End of training

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.4861165901356519,
-    "eval_loss": 2.557232141494751,
-    "eval_runtime": 4.0987,
+    "eval_accuracy": 0.4999763306139839,
+    "eval_loss": 2.471221446990967,
+    "eval_runtime": 3.3783,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1095.703,
-    "eval_steps_per_second": 2.196,
-    "perplexity": 12.900062308790385,
-    "train_loss": 2.8881424867390284,
-    "train_runtime": 618.3653,
+    "eval_samples_per_second": 1329.36,
+    "eval_steps_per_second": 10.656,
+    "perplexity": 11.836896166518805,
+    "train_loss": 2.567346385267914,
+    "train_runtime": 749.2453,
     "train_samples": 46845,
-    "train_samples_per_second": 757.562,
-    "train_steps_per_second": 11.838
+    "train_samples_per_second": 625.229,
+    "train_steps_per_second": 39.079
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.4861165901356519,
-    "eval_loss": 2.557232141494751,
-    "eval_runtime": 4.0987,
+    "eval_accuracy": 0.4999763306139839,
+    "eval_loss": 2.471221446990967,
+    "eval_runtime": 3.3783,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1095.703,
-    "eval_steps_per_second": 2.196,
-    "perplexity": 12.900062308790385
+    "eval_samples_per_second": 1329.36,
+    "eval_steps_per_second": 10.656,
+    "perplexity": 11.836896166518805
 }

runs/Mar25_21-32-38_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1711421133.phyl-ling-p01.la.utexas.edu.1056977.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a19cdb863f1e44e44caa750adb69adcc8b2bdf7b962ee7bd67f84aef2d84cdf5
+size 417

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
-    "train_loss": 2.8881424867390284,
-    "train_runtime": 618.3653,
+    "train_loss": 2.567346385267914,
+    "train_runtime": 749.2453,
     "train_samples": 46845,
-    "train_samples_per_second": 757.562,
-    "train_steps_per_second": 11.838
+    "train_samples_per_second": 625.229,
+    "train_steps_per_second": 39.079
 }

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": 1.4338148832321167,
       "learning_rate": 6.25e-05,
-      "loss": 6.1167,
+      "loss": 6.2141,
       "step": 500
     },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9016950130462646,
+      "learning_rate": 0.000125,
+      "loss": 3.618,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.043670892715454,
+      "learning_rate": 0.0001875,
+      "loss": 3.3456,
+      "step": 1500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9938444495201111,
+      "learning_rate": 0.00025,
+      "loss": 3.1599,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9072784781455994,
+      "learning_rate": 0.0003125,
+      "loss": 3.0458,
+      "step": 2500
+    },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.39470717462922345,
-      "eval_loss": 3.535817861557007,
-      "eval_runtime": 3.8077,
-      "eval_samples_per_second": 1179.442,
-      "eval_steps_per_second": 2.364,
-      "step": 732
+      "eval_accuracy": 0.4357148242241263,
+      "eval_loss": 3.02449631690979,
+      "eval_runtime": 3.3172,
+      "eval_samples_per_second": 1353.848,
+      "eval_steps_per_second": 10.852,
+      "step": 2928
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.00283682346344,
+      "learning_rate": 0.000375,
+      "loss": 2.9549,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9476096034049988,
+      "learning_rate": 0.00043750000000000006,
+      "loss": 2.8709,
+      "step": 3500
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000125,
-      "loss": 3.4801,
-      "step": 1000
+      "grad_norm": 0.8852096199989319,
+      "learning_rate": 0.0005,
+      "loss": 2.8262,
+      "step": 4000
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8597184419631958,
+      "learning_rate": 0.0005625000000000001,
+      "loss": 2.782,
+      "step": 4500
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.7898221015930176,
+      "learning_rate": 0.000625,
+      "loss": 2.7496,
+      "step": 5000
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.7760418653488159,
+      "learning_rate": 0.0006875,
+      "loss": 2.71,
+      "step": 5500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.43199785397566787,
-      "eval_loss": 3.1201677322387695,
-      "eval_runtime": 3.9073,
-      "eval_samples_per_second": 1149.375,
-      "eval_steps_per_second": 2.303,
-      "step": 1464
+      "eval_accuracy": 0.4585250991922603,
+      "eval_loss": 2.788106679916382,
+      "eval_runtime": 3.4209,
+      "eval_samples_per_second": 1312.812,
+      "eval_steps_per_second": 10.524,
+      "step": 5856
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001875,
-      "loss": 3.1329,
-      "step": 1500
+      "grad_norm": 0.7942250967025757,
+      "learning_rate": 0.00075,
+      "loss": 2.6615,
+      "step": 6000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.7591288685798645,
+      "learning_rate": 0.0008125,
+      "loss": 2.6299,
+      "step": 6500
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8481829762458801,
+      "learning_rate": 0.0008750000000000001,
+      "loss": 2.6236,
+      "step": 7000
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.6588869094848633,
+      "learning_rate": 0.0009375,
+      "loss": 2.6021,
+      "step": 7500
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00025,
-      "loss": 2.9429,
-      "step": 2000
+      "grad_norm": 0.6927358508110046,
+      "learning_rate": 0.001,
+      "loss": 2.5995,
+      "step": 8000
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.6657132506370544,
+      "learning_rate": 0.0010625,
+      "loss": 2.5918,
+      "step": 8500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.44977794609341165,
-      "eval_loss": 2.915814161300659,
-      "eval_runtime": 4.1629,
-      "eval_samples_per_second": 1078.826,
-      "eval_steps_per_second": 2.162,
-      "step": 2196
+      "eval_accuracy": 0.4682225343074601,
+      "eval_loss": 2.6924259662628174,
+      "eval_runtime": 3.4436,
+      "eval_samples_per_second": 1304.177,
+      "eval_steps_per_second": 10.454,
+      "step": 8784
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.6745225787162781,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 2.5379,
+      "step": 9000
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.585035502910614,
+      "learning_rate": 0.0011875,
+      "loss": 2.5163,
+      "step": 9500
     },
     {
       "epoch": 3.42,
-      "learning_rate": 0.0003125,
-      "loss": 2.8071,
-      "step": 2500
+      "grad_norm": 0.5931078791618347,
+      "learning_rate": 0.00125,
+      "loss": 2.5212,
+      "step": 10000
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.520709216594696,
+      "learning_rate": 0.0013125,
+      "loss": 2.5111,
+      "step": 10500
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.5062893629074097,
+      "learning_rate": 0.001375,
+      "loss": 2.5212,
+      "step": 11000
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.4873392879962921,
+      "learning_rate": 0.0014375000000000002,
+      "loss": 2.5122,
+      "step": 11500
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.460516834193321,
-      "eval_loss": 2.790581464767456,
-      "eval_runtime": 4.0203,
-      "eval_samples_per_second": 1117.085,
-      "eval_steps_per_second": 2.239,
-      "step": 2928
+      "eval_accuracy": 0.4758668693467425,
+      "eval_loss": 2.6471331119537354,
+      "eval_runtime": 3.4093,
+      "eval_samples_per_second": 1317.297,
+      "eval_steps_per_second": 10.559,
+      "step": 11712
     },
     {
       "epoch": 4.1,
-      "learning_rate": 0.000375,
-      "loss": 2.7064,
-      "step": 3000
+      "grad_norm": 0.44189488887786865,
+      "learning_rate": 0.0015,
+      "loss": 2.4578,
+      "step": 12000
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.4324429929256439,
+      "learning_rate": 0.0015625,
+      "loss": 2.4466,
+      "step": 12500
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.47547757625579834,
+      "learning_rate": 0.001625,
+      "loss": 2.456,
+      "step": 13000
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.42384257912635803,
+      "learning_rate": 0.0016875,
+      "loss": 2.458,
+      "step": 13500
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00043750000000000006,
-      "loss": 2.6197,
-      "step": 3500
+      "grad_norm": 0.39622461795806885,
+      "learning_rate": 0.0017500000000000003,
+      "loss": 2.466,
+      "step": 14000
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.42361584305763245,
+      "learning_rate": 0.0018124999999999999,
+      "loss": 2.4623,
+      "step": 14500
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.4701406312186929,
-      "eval_loss": 2.6998250484466553,
-      "eval_runtime": 4.0911,
-      "eval_samples_per_second": 1097.737,
-      "eval_steps_per_second": 2.2,
-      "step": 3660
+      "eval_accuracy": 0.48033600008415783,
+      "eval_loss": 2.6053078174591064,
+      "eval_runtime": 3.4538,
+      "eval_samples_per_second": 1300.288,
+      "eval_steps_per_second": 10.423,
+      "step": 14640
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.3994704782962799,
+      "learning_rate": 0.001875,
+      "loss": 2.4137,
+      "step": 15000
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.4023682177066803,
+      "learning_rate": 0.0019375000000000002,
+      "loss": 2.3975,
+      "step": 15500
     },
     {
       "epoch": 5.46,
-      "learning_rate": 0.0005,
-      "loss": 2.5459,
-      "step": 4000
+      "grad_norm": 0.35793477296829224,
+      "learning_rate": 0.002,
+      "loss": 2.406,
+      "step": 16000
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.38212522864341736,
+      "learning_rate": 0.0020625,
+      "loss": 2.4189,
+      "step": 16500
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.38084688782691956,
+      "learning_rate": 0.002125,
+      "loss": 2.4184,
+      "step": 17000
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.34888383746147156,
+      "learning_rate": 0.0021874999999999998,
+      "loss": 2.4246,
+      "step": 17500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.4759738199057783,
-      "eval_loss": 2.641901969909668,
-      "eval_runtime": 4.078,
-      "eval_samples_per_second": 1101.267,
-      "eval_steps_per_second": 2.207,
-      "step": 4392
+      "eval_accuracy": 0.4823645541301325,
+      "eval_loss": 2.5798158645629883,
+      "eval_runtime": 3.4334,
+      "eval_samples_per_second": 1308.029,
+      "eval_steps_per_second": 10.485,
+      "step": 17568
     },
     {
       "epoch": 6.15,
-      "learning_rate": 0.0005625000000000001,
-      "loss": 2.5015,
-      "step": 4500
+      "grad_norm": 0.3350234925746918,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 2.3509,
+      "step": 18000
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.33967897295951843,
+      "learning_rate": 0.0023125000000000003,
+      "loss": 2.3669,
+      "step": 18500
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.3444729447364807,
+      "learning_rate": 0.002375,
+      "loss": 2.3877,
+      "step": 19000
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.3789096176624298,
+      "learning_rate": 0.0024375,
+      "loss": 2.384,
+      "step": 19500
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.000625,
-      "loss": 2.4492,
-      "step": 5000
+      "grad_norm": 0.3346331715583801,
+      "learning_rate": 0.0025,
+      "loss": 2.3871,
+      "step": 20000
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.4802465824036524,
-      "eval_loss": 2.6036274433135986,
-      "eval_runtime": 4.1331,
-      "eval_samples_per_second": 1086.595,
-      "eval_steps_per_second": 2.178,
-      "step": 5124
+      "eval_accuracy": 0.48582729763989924,
+      "eval_loss": 2.564739227294922,
+      "eval_runtime": 3.4251,
+      "eval_samples_per_second": 1311.202,
+      "eval_steps_per_second": 10.511,
+      "step": 20496
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.31661033630371094,
+      "learning_rate": 0.0025625,
+      "loss": 2.3879,
+      "step": 20500
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.324239045381546,
+      "learning_rate": 0.002625,
+      "loss": 2.3191,
+      "step": 21000
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.3616074323654175,
+      "learning_rate": 0.0026875000000000002,
+      "loss": 2.3356,
+      "step": 21500
     },
     {
       "epoch": 7.51,
-      "learning_rate": 0.0006875,
-      "loss": 2.4065,
-      "step": 5500
+      "grad_norm": 0.37798503041267395,
+      "learning_rate": 0.00275,
+      "loss": 2.3425,
+      "step": 22000
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.3979448974132538,
+      "learning_rate": 0.0028125,
+      "loss": 2.3573,
+      "step": 22500
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.31257864832878113,
+      "learning_rate": 0.0028750000000000004,
+      "loss": 2.3644,
+      "step": 23000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.4823891001600752,
-      "eval_loss": 2.5769591331481934,
-      "eval_runtime": 3.9166,
-      "eval_samples_per_second": 1146.648,
-      "eval_steps_per_second": 2.298,
-      "step": 5856
+      "eval_accuracy": 0.48534514348031144,
+      "eval_loss": 2.5570876598358154,
+      "eval_runtime": 3.4172,
+      "eval_samples_per_second": 1314.225,
+      "eval_steps_per_second": 10.535,
+      "step": 23424
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.2986924648284912,
+      "learning_rate": 0.0029375,
+      "loss": 2.3461,
+      "step": 23500
     },
     {
       "epoch": 8.2,
-      "learning_rate": 0.00075,
-      "loss": 2.3867,
-      "step": 6000
+      "grad_norm": 0.32609128952026367,
+      "learning_rate": 0.003,
+      "loss": 2.2927,
+      "step": 24000
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.3273513615131378,
+      "learning_rate": 0.002715909090909091,
+      "loss": 2.3094,
+      "step": 24500
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.321181982755661,
+      "learning_rate": 0.0024318181818181817,
+      "loss": 2.3011,
+      "step": 25000
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.3118550181388855,
+      "learning_rate": 0.002147727272727273,
+      "loss": 2.2864,
+      "step": 25500
     },
     {
       "epoch": 8.88,
-      "learning_rate": 0.0008125,
-      "loss": 2.3626,
-      "step": 6500
+      "grad_norm": 0.30095553398132324,
+      "learning_rate": 0.0018636363636363638,
+      "loss": 2.2824,
+      "step": 26000
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.4863024386480748,
-      "eval_loss": 2.5622453689575195,
-      "eval_runtime": 3.9696,
-      "eval_samples_per_second": 1131.334,
-      "eval_steps_per_second": 2.267,
-      "step": 6588
+      "eval_accuracy": 0.49336643540799885,
+      "eval_loss": 2.5034055709838867,
+      "eval_runtime": 3.4213,
+      "eval_samples_per_second": 1312.663,
+      "eval_steps_per_second": 10.522,
+      "step": 26352
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.3036137819290161,
+      "learning_rate": 0.0015795454545454546,
+      "loss": 2.2438,
+      "step": 26500
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.32144156098365784,
+      "learning_rate": 0.0012954545454545456,
+      "loss": 2.1689,
+      "step": 27000
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.3083929717540741,
+      "learning_rate": 0.0010113636363636364,
+      "loss": 2.162,
+      "step": 27500
     },
     {
       "epoch": 9.56,
-      "learning_rate": 0.0008750000000000001,
-      "loss": 2.3276,
-      "step": 7000
+      "grad_norm": 0.3136695325374603,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 2.1547,
+      "step": 28000
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 0.3204050362110138,
+      "learning_rate": 0.0004431818181818182,
+      "loss": 2.1479,
+      "step": 28500
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 0.3189040422439575,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 2.1369,
+      "step": 29000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.4861165901356519,
-      "eval_loss": 2.557232141494751,
-      "eval_runtime": 4.198,
-      "eval_samples_per_second": 1069.803,
-      "eval_steps_per_second": 2.144,
-      "step": 7320
+      "eval_accuracy": 0.4999763306139839,
+      "eval_loss": 2.471221446990967,
+      "eval_runtime": 3.3986,
+      "eval_samples_per_second": 1321.414,
+      "eval_steps_per_second": 10.592,
+      "step": 29280
     },
     {
       "epoch": 10.0,
-      "step": 7320,
+      "step": 29280,
       "total_flos": 2273237316403200.0,
-      "train_loss": 2.8881424867390284,
-      "train_runtime": 618.3653,
-      "train_samples_per_second": 757.562,
-      "train_steps_per_second": 11.838
+      "train_loss": 2.567346385267914,
+      "train_runtime": 749.2453,
+      "train_samples_per_second": 625.229,
+      "train_steps_per_second": 39.079
     }
   ],
   "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
 }