End of training

all_results.json

@@ -1,16 +1,16 @@
 {
     "avg_token_length": 41.888755020080325,
     "epoch": 10.0,
-    "eval_accuracy": 0.8357429504394531,
-    "eval_loss": 0.4303068220615387,
-    "eval_runtime": 3.5423,
+    "eval_accuracy": 0.8361445665359497,
+    "eval_loss": 0.4362991154193878,
+    "eval_runtime": 3.7997,
     "eval_samples": 2490,
-    "eval_samples_per_second": 702.935,
-    "eval_steps_per_second": 88.079,
-    "total_flos": 2.6187994846190592e+17,
-    "train_loss": 0.4432659422714054,
-    "train_runtime": 8098.9158,
+    "eval_samples_per_second": 655.309,
+    "eval_steps_per_second": 82.111,
+    "total_flos": 2.6143616931499008e+17,
+    "train_loss": 0.4748868410093066,
+    "train_runtime": 8727.3231,
     "train_samples": 392702,
-    "train_samples_per_second": 484.882,
-    "train_steps_per_second": 15.153
+    "train_samples_per_second": 449.968,
+    "train_steps_per_second": 14.062
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "avg_token_length": 41.888755020080325,
     "epoch": 10.0,
-    "eval_accuracy": 0.8357429504394531,
-    "eval_loss": 0.4303068220615387,
-    "eval_runtime": 3.5423,
+    "eval_accuracy": 0.8361445665359497,
+    "eval_loss": 0.4362991154193878,
+    "eval_runtime": 3.7997,
     "eval_samples": 2490,
-    "eval_samples_per_second": 702.935,
-    "eval_steps_per_second": 88.079
+    "eval_samples_per_second": 655.309,
+    "eval_steps_per_second": 82.111
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 10.0,
-    "total_flos": 2.6187994846190592e+17,
-    "train_loss": 0.4432659422714054,
-    "train_runtime": 8098.9158,
+    "total_flos": 2.6143616931499008e+17,
+    "train_loss": 0.4748868410093066,
+    "train_runtime": 8727.3231,
     "train_samples": 392702,
-    "train_samples_per_second": 484.882,
-    "train_steps_per_second": 15.153
+    "train_samples_per_second": 449.968,
+    "train_steps_per_second": 14.062
 }

trainer_state.json

@@ -10,1817 +10,1817 @@
   "log_history": [
     {
       "epoch": 0.04074315514993481,
-      "grad_norm": 3.142120361328125,
+      "grad_norm": 1.5828578472137451,
       "learning_rate": 0.0002987777053455019,
-      "loss": 0.9832,
+      "loss": 1.0838,
       "step": 500
     },
     {
       "epoch": 0.08148631029986962,
-      "grad_norm": 2.901562213897705,
+      "grad_norm": 2.0423190593719482,
       "learning_rate": 0.0002975554106910039,
-      "loss": 0.7704,
+      "loss": 0.8724,
       "step": 1000
     },
     {
       "epoch": 0.12222946544980444,
-      "grad_norm": 4.939942836761475,
+      "grad_norm": 2.746290683746338,
       "learning_rate": 0.00029633311603650584,
-      "loss": 0.7042,
+      "loss": 0.7804,
       "step": 1500
     },
     {
       "epoch": 0.16297262059973924,
-      "grad_norm": 4.045433044433594,
+      "grad_norm": 2.9885292053222656,
       "learning_rate": 0.0002951108213820078,
-      "loss": 0.6688,
+      "loss": 0.7367,
       "step": 2000
     },
     {
       "epoch": 0.20371577574967406,
-      "grad_norm": 4.076220989227295,
+      "grad_norm": 2.7808330059051514,
       "learning_rate": 0.00029388852672750977,
-      "loss": 0.6432,
+      "loss": 0.7016,
       "step": 2500
     },
     {
       "epoch": 0.24445893089960888,
-      "grad_norm": 3.716000556945801,
+      "grad_norm": 2.4769978523254395,
       "learning_rate": 0.0002926662320730117,
-      "loss": 0.6165,
+      "loss": 0.6759,
       "step": 3000
     },
     {
       "epoch": 0.28520208604954367,
-      "grad_norm": 3.549494981765747,
+      "grad_norm": 2.141312599182129,
       "learning_rate": 0.00029144393741851364,
-      "loss": 0.6124,
+      "loss": 0.6623,
       "step": 3500
     },
     {
       "epoch": 0.3259452411994785,
-      "grad_norm": 4.904158592224121,
+      "grad_norm": 2.169989824295044,
       "learning_rate": 0.0002902216427640156,
-      "loss": 0.587,
+      "loss": 0.6352,
       "step": 4000
     },
     {
       "epoch": 0.3666883963494133,
-      "grad_norm": 3.506251573562622,
+      "grad_norm": 3.238466262817383,
       "learning_rate": 0.0002889993481095176,
-      "loss": 0.5961,
+      "loss": 0.6382,
       "step": 4500
     },
     {
       "epoch": 0.4074315514993481,
-      "grad_norm": 3.886331558227539,
+      "grad_norm": 2.6361501216888428,
       "learning_rate": 0.00028777705345501956,
-      "loss": 0.5834,
+      "loss": 0.6168,
       "step": 5000
     },
     {
       "epoch": 0.44817470664928294,
-      "grad_norm": 5.864737510681152,
+      "grad_norm": 2.7179198265075684,
       "learning_rate": 0.0002865547588005215,
-      "loss": 0.5782,
+      "loss": 0.609,
       "step": 5500
     },
     {
       "epoch": 0.48891786179921776,
-      "grad_norm": 2.608105421066284,
+      "grad_norm": 1.717613935470581,
       "learning_rate": 0.00028533246414602344,
-      "loss": 0.5744,
+      "loss": 0.6023,
       "step": 6000
     },
     {
       "epoch": 0.5296610169491526,
-      "grad_norm": 3.3634021282196045,
+      "grad_norm": 1.5747934579849243,
       "learning_rate": 0.00028411016949152543,
-      "loss": 0.5754,
+      "loss": 0.5995,
       "step": 6500
     },
     {
       "epoch": 0.5704041720990873,
-      "grad_norm": 4.730838775634766,
+      "grad_norm": 2.5513648986816406,
       "learning_rate": 0.00028288787483702737,
-      "loss": 0.5568,
+      "loss": 0.5817,
       "step": 7000
     },
     {
       "epoch": 0.6111473272490222,
-      "grad_norm": 3.3426826000213623,
+      "grad_norm": 2.6356778144836426,
       "learning_rate": 0.0002816655801825293,
-      "loss": 0.5591,
+      "loss": 0.5816,
       "step": 7500
     },
     {
       "epoch": 0.651890482398957,
-      "grad_norm": 4.444048881530762,
+      "grad_norm": 1.4905149936676025,
       "learning_rate": 0.00028044328552803124,
-      "loss": 0.5623,
+      "loss": 0.587,
       "step": 8000
     },
     {
       "epoch": 0.6926336375488917,
-      "grad_norm": 2.7828869819641113,
+      "grad_norm": 2.126027822494507,
       "learning_rate": 0.00027922099087353323,
-      "loss": 0.5518,
+      "loss": 0.5763,
       "step": 8500
     },
     {
       "epoch": 0.7333767926988266,
-      "grad_norm": 4.194392681121826,
+      "grad_norm": 1.642021656036377,
       "learning_rate": 0.00027799869621903517,
-      "loss": 0.5591,
+      "loss": 0.5788,
       "step": 9000
     },
     {
       "epoch": 0.7741199478487614,
-      "grad_norm": 5.856390476226807,
+      "grad_norm": 2.919405460357666,
       "learning_rate": 0.0002767764015645371,
-      "loss": 0.5526,
+      "loss": 0.5756,
       "step": 9500
     },
     {
       "epoch": 0.8148631029986962,
-      "grad_norm": 4.61172342300415,
+      "grad_norm": 2.7401773929595947,
       "learning_rate": 0.0002755541069100391,
-      "loss": 0.54,
+      "loss": 0.561,
       "step": 10000
     },
     {
       "epoch": 0.855606258148631,
-      "grad_norm": 4.518310070037842,
+      "grad_norm": 3.268413543701172,
       "learning_rate": 0.00027433181225554104,
-      "loss": 0.5432,
+      "loss": 0.5627,
       "step": 10500
     },
     {
       "epoch": 0.8963494132985659,
-      "grad_norm": 3.190687417984009,
+      "grad_norm": 1.6420167684555054,
       "learning_rate": 0.000273109517601043,
-      "loss": 0.5424,
+      "loss": 0.5591,
       "step": 11000
     },
     {
       "epoch": 0.9370925684485006,
-      "grad_norm": 5.003259181976318,
+      "grad_norm": 3.6072638034820557,
       "learning_rate": 0.00027188722294654497,
-      "loss": 0.5408,
+      "loss": 0.5593,
       "step": 11500
     },
     {
       "epoch": 0.9778357235984355,
-      "grad_norm": 5.045399188995361,
+      "grad_norm": 1.9567480087280273,
       "learning_rate": 0.0002706649282920469,
-      "loss": 0.5345,
+      "loss": 0.5593,
       "step": 12000
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.810441792011261,
-      "eval_loss": 0.476459264755249,
-      "eval_runtime": 3.5774,
-      "eval_samples_per_second": 696.032,
-      "eval_steps_per_second": 87.214,
+      "eval_accuracy": 0.7959839105606079,
+      "eval_loss": 0.5131802558898926,
+      "eval_runtime": 3.8597,
+      "eval_samples_per_second": 645.123,
+      "eval_steps_per_second": 80.835,
       "step": 12272
     },
     {
       "epoch": 1.0185788787483703,
-      "grad_norm": 3.0707671642303467,
+      "grad_norm": 1.4843103885650635,
       "learning_rate": 0.00026944263363754884,
-      "loss": 0.5299,
+      "loss": 0.5502,
       "step": 12500
     },
     {
       "epoch": 1.0593220338983051,
-      "grad_norm": 3.0093071460723877,
+      "grad_norm": 2.488762855529785,
       "learning_rate": 0.00026822033898305083,
-      "loss": 0.5162,
+      "loss": 0.5391,
       "step": 13000
     },
     {
       "epoch": 1.1000651890482398,
-      "grad_norm": 5.579827785491943,
+      "grad_norm": 2.1324970722198486,
       "learning_rate": 0.00026699804432855277,
-      "loss": 0.5202,
+      "loss": 0.543,
       "step": 13500
     },
     {
       "epoch": 1.1408083441981747,
-      "grad_norm": 4.641988754272461,
+      "grad_norm": 1.2764211893081665,
       "learning_rate": 0.00026577574967405476,
-      "loss": 0.5164,
+      "loss": 0.5397,
       "step": 14000
     },
     {
       "epoch": 1.1815514993481095,
-      "grad_norm": 3.9216480255126953,
+      "grad_norm": 1.6294676065444946,
       "learning_rate": 0.0002645534550195567,
-      "loss": 0.5289,
+      "loss": 0.5487,
       "step": 14500
     },
     {
       "epoch": 1.2222946544980444,
-      "grad_norm": 4.214970588684082,
+      "grad_norm": 1.9771702289581299,
       "learning_rate": 0.00026333116036505864,
-      "loss": 0.5185,
+      "loss": 0.5365,
       "step": 15000
     },
     {
       "epoch": 1.263037809647979,
-      "grad_norm": 5.768319606781006,
+      "grad_norm": 2.3716676235198975,
       "learning_rate": 0.00026210886571056063,
-      "loss": 0.5232,
+      "loss": 0.5367,
       "step": 15500
     },
     {
       "epoch": 1.303780964797914,
-      "grad_norm": 5.26653528213501,
+      "grad_norm": 3.3372445106506348,
       "learning_rate": 0.00026088657105606257,
-      "loss": 0.5182,
+      "loss": 0.5339,
       "step": 16000
     },
     {
       "epoch": 1.3445241199478488,
-      "grad_norm": 3.733856201171875,
+      "grad_norm": 1.6658248901367188,
       "learning_rate": 0.0002596642764015645,
-      "loss": 0.5092,
+      "loss": 0.5306,
       "step": 16500
     },
     {
       "epoch": 1.3852672750977835,
-      "grad_norm": 5.003189563751221,
+      "grad_norm": 3.4499964714050293,
       "learning_rate": 0.0002584419817470665,
-      "loss": 0.525,
+      "loss": 0.5337,
       "step": 17000
     },
     {
       "epoch": 1.4260104302477183,
-      "grad_norm": 5.187269687652588,
+      "grad_norm": 1.6458457708358765,
       "learning_rate": 0.00025721968709256843,
-      "loss": 0.5141,
+      "loss": 0.5288,
       "step": 17500
     },
     {
       "epoch": 1.4667535853976532,
-      "grad_norm": 3.338898181915283,
+      "grad_norm": 1.9066393375396729,
       "learning_rate": 0.00025599739243807037,
-      "loss": 0.5059,
+      "loss": 0.5255,
       "step": 18000
     },
     {
       "epoch": 1.5074967405475879,
-      "grad_norm": 3.167060136795044,
+      "grad_norm": 1.735753059387207,
       "learning_rate": 0.0002547750977835723,
-      "loss": 0.5113,
+      "loss": 0.5282,
       "step": 18500
     },
     {
       "epoch": 1.548239895697523,
-      "grad_norm": 5.613772392272949,
+      "grad_norm": 1.7421000003814697,
       "learning_rate": 0.0002535528031290743,
-      "loss": 0.5002,
+      "loss": 0.5194,
       "step": 19000
     },
     {
       "epoch": 1.5889830508474576,
-      "grad_norm": 3.6054158210754395,
+      "grad_norm": 2.0979628562927246,
       "learning_rate": 0.00025233050847457624,
-      "loss": 0.5108,
+      "loss": 0.5334,
       "step": 19500
     },
     {
       "epoch": 1.6297262059973925,
-      "grad_norm": 3.535848617553711,
+      "grad_norm": 1.8151721954345703,
       "learning_rate": 0.0002511082138200782,
-      "loss": 0.5134,
+      "loss": 0.5283,
       "step": 20000
     },
     {
       "epoch": 1.6704693611473274,
-      "grad_norm": 2.494443655014038,
+      "grad_norm": 1.5404752492904663,
       "learning_rate": 0.00024988591916558017,
-      "loss": 0.5104,
+      "loss": 0.5232,
       "step": 20500
     },
     {
       "epoch": 1.711212516297262,
-      "grad_norm": 4.686164855957031,
+      "grad_norm": 3.021263837814331,
       "learning_rate": 0.0002486636245110821,
-      "loss": 0.4987,
+      "loss": 0.5184,
       "step": 21000
     },
     {
       "epoch": 1.7519556714471969,
-      "grad_norm": 6.178253173828125,
+      "grad_norm": 3.665238380432129,
       "learning_rate": 0.00024744132985658404,
-      "loss": 0.501,
+      "loss": 0.5166,
       "step": 21500
     },
     {
       "epoch": 1.7926988265971318,
-      "grad_norm": 3.4898393154144287,
+      "grad_norm": 1.4015684127807617,
       "learning_rate": 0.00024621903520208603,
-      "loss": 0.504,
+      "loss": 0.5158,
       "step": 22000
     },
     {
       "epoch": 1.8334419817470664,
-      "grad_norm": 5.974262237548828,
+      "grad_norm": 3.794595241546631,
       "learning_rate": 0.000244996740547588,
-      "loss": 0.5112,
+      "loss": 0.5267,
       "step": 22500
     },
     {
       "epoch": 1.8741851368970013,
-      "grad_norm": 4.1161980628967285,
+      "grad_norm": 1.9950809478759766,
       "learning_rate": 0.00024377444589308996,
-      "loss": 0.4934,
+      "loss": 0.5119,
       "step": 23000
     },
     {
       "epoch": 1.9149282920469362,
-      "grad_norm": 3.786954164505005,
+      "grad_norm": 1.5925782918930054,
       "learning_rate": 0.0002425521512385919,
-      "loss": 0.5051,
+      "loss": 0.5203,
       "step": 23500
     },
     {
       "epoch": 1.9556714471968708,
-      "grad_norm": 5.365992546081543,
+      "grad_norm": 2.9431841373443604,
       "learning_rate": 0.00024132985658409386,
-      "loss": 0.5011,
+      "loss": 0.512,
       "step": 24000
     },
     {
       "epoch": 1.996414602346806,
-      "grad_norm": 3.0433201789855957,
+      "grad_norm": 1.6975939273834229,
       "learning_rate": 0.00024010756192959583,
-      "loss": 0.4965,
+      "loss": 0.5079,
       "step": 24500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.8253012299537659,
-      "eval_loss": 0.4476940929889679,
-      "eval_runtime": 3.5371,
-      "eval_samples_per_second": 703.959,
-      "eval_steps_per_second": 88.207,
+      "eval_accuracy": 0.811646580696106,
+      "eval_loss": 0.4724164307117462,
+      "eval_runtime": 3.8494,
+      "eval_samples_per_second": 646.846,
+      "eval_steps_per_second": 81.051,
       "step": 24544
     },
     {
       "epoch": 2.0371577574967406,
-      "grad_norm": 5.630396366119385,
+      "grad_norm": 2.4534783363342285,
       "learning_rate": 0.00023888526727509777,
-      "loss": 0.4787,
+      "loss": 0.5027,
       "step": 25000
     },
     {
       "epoch": 2.077900912646675,
-      "grad_norm": 3.847637891769409,
+      "grad_norm": 2.0405538082122803,
       "learning_rate": 0.00023766297262059973,
-      "loss": 0.4754,
+      "loss": 0.5008,
       "step": 25500
     },
     {
       "epoch": 2.1186440677966103,
-      "grad_norm": 3.24627685546875,
+      "grad_norm": 1.833534836769104,
       "learning_rate": 0.00023644067796610167,
-      "loss": 0.4802,
+      "loss": 0.5043,
       "step": 26000
     },
     {
       "epoch": 2.159387222946545,
-      "grad_norm": 3.7128448486328125,
+      "grad_norm": 1.8968788385391235,
       "learning_rate": 0.00023521838331160363,
-      "loss": 0.483,
+      "loss": 0.5018,
       "step": 26500
     },
     {
       "epoch": 2.2001303780964796,
-      "grad_norm": 3.6413943767547607,
+      "grad_norm": 1.6663825511932373,
       "learning_rate": 0.0002339960886571056,
-      "loss": 0.48,
+      "loss": 0.501,
       "step": 27000
     },
     {
       "epoch": 2.2408735332464147,
-      "grad_norm": 5.281874179840088,
+      "grad_norm": 2.6435630321502686,
       "learning_rate": 0.00023277379400260753,
-      "loss": 0.4739,
+      "loss": 0.4976,
       "step": 27500
     },
     {
       "epoch": 2.2816166883963493,
-      "grad_norm": 3.1675126552581787,
+      "grad_norm": 1.5958212614059448,
       "learning_rate": 0.0002315514993481095,
-      "loss": 0.4868,
+      "loss": 0.5056,
       "step": 28000
     },
     {
       "epoch": 2.322359843546284,
-      "grad_norm": 2.996647596359253,
+      "grad_norm": 1.644378423690796,
       "learning_rate": 0.00023032920469361144,
-      "loss": 0.47,
+      "loss": 0.4856,
       "step": 28500
     },
     {
       "epoch": 2.363102998696219,
-      "grad_norm": 6.074376583099365,
+      "grad_norm": 1.9557281732559204,
       "learning_rate": 0.0002291069100391134,
-      "loss": 0.4747,
+      "loss": 0.4943,
       "step": 29000
     },
     {
       "epoch": 2.4038461538461537,
-      "grad_norm": 4.998762607574463,
+      "grad_norm": 3.529120922088623,
       "learning_rate": 0.00022788461538461537,
-      "loss": 0.4801,
+      "loss": 0.4956,
       "step": 29500
     },
     {
       "epoch": 2.444589308996089,
-      "grad_norm": 4.5490031242370605,
+      "grad_norm": 2.105900287628174,
       "learning_rate": 0.0002266623207301173,
-      "loss": 0.4852,
+      "loss": 0.5032,
       "step": 30000
     },
     {
       "epoch": 2.4853324641460235,
-      "grad_norm": 4.78202486038208,
+      "grad_norm": 2.1510558128356934,
       "learning_rate": 0.00022544002607561927,
-      "loss": 0.4871,
+      "loss": 0.5014,
       "step": 30500
     },
     {
       "epoch": 2.526075619295958,
-      "grad_norm": 4.470094680786133,
+      "grad_norm": 2.0106310844421387,
       "learning_rate": 0.0002242177314211212,
-      "loss": 0.4913,
+      "loss": 0.5047,
       "step": 31000
     },
     {
       "epoch": 2.5668187744458932,
-      "grad_norm": 8.678927421569824,
+      "grad_norm": 3.261289358139038,
       "learning_rate": 0.0002229954367666232,
-      "loss": 0.4852,
+      "loss": 0.5044,
       "step": 31500
     },
     {
       "epoch": 2.607561929595828,
-      "grad_norm": 4.836784362792969,
+      "grad_norm": 1.9700287580490112,
       "learning_rate": 0.00022177314211212516,
-      "loss": 0.4706,
+      "loss": 0.4921,
       "step": 32000
     },
     {
       "epoch": 2.648305084745763,
-      "grad_norm": 3.676816940307617,
+      "grad_norm": 1.7623224258422852,
       "learning_rate": 0.0002205508474576271,
-      "loss": 0.4759,
+      "loss": 0.493,
       "step": 32500
     },
     {
       "epoch": 2.6890482398956976,
-      "grad_norm": 4.985140800476074,
+      "grad_norm": 2.0307579040527344,
       "learning_rate": 0.00021932855280312906,
-      "loss": 0.478,
+      "loss": 0.4944,
       "step": 33000
     },
     {
       "epoch": 2.7297913950456323,
-      "grad_norm": 4.114302635192871,
+      "grad_norm": 1.6054081916809082,
       "learning_rate": 0.00021810625814863103,
-      "loss": 0.4802,
+      "loss": 0.495,
       "step": 33500
     },
     {
       "epoch": 2.770534550195567,
-      "grad_norm": 3.370340585708618,
+      "grad_norm": 1.2301048040390015,
       "learning_rate": 0.00021688396349413296,
-      "loss": 0.488,
+      "loss": 0.5006,
       "step": 34000
     },
     {
       "epoch": 2.811277705345502,
-      "grad_norm": 4.248477935791016,
+      "grad_norm": 2.173492670059204,
       "learning_rate": 0.00021566166883963493,
-      "loss": 0.4807,
+      "loss": 0.4952,
       "step": 34500
     },
     {
       "epoch": 2.8520208604954367,
-      "grad_norm": 4.306929588317871,
+      "grad_norm": 1.8370041847229004,
       "learning_rate": 0.0002144393741851369,
-      "loss": 0.4647,
+      "loss": 0.4774,
       "step": 35000
     },
     {
       "epoch": 2.8927640156453718,
-      "grad_norm": 4.3554253578186035,
+      "grad_norm": 2.2364652156829834,
       "learning_rate": 0.00021321707953063883,
-      "loss": 0.47,
+      "loss": 0.4848,
       "step": 35500
     },
     {
       "epoch": 2.9335071707953064,
-      "grad_norm": 4.758439064025879,
+      "grad_norm": 1.6861385107040405,
       "learning_rate": 0.0002119947848761408,
-      "loss": 0.4796,
+      "loss": 0.4982,
       "step": 36000
     },
     {
       "epoch": 2.974250325945241,
-      "grad_norm": 3.889279842376709,
+      "grad_norm": 1.938073992729187,
       "learning_rate": 0.00021077249022164273,
-      "loss": 0.465,
+      "loss": 0.4815,
       "step": 36500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.7819277048110962,
-      "eval_loss": 0.5279985070228577,
-      "eval_runtime": 3.5081,
-      "eval_samples_per_second": 709.784,
-      "eval_steps_per_second": 88.937,
+      "eval_accuracy": 0.7899598479270935,
+      "eval_loss": 0.5340293049812317,
+      "eval_runtime": 3.8273,
+      "eval_samples_per_second": 650.596,
+      "eval_steps_per_second": 81.521,
       "step": 36816
     },
     {
       "epoch": 3.014993481095176,
-      "grad_norm": 2.9451563358306885,
+      "grad_norm": 1.7496318817138672,
       "learning_rate": 0.0002095501955671447,
-      "loss": 0.4777,
+      "loss": 0.4926,
       "step": 37000
     },
     {
       "epoch": 3.055736636245111,
-      "grad_norm": 4.705958843231201,
+      "grad_norm": 1.6960176229476929,
       "learning_rate": 0.00020832790091264666,
-      "loss": 0.4567,
+      "loss": 0.4807,
       "step": 37500
     },
     {
       "epoch": 3.0964797913950455,
-      "grad_norm": 4.704124450683594,
+      "grad_norm": 1.9474419355392456,
       "learning_rate": 0.0002071056062581486,
-      "loss": 0.451,
+      "loss": 0.4753,
       "step": 38000
     },
     {
       "epoch": 3.1372229465449806,
-      "grad_norm": 3.8129329681396484,
+      "grad_norm": 2.938944101333618,
       "learning_rate": 0.00020588331160365056,
-      "loss": 0.4615,
+      "loss": 0.4869,
       "step": 38500
     },
     {
       "epoch": 3.1779661016949152,
-      "grad_norm": 4.089268207550049,
+      "grad_norm": 1.9578561782836914,
       "learning_rate": 0.0002046610169491525,
-      "loss": 0.45,
+      "loss": 0.473,
       "step": 39000
     },
     {
       "epoch": 3.21870925684485,
-      "grad_norm": 4.320204257965088,
+      "grad_norm": 2.039128541946411,
       "learning_rate": 0.00020343872229465447,
-      "loss": 0.4502,
+      "loss": 0.4744,
       "step": 39500
     },
     {
       "epoch": 3.259452411994785,
-      "grad_norm": 6.090836048126221,
+      "grad_norm": 3.681124687194824,
       "learning_rate": 0.00020221642764015643,
-      "loss": 0.4557,
+      "loss": 0.4799,
       "step": 40000
     },
     {
       "epoch": 3.3001955671447196,
-      "grad_norm": 3.542620897293091,
+      "grad_norm": 1.5929746627807617,
       "learning_rate": 0.0002009941329856584,
-      "loss": 0.4595,
+      "loss": 0.4787,
       "step": 40500
     },
     {
       "epoch": 3.3409387222946547,
-      "grad_norm": 3.8268799781799316,
+      "grad_norm": 1.7738901376724243,
       "learning_rate": 0.00019977183833116036,
-      "loss": 0.4514,
+      "loss": 0.4721,
       "step": 41000
     },
     {
       "epoch": 3.3816818774445894,
-      "grad_norm": 3.9095098972320557,
+      "grad_norm": 1.670168161392212,
       "learning_rate": 0.00019854954367666232,
-      "loss": 0.4637,
+      "loss": 0.4842,
       "step": 41500
     },
     {
       "epoch": 3.422425032594524,
-      "grad_norm": 4.160515785217285,
+      "grad_norm": 2.07709002494812,
       "learning_rate": 0.00019732724902216426,
-      "loss": 0.4458,
+      "loss": 0.4676,
       "step": 42000
     },
     {
       "epoch": 3.463168187744459,
-      "grad_norm": 3.263516426086426,
+      "grad_norm": 1.76602041721344,
       "learning_rate": 0.00019610495436766623,
-      "loss": 0.4592,
+      "loss": 0.478,
       "step": 42500
     },
     {
       "epoch": 3.5039113428943938,
-      "grad_norm": 4.349822044372559,
+      "grad_norm": 1.7546826601028442,
       "learning_rate": 0.00019488265971316816,
-      "loss": 0.463,
+      "loss": 0.481,
       "step": 43000
     },
     {
       "epoch": 3.5446544980443284,
-      "grad_norm": 4.780073642730713,
+      "grad_norm": 1.9292327165603638,
       "learning_rate": 0.00019366036505867013,
-      "loss": 0.4579,
+      "loss": 0.4763,
       "step": 43500
     },
     {
       "epoch": 3.5853976531942635,
-      "grad_norm": 5.213197708129883,
+      "grad_norm": 2.211700916290283,
       "learning_rate": 0.0001924380704041721,
-      "loss": 0.4537,
+      "loss": 0.4736,
       "step": 44000
     },
     {
       "epoch": 3.626140808344198,
-      "grad_norm": 5.990653038024902,
+      "grad_norm": 3.4379701614379883,
       "learning_rate": 0.00019121577574967403,
-      "loss": 0.4495,
+      "loss": 0.4691,
       "step": 44500
     },
     {
       "epoch": 3.666883963494133,
-      "grad_norm": 8.249653816223145,
+      "grad_norm": 3.3809897899627686,
       "learning_rate": 0.000189993481095176,
-      "loss": 0.4628,
+      "loss": 0.4865,
       "step": 45000
     },
     {
       "epoch": 3.707627118644068,
-      "grad_norm": 7.788904666900635,
+      "grad_norm": 2.958820343017578,
       "learning_rate": 0.00018877118644067796,
-      "loss": 0.4583,
+      "loss": 0.4773,
       "step": 45500
     },
     {
       "epoch": 3.7483702737940026,
-      "grad_norm": 4.790676116943359,
+      "grad_norm": 2.081221342086792,
       "learning_rate": 0.0001875488917861799,
-      "loss": 0.4479,
+      "loss": 0.4643,
       "step": 46000
     },
     {
       "epoch": 3.7891134289439377,
-      "grad_norm": 6.014172077178955,
+      "grad_norm": 1.3643114566802979,
       "learning_rate": 0.00018632659713168186,
-      "loss": 0.4482,
+      "loss": 0.4615,
       "step": 46500
     },
     {
       "epoch": 3.8298565840938723,
-      "grad_norm": 4.927757263183594,
+      "grad_norm": 2.669874668121338,
       "learning_rate": 0.0001851043024771838,
-      "loss": 0.4614,
+      "loss": 0.4786,
       "step": 47000
     },
     {
       "epoch": 3.870599739243807,
-      "grad_norm": 3.723923444747925,
+      "grad_norm": 1.4289213418960571,
       "learning_rate": 0.00018388200782268576,
-      "loss": 0.459,
+      "loss": 0.4796,
       "step": 47500
     },
     {
       "epoch": 3.9113428943937416,
-      "grad_norm": 3.931187391281128,
+      "grad_norm": 2.1719110012054443,
       "learning_rate": 0.00018265971316818773,
-      "loss": 0.4635,
+      "loss": 0.4798,
       "step": 48000
     },
     {
       "epoch": 3.9520860495436767,
-      "grad_norm": 3.8654162883758545,
+      "grad_norm": 1.5655969381332397,
       "learning_rate": 0.00018143741851368966,
-      "loss": 0.4427,
+      "loss": 0.4587,
       "step": 48500
     },
     {
       "epoch": 3.9928292046936114,
-      "grad_norm": 5.012121677398682,
+      "grad_norm": 2.242460250854492,
       "learning_rate": 0.00018021512385919163,
-      "loss": 0.4634,
+      "loss": 0.4757,
       "step": 49000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8244979977607727,
-      "eval_loss": 0.4430766999721527,
-      "eval_runtime": 3.5278,
-      "eval_samples_per_second": 705.821,
-      "eval_steps_per_second": 88.44,
+      "eval_accuracy": 0.823293149471283,
+      "eval_loss": 0.4503098726272583,
+      "eval_runtime": 3.7594,
+      "eval_samples_per_second": 662.343,
+      "eval_steps_per_second": 82.992,
       "step": 49088
     },
     {
       "epoch": 4.0335723598435465,
-      "grad_norm": 4.3685173988342285,
+      "grad_norm": 1.2779546976089478,
       "learning_rate": 0.00017899282920469362,
-      "loss": 0.4362,
+      "loss": 0.4646,
       "step": 49500
     },
     {
       "epoch": 4.074315514993481,
-      "grad_norm": 7.587422847747803,
+      "grad_norm": 2.8832106590270996,
       "learning_rate": 0.00017777053455019556,
-      "loss": 0.4204,
+      "loss": 0.4535,
       "step": 50000
     },
     {
       "epoch": 4.115058670143416,
-      "grad_norm": 6.1443963050842285,
+      "grad_norm": 2.621349811553955,
       "learning_rate": 0.00017654823989569752,
-      "loss": 0.4345,
+      "loss": 0.466,
       "step": 50500
     },
     {
       "epoch": 4.15580182529335,
-      "grad_norm": 5.416019439697266,
+      "grad_norm": 2.7702085971832275,
       "learning_rate": 0.00017532594524119946,
-      "loss": 0.4295,
+      "loss": 0.4535,
       "step": 51000
     },
     {
       "epoch": 4.196544980443286,
-      "grad_norm": 5.976702690124512,
+      "grad_norm": 2.0533089637756348,
       "learning_rate": 0.00017410365058670142,
-      "loss": 0.431,
+      "loss": 0.4539,
       "step": 51500
     },
     {
       "epoch": 4.237288135593221,
-      "grad_norm": 5.283466815948486,
+      "grad_norm": 2.3479714393615723,
       "learning_rate": 0.0001728813559322034,
-      "loss": 0.434,
+      "loss": 0.4536,
       "step": 52000
     },
     {
       "epoch": 4.278031290743155,
-      "grad_norm": 6.41757345199585,
+      "grad_norm": 2.5870485305786133,
       "learning_rate": 0.00017165906127770533,
-      "loss": 0.4283,
+      "loss": 0.4586,
       "step": 52500
     },
     {
       "epoch": 4.31877444589309,
-      "grad_norm": 6.220736980438232,
+      "grad_norm": 2.1694679260253906,
       "learning_rate": 0.0001704367666232073,
-      "loss": 0.4407,
+      "loss": 0.468,
       "step": 53000
     },
     {
       "epoch": 4.3595176010430245,
-      "grad_norm": 4.802826881408691,
+      "grad_norm": 2.3384058475494385,
       "learning_rate": 0.00016921447196870926,
-      "loss": 0.4345,
+      "loss": 0.4629,
       "step": 53500
     },
     {
       "epoch": 4.400260756192959,
-      "grad_norm": 3.7236640453338623,
+      "grad_norm": 1.984429121017456,
       "learning_rate": 0.0001679921773142112,
-      "loss": 0.4421,
+      "loss": 0.4609,
       "step": 54000
     },
     {
       "epoch": 4.441003911342895,
-      "grad_norm": 5.5577712059021,
+      "grad_norm": 2.1723473072052,
       "learning_rate": 0.00016676988265971316,
-      "loss": 0.4391,
+      "loss": 0.4569,
       "step": 54500
     },
     {
       "epoch": 4.481747066492829,
-      "grad_norm": 4.09696626663208,
+      "grad_norm": 1.6641265153884888,
       "learning_rate": 0.0001655475880052151,
-      "loss": 0.4376,
+      "loss": 0.4604,
       "step": 55000
     },
     {
       "epoch": 4.522490221642764,
-      "grad_norm": 4.795173645019531,
+      "grad_norm": 2.4631197452545166,
       "learning_rate": 0.00016432529335071706,
-      "loss": 0.4316,
+      "loss": 0.4534,
       "step": 55500
     },
     {
       "epoch": 4.563233376792699,
-      "grad_norm": 2.8243749141693115,
+      "grad_norm": 1.8890092372894287,
       "learning_rate": 0.00016310299869621902,
-      "loss": 0.4341,
+      "loss": 0.4594,
       "step": 56000
     },
     {
       "epoch": 4.603976531942633,
-      "grad_norm": 5.1735029220581055,
+      "grad_norm": 2.240000009536743,
       "learning_rate": 0.00016188070404172096,
-      "loss": 0.4402,
+      "loss": 0.459,
       "step": 56500
     },
     {
       "epoch": 4.644719687092568,
-      "grad_norm": 5.092055320739746,
+      "grad_norm": 2.3564798831939697,
       "learning_rate": 0.00016065840938722293,
-      "loss": 0.4432,
+      "loss": 0.4644,
       "step": 57000
     },
     {
       "epoch": 4.6854628422425035,
-      "grad_norm": 4.940371990203857,
+      "grad_norm": 2.3006091117858887,
       "learning_rate": 0.00015943611473272486,
-      "loss": 0.4393,
+      "loss": 0.4671,
       "step": 57500
     },
     {
       "epoch": 4.726205997392438,
-      "grad_norm": 5.340673446655273,
+      "grad_norm": 2.5902099609375,
       "learning_rate": 0.00015821382007822685,
-      "loss": 0.4337,
+      "loss": 0.4574,
       "step": 58000
     },
     {
       "epoch": 4.766949152542373,
-      "grad_norm": 4.699270248413086,
+      "grad_norm": 2.838531970977783,
       "learning_rate": 0.00015699152542372882,
-      "loss": 0.4385,
+      "loss": 0.4579,
       "step": 58500
     },
     {
       "epoch": 4.8076923076923075,
-      "grad_norm": 3.0422141551971436,
+      "grad_norm": 2.0663444995880127,
       "learning_rate": 0.00015576923076923076,
-      "loss": 0.4451,
+      "loss": 0.475,
       "step": 59000
     },
     {
       "epoch": 4.848435462842242,
-      "grad_norm": 4.510776996612549,
+      "grad_norm": 2.3260273933410645,
       "learning_rate": 0.00015454693611473272,
-      "loss": 0.4322,
+      "loss": 0.4564,
       "step": 59500
     },
     {
       "epoch": 4.889178617992178,
-      "grad_norm": 5.1884565353393555,
+      "grad_norm": 1.6334234476089478,
       "learning_rate": 0.00015332464146023469,
-      "loss": 0.4346,
+      "loss": 0.4582,
       "step": 60000
     },
     {
       "epoch": 4.929921773142112,
-      "grad_norm": 5.881906509399414,
+      "grad_norm": 3.73071026802063,
       "learning_rate": 0.00015210234680573662,
-      "loss": 0.435,
+      "loss": 0.462,
       "step": 60500
     },
     {
       "epoch": 4.970664928292047,
-      "grad_norm": 6.772756099700928,
+      "grad_norm": 2.0201449394226074,
       "learning_rate": 0.0001508800521512386,
-      "loss": 0.436,
+      "loss": 0.4575,
       "step": 61000
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.8345381617546082,
-      "eval_loss": 0.42114126682281494,
-      "eval_runtime": 3.525,
-      "eval_samples_per_second": 706.383,
-      "eval_steps_per_second": 88.511,
+      "eval_accuracy": 0.8321285247802734,
+      "eval_loss": 0.43152591586112976,
+      "eval_runtime": 3.7584,
+      "eval_samples_per_second": 662.509,
+      "eval_steps_per_second": 83.013,
       "step": 61360
     },
     {
       "epoch": 5.011408083441982,
-      "grad_norm": 6.8924736976623535,
+      "grad_norm": 3.175994396209717,
       "learning_rate": 0.00014965775749674052,
-      "loss": 0.4388,
+      "loss": 0.4644,
       "step": 61500
     },
     {
       "epoch": 5.052151238591916,
-      "grad_norm": 7.1412882804870605,
+      "grad_norm": 2.7221341133117676,
       "learning_rate": 0.0001484354628422425,
-      "loss": 0.4125,
+      "loss": 0.4479,
       "step": 62000
     },
     {
       "epoch": 5.092894393741851,
-      "grad_norm": 7.67710542678833,
+      "grad_norm": 2.5768377780914307,
       "learning_rate": 0.00014721316818774445,
-      "loss": 0.4091,
+      "loss": 0.4388,
       "step": 62500
     },
     {
       "epoch": 5.1336375488917865,
-      "grad_norm": 5.673561096191406,
+      "grad_norm": 2.2102932929992676,
       "learning_rate": 0.0001459908735332464,
-      "loss": 0.4138,
+      "loss": 0.4455,
       "step": 63000
     },
     {
       "epoch": 5.174380704041721,
-      "grad_norm": 5.589047908782959,
+      "grad_norm": 2.931208372116089,
       "learning_rate": 0.00014476857887874836,
-      "loss": 0.4174,
+      "loss": 0.4491,
       "step": 63500
     },
     {
       "epoch": 5.215123859191656,
-      "grad_norm": 3.591366767883301,
+      "grad_norm": 1.4483258724212646,
       "learning_rate": 0.00014354628422425032,
-      "loss": 0.4116,
+      "loss": 0.4412,
       "step": 64000
     },
     {
       "epoch": 5.25586701434159,
-      "grad_norm": 5.160641193389893,
+      "grad_norm": 1.9341588020324707,
       "learning_rate": 0.00014232398956975226,
-      "loss": 0.4168,
+      "loss": 0.4485,
       "step": 64500
     },
     {
       "epoch": 5.296610169491525,
-      "grad_norm": 7.442717552185059,
+      "grad_norm": 2.019665241241455,
       "learning_rate": 0.00014110169491525422,
-      "loss": 0.4137,
+      "loss": 0.443,
       "step": 65000
     },
     {
       "epoch": 5.337353324641461,
-      "grad_norm": 5.819495677947998,
+      "grad_norm": 2.324504852294922,
       "learning_rate": 0.0001398794002607562,
-      "loss": 0.4226,
+      "loss": 0.4574,
       "step": 65500
     },
     {
       "epoch": 5.378096479791395,
-      "grad_norm": 5.86204719543457,
+      "grad_norm": 1.5774611234664917,
       "learning_rate": 0.00013865710560625815,
-      "loss": 0.4158,
+      "loss": 0.4433,
       "step": 66000
     },
     {
       "epoch": 5.41883963494133,
-      "grad_norm": 5.929037094116211,
+      "grad_norm": 2.7634074687957764,
       "learning_rate": 0.0001374348109517601,
-      "loss": 0.4139,
+      "loss": 0.439,
       "step": 66500
     },
     {
       "epoch": 5.459582790091265,
-      "grad_norm": 6.2193427085876465,
+      "grad_norm": 3.11040997505188,
       "learning_rate": 0.00013621251629726205,
-      "loss": 0.4278,
+      "loss": 0.461,
       "step": 67000
     },
     {
       "epoch": 5.500325945241199,
-      "grad_norm": 3.6446192264556885,
+      "grad_norm": 1.974280834197998,
       "learning_rate": 0.000134990221642764,
-      "loss": 0.4263,
+      "loss": 0.4529,
       "step": 67500
     },
     {
       "epoch": 5.541069100391134,
-      "grad_norm": 4.629395961761475,
+      "grad_norm": 2.044983386993408,
       "learning_rate": 0.00013376792698826596,
-      "loss": 0.4272,
+      "loss": 0.4569,
       "step": 68000
     },
     {
       "epoch": 5.581812255541069,
-      "grad_norm": 4.330788612365723,
+      "grad_norm": 3.2112531661987305,
       "learning_rate": 0.00013254563233376792,
-      "loss": 0.4189,
+      "loss": 0.4464,
       "step": 68500
     },
     {
       "epoch": 5.622555410691004,
-      "grad_norm": 6.034902572631836,
+      "grad_norm": 3.1195149421691895,
       "learning_rate": 0.00013132333767926986,
-      "loss": 0.4214,
+      "loss": 0.4479,
       "step": 69000
     },
     {
       "epoch": 5.663298565840939,
-      "grad_norm": 4.091671943664551,
+      "grad_norm": 1.791577935218811,
       "learning_rate": 0.00013010104302477182,
-      "loss": 0.4124,
+      "loss": 0.4409,
       "step": 69500
     },
     {
       "epoch": 5.704041720990873,
-      "grad_norm": 5.487693786621094,
+      "grad_norm": 3.3514599800109863,
       "learning_rate": 0.00012887874837027379,
-      "loss": 0.4242,
+      "loss": 0.4477,
       "step": 70000
     },
     {
       "epoch": 5.744784876140808,
-      "grad_norm": 5.3786115646362305,
+      "grad_norm": 1.6975388526916504,
       "learning_rate": 0.00012765645371577575,
-      "loss": 0.4298,
+      "loss": 0.4539,
       "step": 70500
     },
     {
       "epoch": 5.7855280312907436,
-      "grad_norm": 4.010681629180908,
+      "grad_norm": 1.7184616327285767,
       "learning_rate": 0.0001264341590612777,
-      "loss": 0.4253,
+      "loss": 0.4538,
       "step": 71000
     },
     {
       "epoch": 5.826271186440678,
-      "grad_norm": 5.500887870788574,
+      "grad_norm": 2.1258814334869385,
       "learning_rate": 0.00012521186440677965,
-      "loss": 0.4245,
+      "loss": 0.4477,
       "step": 71500
     },
     {
       "epoch": 5.867014341590613,
-      "grad_norm": 5.5760931968688965,
+      "grad_norm": 3.146895170211792,
       "learning_rate": 0.0001239895697522816,
-      "loss": 0.4102,
+      "loss": 0.4393,
       "step": 72000
     },
     {
       "epoch": 5.9077574967405475,
-      "grad_norm": 5.079644203186035,
+      "grad_norm": 3.5936505794525146,
       "learning_rate": 0.00012276727509778355,
-      "loss": 0.4201,
+      "loss": 0.4496,
       "step": 72500
     },
     {
       "epoch": 5.948500651890482,
-      "grad_norm": 5.812402248382568,
+      "grad_norm": 2.3087146282196045,
       "learning_rate": 0.0001215449804432855,
-      "loss": 0.4122,
+      "loss": 0.4469,
       "step": 73000
     },
     {
       "epoch": 5.989243807040417,
-      "grad_norm": 19.5577335357666,
+      "grad_norm": 1.9528945684432983,
       "learning_rate": 0.00012032268578878747,
-      "loss": 0.4143,
+      "loss": 0.4435,
       "step": 73500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.8341365456581116,
-      "eval_loss": 0.4439743757247925,
-      "eval_runtime": 3.5304,
-      "eval_samples_per_second": 705.293,
-      "eval_steps_per_second": 88.374,
+      "eval_accuracy": 0.8369477987289429,
+      "eval_loss": 0.4312504529953003,
+      "eval_runtime": 3.772,
+      "eval_samples_per_second": 660.131,
+      "eval_steps_per_second": 82.715,
       "step": 73632
     },
     {
       "epoch": 6.029986962190352,
-      "grad_norm": 5.7537946701049805,
+      "grad_norm": 2.099862813949585,
       "learning_rate": 0.00011910039113428943,
-      "loss": 0.4095,
+      "loss": 0.4439,
       "step": 74000
     },
     {
       "epoch": 6.070730117340287,
-      "grad_norm": 4.046775817871094,
+      "grad_norm": 1.8885704278945923,
       "learning_rate": 0.00011787809647979139,
-      "loss": 0.404,
+      "loss": 0.4367,
       "step": 74500
     },
     {
       "epoch": 6.111473272490222,
-      "grad_norm": 5.266561508178711,
+      "grad_norm": 2.2419934272766113,
       "learning_rate": 0.00011665580182529335,
-      "loss": 0.3959,
+      "loss": 0.4308,
       "step": 75000
     },
     {
       "epoch": 6.152216427640156,
-      "grad_norm": 4.223605632781982,
+      "grad_norm": 1.9617372751235962,
       "learning_rate": 0.0001154335071707953,
-      "loss": 0.3998,
+      "loss": 0.4325,
       "step": 75500
     },
     {
       "epoch": 6.192959582790091,
-      "grad_norm": 4.90071964263916,
+      "grad_norm": 2.785778522491455,
       "learning_rate": 0.00011421121251629725,
-      "loss": 0.398,
+      "loss": 0.4363,
       "step": 76000
     },
     {
       "epoch": 6.2337027379400265,
-      "grad_norm": 4.251009464263916,
+      "grad_norm": 1.9203062057495117,
       "learning_rate": 0.0001129889178617992,
-      "loss": 0.4121,
+      "loss": 0.444,
       "step": 76500
     },
     {
       "epoch": 6.274445893089961,
-      "grad_norm": 4.648316383361816,
+      "grad_norm": 2.095256805419922,
       "learning_rate": 0.00011176662320730115,
-      "loss": 0.4081,
+      "loss": 0.4443,
       "step": 77000
     },
     {
       "epoch": 6.315189048239896,
-      "grad_norm": 3.8864428997039795,
+      "grad_norm": 3.069437265396118,
       "learning_rate": 0.00011054432855280312,
-      "loss": 0.3964,
+      "loss": 0.4275,
       "step": 77500
     },
     {
       "epoch": 6.3559322033898304,
-      "grad_norm": 6.131734371185303,
+      "grad_norm": 2.62670636177063,
       "learning_rate": 0.00010932203389830507,
-      "loss": 0.4089,
+      "loss": 0.4411,
       "step": 78000
     },
     {
       "epoch": 6.396675358539765,
-      "grad_norm": 6.840878009796143,
+      "grad_norm": 3.285296678543091,
       "learning_rate": 0.00010809973924380703,
-      "loss": 0.4138,
+      "loss": 0.4512,
       "step": 78500
     },
     {
       "epoch": 6.4374185136897,
-      "grad_norm": 4.359363079071045,
+      "grad_norm": 1.5842539072036743,
       "learning_rate": 0.00010687744458930898,
-      "loss": 0.4031,
+      "loss": 0.4358,
       "step": 79000
     },
     {
       "epoch": 6.478161668839635,
-      "grad_norm": 5.549428462982178,
+      "grad_norm": 2.1644914150238037,
       "learning_rate": 0.00010565514993481095,
-      "loss": 0.4112,
+      "loss": 0.4418,
       "step": 79500
     },
     {
       "epoch": 6.51890482398957,
-      "grad_norm": 3.57088041305542,
+      "grad_norm": 2.471137523651123,
       "learning_rate": 0.0001044328552803129,
-      "loss": 0.4033,
+      "loss": 0.4356,
       "step": 80000
     },
     {
       "epoch": 6.559647979139505,
-      "grad_norm": 4.713502883911133,
+      "grad_norm": 2.335623025894165,
       "learning_rate": 0.00010321056062581485,
-      "loss": 0.4129,
+      "loss": 0.4406,
       "step": 80500
     },
     {
       "epoch": 6.600391134289439,
-      "grad_norm": 5.755394458770752,
+      "grad_norm": 4.2139081954956055,
       "learning_rate": 0.0001019882659713168,
-      "loss": 0.4001,
+      "loss": 0.4333,
       "step": 81000
     },
     {
       "epoch": 6.641134289439374,
-      "grad_norm": 5.579432010650635,
+      "grad_norm": 2.9831790924072266,
       "learning_rate": 0.00010076597131681877,
-      "loss": 0.4057,
+      "loss": 0.4409,
       "step": 81500
     },
     {
       "epoch": 6.681877444589309,
-      "grad_norm": 5.298727512359619,
+      "grad_norm": 1.6623711585998535,
       "learning_rate": 9.954367666232072e-05,
-      "loss": 0.4069,
+      "loss": 0.4375,
       "step": 82000
     },
     {
       "epoch": 6.722620599739244,
-      "grad_norm": 6.568725109100342,
+      "grad_norm": 2.8634488582611084,
       "learning_rate": 9.832138200782268e-05,
-      "loss": 0.3974,
+      "loss": 0.4288,
       "step": 82500
     },
     {
       "epoch": 6.763363754889179,
-      "grad_norm": 5.902369022369385,
+      "grad_norm": 3.4033591747283936,
       "learning_rate": 9.709908735332463e-05,
-      "loss": 0.3998,
+      "loss": 0.4357,
       "step": 83000
     },
     {
       "epoch": 6.804106910039113,
-      "grad_norm": 5.710369110107422,
+      "grad_norm": 2.3420987129211426,
       "learning_rate": 9.58767926988266e-05,
-      "loss": 0.4094,
+      "loss": 0.4443,
       "step": 83500
     },
     {
       "epoch": 6.844850065189048,
-      "grad_norm": 7.27541446685791,
+      "grad_norm": 2.943692684173584,
       "learning_rate": 9.465449804432855e-05,
-      "loss": 0.3889,
+      "loss": 0.4233,
       "step": 84000
     },
     {
       "epoch": 6.885593220338983,
-      "grad_norm": 2.905696153640747,
+      "grad_norm": 1.5020798444747925,
       "learning_rate": 9.34322033898305e-05,
-      "loss": 0.3988,
+      "loss": 0.4299,
       "step": 84500
     },
     {
       "epoch": 6.926336375488918,
-      "grad_norm": 4.89232063293457,
+      "grad_norm": 1.8622304201126099,
       "learning_rate": 9.220990873533245e-05,
-      "loss": 0.3921,
+      "loss": 0.4253,
       "step": 85000
     },
     {
       "epoch": 6.967079530638853,
-      "grad_norm": 6.2112274169921875,
+      "grad_norm": 3.1434714794158936,
       "learning_rate": 9.098761408083442e-05,
-      "loss": 0.3907,
+      "loss": 0.4228,
       "step": 85500
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.8321285247802734,
-      "eval_loss": 0.43350040912628174,
-      "eval_runtime": 3.5095,
-      "eval_samples_per_second": 709.513,
-      "eval_steps_per_second": 88.903,
+      "eval_accuracy": 0.8309236764907837,
+      "eval_loss": 0.43813642859458923,
+      "eval_runtime": 3.7707,
+      "eval_samples_per_second": 660.357,
+      "eval_steps_per_second": 82.743,
       "step": 85904
     },
     {
       "epoch": 7.0078226857887875,
-      "grad_norm": 6.440418720245361,
+      "grad_norm": 2.7635395526885986,
       "learning_rate": 8.976531942633637e-05,
-      "loss": 0.3996,
+      "loss": 0.4382,
       "step": 86000
     },
     {
       "epoch": 7.048565840938722,
-      "grad_norm": 5.757605075836182,
+      "grad_norm": 2.407377004623413,
       "learning_rate": 8.854302477183832e-05,
-      "loss": 0.377,
+      "loss": 0.4138,
       "step": 86500
     },
     {
       "epoch": 7.089308996088657,
-      "grad_norm": 5.203352928161621,
+      "grad_norm": 2.3589298725128174,
       "learning_rate": 8.732073011734028e-05,
-      "loss": 0.3928,
+      "loss": 0.4307,
       "step": 87000
     },
     {
       "epoch": 7.130052151238592,
-      "grad_norm": 5.605312347412109,
+      "grad_norm": 2.2621631622314453,
       "learning_rate": 8.609843546284225e-05,
-      "loss": 0.3863,
+      "loss": 0.4195,
       "step": 87500
     },
     {
       "epoch": 7.170795306388527,
-      "grad_norm": 5.28822660446167,
+      "grad_norm": 2.3238487243652344,
       "learning_rate": 8.48761408083442e-05,
-      "loss": 0.3843,
+      "loss": 0.4251,
       "step": 88000
     },
     {
       "epoch": 7.211538461538462,
-      "grad_norm": 4.207997798919678,
+      "grad_norm": 1.8608829975128174,
       "learning_rate": 8.365384615384615e-05,
-      "loss": 0.3886,
+      "loss": 0.4318,
       "step": 88500
     },
     {
       "epoch": 7.252281616688396,
-      "grad_norm": 4.063730239868164,
+      "grad_norm": 1.5675103664398193,
       "learning_rate": 8.24315514993481e-05,
-      "loss": 0.3914,
+      "loss": 0.43,
       "step": 89000
     },
     {
       "epoch": 7.293024771838331,
-      "grad_norm": 6.433591365814209,
+      "grad_norm": 3.2853474617004395,
       "learning_rate": 8.120925684485006e-05,
-      "loss": 0.3926,
+      "loss": 0.4336,
       "step": 89500
     },
     {
       "epoch": 7.333767926988266,
-      "grad_norm": 3.1585044860839844,
+      "grad_norm": 2.1077964305877686,
       "learning_rate": 7.998696219035201e-05,
-      "loss": 0.39,
+      "loss": 0.4304,
       "step": 90000
     },
     {
       "epoch": 7.374511082138201,
-      "grad_norm": 5.051187992095947,
+      "grad_norm": 2.3823015689849854,
       "learning_rate": 7.876466753585397e-05,
-      "loss": 0.3854,
+      "loss": 0.4269,
       "step": 90500
     },
     {
       "epoch": 7.415254237288136,
-      "grad_norm": 5.842960834503174,
+      "grad_norm": 2.7820513248443604,
       "learning_rate": 7.754237288135592e-05,
-      "loss": 0.3748,
+      "loss": 0.4152,
       "step": 91000
     },
     {
       "epoch": 7.4559973924380705,
-      "grad_norm": 3.273320198059082,
+      "grad_norm": 1.7721502780914307,
       "learning_rate": 7.63200782268579e-05,
-      "loss": 0.3885,
+      "loss": 0.4286,
       "step": 91500
     },
     {
       "epoch": 7.496740547588005,
-      "grad_norm": 4.630128860473633,
+      "grad_norm": 1.8405511379241943,
       "learning_rate": 7.509778357235985e-05,
-      "loss": 0.3914,
+      "loss": 0.4298,
       "step": 92000
     },
     {
       "epoch": 7.53748370273794,
-      "grad_norm": 4.444061279296875,
+      "grad_norm": 1.8865771293640137,
       "learning_rate": 7.387548891786178e-05,
-      "loss": 0.3882,
+      "loss": 0.4285,
       "step": 92500
     },
     {
       "epoch": 7.578226857887875,
-      "grad_norm": 4.932531833648682,
+      "grad_norm": 3.2232296466827393,
       "learning_rate": 7.265319426336375e-05,
-      "loss": 0.3886,
+      "loss": 0.4262,
       "step": 93000
     },
     {
       "epoch": 7.61897001303781,
-      "grad_norm": 5.399273872375488,
+      "grad_norm": 2.0036380290985107,
       "learning_rate": 7.14308996088657e-05,
-      "loss": 0.3937,
+      "loss": 0.4351,
       "step": 93500
     },
     {
       "epoch": 7.659713168187745,
-      "grad_norm": 7.287408351898193,
+      "grad_norm": 2.9521334171295166,
       "learning_rate": 7.020860495436766e-05,
-      "loss": 0.3783,
+      "loss": 0.4169,
       "step": 94000
     },
     {
       "epoch": 7.700456323337679,
-      "grad_norm": 6.242406368255615,
+      "grad_norm": 2.6853766441345215,
       "learning_rate": 6.898631029986961e-05,
-      "loss": 0.395,
+      "loss": 0.4318,
       "step": 94500
     },
     {
       "epoch": 7.741199478487614,
-      "grad_norm": 3.794053077697754,
+      "grad_norm": 2.0400960445404053,
       "learning_rate": 6.776401564537158e-05,
-      "loss": 0.3967,
+      "loss": 0.4353,
       "step": 95000
     },
     {
       "epoch": 7.781942633637549,
-      "grad_norm": 7.381150245666504,
+      "grad_norm": 3.423992156982422,
       "learning_rate": 6.654172099087353e-05,
-      "loss": 0.3837,
+      "loss": 0.4236,
       "step": 95500
     },
     {
       "epoch": 7.822685788787483,
-      "grad_norm": 4.904209613800049,
+      "grad_norm": 2.914910078048706,
       "learning_rate": 6.531942633637548e-05,
-      "loss": 0.3874,
+      "loss": 0.4246,
       "step": 96000
     },
     {
       "epoch": 7.863428943937419,
-      "grad_norm": 4.649083137512207,
+      "grad_norm": 2.891737699508667,
       "learning_rate": 6.409713168187743e-05,
-      "loss": 0.389,
+      "loss": 0.4278,
       "step": 96500
     },
     {
       "epoch": 7.904172099087353,
-      "grad_norm": 7.420172691345215,
+      "grad_norm": 4.279940605163574,
       "learning_rate": 6.28748370273794e-05,
-      "loss": 0.3877,
+      "loss": 0.4281,
       "step": 97000
     },
     {
       "epoch": 7.944915254237288,
-      "grad_norm": 4.977739334106445,
+      "grad_norm": 3.536177158355713,
       "learning_rate": 6.165254237288135e-05,
-      "loss": 0.3885,
+      "loss": 0.4283,
       "step": 97500
     },
     {
       "epoch": 7.985658409387223,
-      "grad_norm": 5.967884063720703,
+      "grad_norm": 2.03955078125,
       "learning_rate": 6.0430247718383304e-05,
-      "loss": 0.3899,
+      "loss": 0.4342,
       "step": 98000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.8401606678962708,
-      "eval_loss": 0.421245813369751,
-      "eval_runtime": 3.5351,
-      "eval_samples_per_second": 704.361,
-      "eval_steps_per_second": 88.257,
+      "eval_accuracy": 0.8365461826324463,
+      "eval_loss": 0.4348722994327545,
+      "eval_runtime": 3.765,
+      "eval_samples_per_second": 661.356,
+      "eval_steps_per_second": 82.869,
       "step": 98176
     },
     {
       "epoch": 8.026401564537158,
-      "grad_norm": 3.8275697231292725,
+      "grad_norm": 3.3217365741729736,
       "learning_rate": 5.920795306388526e-05,
-      "loss": 0.3753,
+      "loss": 0.4183,
       "step": 98500
     },
     {
       "epoch": 8.067144719687093,
-      "grad_norm": 5.281260967254639,
+      "grad_norm": 2.7104313373565674,
       "learning_rate": 5.798565840938721e-05,
-      "loss": 0.3657,
+      "loss": 0.4115,
       "step": 99000
     },
     {
       "epoch": 8.107887874837028,
-      "grad_norm": 4.99520206451416,
+      "grad_norm": 1.903761386871338,
       "learning_rate": 5.676336375488918e-05,
-      "loss": 0.365,
+      "loss": 0.4142,
       "step": 99500
     },
     {
       "epoch": 8.148631029986962,
-      "grad_norm": 8.345664978027344,
+      "grad_norm": 3.378157615661621,
       "learning_rate": 5.554106910039113e-05,
-      "loss": 0.3766,
+      "loss": 0.4248,
       "step": 100000
     },
     {
       "epoch": 8.189374185136897,
-      "grad_norm": 7.439541339874268,
+      "grad_norm": 2.3883683681488037,
       "learning_rate": 5.4318774445893086e-05,
-      "loss": 0.3695,
+      "loss": 0.4153,
       "step": 100500
     },
     {
       "epoch": 8.230117340286832,
-      "grad_norm": 18.46923828125,
+      "grad_norm": 3.1941487789154053,
       "learning_rate": 5.309647979139504e-05,
-      "loss": 0.3751,
+      "loss": 0.4202,
       "step": 101000
     },
     {
       "epoch": 8.270860495436766,
-      "grad_norm": 7.099632740020752,
+      "grad_norm": 2.556144952774048,
       "learning_rate": 5.1874185136897e-05,
-      "loss": 0.3747,
+      "loss": 0.4212,
       "step": 101500
     },
     {
       "epoch": 8.3116036505867,
-      "grad_norm": 5.036531925201416,
+      "grad_norm": 2.3176252841949463,
       "learning_rate": 5.065189048239895e-05,
-      "loss": 0.3736,
+      "loss": 0.4153,
       "step": 102000
     },
     {
       "epoch": 8.352346805736635,
-      "grad_norm": 5.257627010345459,
+      "grad_norm": 2.065124750137329,
       "learning_rate": 4.942959582790091e-05,
-      "loss": 0.3718,
+      "loss": 0.4213,
       "step": 102500
     },
     {
       "epoch": 8.393089960886572,
-      "grad_norm": 4.144084453582764,
+      "grad_norm": 2.1476380825042725,
       "learning_rate": 4.820730117340286e-05,
-      "loss": 0.3831,
+      "loss": 0.4282,
       "step": 103000
     },
     {
       "epoch": 8.433833116036507,
-      "grad_norm": 7.010155200958252,
+      "grad_norm": 2.8716719150543213,
       "learning_rate": 4.698500651890482e-05,
-      "loss": 0.3728,
+      "loss": 0.4185,
       "step": 103500
     },
     {
       "epoch": 8.474576271186441,
-      "grad_norm": 4.621219635009766,
+      "grad_norm": 2.4064836502075195,
       "learning_rate": 4.576271186440678e-05,
-      "loss": 0.3708,
+      "loss": 0.4197,
       "step": 104000
     },
     {
       "epoch": 8.515319426336376,
-      "grad_norm": 7.2299394607543945,
+      "grad_norm": 2.5717904567718506,
       "learning_rate": 4.4540417209908735e-05,
-      "loss": 0.3747,
+      "loss": 0.4259,
       "step": 104500
     },
     {
       "epoch": 8.55606258148631,
-      "grad_norm": 7.200563430786133,
+      "grad_norm": 3.6525187492370605,
       "learning_rate": 4.3318122555410686e-05,
-      "loss": 0.3817,
+      "loss": 0.4256,
       "step": 105000
     },
     {
       "epoch": 8.596805736636245,
-      "grad_norm": 4.539661884307861,
+      "grad_norm": 2.679851531982422,
       "learning_rate": 4.2095827900912643e-05,
-      "loss": 0.375,
+      "loss": 0.4168,
       "step": 105500
     },
     {
       "epoch": 8.63754889178618,
-      "grad_norm": 4.076575756072998,
+      "grad_norm": 2.0390915870666504,
       "learning_rate": 4.08735332464146e-05,
-      "loss": 0.3761,
+      "loss": 0.4158,
       "step": 106000
     },
     {
       "epoch": 8.678292046936114,
-      "grad_norm": 5.755791664123535,
+      "grad_norm": 3.6200144290924072,
       "learning_rate": 3.965123859191656e-05,
-      "loss": 0.388,
+      "loss": 0.426,
       "step": 106500
     },
     {
       "epoch": 8.719035202086049,
-      "grad_norm": 4.549001216888428,
+      "grad_norm": 2.4385690689086914,
       "learning_rate": 3.842894393741851e-05,
-      "loss": 0.3794,
+      "loss": 0.4243,
       "step": 107000
     },
     {
       "epoch": 8.759778357235984,
-      "grad_norm": 4.821287155151367,
+      "grad_norm": 3.2193117141723633,
       "learning_rate": 3.720664928292047e-05,
-      "loss": 0.3696,
+      "loss": 0.4155,
       "step": 107500
     },
     {
       "epoch": 8.800521512385918,
-      "grad_norm": 5.3173089027404785,
+      "grad_norm": 3.227754831314087,
       "learning_rate": 3.5984354628422425e-05,
-      "loss": 0.3777,
+      "loss": 0.4224,
       "step": 108000
     },
     {
       "epoch": 8.841264667535853,
-      "grad_norm": 5.278952121734619,
+      "grad_norm": 1.8869520425796509,
       "learning_rate": 3.4762059973924376e-05,
-      "loss": 0.3743,
+      "loss": 0.4178,
       "step": 108500
     },
     {
       "epoch": 8.88200782268579,
-      "grad_norm": 6.104799270629883,
+      "grad_norm": 3.0674214363098145,
       "learning_rate": 3.3539765319426334e-05,
-      "loss": 0.3791,
+      "loss": 0.4216,
       "step": 109000
     },
     {
       "epoch": 8.922750977835724,
-      "grad_norm": 4.673792839050293,
+      "grad_norm": 2.215930461883545,
       "learning_rate": 3.2317470664928285e-05,
-      "loss": 0.3774,
+      "loss": 0.4205,
       "step": 109500
     },
     {
       "epoch": 8.963494132985659,
-      "grad_norm": 6.051920413970947,
+      "grad_norm": 3.146869421005249,
       "learning_rate": 3.109517601043025e-05,
-      "loss": 0.3752,
+      "loss": 0.4236,
       "step": 110000
     },
     {
       "epoch": 9.0,
       "eval_accuracy": 0.8373494148254395,
-      "eval_loss": 0.43148961663246155,
-      "eval_runtime": 3.528,
-      "eval_samples_per_second": 705.779,
-      "eval_steps_per_second": 88.435,
+      "eval_loss": 0.43542519211769104,
+      "eval_runtime": 3.7784,
+      "eval_samples_per_second": 659.012,
+      "eval_steps_per_second": 82.575,
       "step": 110448
     },
     {
       "epoch": 9.004237288135593,
-      "grad_norm": 3.2586278915405273,
+      "grad_norm": 1.5620496273040771,
       "learning_rate": 2.98728813559322e-05,
-      "loss": 0.3684,
+      "loss": 0.4145,
       "step": 110500
     },
     {
       "epoch": 9.044980443285528,
-      "grad_norm": 3.5761301517486572,
+      "grad_norm": 1.8520050048828125,
       "learning_rate": 2.8650586701434158e-05,
-      "loss": 0.3634,
+      "loss": 0.4108,
       "step": 111000
     },
     {
       "epoch": 9.085723598435463,
-      "grad_norm": 8.008382797241211,
+      "grad_norm": 3.6136269569396973,
       "learning_rate": 2.7428292046936113e-05,
-      "loss": 0.3574,
+      "loss": 0.4061,
       "step": 111500
     },
     {
       "epoch": 9.126466753585397,
-      "grad_norm": 3.5123190879821777,
+      "grad_norm": 1.7939780950546265,
       "learning_rate": 2.6205997392438067e-05,
-      "loss": 0.3794,
+      "loss": 0.4247,
       "step": 112000
     },
     {
       "epoch": 9.167209908735332,
-      "grad_norm": 6.476950168609619,
+      "grad_norm": 1.5612313747406006,
       "learning_rate": 2.4983702737940025e-05,
-      "loss": 0.3677,
+      "loss": 0.4174,
       "step": 112500
     },
     {
       "epoch": 9.207953063885267,
-      "grad_norm": 3.6542065143585205,
+      "grad_norm": 2.204183340072632,
       "learning_rate": 2.376140808344198e-05,
-      "loss": 0.3666,
+      "loss": 0.4185,
       "step": 113000
     },
     {
       "epoch": 9.248696219035201,
-      "grad_norm": 5.919882297515869,
+      "grad_norm": 3.115217447280884,
       "learning_rate": 2.2539113428943937e-05,
-      "loss": 0.3605,
+      "loss": 0.4106,
       "step": 113500
     },
     {
       "epoch": 9.289439374185136,
-      "grad_norm": 10.532210350036621,
+      "grad_norm": 2.998296022415161,
       "learning_rate": 2.131681877444589e-05,
-      "loss": 0.3703,
+      "loss": 0.4193,
       "step": 114000
     },
     {
       "epoch": 9.330182529335072,
-      "grad_norm": 4.157111167907715,
+      "grad_norm": 2.2693541049957275,
       "learning_rate": 2.009452411994785e-05,
-      "loss": 0.3712,
+      "loss": 0.4196,
       "step": 114500
     },
     {
       "epoch": 9.370925684485007,
-      "grad_norm": 4.405088424682617,
+      "grad_norm": 1.6824918985366821,
       "learning_rate": 1.8872229465449803e-05,
-      "loss": 0.3595,
+      "loss": 0.4116,
       "step": 115000
     },
     {
       "epoch": 9.411668839634942,
-      "grad_norm": 7.088386535644531,
+      "grad_norm": 3.1412432193756104,
       "learning_rate": 1.7649934810951758e-05,
-      "loss": 0.3637,
+      "loss": 0.4139,
       "step": 115500
     },
     {
       "epoch": 9.452411994784876,
-      "grad_norm": 3.80808162689209,
+      "grad_norm": 1.6451703310012817,
       "learning_rate": 1.6427640156453715e-05,
-      "loss": 0.3662,
+      "loss": 0.4155,
       "step": 116000
     },
     {
       "epoch": 9.493155149934811,
-      "grad_norm": 5.665896415710449,
+      "grad_norm": 2.260284900665283,
       "learning_rate": 1.520534550195567e-05,
-      "loss": 0.369,
+      "loss": 0.4231,
       "step": 116500
     },
     {
       "epoch": 9.533898305084746,
-      "grad_norm": 4.733582496643066,
+      "grad_norm": 2.125267505645752,
       "learning_rate": 1.3983050847457626e-05,
-      "loss": 0.3665,
+      "loss": 0.4216,
       "step": 117000
     },
     {
       "epoch": 9.57464146023468,
-      "grad_norm": 6.934778690338135,
+      "grad_norm": 2.3000893592834473,
       "learning_rate": 1.2760756192959582e-05,
-      "loss": 0.3567,
+      "loss": 0.4078,
       "step": 117500
     },
     {
       "epoch": 9.615384615384615,
-      "grad_norm": 6.5998311042785645,
+      "grad_norm": 2.3258585929870605,
       "learning_rate": 1.1538461538461538e-05,
-      "loss": 0.3612,
+      "loss": 0.4133,
       "step": 118000
     },
     {
       "epoch": 9.65612777053455,
-      "grad_norm": 7.71038293838501,
+      "grad_norm": 3.550931692123413,
       "learning_rate": 1.0316166883963494e-05,
-      "loss": 0.3562,
+      "loss": 0.4075,
       "step": 118500
     },
     {
       "epoch": 9.696870925684484,
-      "grad_norm": 4.586871147155762,
+      "grad_norm": 3.0916647911071777,
       "learning_rate": 9.093872229465448e-06,
-      "loss": 0.3661,
+      "loss": 0.4166,
       "step": 119000
     },
     {
       "epoch": 9.737614080834419,
-      "grad_norm": 5.962682247161865,
+      "grad_norm": 2.898163318634033,
       "learning_rate": 7.871577574967404e-06,
-      "loss": 0.3675,
+      "loss": 0.4154,
       "step": 119500
     },
     {
       "epoch": 9.778357235984355,
-      "grad_norm": 3.3349926471710205,
+      "grad_norm": 1.945553183555603,
       "learning_rate": 6.649282920469361e-06,
-      "loss": 0.3571,
+      "loss": 0.4063,
       "step": 120000
     },
     {
       "epoch": 9.81910039113429,
-      "grad_norm": 5.066919326782227,
+      "grad_norm": 2.7394824028015137,
       "learning_rate": 5.426988265971316e-06,
-      "loss": 0.3558,
+      "loss": 0.4084,
       "step": 120500
     },
     {
       "epoch": 9.859843546284225,
-      "grad_norm": 5.792361259460449,
+      "grad_norm": 2.676351547241211,
       "learning_rate": 4.2046936114732716e-06,
-      "loss": 0.368,
+      "loss": 0.4178,
       "step": 121000
     },
     {
       "epoch": 9.90058670143416,
-      "grad_norm": 6.927473545074463,
+      "grad_norm": 2.577317237854004,
       "learning_rate": 2.982398956975228e-06,
-      "loss": 0.3631,
+      "loss": 0.4139,
       "step": 121500
     },
     {
       "epoch": 9.941329856584094,
-      "grad_norm": 6.935751914978027,
+      "grad_norm": 4.512514114379883,
       "learning_rate": 1.7601043024771837e-06,
-      "loss": 0.3623,
+      "loss": 0.4115,
       "step": 122000
     },
     {
       "epoch": 9.982073011734029,
-      "grad_norm": 1.9150322675704956,
+      "grad_norm": 2.307258129119873,
       "learning_rate": 5.378096479791394e-07,
-      "loss": 0.3639,
+      "loss": 0.4151,
       "step": 122500
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.8357429504394531,
-      "eval_loss": 0.4303068220615387,
-      "eval_runtime": 3.5313,
-      "eval_samples_per_second": 705.119,
-      "eval_steps_per_second": 88.352,
+      "eval_accuracy": 0.8361445665359497,
+      "eval_loss": 0.4362991154193878,
+      "eval_runtime": 3.7734,
+      "eval_samples_per_second": 659.887,
+      "eval_steps_per_second": 82.685,
       "step": 122720
     },
     {
       "epoch": 10.0,
       "step": 122720,
-      "total_flos": 2.6187994846190592e+17,
-      "train_loss": 0.4432659422714054,
-      "train_runtime": 8098.9158,
-      "train_samples_per_second": 484.882,
-      "train_steps_per_second": 15.153
+      "total_flos": 2.6143616931499008e+17,
+      "train_loss": 0.4748868410093066,
+      "train_runtime": 8727.3231,
+      "train_samples_per_second": 449.968,
+      "train_steps_per_second": 14.062
     }
   ],
   "logging_steps": 500,
@@ -1840,7 +1840,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.6187994846190592e+17,
+  "total_flos": 2.6143616931499008e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null