End of training

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.4866723823850676,
-    "eval_loss": 2.553661823272705,
-    "eval_runtime": 4.2435,
+    "eval_accuracy": 0.5001937383077616,
+    "eval_loss": 2.473005771636963,
+    "eval_runtime": 3.3872,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1058.333,
-    "eval_steps_per_second": 2.121,
-    "perplexity": 12.854087103158811,
-    "train_loss": 2.884315123844668,
-    "train_runtime": 615.9463,
+    "eval_samples_per_second": 1325.86,
+    "eval_steps_per_second": 10.628,
+    "perplexity": 11.858035886533791,
+    "train_loss": 2.5663982599810824,
+    "train_runtime": 749.553,
     "train_samples": 46845,
-    "train_samples_per_second": 760.537,
-    "train_steps_per_second": 11.884
+    "train_samples_per_second": 624.972,
+    "train_steps_per_second": 39.063
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.4866723823850676,
-    "eval_loss": 2.553661823272705,
-    "eval_runtime": 4.2435,
+    "eval_accuracy": 0.5001937383077616,
+    "eval_loss": 2.473005771636963,
+    "eval_runtime": 3.3872,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1058.333,
-    "eval_steps_per_second": 2.121,
-    "perplexity": 12.854087103158811
+    "eval_samples_per_second": 1325.86,
+    "eval_steps_per_second": 10.628,
+    "perplexity": 11.858035886533791
 }

runs/Mar25_20-13-06_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1711416361.phyl-ling-p01.la.utexas.edu.1051288.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9535952f619878bd9c695fe1d39444e4725bf849a9ed271cb2e2e888d9690d4a
+size 417

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
-    "train_loss": 2.884315123844668,
-    "train_runtime": 615.9463,
+    "train_loss": 2.5663982599810824,
+    "train_runtime": 749.553,
     "train_samples": 46845,
-    "train_samples_per_second": 760.537,
-    "train_steps_per_second": 11.884
+    "train_samples_per_second": 624.972,
+    "train_steps_per_second": 39.063
 }

trainer_state.json

@@ -3,200 +3,524 @@
   "best_model_checkpoint": null,
   "epoch": 10.0,
   "eval_steps": 500,
-  "global_step": 7320,
+  "global_step": 29280,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.68,
+      "epoch": 0.17,
+      "grad_norm": 1.12882661819458,
       "learning_rate": 6.25e-05,
-      "loss": 6.0764,
+      "loss": 6.1655,
       "step": 500
     },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9656639099121094,
+      "learning_rate": 0.000125,
+      "loss": 3.6235,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9288356304168701,
+      "learning_rate": 0.0001875,
+      "loss": 3.3342,
+      "step": 1500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.94553542137146,
+      "learning_rate": 0.00025,
+      "loss": 3.1718,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9111424088478088,
+      "learning_rate": 0.0003125,
+      "loss": 3.0482,
+      "step": 2500
+    },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.39467386216001554,
-      "eval_loss": 3.536583662033081,
-      "eval_runtime": 3.7627,
-      "eval_samples_per_second": 1193.571,
-      "eval_steps_per_second": 2.392,
-      "step": 732
+      "eval_accuracy": 0.4382991705195167,
+      "eval_loss": 3.0123326778411865,
+      "eval_runtime": 3.3766,
+      "eval_samples_per_second": 1330.021,
+      "eval_steps_per_second": 10.661,
+      "step": 2928
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8901249170303345,
+      "learning_rate": 0.000375,
+      "loss": 2.9639,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9494138956069946,
+      "learning_rate": 0.00043750000000000006,
+      "loss": 2.8729,
+      "step": 3500
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000125,
-      "loss": 3.4854,
-      "step": 1000
+      "grad_norm": 0.9213692545890808,
+      "learning_rate": 0.0005,
+      "loss": 2.8189,
+      "step": 4000
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8084624409675598,
+      "learning_rate": 0.0005625000000000001,
+      "loss": 2.7781,
+      "step": 4500
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8010088205337524,
+      "learning_rate": 0.000625,
+      "loss": 2.7388,
+      "step": 5000
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.8517431616783142,
+      "learning_rate": 0.0006875,
+      "loss": 2.72,
+      "step": 5500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.4325168271801696,
-      "eval_loss": 3.1204140186309814,
-      "eval_runtime": 3.858,
-      "eval_samples_per_second": 1164.071,
-      "eval_steps_per_second": 2.333,
-      "step": 1464
+      "eval_accuracy": 0.46022754169756835,
+      "eval_loss": 2.782708168029785,
+      "eval_runtime": 3.4325,
+      "eval_samples_per_second": 1308.36,
+      "eval_steps_per_second": 10.488,
+      "step": 5856
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001875,
-      "loss": 3.1417,
-      "step": 1500
+      "grad_norm": 0.8044182658195496,
+      "learning_rate": 0.00075,
+      "loss": 2.6633,
+      "step": 6000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.7476066946983337,
+      "learning_rate": 0.0008125,
+      "loss": 2.6251,
+      "step": 6500
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.7374809384346008,
+      "learning_rate": 0.0008750000000000001,
+      "loss": 2.6048,
+      "step": 7000
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.728888750076294,
+      "learning_rate": 0.0009375,
+      "loss": 2.5947,
+      "step": 7500
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00025,
-      "loss": 2.9429,
-      "step": 2000
+      "grad_norm": 0.6759493350982666,
+      "learning_rate": 0.001,
+      "loss": 2.6049,
+      "step": 8000
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.6459181904792786,
+      "learning_rate": 0.0010625,
+      "loss": 2.5878,
+      "step": 8500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.45065283673208184,
-      "eval_loss": 2.9128530025482178,
-      "eval_runtime": 4.0585,
-      "eval_samples_per_second": 1106.568,
-      "eval_steps_per_second": 2.218,
-      "step": 2196
+      "eval_accuracy": 0.469314832639908,
+      "eval_loss": 2.6918864250183105,
+      "eval_runtime": 3.4397,
+      "eval_samples_per_second": 1305.64,
+      "eval_steps_per_second": 10.466,
+      "step": 8784
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.601531445980072,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 2.5422,
+      "step": 9000
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.586744487285614,
+      "learning_rate": 0.0011875,
+      "loss": 2.5161,
+      "step": 9500
     },
     {
       "epoch": 3.42,
-      "learning_rate": 0.0003125,
-      "loss": 2.8056,
-      "step": 2500
+      "grad_norm": 0.5889135599136353,
+      "learning_rate": 0.00125,
+      "loss": 2.518,
+      "step": 10000
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.5179043412208557,
+      "learning_rate": 0.0013125,
+      "loss": 2.5173,
+      "step": 10500
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.48668384552001953,
+      "learning_rate": 0.001375,
+      "loss": 2.5081,
+      "step": 11000
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.46291133761405945,
+      "learning_rate": 0.0014375000000000002,
+      "loss": 2.5096,
+      "step": 11500
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.46146886949752525,
-      "eval_loss": 2.78314208984375,
-      "eval_runtime": 4.0564,
-      "eval_samples_per_second": 1107.139,
-      "eval_steps_per_second": 2.219,
-      "step": 2928
+      "eval_accuracy": 0.4774921671865165,
+      "eval_loss": 2.63217830657959,
+      "eval_runtime": 3.4241,
+      "eval_samples_per_second": 1311.585,
+      "eval_steps_per_second": 10.514,
+      "step": 11712
     },
     {
       "epoch": 4.1,
-      "learning_rate": 0.000375,
-      "loss": 2.7014,
-      "step": 3000
+      "grad_norm": 0.47367745637893677,
+      "learning_rate": 0.0015,
+      "loss": 2.4578,
+      "step": 12000
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.5070286393165588,
+      "learning_rate": 0.0015625,
+      "loss": 2.4408,
+      "step": 12500
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.43498894572257996,
+      "learning_rate": 0.001625,
+      "loss": 2.4561,
+      "step": 13000
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.4298257529735565,
+      "learning_rate": 0.0016875,
+      "loss": 2.463,
+      "step": 13500
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00043750000000000006,
-      "loss": 2.6127,
-      "step": 3500
+      "grad_norm": 0.39968445897102356,
+      "learning_rate": 0.0017500000000000003,
+      "loss": 2.4608,
+      "step": 14000
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.41952285170555115,
+      "learning_rate": 0.0018124999999999999,
+      "loss": 2.4696,
+      "step": 14500
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.4704597296079473,
-      "eval_loss": 2.6986284255981445,
-      "eval_runtime": 4.0805,
-      "eval_samples_per_second": 1100.59,
-      "eval_steps_per_second": 2.206,
-      "step": 3660
+      "eval_accuracy": 0.48033424679630476,
+      "eval_loss": 2.6079657077789307,
+      "eval_runtime": 3.4168,
+      "eval_samples_per_second": 1314.37,
+      "eval_steps_per_second": 10.536,
+      "step": 14640
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.3715464770793915,
+      "learning_rate": 0.001875,
+      "loss": 2.3974,
+      "step": 15000
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.4010851979255676,
+      "learning_rate": 0.0019375000000000002,
+      "loss": 2.3902,
+      "step": 15500
     },
     {
       "epoch": 5.46,
-      "learning_rate": 0.0005,
-      "loss": 2.543,
-      "step": 4000
+      "grad_norm": 0.4189828634262085,
+      "learning_rate": 0.002,
+      "loss": 2.4066,
+      "step": 16000
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.3794609308242798,
+      "learning_rate": 0.0020625,
+      "loss": 2.4223,
+      "step": 16500
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.3260567784309387,
+      "learning_rate": 0.002125,
+      "loss": 2.4273,
+      "step": 17000
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.3328179717063904,
+      "learning_rate": 0.0021874999999999998,
+      "loss": 2.4356,
+      "step": 17500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.47775691365232653,
-      "eval_loss": 2.63608980178833,
-      "eval_runtime": 3.9693,
-      "eval_samples_per_second": 1131.446,
-      "eval_steps_per_second": 2.267,
-      "step": 4392
+      "eval_accuracy": 0.4817894757143333,
+      "eval_loss": 2.5834248065948486,
+      "eval_runtime": 3.4191,
+      "eval_samples_per_second": 1313.516,
+      "eval_steps_per_second": 10.529,
+      "step": 17568
     },
     {
       "epoch": 6.15,
-      "learning_rate": 0.0005625000000000001,
-      "loss": 2.4933,
-      "step": 4500
+      "grad_norm": 0.3008418083190918,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 2.3574,
+      "step": 18000
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.3263770639896393,
+      "learning_rate": 0.0023125000000000003,
+      "loss": 2.365,
+      "step": 18500
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.3040842115879059,
+      "learning_rate": 0.002375,
+      "loss": 2.3788,
+      "step": 19000
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.3506632149219513,
+      "learning_rate": 0.0024375,
+      "loss": 2.3833,
+      "step": 19500
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.000625,
-      "loss": 2.4495,
-      "step": 5000
+      "grad_norm": 0.34952595829963684,
+      "learning_rate": 0.0025,
+      "loss": 2.386,
+      "step": 20000
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.48105835467961294,
-      "eval_loss": 2.6053810119628906,
-      "eval_runtime": 3.956,
-      "eval_samples_per_second": 1135.241,
-      "eval_steps_per_second": 2.275,
-      "step": 5124
+      "eval_accuracy": 0.4853819625252254,
+      "eval_loss": 2.5648064613342285,
+      "eval_runtime": 3.4161,
+      "eval_samples_per_second": 1314.649,
+      "eval_steps_per_second": 10.538,
+      "step": 20496
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.353624165058136,
+      "learning_rate": 0.0025625,
+      "loss": 2.3941,
+      "step": 20500
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.36951765418052673,
+      "learning_rate": 0.002625,
+      "loss": 2.3144,
+      "step": 21000
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.34771400690078735,
+      "learning_rate": 0.0026875000000000002,
+      "loss": 2.3379,
+      "step": 21500
     },
     {
       "epoch": 7.51,
-      "learning_rate": 0.0006875,
-      "loss": 2.4047,
-      "step": 5500
+      "grad_norm": 0.32248654961586,
+      "learning_rate": 0.00275,
+      "loss": 2.3457,
+      "step": 22000
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.3408921957015991,
+      "learning_rate": 0.0028125,
+      "loss": 2.3511,
+      "step": 22500
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.34288591146469116,
+      "learning_rate": 0.0028750000000000004,
+      "loss": 2.3579,
+      "step": 23000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.4840179045755553,
-      "eval_loss": 2.5785861015319824,
-      "eval_runtime": 3.9895,
-      "eval_samples_per_second": 1125.713,
-      "eval_steps_per_second": 2.256,
-      "step": 5856
+      "eval_accuracy": 0.48610782369638667,
+      "eval_loss": 2.5603365898132324,
+      "eval_runtime": 3.3768,
+      "eval_samples_per_second": 1329.97,
+      "eval_steps_per_second": 10.661,
+      "step": 23424
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.3494857847690582,
+      "learning_rate": 0.0029375,
+      "loss": 2.3583,
+      "step": 23500
     },
     {
       "epoch": 8.2,
-      "learning_rate": 0.00075,
-      "loss": 2.383,
-      "step": 6000
+      "grad_norm": 0.2930866777896881,
+      "learning_rate": 0.003,
+      "loss": 2.2877,
+      "step": 24000
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.3050256669521332,
+      "learning_rate": 0.002715909090909091,
+      "loss": 2.3009,
+      "step": 24500
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.30324727296829224,
+      "learning_rate": 0.0024318181818181817,
+      "loss": 2.3073,
+      "step": 25000
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.36902040243148804,
+      "learning_rate": 0.002147727272727273,
+      "loss": 2.296,
+      "step": 25500
     },
     {
       "epoch": 8.88,
-      "learning_rate": 0.0008125,
-      "loss": 2.3641,
-      "step": 6500
+      "grad_norm": 0.2886076867580414,
+      "learning_rate": 0.0018636363636363638,
+      "loss": 2.2817,
+      "step": 26000
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.4853205974503688,
-      "eval_loss": 2.565032482147217,
-      "eval_runtime": 3.7862,
-      "eval_samples_per_second": 1186.157,
-      "eval_steps_per_second": 2.377,
-      "step": 6588
+      "eval_accuracy": 0.49435178318141093,
+      "eval_loss": 2.5002617835998535,
+      "eval_runtime": 3.3753,
+      "eval_samples_per_second": 1330.555,
+      "eval_steps_per_second": 10.666,
+      "step": 26352
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.3240475356578827,
+      "learning_rate": 0.0015795454545454546,
+      "loss": 2.2527,
+      "step": 26500
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.3238734304904938,
+      "learning_rate": 0.0012954545454545456,
+      "loss": 2.1689,
+      "step": 27000
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.3348115086555481,
+      "learning_rate": 0.0010113636363636364,
+      "loss": 2.1655,
+      "step": 27500
     },
     {
       "epoch": 9.56,
-      "learning_rate": 0.0008750000000000001,
-      "loss": 2.3267,
-      "step": 7000
+      "grad_norm": 0.3114456534385681,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 2.1548,
+      "step": 28000
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 0.34318503737449646,
+      "learning_rate": 0.0004431818181818182,
+      "loss": 2.1447,
+      "step": 28500
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 0.2760942578315735,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 2.1431,
+      "step": 29000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.4866723823850676,
-      "eval_loss": 2.553661823272705,
-      "eval_runtime": 3.9949,
-      "eval_samples_per_second": 1124.194,
-      "eval_steps_per_second": 2.253,
-      "step": 7320
+      "eval_accuracy": 0.5001937383077616,
+      "eval_loss": 2.473005771636963,
+      "eval_runtime": 3.375,
+      "eval_samples_per_second": 1330.654,
+      "eval_steps_per_second": 10.667,
+      "step": 29280
     },
     {
       "epoch": 10.0,
-      "step": 7320,
+      "step": 29280,
       "total_flos": 2273237316403200.0,
-      "train_loss": 2.884315123844668,
-      "train_runtime": 615.9463,
-      "train_samples_per_second": 760.537,
-      "train_steps_per_second": 11.884
+      "train_loss": 2.5663982599810824,
+      "train_runtime": 749.553,
+      "train_samples_per_second": 624.972,
+      "train_steps_per_second": 39.063
     }
   ],
   "logging_steps": 500,
-  "max_steps": 7320,
+  "max_steps": 29280,
+  "num_input_tokens_seen": 0,
   "num_train_epochs": 10,
   "save_steps": 2000,
   "total_flos": 2273237316403200.0,
+  "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null
 }