+      "loss": 0.0808,
       "step": 8000
     },
     {
       "epoch": 1.4738393515106853,
-      "eval_loss": 0.46537238359451294,
-      "eval_runtime": 74.5146,
-      "eval_samples_per_second": 16.198,
-      "eval_steps_per_second": 2.026,
+      "eval_loss": 0.6011174917221069,
+      "eval_runtime": 76.2572,
+      "eval_samples_per_second": 15.828,
+      "eval_steps_per_second": 1.98,
       "step": 8000
     },
     {
       "epoch": 1.565954310980103,
-      "grad_norm": 1.3851042240276001e-05,
+      "grad_norm": 0.032347481697797775,
       "learning_rate": 2.390076148366495e-07,
-      "loss": 0.0878,
+      "loss": 0.0536,
       "step": 8500
     },
     {
       "epoch": 1.565954310980103,
-      "eval_loss": 0.4768938720226288,
-      "eval_runtime": 74.5757,
-      "eval_samples_per_second": 16.185,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 0.5893377661705017,
+      "eval_runtime": 75.429,
+      "eval_samples_per_second": 16.002,
+      "eval_steps_per_second": 2.002,
       "step": 8500
     },
     {
       "epoch": 1.658069270449521,
-      "grad_norm": 4.7310441004810855e-06,
+      "grad_norm": 2.7040863415095373e-07,
       "learning_rate": 2.236551215917465e-07,
-      "loss": 0.0369,
+      "loss": 0.0869,
       "step": 9000
     },
     {
       "epoch": 1.658069270449521,
-      "eval_loss": 0.4683688282966614,
-      "eval_runtime": 74.5774,
-      "eval_samples_per_second": 16.185,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 0.59163498878479,
+      "eval_runtime": 75.4027,
+      "eval_samples_per_second": 16.007,
+      "eval_steps_per_second": 2.003,
       "step": 9000
     },
     {
       "epoch": 1.750184229918939,
-      "grad_norm": 0.005283420439809561,
+      "grad_norm": 3.4501523166197146e-11,
       "learning_rate": 2.083026283468435e-07,
-      "loss": 0.1034,
+      "loss": 0.0752,
       "step": 9500
     },
     {
       "epoch": 1.750184229918939,
-      "eval_loss": 0.4715610444545746,
-      "eval_runtime": 74.6527,
-      "eval_samples_per_second": 16.168,
-      "eval_steps_per_second": 2.023,
+      "eval_loss": 0.5707471370697021,
+      "eval_runtime": 76.0154,
+      "eval_samples_per_second": 15.878,
+      "eval_steps_per_second": 1.986,
       "step": 9500
     },
     {
       "epoch": 1.8422991893883567,
-      "grad_norm": 1.703993320465088,
+      "grad_norm": 0.00039183301851153374,
       "learning_rate": 1.9295013510194055e-07,
-      "loss": 0.0852,
+      "loss": 0.0577,
       "step": 10000
     },
     {
       "epoch": 1.8422991893883567,
-      "eval_loss": 0.47203630208969116,
-      "eval_runtime": 74.6773,
-      "eval_samples_per_second": 16.163,
-      "eval_steps_per_second": 2.022,
+      "eval_loss": 0.5678250789642334,
+      "eval_runtime": 75.8084,
+      "eval_samples_per_second": 15.922,
+      "eval_steps_per_second": 1.992,
       "step": 10000
     },
     {
       "epoch": 1.9344141488577744,
-      "grad_norm": 8.647094205116446e-07,
+      "grad_norm": 1.0438248487787405e-08,
       "learning_rate": 1.7759764185703757e-07,
-      "loss": 0.0493,
+      "loss": 0.0891,
       "step": 10500
     },
     {
       "epoch": 1.9344141488577744,
-      "eval_loss": 0.47136709094047546,
-      "eval_runtime": 74.7402,
-      "eval_samples_per_second": 16.149,
-      "eval_steps_per_second": 2.02,
+      "eval_loss": 0.563089907169342,
+      "eval_runtime": 75.3868,
+      "eval_samples_per_second": 16.011,
+      "eval_steps_per_second": 2.003,
       "step": 10500
     },
     {
       "epoch": 2.026529108327192,
-      "grad_norm": 0.0002839279477484524,
+      "grad_norm": 1.162303306045942e-06,
       "learning_rate": 1.6224514861213458e-07,
-      "loss": 0.0603,
+      "loss": 0.0559,
       "step": 11000
     },
     {
       "epoch": 2.026529108327192,
-      "eval_loss": 0.4660906493663788,
-      "eval_runtime": 74.8543,
-      "eval_samples_per_second": 16.125,
-      "eval_steps_per_second": 2.017,
+      "eval_loss": 0.5547806620597839,
+      "eval_runtime": 76.4427,
+      "eval_samples_per_second": 15.79,
+      "eval_steps_per_second": 1.975,
       "step": 11000
     },
     {
       "epoch": 2.1186440677966103,
-      "grad_norm": 0.17852088809013367,
+      "grad_norm": 123.83905029296875,
       "learning_rate": 1.4689265536723165e-07,
-      "loss": 0.0547,
+      "loss": 0.0385,
       "step": 11500
     },
     {
       "epoch": 2.1186440677966103,
-      "eval_loss": 0.4668720066547394,
-      "eval_runtime": 74.8609,
-      "eval_samples_per_second": 16.123,
-      "eval_steps_per_second": 2.017,
+      "eval_loss": 0.5536245703697205,
+      "eval_runtime": 75.6886,
+      "eval_samples_per_second": 15.947,
+      "eval_steps_per_second": 1.995,
       "step": 11500
     },
     {
       "epoch": 2.210759027266028,
-      "grad_norm": 4.607869868777925e-06,
+      "grad_norm": 1.234429121017456,
       "learning_rate": 1.3154016212232866e-07,
-      "loss": 0.0793,
+      "loss": 0.0185,
       "step": 12000
     },
     {
       "epoch": 2.210759027266028,
-      "eval_loss": 0.4664075970649719,
-      "eval_runtime": 74.7262,
-      "eval_samples_per_second": 16.152,
-      "eval_steps_per_second": 2.021,
+      "eval_loss": 0.5518601536750793,
+      "eval_runtime": 75.2276,
+      "eval_samples_per_second": 16.045,
+      "eval_steps_per_second": 2.007,
       "step": 12000
     },
     {
-      "epoch": 2.302449806594216,
-      "grad_norm": 0.00044986297143623233,
-      "learning_rate": 1.1625836556763061e-07,
-      "loss": 0.0415,
+      "epoch": 2.3028739867354457,
+      "grad_norm": 3.730309009552002,
+      "learning_rate": 1.1618766887742569e-07,
+      "loss": 0.0642,
       "step": 12500
     },
     {
-      "epoch": 2.302449806594216,
-      "eval_loss": 0.2888467311859131,
-      "eval_runtime": 78.2614,
-      "eval_samples_per_second": 15.41,
-      "eval_steps_per_second": 1.929,
+      "epoch": 2.3028739867354457,
+      "eval_loss": 0.5504964590072632,
+      "eval_runtime": 76.2015,
+      "eval_samples_per_second": 15.84,
+      "eval_steps_per_second": 1.982,
       "step": 12500
     },
     {
-      "epoch": 2.394547798857985,
-      "grad_norm": 4.1875861001017256e-08,
-      "learning_rate": 1.0090870019033584e-07,
-      "loss": 0.0565,
+      "epoch": 2.3949889462048635,
+      "grad_norm": 0.0008666729554533958,
+      "learning_rate": 1.0083517563252273e-07,
+      "loss": 0.0456,
       "step": 13000
     },
     {
-      "epoch": 2.394547798857985,
-      "eval_loss": 0.29098838567733765,
-      "eval_runtime": 78.3993,
-      "eval_samples_per_second": 15.383,
-      "eval_steps_per_second": 1.926,
+      "epoch": 2.3949889462048635,
+      "eval_loss": 0.5443520545959473,
+      "eval_runtime": 75.4909,
+      "eval_samples_per_second": 15.989,
+      "eval_steps_per_second": 2.0,
       "step": 13000
     },
     {
-      "epoch": 2.4866457911217537,
-      "grad_norm": 1.556837378302589e-05,
-      "learning_rate": 8.555903481304106e-08,
-      "loss": 0.0629,
+      "epoch": 2.4871039056742816,
+      "grad_norm": 0.000371042697224766,
+      "learning_rate": 8.548268238761974e-08,
+      "loss": 0.0476,
       "step": 13500
     },
     {
-      "epoch": 2.4866457911217537,
-      "eval_loss": 0.2888970673084259,
-      "eval_runtime": 76.7751,
-      "eval_samples_per_second": 15.708,
-      "eval_steps_per_second": 1.967,
+      "epoch": 2.4871039056742816,
+      "eval_loss": 0.5395111441612244,
+      "eval_runtime": 75.2501,
+      "eval_samples_per_second": 16.04,
+      "eval_steps_per_second": 2.007,
       "step": 13500
     },
     {
-      "epoch": 2.578743783385522,
-      "grad_norm": 0.12299621850252151,
-      "learning_rate": 7.020936943574628e-08,
-      "loss": 0.0584,
+      "epoch": 2.5792188651436994,
+      "grad_norm": 0.01261068508028984,
+      "learning_rate": 7.013018914271677e-08,
+      "loss": 0.027,
       "step": 14000
     },
     {
-      "epoch": 2.578743783385522,
-      "eval_loss": 0.28743651509284973,
-      "eval_runtime": 76.8684,
-      "eval_samples_per_second": 15.689,
-      "eval_steps_per_second": 1.964,
+      "epoch": 2.5792188651436994,
+      "eval_loss": 0.5361316204071045,
+      "eval_runtime": 76.0447,
+      "eval_samples_per_second": 15.872,
+      "eval_steps_per_second": 1.986,
       "step": 14000
     },
     {
-      "epoch": 2.670841775649291,
-      "grad_norm": 0.25907593965530396,
-      "learning_rate": 5.485970405845152e-08,
-      "loss": 0.0582,
+      "epoch": 2.671333824613117,
+      "grad_norm": 4.373817752423292e-09,
+      "learning_rate": 5.47776958978138e-08,
+      "loss": 0.042,
       "step": 14500
     },
     {
-      "epoch": 2.670841775649291,
-      "eval_loss": 0.2863319218158722,
-      "eval_runtime": 77.0162,
-      "eval_samples_per_second": 15.659,
-      "eval_steps_per_second": 1.961,
+      "epoch": 2.671333824613117,
+      "eval_loss": 0.5355635285377502,
+      "eval_runtime": 75.7231,
+      "eval_samples_per_second": 15.94,
+      "eval_steps_per_second": 1.994,
       "step": 14500
     },
     {
-      "epoch": 2.7629397679130596,
-      "grad_norm": 1.9590417650761083e-05,
-      "learning_rate": 3.951003868115675e-08,
-      "loss": 0.052,
+      "epoch": 2.7634487840825352,
+      "grad_norm": 8.442443686362822e-07,
+      "learning_rate": 3.942520265291083e-08,
+      "loss": 0.0469,
       "step": 15000
     },
     {
-      "epoch": 2.7629397679130596,
-      "eval_loss": 0.2845611274242401,
-      "eval_runtime": 76.9043,
-      "eval_samples_per_second": 15.682,
-      "eval_steps_per_second": 1.963,
+      "epoch": 2.7634487840825352,
+      "eval_loss": 0.5343210697174072,
+      "eval_runtime": 75.4526,
+      "eval_samples_per_second": 15.997,
+      "eval_steps_per_second": 2.001,
       "step": 15000
     },
     {
-      "epoch": 2.8550377601768284,
-      "grad_norm": 0.2627514600753784,
-      "learning_rate": 2.4160373303861975e-08,
-      "loss": 0.0402,
+      "epoch": 2.855563743551953,
+      "grad_norm": 4.8845563010990745e-11,
+      "learning_rate": 2.407270940800786e-08,
+      "loss": 0.0438,
       "step": 15500
     },
     {
-      "epoch": 2.8550377601768284,
-      "eval_loss": 0.28350818157196045,
-      "eval_runtime": 76.8887,
-      "eval_samples_per_second": 15.685,
-      "eval_steps_per_second": 1.964,
+      "epoch": 2.855563743551953,
+      "eval_loss": 0.5331963896751404,
+      "eval_runtime": 76.1088,
+      "eval_samples_per_second": 15.859,
+      "eval_steps_per_second": 1.984,
       "step": 15500
     },
     {
-      "epoch": 2.947135752440597,
-      "grad_norm": 0.00015808168973308057,
-      "learning_rate": 8.8107079265672e-09,
-      "loss": 0.0518,
+      "epoch": 2.9476787030213707,
+      "grad_norm": 4.494071006774902,
+      "learning_rate": 8.720216163104888e-09,
+      "loss": 0.0481,
       "step": 16000
     },
     {
-      "epoch": 2.947135752440597,
-      "eval_loss": 0.28307044506073,
-      "eval_runtime": 76.733,
-      "eval_samples_per_second": 15.717,
-      "eval_steps_per_second": 1.968,
+      "epoch": 2.9476787030213707,
+      "eval_loss": 0.5312151312828064,
+      "eval_runtime": 78.7608,
+      "eval_samples_per_second": 15.325,
+      "eval_steps_per_second": 1.917,
       "step": 16000
     },
     {
       "epoch": 3.0,
-      "step": 16287,
-      "total_flos": 5859923904498510.0,
-      "train_loss": 0.014361034947702911,
-      "train_runtime": 3221.5046,
-      "train_samples_per_second": 10.11,
-      "train_steps_per_second": 5.056
+      "step": 16284,
+      "total_flos": 5859024508620360.0,
+      "train_loss": 0.093766236650771,
+      "train_runtime": 12421.8339,
+      "train_samples_per_second": 2.622,
+      "train_steps_per_second": 1.311
     }
   ],
   "logging_steps": 500,
-  "max_steps": 16287,
+  "max_steps": 16284,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
@@ -515,7 +515,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5859923904498510.0,
+  "total_flos": 5859024508620360.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null