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

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 7.996245306633291,
-    "eval_accuracy": 0.9830832597822889,
-    "eval_loss": 0.0956372618675232,
-    "eval_runtime": 5.6145,
-    "eval_samples_per_second": 1210.799,
-    "eval_steps_per_second": 37.938,
+    "eval_accuracy": 0.9763165636952045,
+    "eval_loss": 0.40624934434890747,
+    "eval_runtime": 5.5217,
+    "eval_samples_per_second": 1231.15,
+    "eval_steps_per_second": 38.575,
     "total_flos": 3.777723239743488e+18,
-    "train_loss": 0.596273283347787,
-    "train_runtime": 640.7753,
-    "train_samples_per_second": 637.902,
-    "train_steps_per_second": 2.484
+    "train_loss": 1.152813568037359,
+    "train_runtime": 637.2674,
+    "train_samples_per_second": 641.414,
+    "train_steps_per_second": 2.498
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 7.996245306633291,
-    "eval_accuracy": 0.9830832597822889,
-    "eval_loss": 0.0956372618675232,
-    "eval_runtime": 5.6145,
-    "eval_samples_per_second": 1210.799,
-    "eval_steps_per_second": 37.938
+    "eval_accuracy": 0.9763165636952045,
+    "eval_loss": 0.40624934434890747,
+    "eval_runtime": 5.5217,
+    "eval_samples_per_second": 1231.15,
+    "eval_steps_per_second": 38.575
 }

runs/May14_16-06-35_cs-Precision-7960-Tower/events.out.tfevents.1747253871.cs-Precision-7960-Tower.128840.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2152ebbe8021d73f3e8a88a5c56305eda145e557c9a4a77848f71d3a815b7b72
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 7.996245306633291,
     "total_flos": 3.777723239743488e+18,
-    "train_loss": 0.596273283347787,
-    "train_runtime": 640.7753,
-    "train_samples_per_second": 637.902,
-    "train_steps_per_second": 2.484
+    "train_loss": 1.152813568037359,
+    "train_runtime": 637.2674,
+    "train_samples_per_second": 641.414,
+    "train_steps_per_second": 2.498
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.9830832597822889,
-  "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1393",
+  "best_metric": 0.9763165636952045,
+  "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1592",
   "epoch": 7.996245306633291,
   "eval_steps": 500,
   "global_step": 1592,
@@ -10,1197 +10,1197 @@
   "log_history": [
     {
       "epoch": 0.05006257822277847,
-      "grad_norm": 2.299422264099121,
-      "learning_rate": 1.875e-06,
-      "loss": 4.1412,
+      "grad_norm": 2.2142410278320312,
+      "learning_rate": 6.25e-07,
+      "loss": 4.0997,
       "step": 10
     },
     {
       "epoch": 0.10012515644555695,
-      "grad_norm": 3.199314832687378,
-      "learning_rate": 3.75e-06,
-      "loss": 4.1637,
+      "grad_norm": 2.4960668087005615,
+      "learning_rate": 1.25e-06,
+      "loss": 4.1744,
       "step": 20
     },
     {
       "epoch": 0.15018773466833543,
-      "grad_norm": 3.3083832263946533,
-      "learning_rate": 5.625e-06,
-      "loss": 4.0438,
+      "grad_norm": 2.9901962280273438,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 4.1686,
       "step": 30
     },
     {
       "epoch": 0.2002503128911139,
-      "grad_norm": 4.229264736175537,
-      "learning_rate": 7.5e-06,
-      "loss": 3.8012,
+      "grad_norm": 2.9945929050445557,
+      "learning_rate": 2.5e-06,
+      "loss": 4.1114,
       "step": 40
     },
     {
       "epoch": 0.2503128911138924,
-      "grad_norm": 5.718367576599121,
-      "learning_rate": 9.375000000000001e-06,
-      "loss": 3.3779,
+      "grad_norm": 3.4015228748321533,
+      "learning_rate": 3.125e-06,
+      "loss": 4.0183,
       "step": 50
     },
     {
       "epoch": 0.30037546933667086,
-      "grad_norm": 6.0788254737854,
-      "learning_rate": 1.125e-05,
-      "loss": 2.8533,
+      "grad_norm": 4.165560722351074,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 3.875,
       "step": 60
     },
     {
       "epoch": 0.3504380475594493,
-      "grad_norm": 5.600748538970947,
-      "learning_rate": 1.3125e-05,
-      "loss": 2.4796,
+      "grad_norm": 4.998468399047852,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 3.6691,
       "step": 70
     },
     {
       "epoch": 0.4005006257822278,
-      "grad_norm": 5.035912990570068,
-      "learning_rate": 1.5e-05,
-      "loss": 2.253,
+      "grad_norm": 5.729038238525391,
+      "learning_rate": 5e-06,
+      "loss": 3.3538,
       "step": 80
     },
     {
       "epoch": 0.45056320400500627,
-      "grad_norm": 4.351953983306885,
-      "learning_rate": 1.6875e-05,
-      "loss": 2.0939,
+      "grad_norm": 5.879266738891602,
+      "learning_rate": 5.625e-06,
+      "loss": 2.9866,
       "step": 90
     },
     {
       "epoch": 0.5006257822277848,
-      "grad_norm": 3.4278855323791504,
-      "learning_rate": 1.8750000000000002e-05,
-      "loss": 1.9864,
+      "grad_norm": 5.562048435211182,
+      "learning_rate": 6.25e-06,
+      "loss": 2.6957,
       "step": 100
     },
     {
       "epoch": 0.5506883604505632,
-      "grad_norm": 2.862748384475708,
-      "learning_rate": 2.0625e-05,
-      "loss": 1.8611,
+      "grad_norm": 5.61751651763916,
+      "learning_rate": 6.875e-06,
+      "loss": 2.4759,
       "step": 110
     },
     {
       "epoch": 0.6007509386733417,
-      "grad_norm": 1.2488276958465576,
-      "learning_rate": 2.25e-05,
-      "loss": 1.8302,
+      "grad_norm": 4.866910457611084,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.358,
       "step": 120
     },
     {
       "epoch": 0.6508135168961201,
-      "grad_norm": 0.6365911960601807,
-      "learning_rate": 2.4375e-05,
-      "loss": 1.7982,
+      "grad_norm": 4.472853183746338,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 2.2576,
       "step": 130
     },
     {
       "epoch": 0.7008760951188986,
-      "grad_norm": 0.5073445439338684,
-      "learning_rate": 2.625e-05,
-      "loss": 1.6792,
+      "grad_norm": 4.833339691162109,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.1076,
       "step": 140
     },
     {
       "epoch": 0.7509386733416771,
-      "grad_norm": 0.9256235361099243,
-      "learning_rate": 2.8125e-05,
-      "loss": 1.7528,
+      "grad_norm": 4.425817966461182,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 2.0808,
       "step": 150
     },
     {
       "epoch": 0.8010012515644556,
-      "grad_norm": 5.669793128967285,
-      "learning_rate": 3e-05,
-      "loss": 1.7688,
+      "grad_norm": 3.747729539871216,
+      "learning_rate": 1e-05,
+      "loss": 2.0467,
       "step": 160
     },
     {
       "epoch": 0.851063829787234,
-      "grad_norm": 10.126972198486328,
-      "learning_rate": 2.979050279329609e-05,
-      "loss": 1.6876,
+      "grad_norm": 3.6987853050231934,
+      "learning_rate": 9.930167597765364e-06,
+      "loss": 1.9358,
       "step": 170
     },
     {
       "epoch": 0.9011264080100125,
-      "grad_norm": 1.5617619752883911,
-      "learning_rate": 2.958100558659218e-05,
-      "loss": 1.6298,
+      "grad_norm": 3.0652871131896973,
+      "learning_rate": 9.860335195530727e-06,
+      "loss": 1.904,
       "step": 180
     },
     {
       "epoch": 0.951188986232791,
-      "grad_norm": 1.5987392663955688,
-      "learning_rate": 2.937150837988827e-05,
-      "loss": 1.6106,
+      "grad_norm": 2.3470098972320557,
+      "learning_rate": 9.79050279329609e-06,
+      "loss": 1.8801,
       "step": 190
     },
     {
       "epoch": 0.9962453066332916,
       "eval_accuracy": 0.6209179170344219,
-      "eval_loss": 1.4251551628112793,
-      "eval_runtime": 4.8605,
-      "eval_samples_per_second": 1398.631,
-      "eval_steps_per_second": 43.823,
+      "eval_loss": 1.7453958988189697,
+      "eval_runtime": 4.9497,
+      "eval_samples_per_second": 1373.405,
+      "eval_steps_per_second": 43.033,
       "step": 199
     },
     {
       "epoch": 1.0050062578222778,
-      "grad_norm": 3.5913760662078857,
-      "learning_rate": 2.9162011173184356e-05,
-      "loss": 1.6815,
+      "grad_norm": 2.7552437782287598,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 1.9461,
       "step": 200
     },
     {
       "epoch": 1.0550688360450564,
-      "grad_norm": 2.8698642253875732,
-      "learning_rate": 2.895251396648045e-05,
-      "loss": 1.4457,
+      "grad_norm": 1.5700269937515259,
+      "learning_rate": 9.650837988826817e-06,
+      "loss": 1.7812,
       "step": 210
     },
     {
       "epoch": 1.1051314142678348,
-      "grad_norm": 2.3613011837005615,
-      "learning_rate": 2.8743016759776535e-05,
-      "loss": 1.4503,
+      "grad_norm": 1.4513113498687744,
+      "learning_rate": 9.581005586592178e-06,
+      "loss": 1.7865,
       "step": 220
     },
     {
       "epoch": 1.1551939924906134,
-      "grad_norm": 3.4527103900909424,
-      "learning_rate": 2.8533519553072625e-05,
-      "loss": 1.2686,
+      "grad_norm": 1.5353403091430664,
+      "learning_rate": 9.511173184357543e-06,
+      "loss": 1.7024,
       "step": 230
     },
     {
       "epoch": 1.2052565707133918,
-      "grad_norm": 4.879206657409668,
-      "learning_rate": 2.8324022346368715e-05,
-      "loss": 1.2226,
+      "grad_norm": 1.4006606340408325,
+      "learning_rate": 9.441340782122905e-06,
+      "loss": 1.7307,
       "step": 240
     },
     {
       "epoch": 1.2553191489361701,
-      "grad_norm": 3.2351438999176025,
-      "learning_rate": 2.8114525139664805e-05,
-      "loss": 1.1545,
+      "grad_norm": 2.397796630859375,
+      "learning_rate": 9.371508379888268e-06,
+      "loss": 1.7121,
       "step": 250
     },
     {
       "epoch": 1.3053817271589487,
-      "grad_norm": 5.1034464836120605,
-      "learning_rate": 2.7905027932960894e-05,
-      "loss": 1.1284,
+      "grad_norm": 2.891803503036499,
+      "learning_rate": 9.301675977653633e-06,
+      "loss": 1.7195,
       "step": 260
     },
     {
       "epoch": 1.355444305381727,
-      "grad_norm": 2.128084421157837,
-      "learning_rate": 2.7695530726256984e-05,
-      "loss": 1.0926,
+      "grad_norm": 13.237679481506348,
+      "learning_rate": 9.231843575418995e-06,
+      "loss": 1.6628,
       "step": 270
     },
     {
       "epoch": 1.4055068836045057,
-      "grad_norm": 5.853870391845703,
-      "learning_rate": 2.7486033519553074e-05,
-      "loss": 1.075,
+      "grad_norm": 4.088449478149414,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 1.6614,
       "step": 280
     },
     {
       "epoch": 1.455569461827284,
-      "grad_norm": 2.4751949310302734,
-      "learning_rate": 2.7276536312849163e-05,
-      "loss": 0.9992,
+      "grad_norm": 4.54097318649292,
+      "learning_rate": 9.092178770949721e-06,
+      "loss": 1.5913,
       "step": 290
     },
     {
       "epoch": 1.5056320400500627,
-      "grad_norm": 3.3400278091430664,
-      "learning_rate": 2.7067039106145253e-05,
-      "loss": 0.9649,
+      "grad_norm": 2.6582536697387695,
+      "learning_rate": 9.022346368715084e-06,
+      "loss": 1.5883,
       "step": 300
     },
     {
       "epoch": 1.555694618272841,
-      "grad_norm": 2.893463611602783,
-      "learning_rate": 2.685754189944134e-05,
-      "loss": 1.0066,
+      "grad_norm": 5.469503402709961,
+      "learning_rate": 8.952513966480448e-06,
+      "loss": 1.6299,
       "step": 310
     },
     {
       "epoch": 1.6057571964956194,
-      "grad_norm": 2.179349660873413,
-      "learning_rate": 2.6648044692737432e-05,
-      "loss": 0.9203,
+      "grad_norm": 2.4134175777435303,
+      "learning_rate": 8.88268156424581e-06,
+      "loss": 1.5386,
       "step": 320
     },
     {
       "epoch": 1.655819774718398,
-      "grad_norm": 4.882504463195801,
-      "learning_rate": 2.643854748603352e-05,
-      "loss": 0.9159,
+      "grad_norm": 5.472087860107422,
+      "learning_rate": 8.812849162011174e-06,
+      "loss": 1.5774,
       "step": 330
     },
     {
       "epoch": 1.7058823529411766,
-      "grad_norm": 4.354543685913086,
-      "learning_rate": 2.622905027932961e-05,
-      "loss": 0.8608,
+      "grad_norm": 3.7888548374176025,
+      "learning_rate": 8.743016759776537e-06,
+      "loss": 1.5314,
       "step": 340
     },
     {
       "epoch": 1.7559449311639548,
-      "grad_norm": 3.870502233505249,
-      "learning_rate": 2.60195530726257e-05,
-      "loss": 0.8231,
+      "grad_norm": 6.527620315551758,
+      "learning_rate": 8.6731843575419e-06,
+      "loss": 1.4678,
       "step": 350
     },
     {
       "epoch": 1.8060075093867334,
-      "grad_norm": 3.579007148742676,
-      "learning_rate": 2.5810055865921788e-05,
-      "loss": 0.7965,
+      "grad_norm": 6.796471118927002,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 1.4769,
       "step": 360
     },
     {
       "epoch": 1.856070087609512,
-      "grad_norm": 4.881648540496826,
-      "learning_rate": 2.5600558659217877e-05,
-      "loss": 0.7647,
+      "grad_norm": 4.604396343231201,
+      "learning_rate": 8.533519553072627e-06,
+      "loss": 1.4285,
       "step": 370
     },
     {
       "epoch": 1.9061326658322904,
-      "grad_norm": 2.9336421489715576,
-      "learning_rate": 2.5391061452513967e-05,
-      "loss": 0.7005,
+      "grad_norm": 5.599514484405518,
+      "learning_rate": 8.463687150837988e-06,
+      "loss": 1.3812,
       "step": 380
     },
     {
       "epoch": 1.9561952440550687,
-      "grad_norm": 2.542874813079834,
-      "learning_rate": 2.5181564245810057e-05,
-      "loss": 0.6495,
+      "grad_norm": 13.588865280151367,
+      "learning_rate": 8.393854748603353e-06,
+      "loss": 1.389,
       "step": 390
     },
     {
       "epoch": 1.9962453066332917,
-      "eval_accuracy": 0.9682259488084731,
-      "eval_loss": 0.5032486915588379,
-      "eval_runtime": 4.9258,
-      "eval_samples_per_second": 1380.074,
-      "eval_steps_per_second": 43.242,
+      "eval_accuracy": 0.6518093556928508,
+      "eval_loss": 1.240417718887329,
+      "eval_runtime": 4.8608,
+      "eval_samples_per_second": 1398.548,
+      "eval_steps_per_second": 43.82,
       "step": 398
     },
     {
       "epoch": 2.0100125156445556,
-      "grad_norm": 1.9056552648544312,
-      "learning_rate": 2.4972067039106143e-05,
-      "loss": 0.6495,
+      "grad_norm": 5.914628028869629,
+      "learning_rate": 8.324022346368715e-06,
+      "loss": 1.4118,
       "step": 400
     },
     {
       "epoch": 2.0600750938673342,
-      "grad_norm": 3.567265033721924,
-      "learning_rate": 2.4762569832402236e-05,
-      "loss": 0.5869,
+      "grad_norm": 9.35991096496582,
+      "learning_rate": 8.254189944134078e-06,
+      "loss": 1.3324,
       "step": 410
     },
     {
       "epoch": 2.110137672090113,
-      "grad_norm": 2.240018844604492,
-      "learning_rate": 2.4553072625698326e-05,
-      "loss": 0.5728,
+      "grad_norm": 7.8319220542907715,
+      "learning_rate": 8.184357541899443e-06,
+      "loss": 1.3393,
       "step": 420
     },
     {
       "epoch": 2.160200250312891,
-      "grad_norm": 2.6313724517822266,
-      "learning_rate": 2.4343575418994412e-05,
-      "loss": 0.5028,
+      "grad_norm": 3.685518264770508,
+      "learning_rate": 8.114525139664805e-06,
+      "loss": 1.3176,
       "step": 430
     },
     {
       "epoch": 2.2102628285356696,
-      "grad_norm": 3.360229015350342,
-      "learning_rate": 2.4134078212290505e-05,
-      "loss": 0.4928,
+      "grad_norm": 4.464322566986084,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 1.2491,
       "step": 440
     },
     {
       "epoch": 2.260325406758448,
-      "grad_norm": 5.249541282653809,
-      "learning_rate": 2.392458100558659e-05,
-      "loss": 0.4773,
+      "grad_norm": 6.714467525482178,
+      "learning_rate": 7.974860335195531e-06,
+      "loss": 1.2923,
       "step": 450
     },
     {
       "epoch": 2.3103879849812268,
-      "grad_norm": 3.6117191314697266,
-      "learning_rate": 2.3715083798882685e-05,
-      "loss": 0.4852,
+      "grad_norm": 3.9189820289611816,
+      "learning_rate": 7.905027932960894e-06,
+      "loss": 1.2584,
       "step": 460
     },
     {
       "epoch": 2.360450563204005,
-      "grad_norm": 4.820945739746094,
-      "learning_rate": 2.350558659217877e-05,
-      "loss": 0.4331,
+      "grad_norm": 10.34443473815918,
+      "learning_rate": 7.835195530726258e-06,
+      "loss": 1.2372,
       "step": 470
     },
     {
       "epoch": 2.4105131414267835,
-      "grad_norm": 4.089610576629639,
-      "learning_rate": 2.329608938547486e-05,
-      "loss": 0.4246,
+      "grad_norm": 3.698418140411377,
+      "learning_rate": 7.76536312849162e-06,
+      "loss": 1.225,
       "step": 480
     },
     {
       "epoch": 2.460575719649562,
-      "grad_norm": 4.083464622497559,
-      "learning_rate": 2.308659217877095e-05,
-      "loss": 0.3752,
+      "grad_norm": 8.271072387695312,
+      "learning_rate": 7.695530726256984e-06,
+      "loss": 1.1353,
       "step": 490
     },
     {
       "epoch": 2.5106382978723403,
-      "grad_norm": 4.422226428985596,
-      "learning_rate": 2.287709497206704e-05,
-      "loss": 0.3916,
+      "grad_norm": 3.2551236152648926,
+      "learning_rate": 7.625698324022347e-06,
+      "loss": 1.1894,
       "step": 500
     },
     {
       "epoch": 2.560700876095119,
-      "grad_norm": 2.952890634536743,
-      "learning_rate": 2.266759776536313e-05,
-      "loss": 0.3973,
+      "grad_norm": 7.808169364929199,
+      "learning_rate": 7.5558659217877105e-06,
+      "loss": 1.2102,
       "step": 510
     },
     {
       "epoch": 2.6107634543178975,
-      "grad_norm": 3.720259428024292,
-      "learning_rate": 2.245810055865922e-05,
-      "loss": 0.3432,
+      "grad_norm": 10.614561080932617,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 1.1493,
       "step": 520
     },
     {
       "epoch": 2.660826032540676,
-      "grad_norm": 4.10168981552124,
-      "learning_rate": 2.224860335195531e-05,
-      "loss": 0.3479,
+      "grad_norm": 4.759502410888672,
+      "learning_rate": 7.416201117318437e-06,
+      "loss": 1.1384,
       "step": 530
     },
     {
       "epoch": 2.710888610763454,
-      "grad_norm": 4.39931058883667,
-      "learning_rate": 2.2039106145251395e-05,
-      "loss": 0.3418,
+      "grad_norm": 8.49401569366455,
+      "learning_rate": 7.346368715083799e-06,
+      "loss": 1.0993,
       "step": 540
     },
     {
       "epoch": 2.760951188986233,
-      "grad_norm": 2.6174728870391846,
-      "learning_rate": 2.182960893854749e-05,
-      "loss": 0.3153,
+      "grad_norm": 4.307897567749023,
+      "learning_rate": 7.2765363128491625e-06,
+      "loss": 1.1151,
       "step": 550
     },
     {
       "epoch": 2.8110137672090114,
-      "grad_norm": 3.489020347595215,
-      "learning_rate": 2.1620111731843575e-05,
-      "loss": 0.3242,
+      "grad_norm": 4.177582740783691,
+      "learning_rate": 7.206703910614526e-06,
+      "loss": 1.1375,
       "step": 560
     },
     {
       "epoch": 2.8610763454317896,
-      "grad_norm": 3.2841830253601074,
-      "learning_rate": 2.1410614525139664e-05,
-      "loss": 0.3016,
+      "grad_norm": 3.8811707496643066,
+      "learning_rate": 7.136871508379889e-06,
+      "loss": 1.1109,
       "step": 570
     },
     {
       "epoch": 2.911138923654568,
-      "grad_norm": 4.06994104385376,
-      "learning_rate": 2.1201117318435754e-05,
-      "loss": 0.3237,
+      "grad_norm": 2.818143129348755,
+      "learning_rate": 7.067039106145251e-06,
+      "loss": 1.1036,
       "step": 580
     },
     {
       "epoch": 2.9612015018773468,
-      "grad_norm": 2.4799962043762207,
-      "learning_rate": 2.0991620111731844e-05,
-      "loss": 0.2978,
+      "grad_norm": 3.4761757850646973,
+      "learning_rate": 6.9972067039106154e-06,
+      "loss": 1.1239,
       "step": 590
     },
     {
       "epoch": 2.9962453066332917,
-      "eval_accuracy": 0.9782288908502501,
-      "eval_loss": 0.19027507305145264,
-      "eval_runtime": 4.8737,
-      "eval_samples_per_second": 1394.823,
-      "eval_steps_per_second": 43.704,
+      "eval_accuracy": 0.7880258899676376,
+      "eval_loss": 1.069029688835144,
+      "eval_runtime": 4.802,
+      "eval_samples_per_second": 1415.653,
+      "eval_steps_per_second": 44.356,
       "step": 597
     },
     {
       "epoch": 3.0150187734668337,
-      "grad_norm": 2.1213083267211914,
-      "learning_rate": 2.0782122905027933e-05,
-      "loss": 0.3231,
+      "grad_norm": 4.441803455352783,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 1.1924,
       "step": 600
     },
     {
       "epoch": 3.065081351689612,
-      "grad_norm": 4.8361945152282715,
-      "learning_rate": 2.0572625698324023e-05,
-      "loss": 0.302,
+      "grad_norm": 6.063540458679199,
+      "learning_rate": 6.857541899441341e-06,
+      "loss": 1.0628,
       "step": 610
     },
     {
       "epoch": 3.1151439299123904,
-      "grad_norm": 3.293104887008667,
-      "learning_rate": 2.0363128491620113e-05,
-      "loss": 0.2883,
+      "grad_norm": 4.163120746612549,
+      "learning_rate": 6.787709497206705e-06,
+      "loss": 1.0444,
       "step": 620
     },
     {
       "epoch": 3.165206508135169,
-      "grad_norm": 3.274291515350342,
-      "learning_rate": 2.01536312849162e-05,
-      "loss": 0.2961,
+      "grad_norm": 6.099351406097412,
+      "learning_rate": 6.7178770949720675e-06,
+      "loss": 1.1028,
       "step": 630
     },
     {
       "epoch": 3.2152690863579476,
-      "grad_norm": 3.7976105213165283,
-      "learning_rate": 1.9944134078212292e-05,
-      "loss": 0.2688,
+      "grad_norm": 4.259471416473389,
+      "learning_rate": 6.648044692737431e-06,
+      "loss": 1.0317,
       "step": 640
     },
     {
       "epoch": 3.2653316645807258,
-      "grad_norm": 2.9893229007720947,
-      "learning_rate": 1.973463687150838e-05,
-      "loss": 0.2446,
+      "grad_norm": 4.617487907409668,
+      "learning_rate": 6.578212290502793e-06,
+      "loss": 1.0323,
       "step": 650
     },
     {
       "epoch": 3.3153942428035044,
-      "grad_norm": 2.2266604900360107,
-      "learning_rate": 1.952513966480447e-05,
-      "loss": 0.2613,
+      "grad_norm": 4.48543119430542,
+      "learning_rate": 6.508379888268157e-06,
+      "loss": 1.0185,
       "step": 660
     },
     {
       "epoch": 3.365456821026283,
-      "grad_norm": 3.621093511581421,
-      "learning_rate": 1.9315642458100558e-05,
-      "loss": 0.2512,
+      "grad_norm": 4.388581275939941,
+      "learning_rate": 6.43854748603352e-06,
+      "loss": 0.9979,
       "step": 670
     },
     {
       "epoch": 3.415519399249061,
-      "grad_norm": 2.952971935272217,
-      "learning_rate": 1.9106145251396648e-05,
-      "loss": 0.2536,
+      "grad_norm": 7.762565612792969,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 0.9999,
       "step": 680
     },
     {
       "epoch": 3.4655819774718397,
-      "grad_norm": 2.7361905574798584,
-      "learning_rate": 1.889664804469274e-05,
-      "loss": 0.2396,
+      "grad_norm": 3.058173179626465,
+      "learning_rate": 6.298882681564247e-06,
+      "loss": 0.9729,
       "step": 690
     },
     {
       "epoch": 3.5156445556946183,
-      "grad_norm": 2.3844313621520996,
-      "learning_rate": 1.8687150837988827e-05,
-      "loss": 0.2518,
+      "grad_norm": 6.832275390625,
+      "learning_rate": 6.229050279329609e-06,
+      "loss": 1.0058,
       "step": 700
     },
     {
       "epoch": 3.565707133917397,
-      "grad_norm": 3.0508193969726562,
-      "learning_rate": 1.8477653631284917e-05,
-      "loss": 0.2325,
+      "grad_norm": 3.4098267555236816,
+      "learning_rate": 6.1592178770949725e-06,
+      "loss": 0.9634,
       "step": 710
     },
     {
       "epoch": 3.615769712140175,
-      "grad_norm": 3.923941135406494,
-      "learning_rate": 1.8268156424581006e-05,
-      "loss": 0.2277,
+      "grad_norm": 5.360341548919678,
+      "learning_rate": 6.089385474860336e-06,
+      "loss": 0.9622,
       "step": 720
     },
     {
       "epoch": 3.6658322903629537,
-      "grad_norm": 2.638787031173706,
-      "learning_rate": 1.8058659217877096e-05,
-      "loss": 0.2292,
+      "grad_norm": 3.9530041217803955,
+      "learning_rate": 6.019553072625699e-06,
+      "loss": 0.9585,
       "step": 730
     },
     {
       "epoch": 3.7158948685857323,
-      "grad_norm": 2.75313138961792,
-      "learning_rate": 1.7849162011173182e-05,
-      "loss": 0.2364,
+      "grad_norm": 8.933693885803223,
+      "learning_rate": 5.949720670391061e-06,
+      "loss": 0.953,
       "step": 740
     },
     {
       "epoch": 3.7659574468085104,
-      "grad_norm": 3.686354398727417,
-      "learning_rate": 1.7639664804469275e-05,
-      "loss": 0.2409,
+      "grad_norm": 3.6110997200012207,
+      "learning_rate": 5.879888268156425e-06,
+      "loss": 0.9176,
       "step": 750
     },
     {
       "epoch": 3.816020025031289,
-      "grad_norm": 4.230103969573975,
-      "learning_rate": 1.7430167597765365e-05,
-      "loss": 0.2293,
+      "grad_norm": 3.877078056335449,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 0.9133,
       "step": 760
     },
     {
       "epoch": 3.8660826032540676,
-      "grad_norm": 4.4972100257873535,
-      "learning_rate": 1.722067039106145e-05,
-      "loss": 0.2431,
+      "grad_norm": 6.531003952026367,
+      "learning_rate": 5.740223463687151e-06,
+      "loss": 0.9288,
       "step": 770
     },
     {
       "epoch": 3.916145181476846,
-      "grad_norm": 3.6224372386932373,
-      "learning_rate": 1.7011173184357544e-05,
-      "loss": 0.2099,
+      "grad_norm": 3.672001600265503,
+      "learning_rate": 5.670391061452515e-06,
+      "loss": 0.8899,
       "step": 780
     },
     {
       "epoch": 3.966207759699625,
-      "grad_norm": 3.072998285293579,
-      "learning_rate": 1.680167597765363e-05,
-      "loss": 0.2273,
+      "grad_norm": 3.4693164825439453,
+      "learning_rate": 5.6005586592178775e-06,
+      "loss": 0.9107,
       "step": 790
     },
     {
       "epoch": 3.9962453066332917,
-      "eval_accuracy": 0.9771991762283024,
-      "eval_loss": 0.14364813268184662,
-      "eval_runtime": 4.9858,
-      "eval_samples_per_second": 1363.468,
-      "eval_steps_per_second": 42.721,
+      "eval_accuracy": 0.8961459252721389,
+      "eval_loss": 0.7700026035308838,
+      "eval_runtime": 4.7597,
+      "eval_samples_per_second": 1428.236,
+      "eval_steps_per_second": 44.751,
       "step": 796
     },
     {
       "epoch": 4.020025031289111,
-      "grad_norm": 3.3532874584198,
-      "learning_rate": 1.659217877094972e-05,
-      "loss": 0.23,
+      "grad_norm": 5.503833293914795,
+      "learning_rate": 5.530726256983241e-06,
+      "loss": 0.9085,
       "step": 800
     },
     {
       "epoch": 4.07008760951189,
-      "grad_norm": 2.384181261062622,
-      "learning_rate": 1.638268156424581e-05,
-      "loss": 0.2157,
+      "grad_norm": 5.365431785583496,
+      "learning_rate": 5.460893854748603e-06,
+      "loss": 0.908,
       "step": 810
     },
     {
       "epoch": 4.1201501877346685,
-      "grad_norm": 4.237877368927002,
-      "learning_rate": 1.61731843575419e-05,
-      "loss": 0.2141,
+      "grad_norm": 4.582799434661865,
+      "learning_rate": 5.391061452513967e-06,
+      "loss": 0.8559,
       "step": 820
     },
     {
       "epoch": 4.170212765957447,
-      "grad_norm": 3.8752825260162354,
-      "learning_rate": 1.5963687150837986e-05,
-      "loss": 0.214,
+      "grad_norm": 3.201195478439331,
+      "learning_rate": 5.3212290502793296e-06,
+      "loss": 0.8901,
       "step": 830
     },
     {
       "epoch": 4.220275344180226,
-      "grad_norm": 2.571617364883423,
-      "learning_rate": 1.575418994413408e-05,
-      "loss": 0.2124,
+      "grad_norm": 6.477810382843018,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 0.88,
       "step": 840
     },
     {
       "epoch": 4.270337922403003,
-      "grad_norm": 1.8986117839813232,
-      "learning_rate": 1.554469273743017e-05,
-      "loss": 0.2142,
+      "grad_norm": 3.525606870651245,
+      "learning_rate": 5.181564245810057e-06,
+      "loss": 0.8099,
       "step": 850
     },
     {
       "epoch": 4.320400500625782,
-      "grad_norm": 3.71712589263916,
-      "learning_rate": 1.533519553072626e-05,
-      "loss": 0.1939,
+      "grad_norm": 3.92795729637146,
+      "learning_rate": 5.111731843575419e-06,
+      "loss": 0.8326,
       "step": 860
     },
     {
       "epoch": 4.370463078848561,
-      "grad_norm": 2.1387696266174316,
-      "learning_rate": 1.5125698324022348e-05,
-      "loss": 0.1905,
+      "grad_norm": 4.019381523132324,
+      "learning_rate": 5.041899441340783e-06,
+      "loss": 0.7902,
       "step": 870
     },
     {
       "epoch": 4.420525657071339,
-      "grad_norm": 3.3053841590881348,
-      "learning_rate": 1.4916201117318435e-05,
-      "loss": 0.2156,
+      "grad_norm": 3.7563273906707764,
+      "learning_rate": 4.972067039106146e-06,
+      "loss": 0.8322,
       "step": 880
     },
     {
       "epoch": 4.470588235294118,
-      "grad_norm": 2.574657440185547,
-      "learning_rate": 1.4706703910614526e-05,
-      "loss": 0.198,
+      "grad_norm": 4.296588897705078,
+      "learning_rate": 4.902234636871509e-06,
+      "loss": 0.7835,
       "step": 890
     },
     {
       "epoch": 4.520650813516896,
-      "grad_norm": 2.290309429168701,
-      "learning_rate": 1.4497206703910616e-05,
-      "loss": 0.1754,
+      "grad_norm": 4.3726277351379395,
+      "learning_rate": 4.832402234636872e-06,
+      "loss": 0.7505,
       "step": 900
     },
     {
       "epoch": 4.570713391739675,
-      "grad_norm": 2.4253950119018555,
-      "learning_rate": 1.4287709497206705e-05,
-      "loss": 0.1971,
+      "grad_norm": 3.2456297874450684,
+      "learning_rate": 4.762569832402235e-06,
+      "loss": 0.7588,
       "step": 910
     },
     {
       "epoch": 4.6207759699624535,
-      "grad_norm": 3.070322275161743,
-      "learning_rate": 1.4078212290502793e-05,
-      "loss": 0.1928,
+      "grad_norm": 4.264461994171143,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 0.7566,
       "step": 920
     },
     {
       "epoch": 4.670838548185231,
-      "grad_norm": 1.5523446798324585,
-      "learning_rate": 1.3868715083798883e-05,
-      "loss": 0.1948,
+      "grad_norm": 3.387613296508789,
+      "learning_rate": 4.622905027932961e-06,
+      "loss": 0.7585,
       "step": 930
     },
     {
       "epoch": 4.72090112640801,
-      "grad_norm": 3.076679229736328,
-      "learning_rate": 1.3659217877094973e-05,
-      "loss": 0.1822,
+      "grad_norm": 4.655552864074707,
+      "learning_rate": 4.553072625698324e-06,
+      "loss": 0.7171,
       "step": 940
     },
     {
       "epoch": 4.7709637046307884,
-      "grad_norm": 3.6825084686279297,
-      "learning_rate": 1.344972067039106e-05,
-      "loss": 0.1954,
+      "grad_norm": 4.884917259216309,
+      "learning_rate": 4.4832402234636875e-06,
+      "loss": 0.789,
       "step": 950
     },
     {
       "epoch": 4.821026282853567,
-      "grad_norm": 4.037261009216309,
-      "learning_rate": 1.324022346368715e-05,
-      "loss": 0.1946,
+      "grad_norm": 5.926691055297852,
+      "learning_rate": 4.413407821229051e-06,
+      "loss": 0.7626,
       "step": 960
     },
     {
       "epoch": 4.871088861076346,
-      "grad_norm": 3.026543378829956,
-      "learning_rate": 1.3030726256983242e-05,
-      "loss": 0.2155,
+      "grad_norm": 5.080448150634766,
+      "learning_rate": 4.343575418994414e-06,
+      "loss": 0.7362,
       "step": 970
     },
     {
       "epoch": 4.921151439299124,
-      "grad_norm": 2.0362164974212646,
-      "learning_rate": 1.282122905027933e-05,
-      "loss": 0.1851,
+      "grad_norm": 3.3272018432617188,
+      "learning_rate": 4.273743016759777e-06,
+      "loss": 0.753,
       "step": 980
     },
     {
       "epoch": 4.971214017521902,
-      "grad_norm": 1.8788719177246094,
-      "learning_rate": 1.261173184357542e-05,
-      "loss": 0.1866,
+      "grad_norm": 4.017044544219971,
+      "learning_rate": 4.20391061452514e-06,
+      "loss": 0.7231,
       "step": 990
     },
     {
       "epoch": 4.996245306633291,
-      "eval_accuracy": 0.981759340982642,
-      "eval_loss": 0.11026876419782639,
-      "eval_runtime": 4.7749,
-      "eval_samples_per_second": 1423.702,
-      "eval_steps_per_second": 44.608,
+      "eval_accuracy": 0.9658723153868785,
+      "eval_loss": 0.6167460680007935,
+      "eval_runtime": 4.7239,
+      "eval_samples_per_second": 1439.062,
+      "eval_steps_per_second": 45.09,
       "step": 995
     },
     {
       "epoch": 5.025031289111389,
-      "grad_norm": 3.612476110458374,
-      "learning_rate": 1.2402234636871509e-05,
-      "loss": 0.1911,
+      "grad_norm": 3.775956392288208,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.7468,
       "step": 1000
     },
     {
       "epoch": 5.075093867334168,
-      "grad_norm": 1.3886641263961792,
-      "learning_rate": 1.2192737430167599e-05,
-      "loss": 0.1768,
+      "grad_norm": 3.734192371368408,
+      "learning_rate": 4.064245810055866e-06,
+      "loss": 0.7145,
       "step": 1010
     },
     {
       "epoch": 5.1251564455569465,
-      "grad_norm": 3.217656135559082,
-      "learning_rate": 1.1983240223463687e-05,
-      "loss": 0.1835,
+      "grad_norm": 5.420441150665283,
+      "learning_rate": 3.994413407821229e-06,
+      "loss": 0.6958,
       "step": 1020
     },
     {
       "epoch": 5.175219023779725,
-      "grad_norm": 2.281695604324341,
-      "learning_rate": 1.1773743016759776e-05,
-      "loss": 0.1858,
+      "grad_norm": 4.203883171081543,
+      "learning_rate": 3.9245810055865924e-06,
+      "loss": 0.714,
       "step": 1030
     },
     {
       "epoch": 5.225281602002503,
-      "grad_norm": 2.7055630683898926,
-      "learning_rate": 1.1564245810055866e-05,
-      "loss": 0.1718,
+      "grad_norm": 4.452067852020264,
+      "learning_rate": 3.854748603351956e-06,
+      "loss": 0.6791,
       "step": 1040
     },
     {
       "epoch": 5.275344180225281,
-      "grad_norm": 2.00937819480896,
-      "learning_rate": 1.1354748603351954e-05,
-      "loss": 0.1479,
+      "grad_norm": 3.5976009368896484,
+      "learning_rate": 3.7849162011173185e-06,
+      "loss": 0.656,
       "step": 1050
     },
     {
       "epoch": 5.32540675844806,
-      "grad_norm": 2.65446138381958,
-      "learning_rate": 1.1145251396648046e-05,
-      "loss": 0.1664,
+      "grad_norm": 3.9160947799682617,
+      "learning_rate": 3.715083798882682e-06,
+      "loss": 0.6789,
       "step": 1060
     },
     {
       "epoch": 5.375469336670839,
-      "grad_norm": 2.499176502227783,
-      "learning_rate": 1.0935754189944135e-05,
-      "loss": 0.1882,
+      "grad_norm": 3.4226489067077637,
+      "learning_rate": 3.6452513966480454e-06,
+      "loss": 0.6889,
       "step": 1070
     },
     {
       "epoch": 5.425531914893617,
-      "grad_norm": 3.318516492843628,
-      "learning_rate": 1.0726256983240223e-05,
-      "loss": 0.1823,
+      "grad_norm": 5.1838788986206055,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 0.6529,
       "step": 1080
     },
     {
       "epoch": 5.475594493116396,
-      "grad_norm": 2.1236233711242676,
-      "learning_rate": 1.0516759776536313e-05,
-      "loss": 0.1657,
+      "grad_norm": 3.7870497703552246,
+      "learning_rate": 3.5055865921787714e-06,
+      "loss": 0.6201,
       "step": 1090
     },
     {
       "epoch": 5.5256570713391735,
-      "grad_norm": 3.342689037322998,
-      "learning_rate": 1.0307262569832403e-05,
-      "loss": 0.1823,
+      "grad_norm": 5.270695686340332,
+      "learning_rate": 3.435754189944134e-06,
+      "loss": 0.6675,
       "step": 1100
     },
     {
       "epoch": 5.575719649561952,
-      "grad_norm": 2.687920331954956,
-      "learning_rate": 1.0097765363128492e-05,
-      "loss": 0.1888,
+      "grad_norm": 3.8207719326019287,
+      "learning_rate": 3.3659217877094974e-06,
+      "loss": 0.6715,
       "step": 1110
     },
     {
       "epoch": 5.625782227784731,
-      "grad_norm": 3.692422866821289,
-      "learning_rate": 9.88826815642458e-06,
-      "loss": 0.2053,
+      "grad_norm": 3.1934285163879395,
+      "learning_rate": 3.2960893854748607e-06,
+      "loss": 0.6449,
       "step": 1120
     },
     {
       "epoch": 5.675844806007509,
-      "grad_norm": 3.453005790710449,
-      "learning_rate": 9.67877094972067e-06,
-      "loss": 0.179,
+      "grad_norm": 4.760712623596191,
+      "learning_rate": 3.2262569832402235e-06,
+      "loss": 0.6331,
       "step": 1130
     },
     {
       "epoch": 5.725907384230288,
-      "grad_norm": 4.005608081817627,
-      "learning_rate": 9.46927374301676e-06,
-      "loss": 0.1748,
+      "grad_norm": 3.9951508045196533,
+      "learning_rate": 3.1564245810055867e-06,
+      "loss": 0.6289,
       "step": 1140
     },
     {
       "epoch": 5.7759699624530665,
-      "grad_norm": 2.1113505363464355,
-      "learning_rate": 9.25977653631285e-06,
-      "loss": 0.1574,
+      "grad_norm": 5.959083557128906,
+      "learning_rate": 3.0865921787709503e-06,
+      "loss": 0.6101,
       "step": 1150
     },
     {
       "epoch": 5.826032540675845,
-      "grad_norm": 4.529311180114746,
-      "learning_rate": 9.050279329608939e-06,
-      "loss": 0.1599,
+      "grad_norm": 5.146860599517822,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 0.6235,
       "step": 1160
     },
     {
       "epoch": 5.876095118898624,
-      "grad_norm": 1.885956048965454,
-      "learning_rate": 8.840782122905029e-06,
-      "loss": 0.1909,
+      "grad_norm": 4.0103230476379395,
+      "learning_rate": 2.9469273743016764e-06,
+      "loss": 0.6472,
       "step": 1170
     },
     {
       "epoch": 5.926157697121401,
-      "grad_norm": 2.369316816329956,
-      "learning_rate": 8.631284916201118e-06,
-      "loss": 0.1603,
+      "grad_norm": 4.349458694458008,
+      "learning_rate": 2.877094972067039e-06,
+      "loss": 0.5924,
       "step": 1180
     },
     {
       "epoch": 5.97622027534418,
-      "grad_norm": 1.402648687362671,
-      "learning_rate": 8.421787709497206e-06,
-      "loss": 0.1616,
+      "grad_norm": 4.416366100311279,
+      "learning_rate": 2.8072625698324024e-06,
+      "loss": 0.5972,
       "step": 1190
     },
     {
       "epoch": 5.996245306633291,
-      "eval_accuracy": 0.9819064430714917,
-      "eval_loss": 0.0981006771326065,
-      "eval_runtime": 4.775,
-      "eval_samples_per_second": 1423.653,
-      "eval_steps_per_second": 44.607,
+      "eval_accuracy": 0.9735216240070609,
+      "eval_loss": 0.48377424478530884,
+      "eval_runtime": 4.9277,
+      "eval_samples_per_second": 1379.544,
+      "eval_steps_per_second": 43.225,
       "step": 1194
     },
     {
       "epoch": 6.030037546933667,
-      "grad_norm": 1.6693238019943237,
-      "learning_rate": 8.212290502793296e-06,
-      "loss": 0.1931,
+      "grad_norm": 5.7863569259643555,
+      "learning_rate": 2.7374301675977656e-06,
+      "loss": 0.6391,
       "step": 1200
     },
     {
       "epoch": 6.080100125156446,
-      "grad_norm": 2.3462257385253906,
-      "learning_rate": 8.002793296089386e-06,
-      "loss": 0.1506,
+      "grad_norm": 3.5653481483459473,
+      "learning_rate": 2.6675977653631285e-06,
+      "loss": 0.577,
       "step": 1210
     },
     {
       "epoch": 6.130162703379224,
-      "grad_norm": 1.6939945220947266,
-      "learning_rate": 7.793296089385474e-06,
-      "loss": 0.1677,
+      "grad_norm": 2.4277288913726807,
+      "learning_rate": 2.5977653631284917e-06,
+      "loss": 0.5942,
       "step": 1220
     },
     {
       "epoch": 6.180225281602002,
-      "grad_norm": 1.728092908859253,
-      "learning_rate": 7.583798882681565e-06,
-      "loss": 0.1569,
+      "grad_norm": 4.4123406410217285,
+      "learning_rate": 2.5279329608938553e-06,
+      "loss": 0.5952,
       "step": 1230
     },
     {
       "epoch": 6.230287859824781,
-      "grad_norm": 1.6664111614227295,
-      "learning_rate": 7.374301675977653e-06,
-      "loss": 0.1564,
+      "grad_norm": 3.4356672763824463,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 0.5597,
       "step": 1240
     },
     {
       "epoch": 6.280350438047559,
-      "grad_norm": 2.0160274505615234,
-      "learning_rate": 7.164804469273744e-06,
-      "loss": 0.1513,
+      "grad_norm": 3.3665220737457275,
+      "learning_rate": 2.3882681564245814e-06,
+      "loss": 0.586,
       "step": 1250
     },
     {
       "epoch": 6.330413016270338,
-      "grad_norm": 4.013051986694336,
-      "learning_rate": 6.9553072625698325e-06,
-      "loss": 0.1594,
+      "grad_norm": 4.899147987365723,
+      "learning_rate": 2.318435754189944e-06,
+      "loss": 0.5691,
       "step": 1260
     },
     {
       "epoch": 6.380475594493117,
-      "grad_norm": 3.11110258102417,
-      "learning_rate": 6.745810055865922e-06,
-      "loss": 0.1445,
+      "grad_norm": 3.53438663482666,
+      "learning_rate": 2.2486033519553074e-06,
+      "loss": 0.5598,
       "step": 1270
     },
     {
       "epoch": 6.430538172715895,
-      "grad_norm": 3.418999433517456,
-      "learning_rate": 6.536312849162011e-06,
-      "loss": 0.1679,
+      "grad_norm": 2.7354676723480225,
+      "learning_rate": 2.1787709497206706e-06,
+      "loss": 0.5858,
       "step": 1280
     },
     {
       "epoch": 6.480600750938673,
-      "grad_norm": 2.4514362812042236,
-      "learning_rate": 6.326815642458101e-06,
-      "loss": 0.152,
+      "grad_norm": 4.669964790344238,
+      "learning_rate": 2.108938547486034e-06,
+      "loss": 0.5727,
       "step": 1290
     },
     {
       "epoch": 6.5306633291614515,
-      "grad_norm": 3.2242462635040283,
-      "learning_rate": 6.1173184357541904e-06,
-      "loss": 0.1676,
+      "grad_norm": 4.332780361175537,
+      "learning_rate": 2.039106145251397e-06,
+      "loss": 0.5416,
       "step": 1300
     },
     {
       "epoch": 6.58072590738423,
-      "grad_norm": 4.046393871307373,
-      "learning_rate": 5.907821229050279e-06,
-      "loss": 0.1615,
+      "grad_norm": 5.418635845184326,
+      "learning_rate": 1.96927374301676e-06,
+      "loss": 0.5533,
       "step": 1310
     },
     {
       "epoch": 6.630788485607009,
-      "grad_norm": 1.9088122844696045,
-      "learning_rate": 5.698324022346369e-06,
-      "loss": 0.1465,
+      "grad_norm": 2.963872194290161,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 0.5441,
       "step": 1320
     },
     {
       "epoch": 6.680851063829787,
-      "grad_norm": 2.5699033737182617,
-      "learning_rate": 5.488826815642458e-06,
-      "loss": 0.1472,
+      "grad_norm": 3.982882261276245,
+      "learning_rate": 1.8296089385474861e-06,
+      "loss": 0.5522,
       "step": 1330
     },
     {
       "epoch": 6.730913642052566,
-      "grad_norm": 1.9140872955322266,
-      "learning_rate": 5.2793296089385475e-06,
-      "loss": 0.1585,
+      "grad_norm": 3.4341752529144287,
+      "learning_rate": 1.7597765363128494e-06,
+      "loss": 0.5156,
       "step": 1340
     },
     {
       "epoch": 6.7809762202753445,
-      "grad_norm": 2.022095203399658,
-      "learning_rate": 5.069832402234637e-06,
-      "loss": 0.1455,
+      "grad_norm": 3.866319179534912,
+      "learning_rate": 1.6899441340782124e-06,
+      "loss": 0.543,
       "step": 1350
     },
     {
       "epoch": 6.831038798498122,
-      "grad_norm": 2.5366971492767334,
-      "learning_rate": 4.860335195530726e-06,
-      "loss": 0.1542,
+      "grad_norm": 4.521007537841797,
+      "learning_rate": 1.6201117318435754e-06,
+      "loss": 0.5466,
       "step": 1360
     },
     {
       "epoch": 6.881101376720901,
-      "grad_norm": 1.6112697124481201,
-      "learning_rate": 4.650837988826816e-06,
-      "loss": 0.1569,
+      "grad_norm": 4.014106273651123,
+      "learning_rate": 1.5502793296089386e-06,
+      "loss": 0.5131,
       "step": 1370
     },
     {
       "epoch": 6.931163954943679,
-      "grad_norm": 2.8735201358795166,
-      "learning_rate": 4.441340782122905e-06,
-      "loss": 0.1628,
+      "grad_norm": 4.237702369689941,
+      "learning_rate": 1.4804469273743019e-06,
+      "loss": 0.5626,
       "step": 1380
     },
     {
       "epoch": 6.981226533166458,
-      "grad_norm": 2.044304132461548,
-      "learning_rate": 4.231843575418994e-06,
-      "loss": 0.1385,
+      "grad_norm": 3.3088533878326416,
+      "learning_rate": 1.4106145251396649e-06,
+      "loss": 0.5143,
       "step": 1390
     },
     {
       "epoch": 6.996245306633291,
-      "eval_accuracy": 0.9830832597822889,
-      "eval_loss": 0.0956372618675232,
-      "eval_runtime": 4.8597,
-      "eval_samples_per_second": 1398.844,
-      "eval_steps_per_second": 43.83,
+      "eval_accuracy": 0.9761694616063548,
+      "eval_loss": 0.4227001368999481,
+      "eval_runtime": 4.8947,
+      "eval_samples_per_second": 1388.847,
+      "eval_steps_per_second": 43.516,
       "step": 1393
     },
     {
       "epoch": 7.035043804755945,
-      "grad_norm": 1.4923343658447266,
-      "learning_rate": 4.022346368715084e-06,
-      "loss": 0.1558,
+      "grad_norm": 3.7986509799957275,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 0.574,
       "step": 1400
     },
     {
       "epoch": 7.085106382978723,
-      "grad_norm": 3.339181423187256,
-      "learning_rate": 3.812849162011173e-06,
-      "loss": 0.1391,
+      "grad_norm": 4.211392879486084,
+      "learning_rate": 1.2709497206703911e-06,
+      "loss": 0.5194,
       "step": 1410
     },
     {
       "epoch": 7.135168961201502,
-      "grad_norm": 2.091777801513672,
-      "learning_rate": 3.6033519553072625e-06,
-      "loss": 0.1541,
+      "grad_norm": 3.6076323986053467,
+      "learning_rate": 1.2011173184357544e-06,
+      "loss": 0.5085,
       "step": 1420
     },
     {
       "epoch": 7.18523153942428,
-      "grad_norm": 2.6580100059509277,
-      "learning_rate": 3.393854748603352e-06,
-      "loss": 0.1379,
+      "grad_norm": 5.054622650146484,
+      "learning_rate": 1.1312849162011174e-06,
+      "loss": 0.5059,
       "step": 1430
     },
     {
       "epoch": 7.235294117647059,
-      "grad_norm": 2.4065537452697754,
-      "learning_rate": 3.1843575418994414e-06,
-      "loss": 0.1475,
+      "grad_norm": 5.9564313888549805,
+      "learning_rate": 1.0614525139664806e-06,
+      "loss": 0.509,
       "step": 1440
     },
     {
       "epoch": 7.2853566958698375,
-      "grad_norm": 2.618218183517456,
-      "learning_rate": 2.974860335195531e-06,
-      "loss": 0.1828,
+      "grad_norm": 5.2771220207214355,
+      "learning_rate": 9.916201117318436e-07,
+      "loss": 0.5613,
       "step": 1450
     },
     {
       "epoch": 7.335419274092616,
-      "grad_norm": 3.5904743671417236,
-      "learning_rate": 2.7653631284916204e-06,
-      "loss": 0.1365,
+      "grad_norm": 3.4556643962860107,
+      "learning_rate": 9.217877094972068e-07,
+      "loss": 0.5164,
       "step": 1460
     },
     {
       "epoch": 7.385481852315394,
-      "grad_norm": 2.245260000228882,
-      "learning_rate": 2.555865921787709e-06,
-      "loss": 0.1394,
+      "grad_norm": 4.0196003913879395,
+      "learning_rate": 8.519553072625699e-07,
+      "loss": 0.5255,
       "step": 1470
     },
     {
       "epoch": 7.435544430538172,
-      "grad_norm": 2.558086395263672,
-      "learning_rate": 2.346368715083799e-06,
-      "loss": 0.1533,
+      "grad_norm": 2.3283958435058594,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 0.5133,
       "step": 1480
     },
     {
       "epoch": 7.485607008760951,
-      "grad_norm": 2.851020097732544,
-      "learning_rate": 2.136871508379888e-06,
-      "loss": 0.1313,
+      "grad_norm": 3.3878517150878906,
+      "learning_rate": 7.122905027932961e-07,
+      "loss": 0.5028,
       "step": 1490
     },
     {
       "epoch": 7.53566958698373,
-      "grad_norm": 1.7011760473251343,
-      "learning_rate": 1.927374301675978e-06,
-      "loss": 0.1509,
+      "grad_norm": 4.05161190032959,
+      "learning_rate": 6.424581005586592e-07,
+      "loss": 0.5191,
       "step": 1500
     },
     {
       "epoch": 7.585732165206508,
-      "grad_norm": 2.6264467239379883,
-      "learning_rate": 1.717877094972067e-06,
-      "loss": 0.1515,
+      "grad_norm": 3.4434776306152344,
+      "learning_rate": 5.726256983240224e-07,
+      "loss": 0.5347,
       "step": 1510
     },
     {
       "epoch": 7.635794743429287,
-      "grad_norm": 1.6332521438598633,
-      "learning_rate": 1.5083798882681566e-06,
-      "loss": 0.1489,
+      "grad_norm": 3.151704788208008,
+      "learning_rate": 5.027932960893855e-07,
+      "loss": 0.5233,
       "step": 1520
     },
     {
       "epoch": 7.685857321652065,
-      "grad_norm": 2.0622401237487793,
-      "learning_rate": 1.2988826815642458e-06,
-      "loss": 0.1594,
+      "grad_norm": 4.1916046142578125,
+      "learning_rate": 4.3296089385474867e-07,
+      "loss": 0.5469,
       "step": 1530
     },
     {
       "epoch": 7.735919899874844,
-      "grad_norm": 2.3861618041992188,
-      "learning_rate": 1.0893854748603353e-06,
-      "loss": 0.1669,
+      "grad_norm": 5.041410446166992,
+      "learning_rate": 3.631284916201118e-07,
+      "loss": 0.5272,
       "step": 1540
     },
     {
       "epoch": 7.785982478097622,
-      "grad_norm": 4.30822229385376,
-      "learning_rate": 8.798882681564246e-07,
-      "loss": 0.1759,
+      "grad_norm": 4.148556709289551,
+      "learning_rate": 2.932960893854749e-07,
+      "loss": 0.5222,
       "step": 1550
     },
     {
       "epoch": 7.8360450563204,
-      "grad_norm": 1.4631046056747437,
-      "learning_rate": 6.70391061452514e-07,
-      "loss": 0.1645,
+      "grad_norm": 2.708613872528076,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 0.5405,
       "step": 1560
     },
     {
       "epoch": 7.886107634543179,
-      "grad_norm": 2.452613115310669,
-      "learning_rate": 4.608938547486033e-07,
-      "loss": 0.1575,
+      "grad_norm": 3.7867846488952637,
+      "learning_rate": 1.5363128491620113e-07,
+      "loss": 0.5249,
       "step": 1570
     },
     {
       "epoch": 7.9361702127659575,
-      "grad_norm": 1.279895305633545,
-      "learning_rate": 2.5139664804469275e-07,
-      "loss": 0.149,
+      "grad_norm": 3.1069061756134033,
+      "learning_rate": 8.379888268156426e-08,
+      "loss": 0.52,
       "step": 1580
     },
     {
       "epoch": 7.986232790988736,
-      "grad_norm": 2.141481399536133,
-      "learning_rate": 4.189944134078212e-08,
-      "loss": 0.1524,
+      "grad_norm": 4.343489646911621,
+      "learning_rate": 1.3966480446927376e-08,
+      "loss": 0.5159,
       "step": 1590
     },
     {
       "epoch": 7.996245306633291,
-      "eval_accuracy": 0.9824948514268903,
-      "eval_loss": 0.09257339686155319,
-      "eval_runtime": 5.6151,
-      "eval_samples_per_second": 1210.67,
-      "eval_steps_per_second": 37.934,
+      "eval_accuracy": 0.9763165636952045,
+      "eval_loss": 0.40624934434890747,
+      "eval_runtime": 5.5236,
+      "eval_samples_per_second": 1230.722,
+      "eval_steps_per_second": 38.562,
       "step": 1592
     },
     {
       "epoch": 7.996245306633291,
       "step": 1592,
       "total_flos": 3.777723239743488e+18,
-      "train_loss": 0.596273283347787,
-      "train_runtime": 640.7753,
-      "train_samples_per_second": 637.902,
-      "train_steps_per_second": 2.484
+      "train_loss": 1.152813568037359,
+      "train_runtime": 637.2674,
+      "train_samples_per_second": 641.414,
+      "train_steps_per_second": 2.498
     }
   ],
   "logging_steps": 10,