End of training

README.md

@@ -3,6 +3,7 @@ library_name: transformers
 license: apache-2.0
 base_model: facebook/wav2vec2-base
 tags:
+- audio-classification
 - generated_from_trainer
 datasets:
 - superb
@@ -33,7 +34,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the superb dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.1206
+- Loss: 0.1244
 - Accuracy: 0.9815
 
 ## Model description

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 7.996245306633291,
-    "eval_accuracy": 0.9773462783171522,
-    "eval_loss": 0.11222562193870544,
-    "eval_runtime": 5.6077,
-    "eval_samples_per_second": 1212.264,
-    "eval_steps_per_second": 37.984,
+    "eval_accuracy": 0.9814651368049426,
+    "eval_loss": 0.12438357621431351,
+    "eval_runtime": 5.5465,
+    "eval_samples_per_second": 1225.644,
+    "eval_steps_per_second": 38.403,
     "total_flos": 3.777723239743488e+18,
-    "train_loss": 0.6284143277373745,
-    "train_runtime": 634.1755,
-    "train_samples_per_second": 644.541,
-    "train_steps_per_second": 2.51
+    "train_loss": 0.726146658611058,
+    "train_runtime": 635.8846,
+    "train_samples_per_second": 642.808,
+    "train_steps_per_second": 2.504
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 7.996245306633291,
-    "eval_accuracy": 0.9773462783171522,
-    "eval_loss": 0.11222562193870544,
-    "eval_runtime": 5.6077,
-    "eval_samples_per_second": 1212.264,
-    "eval_steps_per_second": 37.984
+    "eval_accuracy": 0.9814651368049426,
+    "eval_loss": 0.12438357621431351,
+    "eval_runtime": 5.5465,
+    "eval_samples_per_second": 1225.644,
+    "eval_steps_per_second": 38.403
 }

runs/May15_00-55-30_cs-Precision-7960-Tower/events.out.tfevents.1747285605.cs-Precision-7960-Tower.141414.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:377b4458bfcf1e26e5f48fc363b72592e31a5d1563296edc5979737602f7404c
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 7.996245306633291,
     "total_flos": 3.777723239743488e+18,
-    "train_loss": 0.6284143277373745,
-    "train_runtime": 634.1755,
-    "train_samples_per_second": 644.541,
-    "train_steps_per_second": 2.51
+    "train_loss": 0.726146658611058,
+    "train_runtime": 635.8846,
+    "train_samples_per_second": 642.808,
+    "train_steps_per_second": 2.504
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.9773462783171522,
-  "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1592",
+  "best_metric": 0.9814651368049426,
+  "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1393",
   "epoch": 7.996245306633291,
   "eval_steps": 500,
   "global_step": 1592,
@@ -10,1197 +10,1197 @@
   "log_history": [
     {
       "epoch": 0.05006257822277847,
-      "grad_norm": 2.5059146881103516,
-      "learning_rate": 3.125e-06,
-      "loss": 4.1393,
+      "grad_norm": 2.2161571979522705,
+      "learning_rate": 1.25e-06,
+      "loss": 4.0993,
       "step": 10
     },
     {
       "epoch": 0.10012515644555695,
-      "grad_norm": 3.3145341873168945,
-      "learning_rate": 6.25e-06,
-      "loss": 4.1209,
+      "grad_norm": 3.068307399749756,
+      "learning_rate": 2.5e-06,
+      "loss": 4.1801,
       "step": 20
     },
     {
       "epoch": 0.15018773466833543,
-      "grad_norm": 3.9769318103790283,
-      "learning_rate": 9.375000000000001e-06,
-      "loss": 3.8903,
+      "grad_norm": 3.121358871459961,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 4.11,
       "step": 30
     },
     {
       "epoch": 0.2002503128911139,
-      "grad_norm": 5.446993350982666,
-      "learning_rate": 1.25e-05,
-      "loss": 3.4194,
+      "grad_norm": 3.456892728805542,
+      "learning_rate": 5e-06,
+      "loss": 3.9697,
       "step": 40
     },
     {
       "epoch": 0.2503128911138924,
-      "grad_norm": 5.676177978515625,
-      "learning_rate": 1.5625e-05,
-      "loss": 2.7891,
+      "grad_norm": 4.699773788452148,
+      "learning_rate": 6.25e-06,
+      "loss": 3.7271,
       "step": 50
     },
     {
       "epoch": 0.30037546933667086,
-      "grad_norm": 5.550710678100586,
-      "learning_rate": 1.8750000000000002e-05,
-      "loss": 2.3951,
+      "grad_norm": 6.0911078453063965,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.3436,
       "step": 60
     },
     {
       "epoch": 0.3504380475594493,
-      "grad_norm": 4.72838020324707,
-      "learning_rate": 2.1875e-05,
-      "loss": 2.1523,
+      "grad_norm": 5.996876239776611,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.8803,
       "step": 70
     },
     {
       "epoch": 0.4005006257822278,
-      "grad_norm": 3.6451964378356934,
-      "learning_rate": 2.5e-05,
-      "loss": 1.9811,
+      "grad_norm": 5.593225479125977,
+      "learning_rate": 1e-05,
+      "loss": 2.54,
       "step": 80
     },
     {
       "epoch": 0.45056320400500627,
-      "grad_norm": 2.2974889278411865,
-      "learning_rate": 2.8125000000000003e-05,
-      "loss": 1.8507,
+      "grad_norm": 5.177072525024414,
+      "learning_rate": 1.125e-05,
+      "loss": 2.326,
       "step": 90
     },
     {
       "epoch": 0.5006257822277848,
-      "grad_norm": 0.8505929708480835,
-      "learning_rate": 3.125e-05,
-      "loss": 1.7874,
+      "grad_norm": 4.610163688659668,
+      "learning_rate": 1.25e-05,
+      "loss": 2.1907,
       "step": 100
     },
     {
       "epoch": 0.5506883604505632,
-      "grad_norm": 0.5304011702537537,
-      "learning_rate": 3.4375e-05,
-      "loss": 1.7174,
+      "grad_norm": 4.401670932769775,
+      "learning_rate": 1.375e-05,
+      "loss": 2.054,
       "step": 110
     },
     {
       "epoch": 0.6007509386733417,
-      "grad_norm": 1.0286344289779663,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 1.7567,
+      "grad_norm": 3.190901041030884,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.994,
       "step": 120
     },
     {
       "epoch": 0.6508135168961201,
-      "grad_norm": 0.9384429454803467,
-      "learning_rate": 4.0625000000000005e-05,
-      "loss": 1.7851,
+      "grad_norm": 2.314962148666382,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.9256,
       "step": 130
     },
     {
       "epoch": 0.7008760951188986,
-      "grad_norm": 2.4175920486450195,
-      "learning_rate": 4.375e-05,
-      "loss": 1.6505,
+      "grad_norm": 2.310375690460205,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.7774,
       "step": 140
     },
     {
       "epoch": 0.7509386733416771,
-      "grad_norm": 1.0501582622528076,
-      "learning_rate": 4.6875e-05,
-      "loss": 1.672,
+      "grad_norm": 1.4242396354675293,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.7961,
       "step": 150
     },
     {
       "epoch": 0.8010012515644556,
-      "grad_norm": 1.9757583141326904,
-      "learning_rate": 5e-05,
-      "loss": 1.6418,
+      "grad_norm": 0.6981024742126465,
+      "learning_rate": 2e-05,
+      "loss": 1.8108,
       "step": 160
     },
     {
       "epoch": 0.851063829787234,
-      "grad_norm": 2.322604179382324,
-      "learning_rate": 4.965083798882682e-05,
-      "loss": 1.5421,
+      "grad_norm": 0.7541513442993164,
+      "learning_rate": 1.9860335195530728e-05,
+      "loss": 1.733,
       "step": 170
     },
     {
       "epoch": 0.9011264080100125,
-      "grad_norm": 1.9013667106628418,
-      "learning_rate": 4.9301675977653635e-05,
-      "loss": 1.5342,
+      "grad_norm": 1.1699227094650269,
+      "learning_rate": 1.9720670391061455e-05,
+      "loss": 1.759,
       "step": 180
     },
     {
       "epoch": 0.951188986232791,
-      "grad_norm": 2.9960286617279053,
-      "learning_rate": 4.8952513966480454e-05,
-      "loss": 1.4829,
+      "grad_norm": 2.1834521293640137,
+      "learning_rate": 1.958100558659218e-05,
+      "loss": 1.7191,
       "step": 190
     },
     {
       "epoch": 0.9962453066332916,
-      "eval_accuracy": 0.6456310679611651,
-      "eval_loss": 1.292258620262146,
-      "eval_runtime": 5.0432,
-      "eval_samples_per_second": 1347.951,
-      "eval_steps_per_second": 42.235,
+      "eval_accuracy": 0.6209179170344219,
+      "eval_loss": 1.5815445184707642,
+      "eval_runtime": 4.9783,
+      "eval_samples_per_second": 1365.528,
+      "eval_steps_per_second": 42.786,
       "step": 199
     },
     {
       "epoch": 1.0050062578222778,
-      "grad_norm": 2.5933773517608643,
-      "learning_rate": 4.860335195530727e-05,
-      "loss": 1.4948,
+      "grad_norm": 3.1251320838928223,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 1.7482,
       "step": 200
     },
     {
       "epoch": 1.0550688360450564,
-      "grad_norm": 2.091658115386963,
-      "learning_rate": 4.825418994413408e-05,
-      "loss": 1.3361,
+      "grad_norm": 4.351515293121338,
+      "learning_rate": 1.9301675977653634e-05,
+      "loss": 1.5606,
       "step": 210
     },
     {
       "epoch": 1.1051314142678348,
-      "grad_norm": 7.571577548980713,
-      "learning_rate": 4.790502793296089e-05,
-      "loss": 1.4043,
+      "grad_norm": 2.624279022216797,
+      "learning_rate": 1.9162011173184357e-05,
+      "loss": 1.5657,
       "step": 220
     },
     {
       "epoch": 1.1551939924906134,
-      "grad_norm": 1.8106225728988647,
-      "learning_rate": 4.755586592178771e-05,
-      "loss": 1.2421,
+      "grad_norm": 2.795576333999634,
+      "learning_rate": 1.9022346368715087e-05,
+      "loss": 1.4578,
       "step": 230
     },
     {
       "epoch": 1.2052565707133918,
-      "grad_norm": 1.4783036708831787,
-      "learning_rate": 4.7206703910614525e-05,
-      "loss": 1.2293,
+      "grad_norm": 2.7234373092651367,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 1.449,
       "step": 240
     },
     {
       "epoch": 1.2553191489361701,
-      "grad_norm": 3.992377996444702,
-      "learning_rate": 4.6857541899441344e-05,
-      "loss": 1.1909,
+      "grad_norm": 5.743226528167725,
+      "learning_rate": 1.8743016759776536e-05,
+      "loss": 1.4041,
       "step": 250
     },
     {
       "epoch": 1.3053817271589487,
-      "grad_norm": 3.2013051509857178,
-      "learning_rate": 4.6508379888268164e-05,
-      "loss": 1.182,
+      "grad_norm": 6.164392471313477,
+      "learning_rate": 1.8603351955307266e-05,
+      "loss": 1.3667,
       "step": 260
     },
     {
       "epoch": 1.355444305381727,
-      "grad_norm": 2.9630327224731445,
-      "learning_rate": 4.615921787709498e-05,
-      "loss": 1.1559,
+      "grad_norm": 3.474940299987793,
+      "learning_rate": 1.846368715083799e-05,
+      "loss": 1.2949,
       "step": 270
     },
     {
       "epoch": 1.4055068836045057,
-      "grad_norm": 3.330141544342041,
-      "learning_rate": 4.581005586592179e-05,
-      "loss": 1.1693,
+      "grad_norm": 4.262696266174316,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 1.2814,
       "step": 280
     },
     {
       "epoch": 1.455569461827284,
-      "grad_norm": 2.7530879974365234,
-      "learning_rate": 4.54608938547486e-05,
-      "loss": 1.1137,
+      "grad_norm": 7.003787994384766,
+      "learning_rate": 1.8184357541899442e-05,
+      "loss": 1.2096,
       "step": 290
     },
     {
       "epoch": 1.5056320400500627,
-      "grad_norm": 1.702641248703003,
-      "learning_rate": 4.511173184357542e-05,
-      "loss": 1.1002,
+      "grad_norm": 3.957937240600586,
+      "learning_rate": 1.804469273743017e-05,
+      "loss": 1.183,
       "step": 300
     },
     {
       "epoch": 1.555694618272841,
-      "grad_norm": 2.645132303237915,
-      "learning_rate": 4.4762569832402234e-05,
-      "loss": 1.1707,
+      "grad_norm": 4.9982991218566895,
+      "learning_rate": 1.7905027932960895e-05,
+      "loss": 1.2105,
       "step": 310
     },
     {
       "epoch": 1.6057571964956194,
-      "grad_norm": 2.5694878101348877,
-      "learning_rate": 4.4413407821229054e-05,
-      "loss": 1.0454,
+      "grad_norm": 3.900266647338867,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 1.1167,
       "step": 320
     },
     {
       "epoch": 1.655819774718398,
-      "grad_norm": 3.6792080402374268,
-      "learning_rate": 4.406424581005587e-05,
-      "loss": 1.0544,
+      "grad_norm": 4.397927284240723,
+      "learning_rate": 1.7625698324022348e-05,
+      "loss": 1.144,
       "step": 330
     },
     {
       "epoch": 1.7058823529411766,
-      "grad_norm": 6.041355609893799,
-      "learning_rate": 4.3715083798882686e-05,
-      "loss": 1.0394,
+      "grad_norm": 3.0843544006347656,
+      "learning_rate": 1.7486033519553075e-05,
+      "loss": 1.101,
       "step": 340
     },
     {
       "epoch": 1.7559449311639548,
-      "grad_norm": 3.808980941772461,
-      "learning_rate": 4.33659217877095e-05,
-      "loss": 0.9862,
+      "grad_norm": 3.6377310752868652,
+      "learning_rate": 1.73463687150838e-05,
+      "loss": 1.0482,
       "step": 350
     },
     {
       "epoch": 1.8060075093867334,
-      "grad_norm": 2.6217703819274902,
-      "learning_rate": 4.301675977653631e-05,
-      "loss": 0.9754,
+      "grad_norm": 2.659217357635498,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 1.0428,
       "step": 360
     },
     {
       "epoch": 1.856070087609512,
-      "grad_norm": 3.5803720951080322,
-      "learning_rate": 4.266759776536313e-05,
-      "loss": 0.932,
+      "grad_norm": 4.358920097351074,
+      "learning_rate": 1.7067039106145254e-05,
+      "loss": 1.0483,
       "step": 370
     },
     {
       "epoch": 1.9061326658322904,
-      "grad_norm": 2.712857961654663,
-      "learning_rate": 4.2318435754189944e-05,
-      "loss": 0.8815,
+      "grad_norm": 4.7584452629089355,
+      "learning_rate": 1.6927374301675977e-05,
+      "loss": 0.9886,
       "step": 380
     },
     {
       "epoch": 1.9561952440550687,
-      "grad_norm": 4.25401496887207,
-      "learning_rate": 4.1969273743016764e-05,
-      "loss": 0.8844,
+      "grad_norm": 3.638371229171753,
+      "learning_rate": 1.6787709497206707e-05,
+      "loss": 1.0088,
       "step": 390
     },
     {
       "epoch": 1.9962453066332917,
-      "eval_accuracy": 0.7917034421888791,
-      "eval_loss": 0.6811065673828125,
-      "eval_runtime": 4.8549,
-      "eval_samples_per_second": 1400.226,
-      "eval_steps_per_second": 43.873,
+      "eval_accuracy": 0.83480435422183,
+      "eval_loss": 0.9595487713813782,
+      "eval_runtime": 5.0807,
+      "eval_samples_per_second": 1338.014,
+      "eval_steps_per_second": 41.924,
       "step": 398
     },
     {
       "epoch": 2.0100125156445556,
-      "grad_norm": 3.6374008655548096,
-      "learning_rate": 4.1620111731843576e-05,
-      "loss": 0.8524,
+      "grad_norm": 2.447006940841675,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 1.0295,
       "step": 400
     },
     {
       "epoch": 2.0600750938673342,
-      "grad_norm": 5.184952259063721,
-      "learning_rate": 4.1270949720670396e-05,
-      "loss": 0.8132,
+      "grad_norm": 3.891425132751465,
+      "learning_rate": 1.6508379888268156e-05,
+      "loss": 0.9607,
       "step": 410
     },
     {
       "epoch": 2.110137672090113,
-      "grad_norm": 4.079555988311768,
-      "learning_rate": 4.092178770949721e-05,
-      "loss": 0.7761,
+      "grad_norm": 3.9283361434936523,
+      "learning_rate": 1.6368715083798886e-05,
+      "loss": 0.9797,
       "step": 420
     },
     {
       "epoch": 2.160200250312891,
-      "grad_norm": 2.930490016937256,
-      "learning_rate": 4.057262569832402e-05,
-      "loss": 0.7665,
+      "grad_norm": 3.9875011444091797,
+      "learning_rate": 1.622905027932961e-05,
+      "loss": 0.923,
       "step": 430
     },
     {
       "epoch": 2.2102628285356696,
-      "grad_norm": 3.523425579071045,
-      "learning_rate": 4.022346368715084e-05,
-      "loss": 0.6984,
+      "grad_norm": 2.9517266750335693,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 0.872,
       "step": 440
     },
     {
       "epoch": 2.260325406758448,
-      "grad_norm": 2.8467819690704346,
-      "learning_rate": 3.9874301675977654e-05,
-      "loss": 0.6988,
+      "grad_norm": 6.958038330078125,
+      "learning_rate": 1.5949720670391062e-05,
+      "loss": 0.8477,
       "step": 450
     },
     {
       "epoch": 2.3103879849812268,
-      "grad_norm": 3.579084634780884,
-      "learning_rate": 3.952513966480447e-05,
-      "loss": 0.6424,
+      "grad_norm": 2.592712640762329,
+      "learning_rate": 1.581005586592179e-05,
+      "loss": 0.803,
       "step": 460
     },
     {
       "epoch": 2.360450563204005,
-      "grad_norm": 4.213118076324463,
-      "learning_rate": 3.9175977653631286e-05,
-      "loss": 0.607,
+      "grad_norm": 3.0653076171875,
+      "learning_rate": 1.5670391061452515e-05,
+      "loss": 0.7824,
       "step": 470
     },
     {
       "epoch": 2.4105131414267835,
-      "grad_norm": 2.531745672225952,
-      "learning_rate": 3.8826815642458106e-05,
-      "loss": 0.608,
+      "grad_norm": 4.134679794311523,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 0.7681,
       "step": 480
     },
     {
       "epoch": 2.460575719649562,
-      "grad_norm": 2.7486679553985596,
-      "learning_rate": 3.847765363128492e-05,
-      "loss": 0.5296,
+      "grad_norm": 2.6600096225738525,
+      "learning_rate": 1.5391061452513968e-05,
+      "loss": 0.6916,
       "step": 490
     },
     {
       "epoch": 2.5106382978723403,
-      "grad_norm": 4.085735321044922,
-      "learning_rate": 3.812849162011173e-05,
-      "loss": 0.5622,
+      "grad_norm": 2.778372287750244,
+      "learning_rate": 1.5251396648044694e-05,
+      "loss": 0.7158,
       "step": 500
     },
     {
       "epoch": 2.560700876095119,
-      "grad_norm": 4.96565580368042,
-      "learning_rate": 3.777932960893855e-05,
-      "loss": 0.524,
+      "grad_norm": 2.7881994247436523,
+      "learning_rate": 1.5111731843575421e-05,
+      "loss": 0.689,
       "step": 510
     },
     {
       "epoch": 2.6107634543178975,
-      "grad_norm": 4.465268611907959,
-      "learning_rate": 3.743016759776536e-05,
-      "loss": 0.4994,
+      "grad_norm": 3.771796226501465,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 0.6383,
       "step": 520
     },
     {
       "epoch": 2.660826032540676,
-      "grad_norm": 3.813931465148926,
-      "learning_rate": 3.708100558659218e-05,
-      "loss": 0.5042,
+      "grad_norm": 3.580981731414795,
+      "learning_rate": 1.4832402234636874e-05,
+      "loss": 0.6086,
       "step": 530
     },
     {
       "epoch": 2.710888610763454,
-      "grad_norm": 2.1845741271972656,
-      "learning_rate": 3.6731843575418996e-05,
-      "loss": 0.4694,
+      "grad_norm": 2.298293352127075,
+      "learning_rate": 1.4692737430167599e-05,
+      "loss": 0.581,
       "step": 540
     },
     {
       "epoch": 2.760951188986233,
-      "grad_norm": 5.71811580657959,
-      "learning_rate": 3.6382681564245815e-05,
-      "loss": 0.4558,
+      "grad_norm": 3.767306327819824,
+      "learning_rate": 1.4553072625698325e-05,
+      "loss": 0.5556,
       "step": 550
     },
     {
       "epoch": 2.8110137672090114,
-      "grad_norm": 4.474643230438232,
-      "learning_rate": 3.603351955307263e-05,
-      "loss": 0.4703,
+      "grad_norm": 3.712104558944702,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 0.5637,
       "step": 560
     },
     {
       "epoch": 2.8610763454317896,
-      "grad_norm": 3.506659746170044,
-      "learning_rate": 3.568435754189944e-05,
-      "loss": 0.4196,
+      "grad_norm": 2.500260829925537,
+      "learning_rate": 1.4273743016759778e-05,
+      "loss": 0.525,
       "step": 570
     },
     {
       "epoch": 2.911138923654568,
-      "grad_norm": 3.286609172821045,
-      "learning_rate": 3.5335195530726253e-05,
-      "loss": 0.433,
+      "grad_norm": 3.4421885013580322,
+      "learning_rate": 1.4134078212290503e-05,
+      "loss": 0.5235,
       "step": 580
     },
     {
       "epoch": 2.9612015018773468,
-      "grad_norm": 3.182314872741699,
-      "learning_rate": 3.498603351955307e-05,
-      "loss": 0.4241,
+      "grad_norm": 3.3740108013153076,
+      "learning_rate": 1.3994413407821231e-05,
+      "loss": 0.4964,
       "step": 590
     },
     {
       "epoch": 2.9962453066332917,
-      "eval_accuracy": 0.9267431597528685,
-      "eval_loss": 0.2847038805484772,
-      "eval_runtime": 4.8683,
-      "eval_samples_per_second": 1396.372,
-      "eval_steps_per_second": 43.752,
+      "eval_accuracy": 0.9727861135628126,
+      "eval_loss": 0.37298157811164856,
+      "eval_runtime": 4.944,
+      "eval_samples_per_second": 1374.987,
+      "eval_steps_per_second": 43.082,
       "step": 597
     },
     {
       "epoch": 3.0150187734668337,
-      "grad_norm": 5.002447128295898,
-      "learning_rate": 3.463687150837989e-05,
-      "loss": 0.4553,
+      "grad_norm": 5.472856521606445,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 0.5243,
       "step": 600
     },
     {
       "epoch": 3.065081351689612,
-      "grad_norm": 4.032111644744873,
-      "learning_rate": 3.4287709497206705e-05,
-      "loss": 0.384,
+      "grad_norm": 3.919224977493286,
+      "learning_rate": 1.3715083798882682e-05,
+      "loss": 0.4697,
       "step": 610
     },
     {
       "epoch": 3.1151439299123904,
-      "grad_norm": 1.9067034721374512,
-      "learning_rate": 3.3938547486033525e-05,
-      "loss": 0.3805,
+      "grad_norm": 2.8592793941497803,
+      "learning_rate": 1.357541899441341e-05,
+      "loss": 0.4581,
       "step": 620
     },
     {
       "epoch": 3.165206508135169,
-      "grad_norm": 3.4726545810699463,
-      "learning_rate": 3.358938547486034e-05,
-      "loss": 0.4047,
+      "grad_norm": 3.517875909805298,
+      "learning_rate": 1.3435754189944135e-05,
+      "loss": 0.4609,
       "step": 630
     },
     {
       "epoch": 3.2152690863579476,
-      "grad_norm": 2.600872039794922,
-      "learning_rate": 3.324022346368715e-05,
-      "loss": 0.3692,
+      "grad_norm": 3.0219643115997314,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 0.4251,
       "step": 640
     },
     {
       "epoch": 3.2653316645807258,
-      "grad_norm": 2.484164237976074,
-      "learning_rate": 3.289106145251396e-05,
-      "loss": 0.3144,
+      "grad_norm": 4.166126728057861,
+      "learning_rate": 1.3156424581005586e-05,
+      "loss": 0.4026,
       "step": 650
     },
     {
       "epoch": 3.3153942428035044,
-      "grad_norm": 2.312174081802368,
-      "learning_rate": 3.254189944134078e-05,
-      "loss": 0.3164,
+      "grad_norm": 2.9157118797302246,
+      "learning_rate": 1.3016759776536314e-05,
+      "loss": 0.4101,
       "step": 660
     },
     {
       "epoch": 3.365456821026283,
-      "grad_norm": 3.2117793560028076,
-      "learning_rate": 3.2192737430167595e-05,
-      "loss": 0.3156,
+      "grad_norm": 2.605928897857666,
+      "learning_rate": 1.287709497206704e-05,
+      "loss": 0.3951,
       "step": 670
     },
     {
       "epoch": 3.415519399249061,
-      "grad_norm": 3.4549670219421387,
-      "learning_rate": 3.1843575418994415e-05,
-      "loss": 0.3241,
+      "grad_norm": 2.282757043838501,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 0.3876,
       "step": 680
     },
     {
       "epoch": 3.4655819774718397,
-      "grad_norm": 2.3556506633758545,
-      "learning_rate": 3.1494413407821234e-05,
-      "loss": 0.3257,
+      "grad_norm": 2.8429336547851562,
+      "learning_rate": 1.2597765363128494e-05,
+      "loss": 0.3735,
       "step": 690
     },
     {
       "epoch": 3.5156445556946183,
-      "grad_norm": 2.6974313259124756,
-      "learning_rate": 3.114525139664805e-05,
-      "loss": 0.3244,
+      "grad_norm": 3.3356730937957764,
+      "learning_rate": 1.2458100558659219e-05,
+      "loss": 0.3873,
       "step": 700
     },
     {
       "epoch": 3.565707133917397,
-      "grad_norm": 1.642268419265747,
-      "learning_rate": 3.079608938547486e-05,
-      "loss": 0.2919,
+      "grad_norm": 3.1262238025665283,
+      "learning_rate": 1.2318435754189945e-05,
+      "loss": 0.3572,
       "step": 710
     },
     {
       "epoch": 3.615769712140175,
-      "grad_norm": 3.0940229892730713,
-      "learning_rate": 3.0446927374301676e-05,
-      "loss": 0.2842,
+      "grad_norm": 3.589921712875366,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 0.3558,
       "step": 720
     },
     {
       "epoch": 3.6658322903629537,
-      "grad_norm": 2.1800785064697266,
-      "learning_rate": 3.0097765363128492e-05,
-      "loss": 0.2938,
+      "grad_norm": 3.085702896118164,
+      "learning_rate": 1.2039106145251398e-05,
+      "loss": 0.3499,
       "step": 730
     },
     {
       "epoch": 3.7158948685857323,
-      "grad_norm": 3.684602737426758,
-      "learning_rate": 2.9748603351955305e-05,
-      "loss": 0.2789,
+      "grad_norm": 3.2283403873443604,
+      "learning_rate": 1.1899441340782123e-05,
+      "loss": 0.3428,
       "step": 740
     },
     {
       "epoch": 3.7659574468085104,
-      "grad_norm": 4.337268829345703,
-      "learning_rate": 2.9399441340782125e-05,
-      "loss": 0.2898,
+      "grad_norm": 2.824887990951538,
+      "learning_rate": 1.175977653631285e-05,
+      "loss": 0.3399,
       "step": 750
     },
     {
       "epoch": 3.816020025031289,
-      "grad_norm": 2.6104233264923096,
-      "learning_rate": 2.9050279329608944e-05,
-      "loss": 0.2588,
+      "grad_norm": 3.595198392868042,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 0.3264,
       "step": 760
     },
     {
       "epoch": 3.8660826032540676,
-      "grad_norm": 3.2375924587249756,
-      "learning_rate": 2.8701117318435757e-05,
-      "loss": 0.2831,
+      "grad_norm": 4.541857719421387,
+      "learning_rate": 1.1480446927374302e-05,
+      "loss": 0.3322,
       "step": 770
     },
     {
       "epoch": 3.916145181476846,
-      "grad_norm": 2.4927661418914795,
-      "learning_rate": 2.8351955307262573e-05,
-      "loss": 0.2541,
+      "grad_norm": 3.149718999862671,
+      "learning_rate": 1.134078212290503e-05,
+      "loss": 0.3032,
       "step": 780
     },
     {
       "epoch": 3.966207759699625,
-      "grad_norm": 3.670045852661133,
-      "learning_rate": 2.8002793296089386e-05,
-      "loss": 0.2724,
+      "grad_norm": 3.3817384243011475,
+      "learning_rate": 1.1201117318435755e-05,
+      "loss": 0.3263,
       "step": 790
     },
     {
       "epoch": 3.9962453066332917,
-      "eval_accuracy": 0.973080317740512,
-      "eval_loss": 0.15457233786582947,
-      "eval_runtime": 4.7958,
-      "eval_samples_per_second": 1417.478,
-      "eval_steps_per_second": 44.413,
+      "eval_accuracy": 0.9783759929390997,
+      "eval_loss": 0.21611753106117249,
+      "eval_runtime": 4.9235,
+      "eval_samples_per_second": 1380.729,
+      "eval_steps_per_second": 43.262,
       "step": 796
     },
     {
       "epoch": 4.020025031289111,
-      "grad_norm": 2.9852592945098877,
-      "learning_rate": 2.7653631284916205e-05,
-      "loss": 0.2705,
+      "grad_norm": 3.4890668392181396,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 0.3262,
       "step": 800
     },
     {
       "epoch": 4.07008760951189,
-      "grad_norm": 2.433262825012207,
-      "learning_rate": 2.7304469273743018e-05,
-      "loss": 0.2544,
+      "grad_norm": 2.202880382537842,
+      "learning_rate": 1.0921787709497206e-05,
+      "loss": 0.3101,
       "step": 810
     },
     {
       "epoch": 4.1201501877346685,
-      "grad_norm": 3.54846453666687,
-      "learning_rate": 2.6955307262569834e-05,
-      "loss": 0.2586,
+      "grad_norm": 2.4988725185394287,
+      "learning_rate": 1.0782122905027934e-05,
+      "loss": 0.3022,
       "step": 820
     },
     {
       "epoch": 4.170212765957447,
-      "grad_norm": 2.550708770751953,
-      "learning_rate": 2.6606145251396647e-05,
-      "loss": 0.2406,
+      "grad_norm": 3.495452404022217,
+      "learning_rate": 1.0642458100558659e-05,
+      "loss": 0.2997,
       "step": 830
     },
     {
       "epoch": 4.220275344180226,
-      "grad_norm": 4.386864185333252,
-      "learning_rate": 2.6256983240223466e-05,
-      "loss": 0.2686,
+      "grad_norm": 2.8825905323028564,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 0.3182,
       "step": 840
     },
     {
       "epoch": 4.270337922403003,
-      "grad_norm": 2.19413161277771,
-      "learning_rate": 2.5907821229050283e-05,
-      "loss": 0.2452,
+      "grad_norm": 2.3349449634552,
+      "learning_rate": 1.0363128491620114e-05,
+      "loss": 0.2884,
       "step": 850
     },
     {
       "epoch": 4.320400500625782,
-      "grad_norm": 2.9501962661743164,
-      "learning_rate": 2.5558659217877095e-05,
-      "loss": 0.2245,
+      "grad_norm": 3.996652603149414,
+      "learning_rate": 1.0223463687150838e-05,
+      "loss": 0.29,
       "step": 860
     },
     {
       "epoch": 4.370463078848561,
-      "grad_norm": 3.0304136276245117,
-      "learning_rate": 2.5209497206703915e-05,
-      "loss": 0.2344,
+      "grad_norm": 1.9067742824554443,
+      "learning_rate": 1.0083798882681567e-05,
+      "loss": 0.2703,
       "step": 870
     },
     {
       "epoch": 4.420525657071339,
-      "grad_norm": 2.301668167114258,
-      "learning_rate": 2.4860335195530728e-05,
-      "loss": 0.2682,
+      "grad_norm": 3.410153388977051,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.2864,
       "step": 880
     },
     {
       "epoch": 4.470588235294118,
-      "grad_norm": 2.6299359798431396,
-      "learning_rate": 2.4511173184357544e-05,
-      "loss": 0.2461,
+      "grad_norm": 2.604459285736084,
+      "learning_rate": 9.804469273743018e-06,
+      "loss": 0.2814,
       "step": 890
     },
     {
       "epoch": 4.520650813516896,
-      "grad_norm": 1.7917237281799316,
-      "learning_rate": 2.416201117318436e-05,
-      "loss": 0.2071,
+      "grad_norm": 1.9956718683242798,
+      "learning_rate": 9.664804469273744e-06,
+      "loss": 0.2501,
       "step": 900
     },
     {
       "epoch": 4.570713391739675,
-      "grad_norm": 1.8918017148971558,
-      "learning_rate": 2.3812849162011176e-05,
-      "loss": 0.2219,
+      "grad_norm": 3.1284875869750977,
+      "learning_rate": 9.52513966480447e-06,
+      "loss": 0.2584,
       "step": 910
     },
     {
       "epoch": 4.6207759699624535,
-      "grad_norm": 2.058867931365967,
-      "learning_rate": 2.346368715083799e-05,
-      "loss": 0.2307,
+      "grad_norm": 3.3868837356567383,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 0.2486,
       "step": 920
     },
     {
       "epoch": 4.670838548185231,
-      "grad_norm": 1.6558533906936646,
-      "learning_rate": 2.3114525139664805e-05,
-      "loss": 0.2127,
+      "grad_norm": 2.5720443725585938,
+      "learning_rate": 9.245810055865922e-06,
+      "loss": 0.2624,
       "step": 930
     },
     {
       "epoch": 4.72090112640801,
-      "grad_norm": 1.8600987195968628,
-      "learning_rate": 2.276536312849162e-05,
-      "loss": 0.2245,
+      "grad_norm": 3.343977689743042,
+      "learning_rate": 9.106145251396648e-06,
+      "loss": 0.2506,
       "step": 940
     },
     {
       "epoch": 4.7709637046307884,
-      "grad_norm": 3.023834466934204,
-      "learning_rate": 2.2416201117318437e-05,
-      "loss": 0.254,
+      "grad_norm": 3.791302442550659,
+      "learning_rate": 8.966480446927375e-06,
+      "loss": 0.2712,
       "step": 950
     },
     {
       "epoch": 4.821026282853567,
-      "grad_norm": 3.464167833328247,
-      "learning_rate": 2.206703910614525e-05,
-      "loss": 0.2364,
+      "grad_norm": 3.8650593757629395,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 0.2508,
       "step": 960
     },
     {
       "epoch": 4.871088861076346,
-      "grad_norm": 2.8578057289123535,
-      "learning_rate": 2.171787709497207e-05,
-      "loss": 0.2414,
+      "grad_norm": 3.790891408920288,
+      "learning_rate": 8.687150837988828e-06,
+      "loss": 0.2653,
       "step": 970
     },
     {
       "epoch": 4.921151439299124,
-      "grad_norm": 1.949834942817688,
-      "learning_rate": 2.1368715083798886e-05,
-      "loss": 0.2368,
+      "grad_norm": 3.150134325027466,
+      "learning_rate": 8.547486033519554e-06,
+      "loss": 0.2431,
       "step": 980
     },
     {
       "epoch": 4.971214017521902,
-      "grad_norm": 3.3509461879730225,
-      "learning_rate": 2.10195530726257e-05,
-      "loss": 0.2362,
+      "grad_norm": 3.999809741973877,
+      "learning_rate": 8.40782122905028e-06,
+      "loss": 0.2512,
       "step": 990
     },
     {
       "epoch": 4.996245306633291,
-      "eval_accuracy": 0.9760223595175052,
-      "eval_loss": 0.1284540593624115,
-      "eval_runtime": 4.9228,
-      "eval_samples_per_second": 1380.926,
-      "eval_steps_per_second": 43.268,
+      "eval_accuracy": 0.979552809649897,
+      "eval_loss": 0.16172775626182556,
+      "eval_runtime": 5.0947,
+      "eval_samples_per_second": 1334.339,
+      "eval_steps_per_second": 41.809,
       "step": 995
     },
     {
       "epoch": 5.025031289111389,
-      "grad_norm": 2.664990186691284,
-      "learning_rate": 2.0670391061452515e-05,
-      "loss": 0.2298,
+      "grad_norm": 3.607999563217163,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 0.261,
       "step": 1000
     },
     {
       "epoch": 5.075093867334168,
-      "grad_norm": 1.4740562438964844,
-      "learning_rate": 2.032122905027933e-05,
-      "loss": 0.2125,
+      "grad_norm": 2.384503126144409,
+      "learning_rate": 8.128491620111732e-06,
+      "loss": 0.2301,
       "step": 1010
     },
     {
       "epoch": 5.1251564455569465,
-      "grad_norm": 3.666245698928833,
-      "learning_rate": 1.9972067039106147e-05,
-      "loss": 0.186,
+      "grad_norm": 3.3627846240997314,
+      "learning_rate": 7.988826815642458e-06,
+      "loss": 0.2363,
       "step": 1020
     },
     {
       "epoch": 5.175219023779725,
-      "grad_norm": 2.3000710010528564,
-      "learning_rate": 1.962290502793296e-05,
-      "loss": 0.2225,
+      "grad_norm": 3.3233959674835205,
+      "learning_rate": 7.849162011173185e-06,
+      "loss": 0.2275,
       "step": 1030
     },
     {
       "epoch": 5.225281602002503,
-      "grad_norm": 2.916823625564575,
-      "learning_rate": 1.9273743016759776e-05,
-      "loss": 0.1999,
+      "grad_norm": 3.21708083152771,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 0.2247,
       "step": 1040
     },
     {
       "epoch": 5.275344180225281,
-      "grad_norm": 2.1694273948669434,
-      "learning_rate": 1.8924581005586592e-05,
-      "loss": 0.1737,
+      "grad_norm": 2.4232187271118164,
+      "learning_rate": 7.569832402234637e-06,
+      "loss": 0.2044,
       "step": 1050
     },
     {
       "epoch": 5.32540675844806,
-      "grad_norm": 2.1091065406799316,
-      "learning_rate": 1.857541899441341e-05,
-      "loss": 0.1898,
+      "grad_norm": 2.9417362213134766,
+      "learning_rate": 7.430167597765364e-06,
+      "loss": 0.2156,
       "step": 1060
     },
     {
       "epoch": 5.375469336670839,
-      "grad_norm": 2.4666635990142822,
-      "learning_rate": 1.8226256983240224e-05,
-      "loss": 0.2115,
+      "grad_norm": 5.781579971313477,
+      "learning_rate": 7.290502793296091e-06,
+      "loss": 0.2479,
       "step": 1070
     },
     {
       "epoch": 5.425531914893617,
-      "grad_norm": 2.0962629318237305,
-      "learning_rate": 1.787709497206704e-05,
-      "loss": 0.2189,
+      "grad_norm": 2.417659044265747,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 0.2349,
       "step": 1080
     },
     {
       "epoch": 5.475594493116396,
-      "grad_norm": 1.7196861505508423,
-      "learning_rate": 1.7527932960893857e-05,
-      "loss": 0.2013,
+      "grad_norm": 3.229393482208252,
+      "learning_rate": 7.011173184357543e-06,
+      "loss": 0.218,
       "step": 1090
     },
     {
       "epoch": 5.5256570713391735,
-      "grad_norm": 2.7366960048675537,
-      "learning_rate": 1.7178770949720673e-05,
-      "loss": 0.2257,
+      "grad_norm": 3.5707740783691406,
+      "learning_rate": 6.871508379888268e-06,
+      "loss": 0.2287,
       "step": 1100
     },
     {
       "epoch": 5.575719649561952,
-      "grad_norm": 2.12593412399292,
-      "learning_rate": 1.6829608938547485e-05,
-      "loss": 0.2162,
+      "grad_norm": 2.99178147315979,
+      "learning_rate": 6.731843575418995e-06,
+      "loss": 0.2474,
       "step": 1110
     },
     {
       "epoch": 5.625782227784731,
-      "grad_norm": 1.8080042600631714,
-      "learning_rate": 1.64804469273743e-05,
-      "loss": 0.2295,
+      "grad_norm": 4.041738986968994,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 0.254,
       "step": 1120
     },
     {
       "epoch": 5.675844806007509,
-      "grad_norm": 2.192559242248535,
-      "learning_rate": 1.6131284916201118e-05,
-      "loss": 0.2077,
+      "grad_norm": 3.7862696647644043,
+      "learning_rate": 6.452513966480447e-06,
+      "loss": 0.2336,
       "step": 1130
     },
     {
       "epoch": 5.725907384230288,
-      "grad_norm": 2.823213577270508,
-      "learning_rate": 1.5782122905027934e-05,
-      "loss": 0.2017,
+      "grad_norm": 3.772305488586426,
+      "learning_rate": 6.312849162011173e-06,
+      "loss": 0.2366,
       "step": 1140
     },
     {
       "epoch": 5.7759699624530665,
-      "grad_norm": 3.1835837364196777,
-      "learning_rate": 1.543296089385475e-05,
-      "loss": 0.1741,
+      "grad_norm": 3.721914291381836,
+      "learning_rate": 6.173184357541901e-06,
+      "loss": 0.2056,
       "step": 1150
     },
     {
       "epoch": 5.826032540675845,
-      "grad_norm": 3.540625810623169,
-      "learning_rate": 1.5083798882681566e-05,
-      "loss": 0.1902,
+      "grad_norm": 3.321104049682617,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 0.2083,
       "step": 1160
     },
     {
       "epoch": 5.876095118898624,
-      "grad_norm": 2.113506555557251,
-      "learning_rate": 1.473463687150838e-05,
-      "loss": 0.2406,
+      "grad_norm": 2.396998405456543,
+      "learning_rate": 5.893854748603353e-06,
+      "loss": 0.2413,
       "step": 1170
     },
     {
       "epoch": 5.926157697121401,
-      "grad_norm": 3.060197353363037,
-      "learning_rate": 1.4385474860335197e-05,
-      "loss": 0.1803,
+      "grad_norm": 2.6875176429748535,
+      "learning_rate": 5.754189944134078e-06,
+      "loss": 0.2049,
       "step": 1180
     },
     {
       "epoch": 5.97622027534418,
-      "grad_norm": 1.7786002159118652,
-      "learning_rate": 1.4036312849162011e-05,
-      "loss": 0.1729,
+      "grad_norm": 3.377181053161621,
+      "learning_rate": 5.614525139664805e-06,
+      "loss": 0.214,
       "step": 1190
     },
     {
       "epoch": 5.996245306633291,
-      "eval_accuracy": 0.9748455428067079,
-      "eval_loss": 0.12367700785398483,
-      "eval_runtime": 5.057,
-      "eval_samples_per_second": 1344.286,
-      "eval_steps_per_second": 42.12,
+      "eval_accuracy": 0.9807296263606943,
+      "eval_loss": 0.13626398146152496,
+      "eval_runtime": 4.9535,
+      "eval_samples_per_second": 1372.375,
+      "eval_steps_per_second": 43.0,
       "step": 1194
     },
     {
       "epoch": 6.030037546933667,
-      "grad_norm": 1.7793939113616943,
-      "learning_rate": 1.3687150837988827e-05,
-      "loss": 0.2018,
+      "grad_norm": 2.5127053260803223,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 0.2374,
       "step": 1200
     },
     {
       "epoch": 6.080100125156446,
-      "grad_norm": 2.2340643405914307,
-      "learning_rate": 1.3337988826815642e-05,
-      "loss": 0.1695,
+      "grad_norm": 2.9415111541748047,
+      "learning_rate": 5.335195530726257e-06,
+      "loss": 0.203,
       "step": 1210
     },
     {
       "epoch": 6.130162703379224,
-      "grad_norm": 1.621952772140503,
-      "learning_rate": 1.2988826815642458e-05,
-      "loss": 0.1791,
+      "grad_norm": 1.5115219354629517,
+      "learning_rate": 5.195530726256983e-06,
+      "loss": 0.2151,
       "step": 1220
     },
     {
       "epoch": 6.180225281602002,
-      "grad_norm": 2.7437918186187744,
-      "learning_rate": 1.2639664804469276e-05,
-      "loss": 0.1956,
+      "grad_norm": 1.8936338424682617,
+      "learning_rate": 5.055865921787711e-06,
+      "loss": 0.2159,
       "step": 1230
     },
     {
       "epoch": 6.230287859824781,
-      "grad_norm": 2.8166093826293945,
-      "learning_rate": 1.229050279329609e-05,
-      "loss": 0.1882,
+      "grad_norm": 2.6654956340789795,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 0.2013,
       "step": 1240
     },
     {
       "epoch": 6.280350438047559,
-      "grad_norm": 1.636942982673645,
-      "learning_rate": 1.1941340782122906e-05,
-      "loss": 0.186,
+      "grad_norm": 3.0140798091888428,
+      "learning_rate": 4.776536312849163e-06,
+      "loss": 0.1985,
       "step": 1250
     },
     {
       "epoch": 6.330413016270338,
-      "grad_norm": 2.7344369888305664,
-      "learning_rate": 1.1592178770949721e-05,
-      "loss": 0.1791,
+      "grad_norm": 3.579763412475586,
+      "learning_rate": 4.636871508379888e-06,
+      "loss": 0.207,
       "step": 1260
     },
     {
       "epoch": 6.380475594493117,
-      "grad_norm": 2.310530662536621,
-      "learning_rate": 1.1243016759776537e-05,
-      "loss": 0.1743,
+      "grad_norm": 2.5355074405670166,
+      "learning_rate": 4.497206703910615e-06,
+      "loss": 0.193,
       "step": 1270
     },
     {
       "epoch": 6.430538172715895,
-      "grad_norm": 2.937742233276367,
-      "learning_rate": 1.0893854748603351e-05,
-      "loss": 0.1898,
+      "grad_norm": 2.3967621326446533,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 0.2139,
       "step": 1280
     },
     {
       "epoch": 6.480600750938673,
-      "grad_norm": 2.7743396759033203,
-      "learning_rate": 1.054469273743017e-05,
-      "loss": 0.182,
+      "grad_norm": 2.9278903007507324,
+      "learning_rate": 4.217877094972068e-06,
+      "loss": 0.2048,
       "step": 1290
     },
     {
       "epoch": 6.5306633291614515,
-      "grad_norm": 2.591944694519043,
-      "learning_rate": 1.0195530726256984e-05,
-      "loss": 0.1809,
+      "grad_norm": 5.488080024719238,
+      "learning_rate": 4.078212290502794e-06,
+      "loss": 0.2098,
       "step": 1300
     },
     {
       "epoch": 6.58072590738423,
-      "grad_norm": 2.280179977416992,
-      "learning_rate": 9.8463687150838e-06,
-      "loss": 0.1882,
+      "grad_norm": 4.781853199005127,
+      "learning_rate": 3.93854748603352e-06,
+      "loss": 0.208,
       "step": 1310
     },
     {
       "epoch": 6.630788485607009,
-      "grad_norm": 2.2294774055480957,
-      "learning_rate": 9.497206703910614e-06,
-      "loss": 0.1624,
+      "grad_norm": 2.9733641147613525,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 0.1923,
       "step": 1320
     },
     {
       "epoch": 6.680851063829787,
-      "grad_norm": 2.2053873538970947,
-      "learning_rate": 9.14804469273743e-06,
-      "loss": 0.1701,
+      "grad_norm": 2.823633909225464,
+      "learning_rate": 3.6592178770949723e-06,
+      "loss": 0.185,
       "step": 1330
     },
     {
       "epoch": 6.730913642052566,
-      "grad_norm": 1.1514060497283936,
-      "learning_rate": 8.798882681564247e-06,
-      "loss": 0.1709,
+      "grad_norm": 2.7353272438049316,
+      "learning_rate": 3.5195530726256988e-06,
+      "loss": 0.1977,
       "step": 1340
     },
     {
       "epoch": 6.7809762202753445,
-      "grad_norm": 1.625301718711853,
-      "learning_rate": 8.449720670391061e-06,
-      "loss": 0.1803,
+      "grad_norm": 2.489300489425659,
+      "learning_rate": 3.3798882681564248e-06,
+      "loss": 0.1999,
       "step": 1350
     },
     {
       "epoch": 6.831038798498122,
-      "grad_norm": 1.733961582183838,
-      "learning_rate": 8.100558659217877e-06,
-      "loss": 0.1792,
+      "grad_norm": 2.8995063304901123,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 0.2111,
       "step": 1360
     },
     {
       "epoch": 6.881101376720901,
-      "grad_norm": 2.374742269515991,
-      "learning_rate": 7.751396648044693e-06,
-      "loss": 0.1711,
+      "grad_norm": 2.4604063034057617,
+      "learning_rate": 3.1005586592178773e-06,
+      "loss": 0.2043,
       "step": 1370
     },
     {
       "epoch": 6.931163954943679,
-      "grad_norm": 2.3548731803894043,
-      "learning_rate": 7.4022346368715096e-06,
-      "loss": 0.1864,
+      "grad_norm": 2.1975669860839844,
+      "learning_rate": 2.9608938547486037e-06,
+      "loss": 0.203,
       "step": 1380
     },
     {
       "epoch": 6.981226533166458,
-      "grad_norm": 1.3292882442474365,
-      "learning_rate": 7.053072625698325e-06,
-      "loss": 0.1632,
+      "grad_norm": 2.8632118701934814,
+      "learning_rate": 2.8212290502793298e-06,
+      "loss": 0.1826,
       "step": 1390
     },
     {
       "epoch": 6.996245306633291,
-      "eval_accuracy": 0.9767578699617535,
-      "eval_loss": 0.10951310396194458,
-      "eval_runtime": 5.0652,
-      "eval_samples_per_second": 1342.105,
-      "eval_steps_per_second": 42.052,
+      "eval_accuracy": 0.9814651368049426,
+      "eval_loss": 0.12438357621431351,
+      "eval_runtime": 4.9437,
+      "eval_samples_per_second": 1375.072,
+      "eval_steps_per_second": 43.085,
       "step": 1393
     },
     {
       "epoch": 7.035043804755945,
-      "grad_norm": 2.334541082382202,
-      "learning_rate": 6.70391061452514e-06,
-      "loss": 0.1726,
+      "grad_norm": 2.0548951625823975,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 0.2089,
       "step": 1400
     },
     {
       "epoch": 7.085106382978723,
-      "grad_norm": 2.8630871772766113,
-      "learning_rate": 6.3547486033519555e-06,
-      "loss": 0.1775,
+      "grad_norm": 2.192319631576538,
+      "learning_rate": 2.5418994413407823e-06,
+      "loss": 0.1803,
       "step": 1410
     },
     {
       "epoch": 7.135168961201502,
-      "grad_norm": 1.685805320739746,
-      "learning_rate": 6.005586592178771e-06,
-      "loss": 0.1886,
+      "grad_norm": 3.190675973892212,
+      "learning_rate": 2.4022346368715087e-06,
+      "loss": 0.1985,
       "step": 1420
     },
     {
       "epoch": 7.18523153942428,
-      "grad_norm": 2.9282922744750977,
-      "learning_rate": 5.656424581005587e-06,
-      "loss": 0.1477,
+      "grad_norm": 3.319995880126953,
+      "learning_rate": 2.2625698324022348e-06,
+      "loss": 0.1815,
       "step": 1430
     },
     {
       "epoch": 7.235294117647059,
-      "grad_norm": 2.2304320335388184,
-      "learning_rate": 5.307262569832402e-06,
-      "loss": 0.1692,
+      "grad_norm": 3.1613974571228027,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 0.1874,
       "step": 1440
     },
     {
       "epoch": 7.2853566958698375,
-      "grad_norm": 2.634939193725586,
-      "learning_rate": 4.958100558659218e-06,
-      "loss": 0.1998,
+      "grad_norm": 5.030276298522949,
+      "learning_rate": 1.9832402234636873e-06,
+      "loss": 0.2316,
       "step": 1450
     },
     {
       "epoch": 7.335419274092616,
-      "grad_norm": 2.268406629562378,
-      "learning_rate": 4.608938547486034e-06,
-      "loss": 0.1655,
+      "grad_norm": 4.310720443725586,
+      "learning_rate": 1.8435754189944135e-06,
+      "loss": 0.187,
       "step": 1460
     },
     {
       "epoch": 7.385481852315394,
-      "grad_norm": 1.96724534034729,
-      "learning_rate": 4.259776536312849e-06,
-      "loss": 0.1619,
+      "grad_norm": 3.277670383453369,
+      "learning_rate": 1.7039106145251397e-06,
+      "loss": 0.189,
       "step": 1470
     },
     {
       "epoch": 7.435544430538172,
-      "grad_norm": 2.0757083892822266,
-      "learning_rate": 3.910614525139665e-06,
-      "loss": 0.1676,
+      "grad_norm": 2.657017707824707,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 0.1963,
       "step": 1480
     },
     {
       "epoch": 7.485607008760951,
-      "grad_norm": 2.4496700763702393,
-      "learning_rate": 3.5614525139664804e-06,
-      "loss": 0.1558,
+      "grad_norm": 1.9806122779846191,
+      "learning_rate": 1.4245810055865922e-06,
+      "loss": 0.1736,
       "step": 1490
     },
     {
       "epoch": 7.53566958698373,
-      "grad_norm": 3.963124990463257,
-      "learning_rate": 3.212290502793296e-06,
-      "loss": 0.1703,
+      "grad_norm": 2.318047285079956,
+      "learning_rate": 1.2849162011173185e-06,
+      "loss": 0.192,
       "step": 1500
     },
     {
       "epoch": 7.585732165206508,
-      "grad_norm": 3.033352851867676,
-      "learning_rate": 2.863128491620112e-06,
-      "loss": 0.1643,
+      "grad_norm": 1.8546510934829712,
+      "learning_rate": 1.1452513966480447e-06,
+      "loss": 0.1915,
       "step": 1510
     },
     {
       "epoch": 7.635794743429287,
-      "grad_norm": 2.1735336780548096,
-      "learning_rate": 2.5139664804469276e-06,
-      "loss": 0.176,
+      "grad_norm": 2.3725483417510986,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 0.1924,
       "step": 1520
     },
     {
       "epoch": 7.685857321652065,
-      "grad_norm": 2.0125844478607178,
-      "learning_rate": 2.1648044692737433e-06,
-      "loss": 0.1829,
+      "grad_norm": 2.3467774391174316,
+      "learning_rate": 8.659217877094973e-07,
+      "loss": 0.1997,
       "step": 1530
     },
     {
       "epoch": 7.735919899874844,
-      "grad_norm": 1.9535727500915527,
-      "learning_rate": 1.815642458100559e-06,
-      "loss": 0.1864,
+      "grad_norm": 2.653728485107422,
+      "learning_rate": 7.262569832402236e-07,
+      "loss": 0.2081,
       "step": 1540
     },
     {
       "epoch": 7.785982478097622,
-      "grad_norm": 2.61956524848938,
-      "learning_rate": 1.4664804469273743e-06,
-      "loss": 0.1847,
+      "grad_norm": 3.015143394470215,
+      "learning_rate": 5.865921787709498e-07,
+      "loss": 0.2194,
       "step": 1550
     },
     {
       "epoch": 7.8360450563204,
-      "grad_norm": 1.4061027765274048,
-      "learning_rate": 1.11731843575419e-06,
-      "loss": 0.1786,
+      "grad_norm": 1.836916446685791,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 0.2127,
       "step": 1560
     },
     {
       "epoch": 7.886107634543179,
-      "grad_norm": 2.1694343090057373,
-      "learning_rate": 7.681564245810056e-07,
-      "loss": 0.1745,
+      "grad_norm": 1.9544241428375244,
+      "learning_rate": 3.0726256983240227e-07,
+      "loss": 0.1995,
       "step": 1570
     },
     {
       "epoch": 7.9361702127659575,
-      "grad_norm": 1.2118408679962158,
-      "learning_rate": 4.1899441340782126e-07,
-      "loss": 0.1668,
+      "grad_norm": 1.683449625968933,
+      "learning_rate": 1.6759776536312851e-07,
+      "loss": 0.1974,
       "step": 1580
     },
     {
       "epoch": 7.986232790988736,
-      "grad_norm": 1.914755940437317,
-      "learning_rate": 6.983240223463688e-08,
-      "loss": 0.1642,
+      "grad_norm": 2.348433494567871,
+      "learning_rate": 2.793296089385475e-08,
+      "loss": 0.1982,
       "step": 1590
     },
     {
       "epoch": 7.996245306633291,
-      "eval_accuracy": 0.9773462783171522,
-      "eval_loss": 0.11222562193870544,
-      "eval_runtime": 5.4676,
-      "eval_samples_per_second": 1243.322,
-      "eval_steps_per_second": 38.957,
+      "eval_accuracy": 0.9814651368049426,
+      "eval_loss": 0.12064122408628464,
+      "eval_runtime": 5.5563,
+      "eval_samples_per_second": 1223.467,
+      "eval_steps_per_second": 38.335,
       "step": 1592
     },
     {
       "epoch": 7.996245306633291,
       "step": 1592,
       "total_flos": 3.777723239743488e+18,
-      "train_loss": 0.6284143277373745,
-      "train_runtime": 634.1755,
-      "train_samples_per_second": 644.541,
-      "train_steps_per_second": 2.51
+      "train_loss": 0.726146658611058,
+      "train_runtime": 635.8846,
+      "train_samples_per_second": 642.808,
+      "train_steps_per_second": 2.504
     }
   ],
   "logging_steps": 10,