End of training

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.48671446129354073,
-    "eval_loss": 2.5570006370544434,
-    "eval_runtime": 4.0107,
+    "eval_accuracy": 0.5000412022645466,
+    "eval_loss": 2.4759199619293213,
+    "eval_runtime": 3.3905,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1119.749,
-    "eval_steps_per_second": 2.244,
-    "perplexity": 12.897076232743924,
-    "train_loss": 2.8811990956791114,
-    "train_runtime": 625.422,
+    "eval_samples_per_second": 1324.582,
+    "eval_steps_per_second": 10.618,
+    "perplexity": 11.892642860763285,
+    "train_loss": 2.5673805945557975,
+    "train_runtime": 750.0006,
     "train_samples": 46845,
-    "train_samples_per_second": 749.014,
-    "train_steps_per_second": 11.704
+    "train_samples_per_second": 624.6,
+    "train_steps_per_second": 39.04
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.48671446129354073,
-    "eval_loss": 2.5570006370544434,
-    "eval_runtime": 4.0107,
+    "eval_accuracy": 0.5000412022645466,
+    "eval_loss": 2.4759199619293213,
+    "eval_runtime": 3.3905,
     "eval_samples": 4491,
-    "eval_samples_per_second": 1119.749,
-    "eval_steps_per_second": 2.244,
-    "perplexity": 12.897076232743924
+    "eval_samples_per_second": 1324.582,
+    "eval_steps_per_second": 10.618,
+    "perplexity": 11.892642860763285
 }

runs/Mar25_19-52-30_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1711415108.phyl-ling-p01.la.utexas.edu.1050359.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbbc4b0c26111b10b7e950e148b91ba594f18f34a44c3fb484455b37804618ad
+size 417

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
-    "train_loss": 2.8811990956791114,
-    "train_runtime": 625.422,
+    "train_loss": 2.5673805945557975,
+    "train_runtime": 750.0006,
     "train_samples": 46845,
-    "train_samples_per_second": 749.014,
-    "train_steps_per_second": 11.704
+    "train_samples_per_second": 624.6,
+    "train_steps_per_second": 39.04
 }

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": 1.1428395509719849,
       "learning_rate": 6.25e-05,
-      "loss": 6.078,
+      "loss": 6.1643,
       "step": 500
     },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8605946898460388,
+      "learning_rate": 0.000125,
+      "loss": 3.6177,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0353199243545532,
+      "learning_rate": 0.0001875,
+      "loss": 3.3423,
+      "step": 1500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0619558095932007,
+      "learning_rate": 0.00025,
+      "loss": 3.174,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8885842561721802,
+      "learning_rate": 0.0003125,
+      "loss": 3.0508,
+      "step": 2500
+    },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.3944669741933561,
-      "eval_loss": 3.533393621444702,
-      "eval_runtime": 3.6343,
-      "eval_samples_per_second": 1235.73,
-      "eval_steps_per_second": 2.476,
-      "step": 732
+      "eval_accuracy": 0.4366703661040366,
+      "eval_loss": 3.018024444580078,
+      "eval_runtime": 3.3274,
+      "eval_samples_per_second": 1349.7,
+      "eval_steps_per_second": 10.819,
+      "step": 2928
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.9355678558349609,
+      "learning_rate": 0.000375,
+      "loss": 2.9588,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9893432259559631,
+      "learning_rate": 0.00043750000000000006,
+      "loss": 2.8647,
+      "step": 3500
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000125,
-      "loss": 3.4808,
-      "step": 1000
+      "grad_norm": 0.8901029229164124,
+      "learning_rate": 0.0005,
+      "loss": 2.8204,
+      "step": 4000
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8454930186271667,
+      "learning_rate": 0.0005625000000000001,
+      "loss": 2.7743,
+      "step": 4500
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.7920435667037964,
+      "learning_rate": 0.000625,
+      "loss": 2.7475,
+      "step": 5000
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.862263023853302,
+      "learning_rate": 0.0006875,
+      "loss": 2.7062,
+      "step": 5500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.4324659818324313,
-      "eval_loss": 3.121781349182129,
-      "eval_runtime": 3.7953,
-      "eval_samples_per_second": 1183.308,
-      "eval_steps_per_second": 2.371,
-      "step": 1464
+      "eval_accuracy": 0.45998208139814184,
+      "eval_loss": 2.785717725753784,
+      "eval_runtime": 3.419,
+      "eval_samples_per_second": 1313.535,
+      "eval_steps_per_second": 10.529,
+      "step": 5856
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001875,
-      "loss": 3.1425,
-      "step": 1500
+      "grad_norm": 0.8128435611724854,
+      "learning_rate": 0.00075,
+      "loss": 2.667,
+      "step": 6000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.7912793755531311,
+      "learning_rate": 0.0008125,
+      "loss": 2.6257,
+      "step": 6500
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.7729538679122925,
+      "learning_rate": 0.0008750000000000001,
+      "loss": 2.622,
+      "step": 7000
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.7662726640701294,
+      "learning_rate": 0.0009375,
+      "loss": 2.5976,
+      "step": 7500
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00025,
-      "loss": 2.9363,
-      "step": 2000
+      "grad_norm": 0.7259579300880432,
+      "learning_rate": 0.001,
+      "loss": 2.6013,
+      "step": 8000
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.624014675617218,
+      "learning_rate": 0.0010625,
+      "loss": 2.5923,
+      "step": 8500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.45024958052588115,
-      "eval_loss": 2.9133858680725098,
-      "eval_runtime": 3.764,
-      "eval_samples_per_second": 1193.139,
-      "eval_steps_per_second": 2.391,
-      "step": 2196
+      "eval_accuracy": 0.470033680659657,
+      "eval_loss": 2.690037488937378,
+      "eval_runtime": 3.4246,
+      "eval_samples_per_second": 1311.398,
+      "eval_steps_per_second": 10.512,
+      "step": 8784
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.6166909337043762,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 2.545,
+      "step": 9000
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.6117091774940491,
+      "learning_rate": 0.0011875,
+      "loss": 2.5121,
+      "step": 9500
     },
     {
       "epoch": 3.42,
-      "learning_rate": 0.0003125,
-      "loss": 2.803,
-      "step": 2500
+      "grad_norm": 0.5614470839500427,
+      "learning_rate": 0.00125,
+      "loss": 2.5241,
+      "step": 10000
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.5524864792823792,
+      "learning_rate": 0.0013125,
+      "loss": 2.5018,
+      "step": 10500
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.488916277885437,
+      "learning_rate": 0.001375,
+      "loss": 2.5242,
+      "step": 11000
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.49240800738334656,
+      "learning_rate": 0.0014375000000000002,
+      "loss": 2.5183,
+      "step": 11500
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.46225434245569,
-      "eval_loss": 2.783372402191162,
-      "eval_runtime": 3.9861,
-      "eval_samples_per_second": 1126.66,
-      "eval_steps_per_second": 2.258,
-      "step": 2928
+      "eval_accuracy": 0.4760229119656636,
+      "eval_loss": 2.6405022144317627,
+      "eval_runtime": 3.3846,
+      "eval_samples_per_second": 1326.91,
+      "eval_steps_per_second": 10.637,
+      "step": 11712
     },
     {
       "epoch": 4.1,
-      "learning_rate": 0.000375,
-      "loss": 2.6954,
-      "step": 3000
+      "grad_norm": 0.4792906641960144,
+      "learning_rate": 0.0015,
+      "loss": 2.4636,
+      "step": 12000
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.44715985655784607,
+      "learning_rate": 0.0015625,
+      "loss": 2.4505,
+      "step": 12500
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.4706518054008484,
+      "learning_rate": 0.001625,
+      "loss": 2.4493,
+      "step": 13000
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.4098159372806549,
+      "learning_rate": 0.0016875,
+      "loss": 2.4672,
+      "step": 13500
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00043750000000000006,
-      "loss": 2.6112,
-      "step": 3500
+      "grad_norm": 0.4300088584423065,
+      "learning_rate": 0.0017500000000000003,
+      "loss": 2.4597,
+      "step": 14000
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.4115521013736725,
+      "learning_rate": 0.0018124999999999999,
+      "loss": 2.4632,
+      "step": 14500
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.4700792661438362,
-      "eval_loss": 2.6976747512817383,
-      "eval_runtime": 4.0708,
-      "eval_samples_per_second": 1103.233,
-      "eval_steps_per_second": 2.211,
-      "step": 3660
+      "eval_accuracy": 0.4799397570293693,
+      "eval_loss": 2.610959768295288,
+      "eval_runtime": 3.4241,
+      "eval_samples_per_second": 1311.574,
+      "eval_steps_per_second": 10.514,
+      "step": 14640
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.36413952708244324,
+      "learning_rate": 0.001875,
+      "loss": 2.4085,
+      "step": 15000
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.4081680178642273,
+      "learning_rate": 0.0019375000000000002,
+      "loss": 2.408,
+      "step": 15500
     },
     {
       "epoch": 5.46,
-      "learning_rate": 0.0005,
-      "loss": 2.5369,
-      "step": 4000
+      "grad_norm": 0.3759631812572479,
+      "learning_rate": 0.002,
+      "loss": 2.4168,
+      "step": 16000
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.3406212627887726,
+      "learning_rate": 0.0020625,
+      "loss": 2.419,
+      "step": 16500
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.3810693621635437,
+      "learning_rate": 0.002125,
+      "loss": 2.4227,
+      "step": 17000
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.40893515944480896,
+      "learning_rate": 0.0021874999999999998,
+      "loss": 2.4241,
+      "step": 17500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.4759703133300722,
-      "eval_loss": 2.644042491912842,
-      "eval_runtime": 4.218,
-      "eval_samples_per_second": 1064.732,
-      "eval_steps_per_second": 2.134,
-      "step": 4392
+      "eval_accuracy": 0.48350945109817184,
+      "eval_loss": 2.5840396881103516,
+      "eval_runtime": 3.3805,
+      "eval_samples_per_second": 1328.515,
+      "eval_steps_per_second": 10.649,
+      "step": 17568
     },
     {
       "epoch": 6.15,
-      "learning_rate": 0.0005625000000000001,
-      "loss": 2.4935,
-      "step": 4500
+      "grad_norm": 0.332253098487854,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 2.3589,
+      "step": 18000
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.3078455924987793,
+      "learning_rate": 0.0023125000000000003,
+      "loss": 2.3668,
+      "step": 18500
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.3293627202510834,
+      "learning_rate": 0.002375,
+      "loss": 2.3857,
+      "step": 19000
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.3553139567375183,
+      "learning_rate": 0.0024375,
+      "loss": 2.3861,
+      "step": 19500
     },
     {
       "epoch": 6.83,
-      "learning_rate": 0.000625,
-      "loss": 2.4479,
-      "step": 5000
+      "grad_norm": 0.35184377431869507,
+      "learning_rate": 0.0025,
+      "loss": 2.3815,
+      "step": 20000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.48506812399953014,
+      "eval_loss": 2.572838306427002,
+      "eval_runtime": 3.4336,
+      "eval_samples_per_second": 1307.964,
+      "eval_steps_per_second": 10.485,
+      "step": 20496
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.4810758875581434,
-      "eval_loss": 2.601984739303589,
-      "eval_runtime": 4.273,
-      "eval_samples_per_second": 1051.027,
-      "eval_steps_per_second": 2.106,
-      "step": 5124
+      "grad_norm": 0.3210340440273285,
+      "learning_rate": 0.0025625,
+      "loss": 2.3912,
+      "step": 20500
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.37861162424087524,
+      "learning_rate": 0.002625,
+      "loss": 2.315,
+      "step": 21000
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.368950217962265,
+      "learning_rate": 0.0026875000000000002,
+      "loss": 2.3418,
+      "step": 21500
     },
     {
       "epoch": 7.51,
-      "learning_rate": 0.0006875,
-      "loss": 2.3999,
-      "step": 5500
+      "grad_norm": 0.3538954257965088,
+      "learning_rate": 0.00275,
+      "loss": 2.3559,
+      "step": 22000
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.33648544549942017,
+      "learning_rate": 0.0028125,
+      "loss": 2.3544,
+      "step": 22500
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.2958958148956299,
+      "learning_rate": 0.0028750000000000004,
+      "loss": 2.3595,
+      "step": 23000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.483818029760308,
-      "eval_loss": 2.581047773361206,
-      "eval_runtime": 4.1042,
-      "eval_samples_per_second": 1094.257,
-      "eval_steps_per_second": 2.193,
-      "step": 5856
+      "eval_accuracy": 0.48671446129354073,
+      "eval_loss": 2.558120012283325,
+      "eval_runtime": 3.4225,
+      "eval_samples_per_second": 1312.192,
+      "eval_steps_per_second": 10.519,
+      "step": 23424
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.3472735285758972,
+      "learning_rate": 0.0029375,
+      "loss": 2.3436,
+      "step": 23500
     },
     {
       "epoch": 8.2,
-      "learning_rate": 0.00075,
-      "loss": 2.3804,
-      "step": 6000
+      "grad_norm": 0.3491041660308838,
+      "learning_rate": 0.003,
+      "loss": 2.2902,
+      "step": 24000
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.36862629652023315,
+      "learning_rate": 0.002715909090909091,
+      "loss": 2.301,
+      "step": 24500
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.3976895809173584,
+      "learning_rate": 0.0024318181818181817,
+      "loss": 2.3072,
+      "step": 25000
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.33942127227783203,
+      "learning_rate": 0.002147727272727273,
+      "loss": 2.2934,
+      "step": 25500
     },
     {
       "epoch": 8.88,
-      "learning_rate": 0.0008125,
-      "loss": 2.3566,
-      "step": 6500
+      "grad_norm": 0.3659175932407379,
+      "learning_rate": 0.0018636363636363638,
+      "loss": 2.2838,
+      "step": 26000
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.48580275160995656,
-      "eval_loss": 2.5657148361206055,
-      "eval_runtime": 4.0599,
-      "eval_samples_per_second": 1106.186,
-      "eval_steps_per_second": 2.217,
-      "step": 6588
+      "eval_accuracy": 0.49492686159721017,
+      "eval_loss": 2.5014333724975586,
+      "eval_runtime": 3.4108,
+      "eval_samples_per_second": 1316.695,
+      "eval_steps_per_second": 10.555,
+      "step": 26352
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.3276906907558441,
+      "learning_rate": 0.0015795454545454546,
+      "loss": 2.2536,
+      "step": 26500
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.29987043142318726,
+      "learning_rate": 0.0012954545454545456,
+      "loss": 2.1713,
+      "step": 27000
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.29404929280281067,
+      "learning_rate": 0.0010113636363636364,
+      "loss": 2.1643,
+      "step": 27500
     },
     {
       "epoch": 9.56,
-      "learning_rate": 0.0008750000000000001,
-      "loss": 2.3251,
-      "step": 7000
+      "grad_norm": 0.26416483521461487,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 2.1585,
+      "step": 28000
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 0.3155469596385956,
+      "learning_rate": 0.0004431818181818182,
+      "loss": 2.1498,
+      "step": 28500
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 0.35845381021499634,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 2.1364,
+      "step": 29000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.48671446129354073,
-      "eval_loss": 2.5570006370544434,
-      "eval_runtime": 4.3421,
-      "eval_samples_per_second": 1034.281,
-      "eval_steps_per_second": 2.073,
-      "step": 7320
+      "eval_accuracy": 0.5000412022645466,
+      "eval_loss": 2.4759199619293213,
+      "eval_runtime": 3.4484,
+      "eval_samples_per_second": 1302.334,
+      "eval_steps_per_second": 10.44,
+      "step": 29280
     },
     {
       "epoch": 10.0,
-      "step": 7320,
+      "step": 29280,
       "total_flos": 2273237316403200.0,
-      "train_loss": 2.8811990956791114,
-      "train_runtime": 625.422,
-      "train_samples_per_second": 749.014,
-      "train_steps_per_second": 11.704
+      "train_loss": 2.5673805945557975,
+      "train_runtime": 750.0006,
+      "train_samples_per_second": 624.6,
+      "train_steps_per_second": 39.04
     }
   ],
   "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
 }