Upload 11 files

all_results.json

@@ -1,15 +1,17 @@
 {
     "epoch": 3.0,
-    "eval_loss": 0.1280035674571991,
-    "eval_mse": 0.12800357063188794,
-    "eval_runtime": 9.6254,
+    "eval_loss": 0.09235269576311111,
+    "eval_mse": 0.09235269895339973,
+    "eval_runtime": 9.5082,
     "eval_samples": 10000,
-    "eval_samples_per_second": 1038.92,
-    "eval_steps_per_second": 129.865,
+    "eval_samples_per_second": 1051.725,
+    "eval_steps_per_second": 131.466,
+    "num_input_tokens_seen": 34560000,
     "total_flos": 4446488701440000.0,
-    "train_loss": 0.12363309427897136,
-    "train_runtime": 1375.441,
+    "train_loss": 0.08211795973601164,
+    "train_runtime": 1433.1905,
     "train_samples": 90000,
-    "train_samples_per_second": 196.301,
-    "train_steps_per_second": 24.538
+    "train_samples_per_second": 188.391,
+    "train_steps_per_second": 23.549,
+    "train_tokens_per_second": 24114.032
 }

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "deberta-v3-xsmall-quality-pretrain",
+  "_name_or_path": "/media/user/Expansion/Data/Quality/pretrained-deberta-xs",
   "architectures": [
     "DebertaV2ForSequenceClassification"
   ],
@@ -37,7 +37,7 @@
   "relative_attention": true,
   "share_att_key": true,
   "torch_dtype": "float32",
-  "transformers_version": "4.44.2",
+  "transformers_version": "4.45.1",
   "type_vocab_size": 0,
   "vocab_size": 128100
 }

eval_results.json

@@ -1,9 +1,10 @@
 {
     "epoch": 3.0,
-    "eval_loss": 0.1280035674571991,
-    "eval_mse": 0.12800357063188794,
-    "eval_runtime": 9.6254,
+    "eval_loss": 0.09235269576311111,
+    "eval_mse": 0.09235269895339973,
+    "eval_runtime": 9.5082,
     "eval_samples": 10000,
-    "eval_samples_per_second": 1038.92,
-    "eval_steps_per_second": 129.865
+    "eval_samples_per_second": 1051.725,
+    "eval_steps_per_second": 131.466,
+    "num_input_tokens_seen": 34560000
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa511c11ec7e034ea02c0d29d8b93bfab1a3175bd0502ff717634440564f2031
+oid sha256:fdb1d2be17ae696c75e5981d9fdcdc61e0794d679d4c1326fed3f6098e74fe9e
 size 283345892

train_results.json

@@ -1,9 +1,11 @@
 {
     "epoch": 3.0,
+    "num_input_tokens_seen": 34560000,
     "total_flos": 4446488701440000.0,
-    "train_loss": 0.12363309427897136,
-    "train_runtime": 1375.441,
+    "train_loss": 0.08211795973601164,
+    "train_runtime": 1433.1905,
     "train_samples": 90000,
-    "train_samples_per_second": 196.301,
-    "train_steps_per_second": 24.538
+    "train_samples_per_second": 188.391,
+    "train_steps_per_second": 23.549,
+    "train_tokens_per_second": 24114.032
 }

trainer_state.json

@@ -10,486 +10,555 @@
   "log_history": [
     {
       "epoch": 0.044444444444444446,
-      "grad_norm": 3.210815906524658,
+      "grad_norm": 4.640678405761719,
       "learning_rate": 4.925925925925926e-05,
-      "loss": 0.4315,
+      "loss": 0.2464,
+      "num_input_tokens_seen": 512000,
       "step": 500
     },
     {
       "epoch": 0.08888888888888889,
-      "grad_norm": 8.126620292663574,
+      "grad_norm": 2.8545427322387695,
       "learning_rate": 4.851851851851852e-05,
-      "loss": 0.2862,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 1024000,
       "step": 1000
     },
     {
       "epoch": 0.13333333333333333,
-      "grad_norm": 4.665143013000488,
+      "grad_norm": 1.1847012042999268,
       "learning_rate": 4.7777777777777784e-05,
-      "loss": 0.237,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 1536000,
       "step": 1500
     },
     {
       "epoch": 0.17777777777777778,
-      "grad_norm": 2.0566887855529785,
+      "grad_norm": 3.767167568206787,
       "learning_rate": 4.703703703703704e-05,
-      "loss": 0.2251,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 2048000,
       "step": 2000
     },
     {
       "epoch": 0.2222222222222222,
-      "grad_norm": 7.968118667602539,
+      "grad_norm": 3.5827574729919434,
       "learning_rate": 4.62962962962963e-05,
-      "loss": 0.2239,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 2560000,
       "step": 2500
     },
     {
       "epoch": 0.26666666666666666,
-      "grad_norm": 1.2121827602386475,
+      "grad_norm": 2.042477607727051,
       "learning_rate": 4.555555555555556e-05,
-      "loss": 0.2203,
+      "loss": 0.1412,
+      "num_input_tokens_seen": 3072000,
       "step": 3000
     },
     {
       "epoch": 0.3111111111111111,
-      "grad_norm": 3.395099401473999,
+      "grad_norm": 1.130966305732727,
       "learning_rate": 4.481481481481482e-05,
-      "loss": 0.2111,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 3584000,
       "step": 3500
     },
     {
       "epoch": 0.35555555555555557,
-      "grad_norm": 1.9526039361953735,
+      "grad_norm": 3.1181435585021973,
       "learning_rate": 4.4074074074074076e-05,
-      "loss": 0.193,
+      "loss": 0.131,
+      "num_input_tokens_seen": 4096000,
       "step": 4000
     },
     {
       "epoch": 0.4,
-      "grad_norm": 3.2926955223083496,
+      "grad_norm": 2.284423351287842,
       "learning_rate": 4.3333333333333334e-05,
-      "loss": 0.1853,
+      "loss": 0.135,
+      "num_input_tokens_seen": 4608000,
       "step": 4500
     },
     {
       "epoch": 0.4444444444444444,
-      "grad_norm": 2.8749849796295166,
+      "grad_norm": 2.6171817779541016,
       "learning_rate": 4.259259259259259e-05,
-      "loss": 0.1675,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 5120000,
       "step": 5000
     },
     {
       "epoch": 0.4888888888888889,
-      "grad_norm": 6.5106706619262695,
+      "grad_norm": 2.712770462036133,
       "learning_rate": 4.185185185185185e-05,
-      "loss": 0.1889,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 5632000,
       "step": 5500
     },
     {
       "epoch": 0.5333333333333333,
-      "grad_norm": 1.893872618675232,
+      "grad_norm": 1.8071026802062988,
       "learning_rate": 4.111111111111111e-05,
-      "loss": 0.2002,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 6144000,
       "step": 6000
     },
     {
       "epoch": 0.5777777777777777,
-      "grad_norm": 4.822785377502441,
+      "grad_norm": 2.124100685119629,
       "learning_rate": 4.0370370370370374e-05,
-      "loss": 0.1802,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 6656000,
       "step": 6500
     },
     {
       "epoch": 0.6222222222222222,
-      "grad_norm": 2.410318374633789,
+      "grad_norm": 2.7214131355285645,
       "learning_rate": 3.962962962962963e-05,
-      "loss": 0.1578,
+      "loss": 0.118,
+      "num_input_tokens_seen": 7168000,
       "step": 7000
     },
     {
       "epoch": 0.6666666666666666,
-      "grad_norm": 7.345259189605713,
+      "grad_norm": 2.097531318664551,
       "learning_rate": 3.888888888888889e-05,
-      "loss": 0.1668,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 7680000,
       "step": 7500
     },
     {
       "epoch": 0.7111111111111111,
-      "grad_norm": 2.6905734539031982,
+      "grad_norm": 2.9476003646850586,
       "learning_rate": 3.814814814814815e-05,
-      "loss": 0.1726,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 8192000,
       "step": 8000
     },
     {
       "epoch": 0.7555555555555555,
-      "grad_norm": 7.399478912353516,
+      "grad_norm": 2.068228006362915,
       "learning_rate": 3.740740740740741e-05,
-      "loss": 0.1586,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 8704000,
       "step": 8500
     },
     {
       "epoch": 0.8,
-      "grad_norm": 1.8845775127410889,
+      "grad_norm": 1.8230115175247192,
       "learning_rate": 3.6666666666666666e-05,
-      "loss": 0.1629,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 9216000,
       "step": 9000
     },
     {
       "epoch": 0.8444444444444444,
-      "grad_norm": 3.4357192516326904,
+      "grad_norm": 2.0530920028686523,
       "learning_rate": 3.592592592592593e-05,
-      "loss": 0.1674,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 9728000,
       "step": 9500
     },
     {
       "epoch": 0.8888888888888888,
-      "grad_norm": 1.8075140714645386,
+      "grad_norm": 1.2028056383132935,
       "learning_rate": 3.518518518518519e-05,
-      "loss": 0.1671,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 10240000,
       "step": 10000
     },
     {
       "epoch": 0.9333333333333333,
-      "grad_norm": 2.545607089996338,
+      "grad_norm": 1.908441185951233,
       "learning_rate": 3.444444444444445e-05,
-      "loss": 0.1595,
+      "loss": 0.102,
+      "num_input_tokens_seen": 10752000,
       "step": 10500
     },
     {
       "epoch": 0.9777777777777777,
-      "grad_norm": 2.33050799369812,
+      "grad_norm": 1.4811742305755615,
       "learning_rate": 3.3703703703703706e-05,
-      "loss": 0.1655,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 11264000,
       "step": 11000
     },
     {
       "epoch": 1.0222222222222221,
-      "grad_norm": 1.9340065717697144,
+      "grad_norm": 2.421898126602173,
       "learning_rate": 3.2962962962962964e-05,
-      "loss": 0.1349,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 11776000,
       "step": 11500
     },
     {
       "epoch": 1.0666666666666667,
-      "grad_norm": 1.3230476379394531,
+      "grad_norm": 1.024445652961731,
       "learning_rate": 3.222222222222223e-05,
-      "loss": 0.1069,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 12288000,
       "step": 12000
     },
     {
       "epoch": 1.1111111111111112,
-      "grad_norm": 4.054622650146484,
+      "grad_norm": 2.516057014465332,
       "learning_rate": 3.148148148148148e-05,
-      "loss": 0.1098,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 12800000,
       "step": 12500
     },
     {
       "epoch": 1.1555555555555554,
-      "grad_norm": 4.885370254516602,
+      "grad_norm": 3.101442813873291,
       "learning_rate": 3.074074074074074e-05,
-      "loss": 0.1162,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 13312000,
       "step": 13000
     },
     {
       "epoch": 1.2,
-      "grad_norm": 3.5849854946136475,
+      "grad_norm": 1.0181483030319214,
       "learning_rate": 3e-05,
-      "loss": 0.1077,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 13824000,
       "step": 13500
     },
     {
       "epoch": 1.2444444444444445,
-      "grad_norm": 9.765403747558594,
+      "grad_norm": 2.7126169204711914,
       "learning_rate": 2.925925925925926e-05,
-      "loss": 0.1181,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 14336000,
       "step": 14000
     },
     {
       "epoch": 1.2888888888888888,
-      "grad_norm": 3.879992961883545,
+      "grad_norm": 2.1057209968566895,
       "learning_rate": 2.851851851851852e-05,
-      "loss": 0.1011,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 14848000,
       "step": 14500
     },
     {
       "epoch": 1.3333333333333333,
-      "grad_norm": 2.2347288131713867,
+      "grad_norm": 2.226621150970459,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.1175,
+      "loss": 0.081,
+      "num_input_tokens_seen": 15360000,
       "step": 15000
     },
     {
       "epoch": 1.3777777777777778,
-      "grad_norm": 7.474192142486572,
+      "grad_norm": 1.1672347784042358,
       "learning_rate": 2.7037037037037037e-05,
-      "loss": 0.1073,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 15872000,
       "step": 15500
     },
     {
       "epoch": 1.4222222222222223,
-      "grad_norm": 4.1693291664123535,
+      "grad_norm": 1.556462287902832,
       "learning_rate": 2.6296296296296296e-05,
-      "loss": 0.0995,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 16384000,
       "step": 16000
     },
     {
       "epoch": 1.4666666666666668,
-      "grad_norm": 1.8383170366287231,
+      "grad_norm": 1.4660066366195679,
       "learning_rate": 2.5555555555555554e-05,
-      "loss": 0.1145,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 16896000,
       "step": 16500
     },
     {
       "epoch": 1.511111111111111,
-      "grad_norm": 0.694262683391571,
+      "grad_norm": 1.3551133871078491,
       "learning_rate": 2.4814814814814816e-05,
-      "loss": 0.115,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 17408000,
       "step": 17000
     },
     {
       "epoch": 1.5555555555555556,
-      "grad_norm": 0.9633584022521973,
+      "grad_norm": 1.1325716972351074,
       "learning_rate": 2.4074074074074074e-05,
-      "loss": 0.1024,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 17920000,
       "step": 17500
     },
     {
       "epoch": 1.6,
-      "grad_norm": 2.0199620723724365,
+      "grad_norm": 1.0320031642913818,
       "learning_rate": 2.3333333333333336e-05,
-      "loss": 0.1011,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 18432000,
       "step": 18000
     },
     {
       "epoch": 1.6444444444444444,
-      "grad_norm": 1.8132920265197754,
+      "grad_norm": 2.1350128650665283,
       "learning_rate": 2.2592592592592594e-05,
-      "loss": 0.1041,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 18944000,
       "step": 18500
     },
     {
       "epoch": 1.6888888888888889,
-      "grad_norm": 8.3416748046875,
+      "grad_norm": 1.1131097078323364,
       "learning_rate": 2.1851851851851852e-05,
-      "loss": 0.1144,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 19456000,
       "step": 19000
     },
     {
       "epoch": 1.7333333333333334,
-      "grad_norm": 1.4581434726715088,
+      "grad_norm": 1.8592647314071655,
       "learning_rate": 2.111111111111111e-05,
-      "loss": 0.0955,
+      "loss": 0.07,
+      "num_input_tokens_seen": 19968000,
       "step": 19500
     },
     {
       "epoch": 1.7777777777777777,
-      "grad_norm": 3.9014599323272705,
+      "grad_norm": 1.7992678880691528,
       "learning_rate": 2.037037037037037e-05,
-      "loss": 0.104,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 20480000,
       "step": 20000
     },
     {
       "epoch": 1.8222222222222222,
-      "grad_norm": 2.070230007171631,
+      "grad_norm": 1.4641705751419067,
       "learning_rate": 1.962962962962963e-05,
-      "loss": 0.0995,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 20992000,
       "step": 20500
     },
     {
       "epoch": 1.8666666666666667,
-      "grad_norm": 0.8812291026115417,
+      "grad_norm": 1.758195161819458,
       "learning_rate": 1.888888888888889e-05,
-      "loss": 0.0982,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 21504000,
       "step": 21000
     },
     {
       "epoch": 1.911111111111111,
-      "grad_norm": 1.1036711931228638,
+      "grad_norm": 1.9842095375061035,
       "learning_rate": 1.814814814814815e-05,
-      "loss": 0.0987,
+      "loss": 0.068,
+      "num_input_tokens_seen": 22016000,
       "step": 21500
     },
     {
       "epoch": 1.9555555555555557,
-      "grad_norm": 0.6822977066040039,
+      "grad_norm": 1.4546846151351929,
       "learning_rate": 1.740740740740741e-05,
-      "loss": 0.101,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 22528000,
       "step": 22000
     },
     {
       "epoch": 2.0,
-      "grad_norm": 4.314443111419678,
+      "grad_norm": 3.412598133087158,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.0979,
+      "loss": 0.068,
+      "num_input_tokens_seen": 23040000,
       "step": 22500
     },
     {
       "epoch": 2.0444444444444443,
-      "grad_norm": 3.5247573852539062,
+      "grad_norm": 2.1435558795928955,
       "learning_rate": 1.5925925925925926e-05,
-      "loss": 0.0733,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 23552000,
       "step": 23000
     },
     {
       "epoch": 2.088888888888889,
-      "grad_norm": 0.6801110506057739,
+      "grad_norm": 1.800618290901184,
       "learning_rate": 1.5185185185185186e-05,
-      "loss": 0.0651,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 24064000,
       "step": 23500
     },
     {
       "epoch": 2.1333333333333333,
-      "grad_norm": 4.516416072845459,
+      "grad_norm": 1.1772105693817139,
       "learning_rate": 1.4444444444444444e-05,
-      "loss": 0.0765,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 24576000,
       "step": 24000
     },
     {
       "epoch": 2.1777777777777776,
-      "grad_norm": 1.9308606386184692,
+      "grad_norm": 0.9198475480079651,
       "learning_rate": 1.3703703703703704e-05,
-      "loss": 0.0707,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 25088000,
       "step": 24500
     },
     {
       "epoch": 2.2222222222222223,
-      "grad_norm": 1.1442885398864746,
+      "grad_norm": 1.3264068365097046,
       "learning_rate": 1.2962962962962962e-05,
-      "loss": 0.0617,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 25600000,
       "step": 25000
     },
     {
       "epoch": 2.2666666666666666,
-      "grad_norm": 5.30832576751709,
+      "grad_norm": 2.010927438735962,
       "learning_rate": 1.2222222222222222e-05,
-      "loss": 0.0693,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 26112000,
       "step": 25500
     },
     {
       "epoch": 2.311111111111111,
-      "grad_norm": 0.5708422660827637,
+      "grad_norm": 1.620568871498108,
       "learning_rate": 1.1481481481481482e-05,
-      "loss": 0.0697,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 26624000,
       "step": 26000
     },
     {
       "epoch": 2.3555555555555556,
-      "grad_norm": 0.680268406867981,
+      "grad_norm": 1.1650017499923706,
       "learning_rate": 1.074074074074074e-05,
-      "loss": 0.066,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 27136000,
       "step": 26500
     },
     {
       "epoch": 2.4,
-      "grad_norm": 0.72934889793396,
+      "grad_norm": 1.404801607131958,
       "learning_rate": 1e-05,
-      "loss": 0.0644,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 27648000,
       "step": 27000
     },
     {
       "epoch": 2.4444444444444446,
-      "grad_norm": 4.663362503051758,
+      "grad_norm": 1.6808654069900513,
       "learning_rate": 9.259259259259259e-06,
-      "loss": 0.0667,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 28160000,
       "step": 27500
     },
     {
       "epoch": 2.488888888888889,
-      "grad_norm": 1.3758256435394287,
+      "grad_norm": 1.0057835578918457,
       "learning_rate": 8.518518518518519e-06,
-      "loss": 0.0575,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 28672000,
       "step": 28000
     },
     {
       "epoch": 2.533333333333333,
-      "grad_norm": 0.8022651672363281,
+      "grad_norm": 1.2674041986465454,
       "learning_rate": 7.777777777777777e-06,
-      "loss": 0.0621,
+      "loss": 0.044,
+      "num_input_tokens_seen": 29184000,
       "step": 28500
     },
     {
       "epoch": 2.5777777777777775,
-      "grad_norm": 1.576392412185669,
+      "grad_norm": 1.236249327659607,
       "learning_rate": 7.0370370370370375e-06,
-      "loss": 0.0586,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 29696000,
       "step": 29000
     },
     {
       "epoch": 2.6222222222222222,
-      "grad_norm": 1.166567087173462,
+      "grad_norm": 1.3907105922698975,
       "learning_rate": 6.296296296296296e-06,
-      "loss": 0.0662,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 30208000,
       "step": 29500
     },
     {
       "epoch": 2.6666666666666665,
-      "grad_norm": 2.4423012733459473,
+      "grad_norm": 1.8549304008483887,
       "learning_rate": 5.555555555555556e-06,
-      "loss": 0.0645,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 30720000,
       "step": 30000
     },
     {
       "epoch": 2.7111111111111112,
-      "grad_norm": 3.1122899055480957,
+      "grad_norm": 1.3427213430404663,
       "learning_rate": 4.814814814814815e-06,
-      "loss": 0.0642,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 31232000,
       "step": 30500
     },
     {
       "epoch": 2.7555555555555555,
-      "grad_norm": 3.2840280532836914,
+      "grad_norm": 1.9282052516937256,
       "learning_rate": 4.074074074074075e-06,
-      "loss": 0.0617,
+      "loss": 0.046,
+      "num_input_tokens_seen": 31744000,
       "step": 31000
     },
     {
       "epoch": 2.8,
-      "grad_norm": 0.6285051703453064,
+      "grad_norm": 1.3202067613601685,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.0566,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 32256000,
       "step": 31500
     },
     {
       "epoch": 2.8444444444444446,
-      "grad_norm": 3.462080240249634,
+      "grad_norm": 0.9947272539138794,
       "learning_rate": 2.5925925925925925e-06,
-      "loss": 0.0614,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 32768000,
       "step": 32000
     },
     {
       "epoch": 2.888888888888889,
-      "grad_norm": 2.2216219902038574,
+      "grad_norm": 1.1175397634506226,
       "learning_rate": 1.8518518518518519e-06,
-      "loss": 0.0647,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 33280000,
       "step": 32500
     },
     {
       "epoch": 2.9333333333333336,
-      "grad_norm": 1.6926839351654053,
+      "grad_norm": 1.4308379888534546,
       "learning_rate": 1.1111111111111112e-06,
-      "loss": 0.0611,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 33792000,
       "step": 33000
     },
     {
       "epoch": 2.977777777777778,
-      "grad_norm": 4.723779678344727,
+      "grad_norm": 1.3511176109313965,
       "learning_rate": 3.703703703703704e-07,
-      "loss": 0.0597,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 34304000,
       "step": 33500
     },
     {
       "epoch": 3.0,
+      "num_input_tokens_seen": 34560000,
       "step": 33750,
       "total_flos": 4446488701440000.0,
-      "train_loss": 0.12363309427897136,
-      "train_runtime": 1375.441,
-      "train_samples_per_second": 196.301,
-      "train_steps_per_second": 24.538
+      "train_loss": 0.08211795973601164,
+      "train_runtime": 1433.1905,
+      "train_samples_per_second": 188.391,
+      "train_steps_per_second": 23.549,
+      "train_tokens_per_second": 24114.032
     }
   ],
   "logging_steps": 500,
   "max_steps": 33750,
-  "num_input_tokens_seen": 0,
+  "num_input_tokens_seen": 34560000,
   "num_train_epochs": 3,
   "save_steps": 500,
   "stateful_callbacks": {