End of training

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.09032258064516,
-    "eval_accuracy": 0.8,
-    "eval_loss": 0.7193946838378906,
-    "eval_runtime": 151.4396,
-    "eval_samples_per_second": 2.641,
-    "eval_steps_per_second": 0.33
+    "epoch": 14.061648745519713,
+    "eval_accuracy": 0.81875,
+    "eval_loss": 0.6550884246826172,
+    "eval_runtime": 306.835,
+    "eval_samples_per_second": 2.607,
+    "eval_steps_per_second": 0.326
 }

runs/Jul21_20-42-21_a62a56ce0424/events.out.tfevents.1721608783.a62a56ce0424.5680.1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a63e23e4ab1595b272dd3222027f64ca94f304dd09bf39740131b669355f362
-size 411
+oid sha256:31b18516b93fba2d4fea00053cdb8e5d832486101d4017aef5348963859ad83f
+size 734

test_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.09032258064516,
-    "eval_accuracy": 0.8,
-    "eval_loss": 0.7193946838378906,
-    "eval_runtime": 151.4396,
-    "eval_samples_per_second": 2.641,
-    "eval_steps_per_second": 0.33
+    "epoch": 14.061648745519713,
+    "eval_accuracy": 0.81875,
+    "eval_loss": 0.6550884246826172,
+    "eval_runtime": 306.835,
+    "eval_samples_per_second": 2.607,
+    "eval_steps_per_second": 0.326
 }

trainer_state.json

@@ -1,784 +1,2131 @@
 {
-  "best_metric": 0.9473684210526315,
-  "best_model_checkpoint": "videomae-base-finetuned-rwf2000-subset/checkpoint-930",
-  "epoch": 9.09032258064516,
+  "best_metric": 0.8782051282051282,
+  "best_model_checkpoint": "videomae-base-finetuned-rwf2000-subset/checkpoint-1870",
+  "epoch": 14.061648745519713,
   "eval_steps": 500,
-  "global_step": 930,
+  "global_step": 2790,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.010752688172043012,
-      "grad_norm": 5.772064208984375,
-      "learning_rate": 5.376344086021506e-06,
-      "loss": 0.8938,
+      "epoch": 0.0035842293906810036,
+      "grad_norm": 10.333197593688965,
+      "learning_rate": 1.7921146953405017e-06,
+      "loss": 0.4614,
       "step": 10
     },
     {
-      "epoch": 0.021505376344086023,
-      "grad_norm": 6.392468452453613,
-      "learning_rate": 1.0752688172043012e-05,
-      "loss": 0.7152,
+      "epoch": 0.007168458781362007,
+      "grad_norm": 10.418760299682617,
+      "learning_rate": 3.5842293906810035e-06,
+      "loss": 0.5532,
       "step": 20
     },
     {
-      "epoch": 0.03225806451612903,
-      "grad_norm": 5.6630167961120605,
-      "learning_rate": 1.6129032258064517e-05,
-      "loss": 0.7008,
+      "epoch": 0.010752688172043012,
+      "grad_norm": 7.5556135177612305,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 0.5127,
       "step": 30
     },
     {
-      "epoch": 0.043010752688172046,
-      "grad_norm": 7.545820713043213,
-      "learning_rate": 2.1505376344086024e-05,
-      "loss": 0.661,
+      "epoch": 0.014336917562724014,
+      "grad_norm": 7.2956719398498535,
+      "learning_rate": 7.168458781362007e-06,
+      "loss": 0.4354,
       "step": 40
     },
     {
-      "epoch": 0.053763440860215055,
-      "grad_norm": 4.271369457244873,
-      "learning_rate": 2.6881720430107527e-05,
-      "loss": 0.6729,
+      "epoch": 0.017921146953405017,
+      "grad_norm": 5.666509628295898,
+      "learning_rate": 8.960573476702509e-06,
+      "loss": 0.3259,
       "step": 50
     },
     {
-      "epoch": 0.06451612903225806,
-      "grad_norm": 6.001001358032227,
-      "learning_rate": 3.2258064516129034e-05,
-      "loss": 0.6542,
+      "epoch": 0.021505376344086023,
+      "grad_norm": 11.845425605773926,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.4544,
       "step": 60
     },
     {
-      "epoch": 0.07526881720430108,
-      "grad_norm": 5.876163005828857,
-      "learning_rate": 3.763440860215054e-05,
-      "loss": 0.6541,
+      "epoch": 0.025089605734767026,
+      "grad_norm": 4.353142738342285,
+      "learning_rate": 1.2544802867383513e-05,
+      "loss": 0.4147,
       "step": 70
     },
     {
-      "epoch": 0.08602150537634409,
-      "grad_norm": 4.0299906730651855,
-      "learning_rate": 4.301075268817205e-05,
-      "loss": 0.6077,
+      "epoch": 0.02867383512544803,
+      "grad_norm": 8.852384567260742,
+      "learning_rate": 1.4336917562724014e-05,
+      "loss": 0.3996,
       "step": 80
     },
     {
-      "epoch": 0.0967741935483871,
-      "grad_norm": 5.537106990814209,
-      "learning_rate": 4.8387096774193554e-05,
-      "loss": 0.5577,
+      "epoch": 0.03225806451612903,
+      "grad_norm": 8.823715209960938,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.4497,
       "step": 90
     },
     {
-      "epoch": 0.1010752688172043,
-      "eval_accuracy": 0.7236842105263158,
-      "eval_loss": 0.4835163354873657,
-      "eval_runtime": 26.2728,
-      "eval_samples_per_second": 2.893,
-      "eval_steps_per_second": 0.381,
-      "step": 94
-    },
-    {
-      "epoch": 1.0064516129032257,
-      "grad_norm": 4.136658191680908,
-      "learning_rate": 4.9581839904420555e-05,
-      "loss": 0.5856,
+      "epoch": 0.035842293906810034,
+      "grad_norm": 6.714145660400391,
+      "learning_rate": 1.7921146953405018e-05,
+      "loss": 0.2831,
       "step": 100
     },
     {
-      "epoch": 1.0172043010752687,
-      "grad_norm": 7.560351848602295,
-      "learning_rate": 4.898446833930705e-05,
-      "loss": 0.5146,
+      "epoch": 0.03942652329749104,
+      "grad_norm": 10.755260467529297,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 0.4612,
       "step": 110
     },
     {
-      "epoch": 1.027956989247312,
-      "grad_norm": 3.6854827404022217,
-      "learning_rate": 4.8387096774193554e-05,
-      "loss": 0.6122,
+      "epoch": 0.043010752688172046,
+      "grad_norm": 11.755001068115234,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.539,
       "step": 120
     },
     {
-      "epoch": 1.038709677419355,
-      "grad_norm": 3.7454755306243896,
-      "learning_rate": 4.778972520908005e-05,
-      "loss": 0.6283,
+      "epoch": 0.04659498207885305,
+      "grad_norm": 4.715596675872803,
+      "learning_rate": 2.3297491039426525e-05,
+      "loss": 0.4302,
       "step": 130
     },
     {
-      "epoch": 1.049462365591398,
-      "grad_norm": 3.77704119682312,
-      "learning_rate": 4.7192353643966546e-05,
-      "loss": 0.5195,
+      "epoch": 0.05017921146953405,
+      "grad_norm": 10.882904052734375,
+      "learning_rate": 2.5089605734767026e-05,
+      "loss": 0.429,
       "step": 140
     },
     {
-      "epoch": 1.060215053763441,
-      "grad_norm": 6.745356559753418,
-      "learning_rate": 4.659498207885305e-05,
-      "loss": 0.605,
+      "epoch": 0.053763440860215055,
+      "grad_norm": 17.894540786743164,
+      "learning_rate": 2.6881720430107527e-05,
+      "loss": 0.4703,
       "step": 150
     },
     {
-      "epoch": 1.070967741935484,
-      "grad_norm": 5.097397327423096,
-      "learning_rate": 4.5997610513739546e-05,
-      "loss": 0.4275,
+      "epoch": 0.05734767025089606,
+      "grad_norm": 9.539034843444824,
+      "learning_rate": 2.8673835125448028e-05,
+      "loss": 0.4972,
       "step": 160
     },
     {
-      "epoch": 1.081720430107527,
-      "grad_norm": 11.386567115783691,
-      "learning_rate": 4.540023894862604e-05,
-      "loss": 0.4862,
+      "epoch": 0.06093189964157706,
+      "grad_norm": 6.870086193084717,
+      "learning_rate": 3.046594982078853e-05,
+      "loss": 0.37,
       "step": 170
     },
     {
-      "epoch": 1.09247311827957,
-      "grad_norm": 2.083325147628784,
-      "learning_rate": 4.4802867383512545e-05,
-      "loss": 0.6696,
+      "epoch": 0.06451612903225806,
+      "grad_norm": 7.155197620391846,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.3956,
       "step": 180
     },
     {
-      "epoch": 1.1010752688172043,
-      "eval_accuracy": 0.8421052631578947,
-      "eval_loss": 0.43573570251464844,
-      "eval_runtime": 21.5443,
-      "eval_samples_per_second": 3.528,
-      "eval_steps_per_second": 0.464,
-      "step": 188
+      "epoch": 0.06702508960573476,
+      "eval_accuracy": 0.717948717948718,
+      "eval_loss": 0.7066396474838257,
+      "eval_runtime": 51.1495,
+      "eval_samples_per_second": 3.05,
+      "eval_steps_per_second": 0.391,
+      "step": 187
     },
     {
-      "epoch": 2.0021505376344084,
-      "grad_norm": 8.380154609680176,
-      "learning_rate": 4.420549581839905e-05,
-      "loss": 0.5353,
+      "epoch": 1.0010752688172042,
+      "grad_norm": 23.43296241760254,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 0.3894,
       "step": 190
     },
     {
-      "epoch": 2.0129032258064514,
-      "grad_norm": 9.097419738769531,
-      "learning_rate": 4.360812425328555e-05,
-      "loss": 0.4885,
+      "epoch": 1.0046594982078854,
+      "grad_norm": 6.450241565704346,
+      "learning_rate": 3.5842293906810036e-05,
+      "loss": 0.5151,
       "step": 200
     },
     {
-      "epoch": 2.0236559139784944,
-      "grad_norm": 3.875176191329956,
-      "learning_rate": 4.301075268817205e-05,
-      "loss": 0.505,
+      "epoch": 1.0082437275985663,
+      "grad_norm": 5.017886161804199,
+      "learning_rate": 3.763440860215054e-05,
+      "loss": 0.4517,
       "step": 210
     },
     {
-      "epoch": 2.0344086021505374,
-      "grad_norm": 8.177785873413086,
-      "learning_rate": 4.241338112305854e-05,
-      "loss": 0.6089,
+      "epoch": 1.0118279569892472,
+      "grad_norm": 20.295490264892578,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 0.5027,
       "step": 220
     },
     {
-      "epoch": 2.0451612903225804,
-      "grad_norm": 14.289607048034668,
-      "learning_rate": 4.1816009557945046e-05,
-      "loss": 0.5993,
+      "epoch": 1.0154121863799284,
+      "grad_norm": 14.873997688293457,
+      "learning_rate": 4.121863799283154e-05,
+      "loss": 0.6656,
       "step": 230
     },
     {
-      "epoch": 2.055913978494624,
-      "grad_norm": 4.532449722290039,
-      "learning_rate": 4.121863799283154e-05,
-      "loss": 0.7074,
+      "epoch": 1.0189964157706093,
+      "grad_norm": 1.4430453777313232,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 0.3837,
       "step": 240
     },
     {
-      "epoch": 2.066666666666667,
-      "grad_norm": 4.093668460845947,
-      "learning_rate": 4.062126642771804e-05,
-      "loss": 0.4219,
+      "epoch": 1.0225806451612902,
+      "grad_norm": 3.686954975128174,
+      "learning_rate": 4.4802867383512545e-05,
+      "loss": 0.4077,
       "step": 250
     },
     {
-      "epoch": 2.07741935483871,
-      "grad_norm": 14.380650520324707,
-      "learning_rate": 4.002389486260454e-05,
-      "loss": 0.4474,
+      "epoch": 1.0261648745519714,
+      "grad_norm": 8.648886680603027,
+      "learning_rate": 4.659498207885305e-05,
+      "loss": 0.4828,
       "step": 260
     },
     {
-      "epoch": 2.088172043010753,
-      "grad_norm": 6.174684047698975,
-      "learning_rate": 3.9426523297491045e-05,
-      "loss": 0.5363,
+      "epoch": 1.0297491039426523,
+      "grad_norm": 13.670947074890137,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 0.4882,
       "step": 270
     },
     {
-      "epoch": 2.098924731182796,
-      "grad_norm": 11.756628036499023,
-      "learning_rate": 3.882915173237754e-05,
-      "loss": 0.5608,
+      "epoch": 1.0333333333333334,
+      "grad_norm": 1.407135248184204,
+      "learning_rate": 4.998008761449622e-05,
+      "loss": 0.4301,
       "step": 280
     },
     {
-      "epoch": 2.1010752688172043,
-      "eval_accuracy": 0.8289473684210527,
-      "eval_loss": 0.479496568441391,
-      "eval_runtime": 21.9809,
-      "eval_samples_per_second": 3.458,
-      "eval_steps_per_second": 0.455,
-      "step": 282
-    },
-    {
-      "epoch": 3.0086021505376346,
-      "grad_norm": 2.0791425704956055,
-      "learning_rate": 3.8231780167264044e-05,
-      "loss": 0.5157,
+      "epoch": 1.0369175627240144,
+      "grad_norm": 11.154088020324707,
+      "learning_rate": 4.978096375945839e-05,
+      "loss": 0.4001,
       "step": 290
     },
     {
-      "epoch": 3.0193548387096776,
-      "grad_norm": 6.193145751953125,
-      "learning_rate": 3.763440860215054e-05,
-      "loss": 0.4626,
+      "epoch": 1.0405017921146953,
+      "grad_norm": 6.446365833282471,
+      "learning_rate": 4.9581839904420555e-05,
+      "loss": 0.6634,
       "step": 300
     },
     {
-      "epoch": 3.0301075268817206,
-      "grad_norm": 8.216785430908203,
-      "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.6098,
+      "epoch": 1.0440860215053764,
+      "grad_norm": 8.026636123657227,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 0.2529,
       "step": 310
     },
     {
-      "epoch": 3.0408602150537636,
-      "grad_norm": 3.388540267944336,
-      "learning_rate": 3.643966547192354e-05,
-      "loss": 0.3466,
+      "epoch": 1.0476702508960574,
+      "grad_norm": 13.86358642578125,
+      "learning_rate": 4.9183592194344885e-05,
+      "loss": 0.6586,
       "step": 320
     },
     {
-      "epoch": 3.0516129032258066,
-      "grad_norm": 3.5496809482574463,
-      "learning_rate": 3.5842293906810036e-05,
-      "loss": 0.3225,
+      "epoch": 1.0512544802867383,
+      "grad_norm": 6.193373680114746,
+      "learning_rate": 4.898446833930705e-05,
+      "loss": 0.6053,
       "step": 330
     },
     {
-      "epoch": 3.0623655913978496,
-      "grad_norm": 0.8877809047698975,
-      "learning_rate": 3.524492234169653e-05,
-      "loss": 0.513,
+      "epoch": 1.0548387096774194,
+      "grad_norm": 3.7244551181793213,
+      "learning_rate": 4.8785344484269216e-05,
+      "loss": 0.4128,
       "step": 340
     },
     {
-      "epoch": 3.0731182795698926,
-      "grad_norm": 7.285158634185791,
-      "learning_rate": 3.4647550776583035e-05,
-      "loss": 0.6786,
+      "epoch": 1.0584229390681004,
+      "grad_norm": 4.799228668212891,
+      "learning_rate": 4.858622062923138e-05,
+      "loss": 0.429,
       "step": 350
     },
     {
-      "epoch": 3.0838709677419356,
-      "grad_norm": 9.199494361877441,
-      "learning_rate": 3.405017921146954e-05,
-      "loss": 0.575,
+      "epoch": 1.0620071684587813,
+      "grad_norm": 13.368525505065918,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 0.5007,
       "step": 360
     },
     {
-      "epoch": 3.0946236559139786,
-      "grad_norm": 2.362581253051758,
-      "learning_rate": 3.3452807646356034e-05,
-      "loss": 0.3918,
+      "epoch": 1.0655913978494624,
+      "grad_norm": 7.234383583068848,
+      "learning_rate": 4.818797291915572e-05,
+      "loss": 0.4019,
       "step": 370
     },
     {
-      "epoch": 3.1010752688172043,
-      "eval_accuracy": 0.7894736842105263,
-      "eval_loss": 0.46907052397727966,
-      "eval_runtime": 21.3364,
-      "eval_samples_per_second": 3.562,
-      "eval_steps_per_second": 0.469,
-      "step": 376
+      "epoch": 1.0670250896057347,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 0.6282246708869934,
+      "eval_runtime": 50.1248,
+      "eval_samples_per_second": 3.112,
+      "eval_steps_per_second": 0.399,
+      "step": 374
     },
     {
-      "epoch": 4.004301075268817,
-      "grad_norm": 3.9972550868988037,
-      "learning_rate": 3.285543608124254e-05,
-      "loss": 0.4378,
+      "epoch": 2.0021505376344084,
+      "grad_norm": 2.577439069747925,
+      "learning_rate": 4.7988849064117885e-05,
+      "loss": 0.494,
       "step": 380
     },
     {
-      "epoch": 4.01505376344086,
-      "grad_norm": 6.704283237457275,
-      "learning_rate": 3.2258064516129034e-05,
-      "loss": 0.4316,
+      "epoch": 2.0057347670250896,
+      "grad_norm": 4.06400203704834,
+      "learning_rate": 4.778972520908005e-05,
+      "loss": 0.3722,
       "step": 390
     },
     {
-      "epoch": 4.025806451612903,
-      "grad_norm": 11.435125350952148,
-      "learning_rate": 3.1660692951015537e-05,
-      "loss": 0.5877,
+      "epoch": 2.0093189964157707,
+      "grad_norm": 8.867179870605469,
+      "learning_rate": 4.759060135404222e-05,
+      "loss": 0.5412,
       "step": 400
     },
     {
-      "epoch": 4.036559139784946,
-      "grad_norm": 7.740775108337402,
-      "learning_rate": 3.106332138590203e-05,
-      "loss": 0.3589,
+      "epoch": 2.0129032258064514,
+      "grad_norm": 2.0079333782196045,
+      "learning_rate": 4.739147749900438e-05,
+      "loss": 0.5593,
       "step": 410
     },
     {
-      "epoch": 4.047311827956989,
-      "grad_norm": 2.672539710998535,
-      "learning_rate": 3.046594982078853e-05,
-      "loss": 0.3523,
+      "epoch": 2.0164874551971326,
+      "grad_norm": 9.170706748962402,
+      "learning_rate": 4.7192353643966546e-05,
+      "loss": 0.3964,
       "step": 420
     },
     {
-      "epoch": 4.058064516129032,
-      "grad_norm": 3.296454429626465,
-      "learning_rate": 2.9868578255675032e-05,
-      "loss": 0.3457,
+      "epoch": 2.0200716845878137,
+      "grad_norm": 9.581748962402344,
+      "learning_rate": 4.699322978892872e-05,
+      "loss": 0.5128,
       "step": 430
     },
     {
-      "epoch": 4.068817204301075,
-      "grad_norm": 11.174155235290527,
-      "learning_rate": 2.9271206690561532e-05,
-      "loss": 0.4028,
+      "epoch": 2.0236559139784944,
+      "grad_norm": 3.3790416717529297,
+      "learning_rate": 4.6794105933890884e-05,
+      "loss": 0.4637,
       "step": 440
     },
     {
-      "epoch": 4.079569892473118,
-      "grad_norm": 2.355884313583374,
-      "learning_rate": 2.8673835125448028e-05,
-      "loss": 0.3813,
+      "epoch": 2.0272401433691756,
+      "grad_norm": 9.568379402160645,
+      "learning_rate": 4.659498207885305e-05,
+      "loss": 0.4377,
       "step": 450
     },
     {
-      "epoch": 4.090322580645161,
-      "grad_norm": 12.960975646972656,
-      "learning_rate": 2.807646356033453e-05,
-      "loss": 0.2315,
+      "epoch": 2.0308243727598567,
+      "grad_norm": 1.0472499132156372,
+      "learning_rate": 4.6395858223815215e-05,
+      "loss": 0.4241,
       "step": 460
     },
     {
-      "epoch": 4.101075268817205,
-      "grad_norm": 14.71526050567627,
-      "learning_rate": 2.747909199522103e-05,
-      "loss": 0.4756,
-      "step": 470
-    },
-    {
-      "epoch": 4.101075268817205,
-      "eval_accuracy": 0.8421052631578947,
-      "eval_loss": 0.44611287117004395,
-      "eval_runtime": 27.2076,
-      "eval_samples_per_second": 2.793,
-      "eval_steps_per_second": 0.368,
+      "epoch": 2.0344086021505374,
+      "grad_norm": 21.286645889282227,
+      "learning_rate": 4.619673436877738e-05,
+      "loss": 0.7317,
       "step": 470
     },
     {
-      "epoch": 5.010752688172043,
-      "grad_norm": 14.609360694885254,
-      "learning_rate": 2.6881720430107527e-05,
-      "loss": 0.5645,
+      "epoch": 2.0379928315412186,
+      "grad_norm": 8.561319351196289,
+      "learning_rate": 4.5997610513739546e-05,
+      "loss": 0.5629,
       "step": 480
     },
     {
-      "epoch": 5.021505376344086,
-      "grad_norm": 3.8693418502807617,
-      "learning_rate": 2.628434886499403e-05,
-      "loss": 0.3794,
+      "epoch": 2.0415770609318997,
+      "grad_norm": 2.304678201675415,
+      "learning_rate": 4.579848665870172e-05,
+      "loss": 0.2492,
       "step": 490
     },
     {
-      "epoch": 5.032258064516129,
-      "grad_norm": 1.618184208869934,
-      "learning_rate": 2.5686977299880526e-05,
-      "loss": 0.3139,
+      "epoch": 2.0451612903225804,
+      "grad_norm": 13.397499084472656,
+      "learning_rate": 4.559936280366388e-05,
+      "loss": 0.4369,
       "step": 500
     },
     {
-      "epoch": 5.043010752688172,
-      "grad_norm": 1.95063054561615,
-      "learning_rate": 2.5089605734767026e-05,
-      "loss": 0.3952,
+      "epoch": 2.0487455197132616,
+      "grad_norm": 1.433841347694397,
+      "learning_rate": 4.540023894862604e-05,
+      "loss": 0.4091,
       "step": 510
     },
     {
-      "epoch": 5.053763440860215,
-      "grad_norm": 9.131407737731934,
-      "learning_rate": 2.4492234169653525e-05,
-      "loss": 0.4587,
+      "epoch": 2.0523297491039427,
+      "grad_norm": 4.154223918914795,
+      "learning_rate": 4.5201115093588214e-05,
+      "loss": 0.3906,
       "step": 520
     },
     {
-      "epoch": 5.064516129032258,
-      "grad_norm": 4.612659931182861,
-      "learning_rate": 2.3894862604540025e-05,
-      "loss": 0.2986,
+      "epoch": 2.055913978494624,
+      "grad_norm": 11.778128623962402,
+      "learning_rate": 4.500199123855038e-05,
+      "loss": 0.4366,
       "step": 530
     },
     {
-      "epoch": 5.075268817204301,
-      "grad_norm": 3.612903594970703,
-      "learning_rate": 2.3297491039426525e-05,
-      "loss": 0.2318,
+      "epoch": 2.0594982078853046,
+      "grad_norm": 0.3377319872379303,
+      "learning_rate": 4.4802867383512545e-05,
+      "loss": 0.464,
       "step": 540
     },
     {
-      "epoch": 5.086021505376344,
-      "grad_norm": 4.401362419128418,
-      "learning_rate": 2.270011947431302e-05,
-      "loss": 0.41,
+      "epoch": 2.0630824372759857,
+      "grad_norm": 6.4462103843688965,
+      "learning_rate": 4.460374352847471e-05,
+      "loss": 0.7451,
       "step": 550
     },
     {
-      "epoch": 5.096774193548387,
-      "grad_norm": 12.280633926391602,
-      "learning_rate": 2.2102747909199524e-05,
-      "loss": 0.3794,
+      "epoch": 2.066666666666667,
+      "grad_norm": 4.705774784088135,
+      "learning_rate": 4.440461967343688e-05,
+      "loss": 0.4473,
       "step": 560
     },
     {
-      "epoch": 5.101075268817205,
-      "eval_accuracy": 0.7368421052631579,
-      "eval_loss": 0.8676896095275879,
-      "eval_runtime": 21.0222,
-      "eval_samples_per_second": 3.615,
-      "eval_steps_per_second": 0.476,
-      "step": 564
+      "epoch": 2.067025089605735,
+      "eval_accuracy": 0.7692307692307693,
+      "eval_loss": 0.43944770097732544,
+      "eval_runtime": 51.1029,
+      "eval_samples_per_second": 3.053,
+      "eval_steps_per_second": 0.391,
+      "step": 561
     },
     {
-      "epoch": 6.006451612903226,
-      "grad_norm": 1.297224998474121,
-      "learning_rate": 2.1505376344086024e-05,
-      "loss": 0.2989,
+      "epoch": 3.003225806451613,
+      "grad_norm": 2.186035633087158,
+      "learning_rate": 4.420549581839905e-05,
+      "loss": 0.3464,
       "step": 570
     },
     {
-      "epoch": 6.017204301075269,
-      "grad_norm": 0.6060790419578552,
-      "learning_rate": 2.0908004778972523e-05,
-      "loss": 0.3388,
+      "epoch": 3.006810035842294,
+      "grad_norm": 7.887436866760254,
+      "learning_rate": 4.400637196336121e-05,
+      "loss": 0.3988,
       "step": 580
     },
     {
-      "epoch": 6.027956989247312,
-      "grad_norm": 20.182498931884766,
-      "learning_rate": 2.031063321385902e-05,
-      "loss": 0.3357,
+      "epoch": 3.010394265232975,
+      "grad_norm": 19.643508911132812,
+      "learning_rate": 4.380724810832338e-05,
+      "loss": 0.3721,
       "step": 590
     },
     {
-      "epoch": 6.038709677419355,
-      "grad_norm": 2.371579170227051,
-      "learning_rate": 1.9713261648745522e-05,
-      "loss": 0.2741,
+      "epoch": 3.013978494623656,
+      "grad_norm": 2.0498669147491455,
+      "learning_rate": 4.360812425328555e-05,
+      "loss": 0.5075,
       "step": 600
     },
     {
-      "epoch": 6.049462365591398,
-      "grad_norm": 11.418489456176758,
-      "learning_rate": 1.9115890083632022e-05,
-      "loss": 0.424,
+      "epoch": 3.017562724014337,
+      "grad_norm": 2.9130897521972656,
+      "learning_rate": 4.340900039824771e-05,
+      "loss": 0.2018,
       "step": 610
     },
     {
-      "epoch": 6.060215053763441,
-      "grad_norm": 2.740072727203369,
-      "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.4774,
+      "epoch": 3.021146953405018,
+      "grad_norm": 3.8146159648895264,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 0.4042,
       "step": 620
     },
     {
-      "epoch": 6.070967741935484,
-      "grad_norm": 14.478816032409668,
-      "learning_rate": 1.7921146953405018e-05,
-      "loss": 0.2869,
+      "epoch": 3.024731182795699,
+      "grad_norm": 10.83116340637207,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 0.4721,
       "step": 630
     },
     {
-      "epoch": 6.081720430107527,
-      "grad_norm": 19.758691787719727,
-      "learning_rate": 1.7323775388291518e-05,
-      "loss": 0.3443,
+      "epoch": 3.02831541218638,
+      "grad_norm": 16.305315017700195,
+      "learning_rate": 4.281162883313421e-05,
+      "loss": 0.403,
       "step": 640
     },
     {
-      "epoch": 6.09247311827957,
-      "grad_norm": 3.96628737449646,
-      "learning_rate": 1.6726403823178017e-05,
-      "loss": 0.3083,
+      "epoch": 3.031899641577061,
+      "grad_norm": 2.8166565895080566,
+      "learning_rate": 4.261250497809638e-05,
+      "loss": 0.3613,
       "step": 650
     },
     {
-      "epoch": 6.101075268817205,
-      "eval_accuracy": 0.868421052631579,
-      "eval_loss": 0.39320939779281616,
-      "eval_runtime": 25.7723,
-      "eval_samples_per_second": 2.949,
-      "eval_steps_per_second": 0.388,
-      "step": 658
-    },
-    {
-      "epoch": 7.002150537634408,
-      "grad_norm": 14.803962707519531,
-      "learning_rate": 1.6129032258064517e-05,
-      "loss": 0.5458,
+      "epoch": 3.035483870967742,
+      "grad_norm": 11.772737503051758,
+      "learning_rate": 4.241338112305854e-05,
+      "loss": 0.3021,
       "step": 660
     },
     {
-      "epoch": 7.012903225806451,
-      "grad_norm": 12.291862487792969,
-      "learning_rate": 1.5531660692951016e-05,
-      "loss": 0.3551,
+      "epoch": 3.039068100358423,
+      "grad_norm": 3.964470624923706,
+      "learning_rate": 4.2214257268020715e-05,
+      "loss": 0.4299,
       "step": 670
     },
     {
-      "epoch": 7.023655913978494,
-      "grad_norm": 16.88749885559082,
-      "learning_rate": 1.4934289127837516e-05,
-      "loss": 0.3371,
+      "epoch": 3.042652329749104,
+      "grad_norm": 12.679352760314941,
+      "learning_rate": 4.2015133412982874e-05,
+      "loss": 0.4539,
       "step": 680
     },
     {
-      "epoch": 7.034408602150537,
-      "grad_norm": 1.4436932802200317,
-      "learning_rate": 1.4336917562724014e-05,
-      "loss": 0.3117,
+      "epoch": 3.046236559139785,
+      "grad_norm": 5.665447235107422,
+      "learning_rate": 4.1816009557945046e-05,
+      "loss": 0.4301,
       "step": 690
     },
     {
-      "epoch": 7.04516129032258,
-      "grad_norm": 1.6931445598602295,
-      "learning_rate": 1.3739545997610515e-05,
-      "loss": 0.1534,
+      "epoch": 3.049820788530466,
+      "grad_norm": 9.17259693145752,
+      "learning_rate": 4.161688570290721e-05,
+      "loss": 0.3725,
       "step": 700
     },
     {
-      "epoch": 7.055913978494623,
-      "grad_norm": 3.7684617042541504,
-      "learning_rate": 1.3142174432497015e-05,
-      "loss": 0.336,
+      "epoch": 3.053405017921147,
+      "grad_norm": 10.35765266418457,
+      "learning_rate": 4.141776184786938e-05,
+      "loss": 0.3407,
       "step": 710
     },
     {
-      "epoch": 7.066666666666666,
-      "grad_norm": 11.843338966369629,
-      "learning_rate": 1.2544802867383513e-05,
-      "loss": 0.3442,
+      "epoch": 3.056989247311828,
+      "grad_norm": 7.472835063934326,
+      "learning_rate": 4.121863799283154e-05,
+      "loss": 0.2796,
       "step": 720
     },
     {
-      "epoch": 7.077419354838709,
-      "grad_norm": 32.888648986816406,
-      "learning_rate": 1.1947431302270013e-05,
-      "loss": 0.6003,
+      "epoch": 3.060573476702509,
+      "grad_norm": 17.262203216552734,
+      "learning_rate": 4.101951413779371e-05,
+      "loss": 0.5121,
       "step": 730
     },
     {
-      "epoch": 7.088172043010752,
-      "grad_norm": 13.602355003356934,
-      "learning_rate": 1.135005973715651e-05,
-      "loss": 0.411,
+      "epoch": 3.06415770609319,
+      "grad_norm": 5.791836261749268,
+      "learning_rate": 4.082039028275588e-05,
+      "loss": 0.3309,
       "step": 740
     },
     {
-      "epoch": 7.098924731182795,
-      "grad_norm": 0.6173302531242371,
-      "learning_rate": 1.0752688172043012e-05,
-      "loss": 0.3336,
-      "step": 750
+      "epoch": 3.067025089605735,
+      "eval_accuracy": 0.782051282051282,
+      "eval_loss": 0.47816720604896545,
+      "eval_runtime": 51.9165,
+      "eval_samples_per_second": 3.005,
+      "eval_steps_per_second": 0.385,
+      "step": 748
     },
     {
-      "epoch": 7.101075268817205,
-      "eval_accuracy": 0.9078947368421053,
-      "eval_loss": 0.32602888345718384,
-      "eval_runtime": 22.1675,
-      "eval_samples_per_second": 3.428,
-      "eval_steps_per_second": 0.451,
-      "step": 752
+      "epoch": 4.000716845878136,
+      "grad_norm": 7.017057418823242,
+      "learning_rate": 4.062126642771804e-05,
+      "loss": 0.5454,
+      "step": 750
     },
     {
-      "epoch": 8.008602150537634,
-      "grad_norm": 9.76364803314209,
-      "learning_rate": 1.015531660692951e-05,
-      "loss": 0.2842,
+      "epoch": 4.004301075268817,
+      "grad_norm": 3.8564045429229736,
+      "learning_rate": 4.042214257268021e-05,
+      "loss": 0.3336,
       "step": 760
     },
     {
-      "epoch": 8.019354838709678,
-      "grad_norm": 9.685689926147461,
-      "learning_rate": 9.557945041816011e-06,
-      "loss": 0.2046,
+      "epoch": 4.007885304659498,
+      "grad_norm": 4.636725902557373,
+      "learning_rate": 4.0223018717642376e-05,
+      "loss": 0.2867,
       "step": 770
     },
     {
-      "epoch": 8.03010752688172,
-      "grad_norm": 4.679247856140137,
-      "learning_rate": 8.960573476702509e-06,
-      "loss": 0.2956,
+      "epoch": 4.011469534050179,
+      "grad_norm": 8.09864616394043,
+      "learning_rate": 4.002389486260454e-05,
+      "loss": 0.2559,
       "step": 780
     },
     {
-      "epoch": 8.040860215053764,
-      "grad_norm": 4.2983078956604,
-      "learning_rate": 8.363201911589009e-06,
-      "loss": 0.1368,
+      "epoch": 4.01505376344086,
+      "grad_norm": 8.088166236877441,
+      "learning_rate": 3.982477100756671e-05,
+      "loss": 0.3084,
       "step": 790
     },
     {
-      "epoch": 8.051612903225806,
-      "grad_norm": 4.975796699523926,
-      "learning_rate": 7.765830346475508e-06,
-      "loss": 0.3849,
+      "epoch": 4.018637992831541,
+      "grad_norm": 13.00614070892334,
+      "learning_rate": 3.962564715252888e-05,
+      "loss": 0.3581,
       "step": 800
     },
     {
-      "epoch": 8.06236559139785,
-      "grad_norm": 23.385377883911133,
-      "learning_rate": 7.168458781362007e-06,
-      "loss": 0.3219,
+      "epoch": 4.022222222222222,
+      "grad_norm": 4.3727641105651855,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 0.3683,
       "step": 810
     },
     {
-      "epoch": 8.073118279569892,
-      "grad_norm": 9.758036613464355,
-      "learning_rate": 6.5710872162485075e-06,
-      "loss": 0.373,
+      "epoch": 4.025806451612903,
+      "grad_norm": 10.826643943786621,
+      "learning_rate": 3.9227399442453203e-05,
+      "loss": 0.439,
       "step": 820
     },
     {
-      "epoch": 8.083870967741936,
-      "grad_norm": 5.094470977783203,
-      "learning_rate": 5.973715651135006e-06,
-      "loss": 0.2467,
+      "epoch": 4.029390681003584,
+      "grad_norm": 8.039769172668457,
+      "learning_rate": 3.9028275587415376e-05,
+      "loss": 0.3645,
       "step": 830
     },
     {
-      "epoch": 8.094623655913978,
-      "grad_norm": 9.906914710998535,
-      "learning_rate": 5.376344086021506e-06,
-      "loss": 0.2514,
+      "epoch": 4.032974910394265,
+      "grad_norm": 2.0027217864990234,
+      "learning_rate": 3.882915173237754e-05,
+      "loss": 0.3318,
       "step": 840
     },
     {
-      "epoch": 8.101075268817205,
-      "eval_accuracy": 0.9210526315789473,
-      "eval_loss": 0.3206616938114166,
-      "eval_runtime": 21.2555,
-      "eval_samples_per_second": 3.576,
-      "eval_steps_per_second": 0.47,
-      "step": 846
-    },
-    {
-      "epoch": 9.004301075268817,
-      "grad_norm": 0.8976335525512695,
-      "learning_rate": 4.7789725209080055e-06,
-      "loss": 0.2381,
+      "epoch": 4.036559139784946,
+      "grad_norm": 6.5030198097229,
+      "learning_rate": 3.8630027877339706e-05,
+      "loss": 0.4075,
       "step": 850
     },
     {
-      "epoch": 9.01505376344086,
-      "grad_norm": 20.732986450195312,
-      "learning_rate": 4.181600955794504e-06,
-      "loss": 0.2403,
+      "epoch": 4.040143369175627,
+      "grad_norm": 9.211589813232422,
+      "learning_rate": 3.843090402230187e-05,
+      "loss": 0.3196,
       "step": 860
     },
     {
-      "epoch": 9.025806451612903,
-      "grad_norm": 0.5518481731414795,
-      "learning_rate": 3.5842293906810035e-06,
-      "loss": 0.1456,
+      "epoch": 4.043727598566308,
+      "grad_norm": 10.966934204101562,
+      "learning_rate": 3.8231780167264044e-05,
+      "loss": 0.1448,
       "step": 870
     },
     {
-      "epoch": 9.036559139784947,
-      "grad_norm": 9.817909240722656,
-      "learning_rate": 2.986857825567503e-06,
-      "loss": 0.1722,
+      "epoch": 4.047311827956989,
+      "grad_norm": 12.808022499084473,
+      "learning_rate": 3.80326563122262e-05,
+      "loss": 0.3627,
       "step": 880
     },
     {
-      "epoch": 9.047311827956989,
-      "grad_norm": 29.461748123168945,
-      "learning_rate": 2.3894862604540028e-06,
-      "loss": 0.1048,
+      "epoch": 4.05089605734767,
+      "grad_norm": 6.971220016479492,
+      "learning_rate": 3.7833532457188375e-05,
+      "loss": 0.5394,
       "step": 890
     },
     {
-      "epoch": 9.058064516129033,
-      "grad_norm": 8.807541847229004,
-      "learning_rate": 1.7921146953405017e-06,
-      "loss": 0.2389,
+      "epoch": 4.054480286738351,
+      "grad_norm": 5.306809425354004,
+      "learning_rate": 3.763440860215054e-05,
+      "loss": 0.2983,
       "step": 900
     },
     {
-      "epoch": 9.068817204301075,
-      "grad_norm": 2.6277596950531006,
-      "learning_rate": 1.1947431302270014e-06,
-      "loss": 0.193,
+      "epoch": 4.058064516129032,
+      "grad_norm": 3.8580007553100586,
+      "learning_rate": 3.7435284747112706e-05,
+      "loss": 0.4265,
       "step": 910
     },
     {
-      "epoch": 9.079569892473119,
-      "grad_norm": 16.283491134643555,
-      "learning_rate": 5.973715651135007e-07,
-      "loss": 0.3097,
+      "epoch": 4.061648745519713,
+      "grad_norm": 14.511212348937988,
+      "learning_rate": 3.723616089207487e-05,
+      "loss": 0.4302,
       "step": 920
     },
     {
-      "epoch": 9.09032258064516,
-      "grad_norm": 21.258689880371094,
-      "learning_rate": 0.0,
-      "loss": 0.3523,
+      "epoch": 4.065232974910394,
+      "grad_norm": 11.154169082641602,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.4007,
       "step": 930
     },
     {
-      "epoch": 9.09032258064516,
-      "eval_accuracy": 0.9473684210526315,
-      "eval_loss": 0.2718675136566162,
-      "eval_runtime": 25.0335,
-      "eval_samples_per_second": 3.036,
-      "eval_steps_per_second": 0.399,
-      "step": 930
+      "epoch": 4.0670250896057345,
+      "eval_accuracy": 0.8461538461538461,
+      "eval_loss": 0.4134625196456909,
+      "eval_runtime": 50.528,
+      "eval_samples_per_second": 3.087,
+      "eval_steps_per_second": 0.396,
+      "step": 935
     },
     {
-      "epoch": 9.09032258064516,
-      "step": 930,
-      "total_flos": 9.225862220448793e+18,
-      "train_loss": 0.4273061529282601,
-      "train_runtime": 3527.5997,
-      "train_samples_per_second": 2.109,
-      "train_steps_per_second": 0.264
+      "epoch": 5.00179211469534,
+      "grad_norm": 1.4756416082382202,
+      "learning_rate": 3.683791318199921e-05,
+      "loss": 0.3005,
+      "step": 940
     },
     {
-      "epoch": 9.09032258064516,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.7193946838378906,
-      "eval_runtime": 163.1453,
-      "eval_samples_per_second": 2.452,
-      "eval_steps_per_second": 0.306,
-      "step": 930
+      "epoch": 5.005376344086022,
+      "grad_norm": 7.019712924957275,
+      "learning_rate": 3.663878932696137e-05,
+      "loss": 0.3069,
+      "step": 950
     },
     {
-      "epoch": 9.09032258064516,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.7193946838378906,
-      "eval_runtime": 151.4396,
-      "eval_samples_per_second": 2.641,
-      "eval_steps_per_second": 0.33,
-      "step": 930
+      "epoch": 5.008960573476703,
+      "grad_norm": 14.922005653381348,
+      "learning_rate": 3.643966547192354e-05,
+      "loss": 0.2685,
+      "step": 960
+    },
+    {
+      "epoch": 5.012544802867383,
+      "grad_norm": 17.716806411743164,
+      "learning_rate": 3.6240541616885705e-05,
+      "loss": 0.3368,
+      "step": 970
+    },
+    {
+      "epoch": 5.016129032258065,
+      "grad_norm": 30.634918212890625,
+      "learning_rate": 3.604141776184787e-05,
+      "loss": 0.4523,
+      "step": 980
+    },
+    {
+      "epoch": 5.019713261648746,
+      "grad_norm": 6.613483905792236,
+      "learning_rate": 3.5842293906810036e-05,
+      "loss": 0.4893,
+      "step": 990
+    },
+    {
+      "epoch": 5.023297491039426,
+      "grad_norm": 9.66140365600586,
+      "learning_rate": 3.564317005177221e-05,
+      "loss": 0.3979,
+      "step": 1000
+    },
+    {
+      "epoch": 5.026881720430108,
+      "grad_norm": 8.311437606811523,
+      "learning_rate": 3.5444046196734373e-05,
+      "loss": 0.2798,
+      "step": 1010
+    },
+    {
+      "epoch": 5.030465949820789,
+      "grad_norm": 3.4359304904937744,
+      "learning_rate": 3.524492234169653e-05,
+      "loss": 0.2007,
+      "step": 1020
+    },
+    {
+      "epoch": 5.034050179211469,
+      "grad_norm": 2.4346439838409424,
+      "learning_rate": 3.5045798486658704e-05,
+      "loss": 0.3485,
+      "step": 1030
+    },
+    {
+      "epoch": 5.037634408602151,
+      "grad_norm": 16.60686683654785,
+      "learning_rate": 3.484667463162087e-05,
+      "loss": 0.3642,
+      "step": 1040
+    },
+    {
+      "epoch": 5.041218637992832,
+      "grad_norm": 6.6109466552734375,
+      "learning_rate": 3.4647550776583035e-05,
+      "loss": 0.4869,
+      "step": 1050
+    },
+    {
+      "epoch": 5.044802867383512,
+      "grad_norm": 14.017187118530273,
+      "learning_rate": 3.44484269215452e-05,
+      "loss": 0.3117,
+      "step": 1060
+    },
+    {
+      "epoch": 5.048387096774194,
+      "grad_norm": 6.0544047355651855,
+      "learning_rate": 3.424930306650737e-05,
+      "loss": 0.3086,
+      "step": 1070
+    },
+    {
+      "epoch": 5.051971326164875,
+      "grad_norm": 12.737698554992676,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 0.3758,
+      "step": 1080
+    },
+    {
+      "epoch": 5.055555555555555,
+      "grad_norm": 3.5636813640594482,
+      "learning_rate": 3.3851055356431704e-05,
+      "loss": 0.3358,
+      "step": 1090
+    },
+    {
+      "epoch": 5.059139784946237,
+      "grad_norm": 15.545931816101074,
+      "learning_rate": 3.365193150139387e-05,
+      "loss": 0.442,
+      "step": 1100
+    },
+    {
+      "epoch": 5.062724014336918,
+      "grad_norm": 26.353408813476562,
+      "learning_rate": 3.3452807646356034e-05,
+      "loss": 0.2054,
+      "step": 1110
+    },
+    {
+      "epoch": 5.066308243727598,
+      "grad_norm": 28.000761032104492,
+      "learning_rate": 3.32536837913182e-05,
+      "loss": 0.3772,
+      "step": 1120
+    },
+    {
+      "epoch": 5.0670250896057345,
+      "eval_accuracy": 0.8461538461538461,
+      "eval_loss": 0.4329083263874054,
+      "eval_runtime": 50.4842,
+      "eval_samples_per_second": 3.09,
+      "eval_steps_per_second": 0.396,
+      "step": 1122
+    },
+    {
+      "epoch": 6.002867383512545,
+      "grad_norm": 6.335020065307617,
+      "learning_rate": 3.3054559936280365e-05,
+      "loss": 0.1473,
+      "step": 1130
+    },
+    {
+      "epoch": 6.006451612903226,
+      "grad_norm": 2.6890759468078613,
+      "learning_rate": 3.285543608124254e-05,
+      "loss": 0.4162,
+      "step": 1140
+    },
+    {
+      "epoch": 6.010035842293907,
+      "grad_norm": 11.375259399414062,
+      "learning_rate": 3.26563122262047e-05,
+      "loss": 0.3748,
+      "step": 1150
+    },
+    {
+      "epoch": 6.013620071684588,
+      "grad_norm": 8.632832527160645,
+      "learning_rate": 3.245718837116687e-05,
+      "loss": 0.2576,
+      "step": 1160
+    },
+    {
+      "epoch": 6.017204301075269,
+      "grad_norm": 6.899137496948242,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4902,
+      "step": 1170
+    },
+    {
+      "epoch": 6.02078853046595,
+      "grad_norm": 6.093740463256836,
+      "learning_rate": 3.2058940661091206e-05,
+      "loss": 0.3972,
+      "step": 1180
+    },
+    {
+      "epoch": 6.024372759856631,
+      "grad_norm": 8.65532398223877,
+      "learning_rate": 3.1859816806053364e-05,
+      "loss": 0.333,
+      "step": 1190
+    },
+    {
+      "epoch": 6.027956989247312,
+      "grad_norm": 4.51920747756958,
+      "learning_rate": 3.1660692951015537e-05,
+      "loss": 0.3342,
+      "step": 1200
+    },
+    {
+      "epoch": 6.031541218637993,
+      "grad_norm": 16.826913833618164,
+      "learning_rate": 3.14615690959777e-05,
+      "loss": 0.2567,
+      "step": 1210
+    },
+    {
+      "epoch": 6.035125448028674,
+      "grad_norm": 1.91347336769104,
+      "learning_rate": 3.126244524093986e-05,
+      "loss": 0.2524,
+      "step": 1220
+    },
+    {
+      "epoch": 6.038709677419355,
+      "grad_norm": 1.6974544525146484,
+      "learning_rate": 3.106332138590203e-05,
+      "loss": 0.4133,
+      "step": 1230
+    },
+    {
+      "epoch": 6.042293906810036,
+      "grad_norm": 6.3330864906311035,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.3822,
+      "step": 1240
+    },
+    {
+      "epoch": 6.045878136200717,
+      "grad_norm": 1.5088220834732056,
+      "learning_rate": 3.0665073675826364e-05,
+      "loss": 0.2523,
+      "step": 1250
+    },
+    {
+      "epoch": 6.049462365591398,
+      "grad_norm": 4.4250946044921875,
+      "learning_rate": 3.046594982078853e-05,
+      "loss": 0.2141,
+      "step": 1260
+    },
+    {
+      "epoch": 6.053046594982079,
+      "grad_norm": 0.5099472403526306,
+      "learning_rate": 3.02668259657507e-05,
+      "loss": 0.1307,
+      "step": 1270
+    },
+    {
+      "epoch": 6.05663082437276,
+      "grad_norm": 10.32909870147705,
+      "learning_rate": 3.0067702110712863e-05,
+      "loss": 0.5991,
+      "step": 1280
+    },
+    {
+      "epoch": 6.060215053763441,
+      "grad_norm": 5.504093647003174,
+      "learning_rate": 2.9868578255675032e-05,
+      "loss": 0.2848,
+      "step": 1290
+    },
+    {
+      "epoch": 6.063799283154122,
+      "grad_norm": 15.00183391571045,
+      "learning_rate": 2.9669454400637198e-05,
+      "loss": 0.4685,
+      "step": 1300
+    },
+    {
+      "epoch": 6.0670250896057345,
+      "eval_accuracy": 0.8653846153846154,
+      "eval_loss": 0.4190846085548401,
+      "eval_runtime": 52.3206,
+      "eval_samples_per_second": 2.982,
+      "eval_steps_per_second": 0.382,
+      "step": 1309
+    },
+    {
+      "epoch": 7.000358422939068,
+      "grad_norm": 20.213930130004883,
+      "learning_rate": 2.9470330545599363e-05,
+      "loss": 0.2573,
+      "step": 1310
+    },
+    {
+      "epoch": 7.003942652329749,
+      "grad_norm": 8.981283187866211,
+      "learning_rate": 2.9271206690561532e-05,
+      "loss": 0.2385,
+      "step": 1320
+    },
+    {
+      "epoch": 7.00752688172043,
+      "grad_norm": 4.789116859436035,
+      "learning_rate": 2.9072082835523694e-05,
+      "loss": 0.4895,
+      "step": 1330
+    },
+    {
+      "epoch": 7.011111111111111,
+      "grad_norm": 16.981555938720703,
+      "learning_rate": 2.8872958980485866e-05,
+      "loss": 0.2175,
+      "step": 1340
+    },
+    {
+      "epoch": 7.014695340501792,
+      "grad_norm": 0.44299283623695374,
+      "learning_rate": 2.8673835125448028e-05,
+      "loss": 0.2241,
+      "step": 1350
+    },
+    {
+      "epoch": 7.018279569892473,
+      "grad_norm": 18.93329429626465,
+      "learning_rate": 2.8474711270410197e-05,
+      "loss": 0.333,
+      "step": 1360
+    },
+    {
+      "epoch": 7.021863799283154,
+      "grad_norm": 10.7378511428833,
+      "learning_rate": 2.8275587415372362e-05,
+      "loss": 0.3464,
+      "step": 1370
+    },
+    {
+      "epoch": 7.025448028673835,
+      "grad_norm": 6.727279186248779,
+      "learning_rate": 2.807646356033453e-05,
+      "loss": 0.4513,
+      "step": 1380
+    },
+    {
+      "epoch": 7.029032258064516,
+      "grad_norm": 0.6868817806243896,
+      "learning_rate": 2.7877339705296696e-05,
+      "loss": 0.3245,
+      "step": 1390
+    },
+    {
+      "epoch": 7.032616487455197,
+      "grad_norm": 2.979949951171875,
+      "learning_rate": 2.7678215850258865e-05,
+      "loss": 0.3958,
+      "step": 1400
+    },
+    {
+      "epoch": 7.036200716845878,
+      "grad_norm": 7.619071960449219,
+      "learning_rate": 2.747909199522103e-05,
+      "loss": 0.2745,
+      "step": 1410
+    },
+    {
+      "epoch": 7.039784946236559,
+      "grad_norm": 6.652284145355225,
+      "learning_rate": 2.7279968140183193e-05,
+      "loss": 0.3702,
+      "step": 1420
+    },
+    {
+      "epoch": 7.04336917562724,
+      "grad_norm": 7.907419204711914,
+      "learning_rate": 2.708084428514536e-05,
+      "loss": 0.2073,
+      "step": 1430
+    },
+    {
+      "epoch": 7.046953405017921,
+      "grad_norm": 10.869131088256836,
+      "learning_rate": 2.6881720430107527e-05,
+      "loss": 0.2441,
+      "step": 1440
+    },
+    {
+      "epoch": 7.050537634408602,
+      "grad_norm": 4.352448463439941,
+      "learning_rate": 2.6682596575069696e-05,
+      "loss": 0.2584,
+      "step": 1450
+    },
+    {
+      "epoch": 7.054121863799283,
+      "grad_norm": 15.832218170166016,
+      "learning_rate": 2.648347272003186e-05,
+      "loss": 0.4251,
+      "step": 1460
+    },
+    {
+      "epoch": 7.057706093189964,
+      "grad_norm": 4.763960361480713,
+      "learning_rate": 2.628434886499403e-05,
+      "loss": 0.1938,
+      "step": 1470
+    },
+    {
+      "epoch": 7.061290322580645,
+      "grad_norm": 0.8033217191696167,
+      "learning_rate": 2.6085225009956192e-05,
+      "loss": 0.1949,
+      "step": 1480
+    },
+    {
+      "epoch": 7.064874551971326,
+      "grad_norm": 18.37034797668457,
+      "learning_rate": 2.5886101154918364e-05,
+      "loss": 0.4056,
+      "step": 1490
+    },
+    {
+      "epoch": 7.0670250896057345,
+      "eval_accuracy": 0.8012820512820513,
+      "eval_loss": 0.5650048851966858,
+      "eval_runtime": 60.9039,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 0.328,
+      "step": 1496
+    },
+    {
+      "epoch": 8.001433691756272,
+      "grad_norm": 15.786755561828613,
+      "learning_rate": 2.5686977299880526e-05,
+      "loss": 0.2818,
+      "step": 1500
+    },
+    {
+      "epoch": 8.005017921146953,
+      "grad_norm": 9.856786727905273,
+      "learning_rate": 2.548785344484269e-05,
+      "loss": 0.243,
+      "step": 1510
+    },
+    {
+      "epoch": 8.008602150537634,
+      "grad_norm": 9.001197814941406,
+      "learning_rate": 2.528872958980486e-05,
+      "loss": 0.308,
+      "step": 1520
+    },
+    {
+      "epoch": 8.012186379928316,
+      "grad_norm": 11.027730941772461,
+      "learning_rate": 2.5089605734767026e-05,
+      "loss": 0.1852,
+      "step": 1530
+    },
+    {
+      "epoch": 8.015770609318997,
+      "grad_norm": 15.640365600585938,
+      "learning_rate": 2.4890481879729195e-05,
+      "loss": 0.3335,
+      "step": 1540
+    },
+    {
+      "epoch": 8.019354838709678,
+      "grad_norm": 20.133060455322266,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 0.3552,
+      "step": 1550
+    },
+    {
+      "epoch": 8.022939068100358,
+      "grad_norm": 0.8335391283035278,
+      "learning_rate": 2.4492234169653525e-05,
+      "loss": 0.3064,
+      "step": 1560
+    },
+    {
+      "epoch": 8.026523297491039,
+      "grad_norm": 0.4149135947227478,
+      "learning_rate": 2.429311031461569e-05,
+      "loss": 0.1344,
+      "step": 1570
+    },
+    {
+      "epoch": 8.03010752688172,
+      "grad_norm": 0.058533698320388794,
+      "learning_rate": 2.409398645957786e-05,
+      "loss": 0.139,
+      "step": 1580
+    },
+    {
+      "epoch": 8.033691756272402,
+      "grad_norm": 0.27755093574523926,
+      "learning_rate": 2.3894862604540025e-05,
+      "loss": 0.3311,
+      "step": 1590
+    },
+    {
+      "epoch": 8.037275985663083,
+      "grad_norm": 3.6542258262634277,
+      "learning_rate": 2.369573874950219e-05,
+      "loss": 0.1689,
+      "step": 1600
+    },
+    {
+      "epoch": 8.040860215053764,
+      "grad_norm": 0.14547933638095856,
+      "learning_rate": 2.349661489446436e-05,
+      "loss": 0.1994,
+      "step": 1610
+    },
+    {
+      "epoch": 8.044444444444444,
+      "grad_norm": 0.17915168404579163,
+      "learning_rate": 2.3297491039426525e-05,
+      "loss": 0.3071,
+      "step": 1620
+    },
+    {
+      "epoch": 8.048028673835125,
+      "grad_norm": 1.3044124841690063,
+      "learning_rate": 2.309836718438869e-05,
+      "loss": 0.2893,
+      "step": 1630
+    },
+    {
+      "epoch": 8.051612903225806,
+      "grad_norm": 7.5866804122924805,
+      "learning_rate": 2.289924332935086e-05,
+      "loss": 0.245,
+      "step": 1640
+    },
+    {
+      "epoch": 8.055197132616488,
+      "grad_norm": 16.545440673828125,
+      "learning_rate": 2.270011947431302e-05,
+      "loss": 0.1421,
+      "step": 1650
+    },
+    {
+      "epoch": 8.058781362007169,
+      "grad_norm": 19.21019744873047,
+      "learning_rate": 2.250099561927519e-05,
+      "loss": 0.2807,
+      "step": 1660
+    },
+    {
+      "epoch": 8.06236559139785,
+      "grad_norm": 7.801510334014893,
+      "learning_rate": 2.2301871764237355e-05,
+      "loss": 0.2407,
+      "step": 1670
+    },
+    {
+      "epoch": 8.06594982078853,
+      "grad_norm": 9.918209075927734,
+      "learning_rate": 2.2102747909199524e-05,
+      "loss": 0.2306,
+      "step": 1680
+    },
+    {
+      "epoch": 8.067025089605735,
+      "eval_accuracy": 0.8076923076923077,
+      "eval_loss": 0.7092758417129517,
+      "eval_runtime": 52.4717,
+      "eval_samples_per_second": 2.973,
+      "eval_steps_per_second": 0.381,
+      "step": 1683
+    },
+    {
+      "epoch": 9.002508960573477,
+      "grad_norm": 9.930907249450684,
+      "learning_rate": 2.190362405416169e-05,
+      "loss": 0.2508,
+      "step": 1690
+    },
+    {
+      "epoch": 9.006093189964158,
+      "grad_norm": 0.4230400621891022,
+      "learning_rate": 2.1704500199123855e-05,
+      "loss": 0.1182,
+      "step": 1700
+    },
+    {
+      "epoch": 9.009677419354839,
+      "grad_norm": 19.38228416442871,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.2058,
+      "step": 1710
+    },
+    {
+      "epoch": 9.01326164874552,
+      "grad_norm": 0.1429557204246521,
+      "learning_rate": 2.130625248904819e-05,
+      "loss": 0.1859,
+      "step": 1720
+    },
+    {
+      "epoch": 9.0168458781362,
+      "grad_norm": 14.09881591796875,
+      "learning_rate": 2.1107128634010358e-05,
+      "loss": 0.165,
+      "step": 1730
+    },
+    {
+      "epoch": 9.020430107526881,
+      "grad_norm": 13.956668853759766,
+      "learning_rate": 2.0908004778972523e-05,
+      "loss": 0.5844,
+      "step": 1740
+    },
+    {
+      "epoch": 9.024014336917563,
+      "grad_norm": 1.542794108390808,
+      "learning_rate": 2.070888092393469e-05,
+      "loss": 0.2968,
+      "step": 1750
+    },
+    {
+      "epoch": 9.027598566308244,
+      "grad_norm": 17.280027389526367,
+      "learning_rate": 2.0509757068896854e-05,
+      "loss": 0.2715,
+      "step": 1760
+    },
+    {
+      "epoch": 9.031182795698925,
+      "grad_norm": 0.40804412961006165,
+      "learning_rate": 2.031063321385902e-05,
+      "loss": 0.2193,
+      "step": 1770
+    },
+    {
+      "epoch": 9.034767025089605,
+      "grad_norm": 0.2087012380361557,
+      "learning_rate": 2.0111509358821188e-05,
+      "loss": 0.3301,
+      "step": 1780
+    },
+    {
+      "epoch": 9.038351254480286,
+      "grad_norm": 12.486738204956055,
+      "learning_rate": 1.9912385503783354e-05,
+      "loss": 0.3881,
+      "step": 1790
+    },
+    {
+      "epoch": 9.041935483870967,
+      "grad_norm": 9.798795700073242,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 0.366,
+      "step": 1800
+    },
+    {
+      "epoch": 9.04551971326165,
+      "grad_norm": 25.9205379486084,
+      "learning_rate": 1.9514137793707688e-05,
+      "loss": 0.3944,
+      "step": 1810
+    },
+    {
+      "epoch": 9.04910394265233,
+      "grad_norm": 14.600288391113281,
+      "learning_rate": 1.9315013938669853e-05,
+      "loss": 0.0899,
+      "step": 1820
+    },
+    {
+      "epoch": 9.05268817204301,
+      "grad_norm": 0.17624612152576447,
+      "learning_rate": 1.9115890083632022e-05,
+      "loss": 0.4656,
+      "step": 1830
+    },
+    {
+      "epoch": 9.056272401433691,
+      "grad_norm": 7.002784252166748,
+      "learning_rate": 1.8916766228594187e-05,
+      "loss": 0.2776,
+      "step": 1840
+    },
+    {
+      "epoch": 9.059856630824372,
+      "grad_norm": 6.815585136413574,
+      "learning_rate": 1.8717642373556353e-05,
+      "loss": 0.1246,
+      "step": 1850
+    },
+    {
+      "epoch": 9.063440860215053,
+      "grad_norm": 25.917802810668945,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.1041,
+      "step": 1860
+    },
+    {
+      "epoch": 9.067025089605735,
+      "grad_norm": 0.36158832907676697,
+      "learning_rate": 1.8319394663480684e-05,
+      "loss": 0.304,
+      "step": 1870
+    },
+    {
+      "epoch": 9.067025089605735,
+      "eval_accuracy": 0.8782051282051282,
+      "eval_loss": 0.39389485120773315,
+      "eval_runtime": 50.2949,
+      "eval_samples_per_second": 3.102,
+      "eval_steps_per_second": 0.398,
+      "step": 1870
+    },
+    {
+      "epoch": 10.00358422939068,
+      "grad_norm": 12.316838264465332,
+      "learning_rate": 1.8120270808442852e-05,
+      "loss": 0.2954,
+      "step": 1880
+    },
+    {
+      "epoch": 10.007168458781361,
+      "grad_norm": 8.945836067199707,
+      "learning_rate": 1.7921146953405018e-05,
+      "loss": 0.3398,
+      "step": 1890
+    },
+    {
+      "epoch": 10.010752688172044,
+      "grad_norm": 9.732643127441406,
+      "learning_rate": 1.7722023098367187e-05,
+      "loss": 0.3115,
+      "step": 1900
+    },
+    {
+      "epoch": 10.014336917562725,
+      "grad_norm": 19.322099685668945,
+      "learning_rate": 1.7522899243329352e-05,
+      "loss": 0.1925,
+      "step": 1910
+    },
+    {
+      "epoch": 10.017921146953405,
+      "grad_norm": 17.751829147338867,
+      "learning_rate": 1.7323775388291518e-05,
+      "loss": 0.1247,
+      "step": 1920
+    },
+    {
+      "epoch": 10.021505376344086,
+      "grad_norm": 0.4507576525211334,
+      "learning_rate": 1.7124651533253686e-05,
+      "loss": 0.183,
+      "step": 1930
+    },
+    {
+      "epoch": 10.025089605734767,
+      "grad_norm": 2.866403341293335,
+      "learning_rate": 1.6925527678215852e-05,
+      "loss": 0.2288,
+      "step": 1940
+    },
+    {
+      "epoch": 10.028673835125447,
+      "grad_norm": 1.3826611042022705,
+      "learning_rate": 1.6726403823178017e-05,
+      "loss": 0.2216,
+      "step": 1950
+    },
+    {
+      "epoch": 10.03225806451613,
+      "grad_norm": 0.2175895720720291,
+      "learning_rate": 1.6527279968140183e-05,
+      "loss": 0.3319,
+      "step": 1960
+    },
+    {
+      "epoch": 10.03584229390681,
+      "grad_norm": 0.07152961939573288,
+      "learning_rate": 1.632815611310235e-05,
+      "loss": 0.1087,
+      "step": 1970
+    },
+    {
+      "epoch": 10.039426523297491,
+      "grad_norm": 12.132889747619629,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.2449,
+      "step": 1980
+    },
+    {
+      "epoch": 10.043010752688172,
+      "grad_norm": 25.55087661743164,
+      "learning_rate": 1.5929908403026682e-05,
+      "loss": 0.1797,
+      "step": 1990
+    },
+    {
+      "epoch": 10.046594982078853,
+      "grad_norm": 1.101890206336975,
+      "learning_rate": 1.573078454798885e-05,
+      "loss": 0.1244,
+      "step": 2000
+    },
+    {
+      "epoch": 10.050179211469533,
+      "grad_norm": 2.2634036540985107,
+      "learning_rate": 1.5531660692951016e-05,
+      "loss": 0.2921,
+      "step": 2010
+    },
+    {
+      "epoch": 10.053763440860216,
+      "grad_norm": 0.08446018397808075,
+      "learning_rate": 1.5332536837913182e-05,
+      "loss": 0.3367,
+      "step": 2020
+    },
+    {
+      "epoch": 10.057347670250897,
+      "grad_norm": 18.109861373901367,
+      "learning_rate": 1.513341298287535e-05,
+      "loss": 0.1402,
+      "step": 2030
+    },
+    {
+      "epoch": 10.060931899641577,
+      "grad_norm": 4.889894485473633,
+      "learning_rate": 1.4934289127837516e-05,
+      "loss": 0.3135,
+      "step": 2040
+    },
+    {
+      "epoch": 10.064516129032258,
+      "grad_norm": 0.5321334004402161,
+      "learning_rate": 1.4735165272799681e-05,
+      "loss": 0.2418,
+      "step": 2050
+    },
+    {
+      "epoch": 10.067025089605735,
+      "eval_accuracy": 0.8333333333333334,
+      "eval_loss": 0.5524674654006958,
+      "eval_runtime": 49.8947,
+      "eval_samples_per_second": 3.127,
+      "eval_steps_per_second": 0.401,
+      "step": 2057
+    },
+    {
+      "epoch": 11.001075268817205,
+      "grad_norm": 8.704042434692383,
+      "learning_rate": 1.4536041417761847e-05,
+      "loss": 0.3554,
+      "step": 2060
+    },
+    {
+      "epoch": 11.004659498207886,
+      "grad_norm": 2.6651906967163086,
+      "learning_rate": 1.4336917562724014e-05,
+      "loss": 0.1467,
+      "step": 2070
+    },
+    {
+      "epoch": 11.008243727598567,
+      "grad_norm": 8.130305290222168,
+      "learning_rate": 1.4137793707686181e-05,
+      "loss": 0.1653,
+      "step": 2080
+    },
+    {
+      "epoch": 11.011827956989247,
+      "grad_norm": 0.1315879374742508,
+      "learning_rate": 1.3938669852648348e-05,
+      "loss": 0.2511,
+      "step": 2090
+    },
+    {
+      "epoch": 11.015412186379928,
+      "grad_norm": 0.27568188309669495,
+      "learning_rate": 1.3739545997610515e-05,
+      "loss": 0.1472,
+      "step": 2100
+    },
+    {
+      "epoch": 11.018996415770609,
+      "grad_norm": 0.6477882862091064,
+      "learning_rate": 1.354042214257268e-05,
+      "loss": 0.2646,
+      "step": 2110
+    },
+    {
+      "epoch": 11.022580645161291,
+      "grad_norm": 18.025819778442383,
+      "learning_rate": 1.3341298287534848e-05,
+      "loss": 0.1733,
+      "step": 2120
+    },
+    {
+      "epoch": 11.026164874551972,
+      "grad_norm": 4.365772724151611,
+      "learning_rate": 1.3142174432497015e-05,
+      "loss": 0.3097,
+      "step": 2130
+    },
+    {
+      "epoch": 11.029749103942653,
+      "grad_norm": 36.36382293701172,
+      "learning_rate": 1.2943050577459182e-05,
+      "loss": 0.3242,
+      "step": 2140
+    },
+    {
+      "epoch": 11.033333333333333,
+      "grad_norm": 0.6666226983070374,
+      "learning_rate": 1.2743926722421346e-05,
+      "loss": 0.0725,
+      "step": 2150
+    },
+    {
+      "epoch": 11.036917562724014,
+      "grad_norm": 17.053176879882812,
+      "learning_rate": 1.2544802867383513e-05,
+      "loss": 0.207,
+      "step": 2160
+    },
+    {
+      "epoch": 11.040501792114695,
+      "grad_norm": 24.50665283203125,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 0.1859,
+      "step": 2170
+    },
+    {
+      "epoch": 11.044086021505377,
+      "grad_norm": 15.833036422729492,
+      "learning_rate": 1.2146555157307845e-05,
+      "loss": 0.2055,
+      "step": 2180
+    },
+    {
+      "epoch": 11.047670250896058,
+      "grad_norm": 4.489932537078857,
+      "learning_rate": 1.1947431302270013e-05,
+      "loss": 0.1705,
+      "step": 2190
+    },
+    {
+      "epoch": 11.051254480286739,
+      "grad_norm": 1.9370172023773193,
+      "learning_rate": 1.174830744723218e-05,
+      "loss": 0.2044,
+      "step": 2200
+    },
+    {
+      "epoch": 11.05483870967742,
+      "grad_norm": 12.393142700195312,
+      "learning_rate": 1.1549183592194345e-05,
+      "loss": 0.1342,
+      "step": 2210
+    },
+    {
+      "epoch": 11.0584229390681,
+      "grad_norm": 16.235197067260742,
+      "learning_rate": 1.135005973715651e-05,
+      "loss": 0.2124,
+      "step": 2220
+    },
+    {
+      "epoch": 11.06200716845878,
+      "grad_norm": 0.11374145746231079,
+      "learning_rate": 1.1150935882118678e-05,
+      "loss": 0.235,
+      "step": 2230
+    },
+    {
+      "epoch": 11.065591397849463,
+      "grad_norm": 41.60282897949219,
+      "learning_rate": 1.0951812027080845e-05,
+      "loss": 0.2089,
+      "step": 2240
+    },
+    {
+      "epoch": 11.067025089605735,
+      "eval_accuracy": 0.8589743589743589,
+      "eval_loss": 0.5139220356941223,
+      "eval_runtime": 50.5308,
+      "eval_samples_per_second": 3.087,
+      "eval_steps_per_second": 0.396,
+      "step": 2244
+    },
+    {
+      "epoch": 12.002150537634408,
+      "grad_norm": 24.850767135620117,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.258,
+      "step": 2250
+    },
+    {
+      "epoch": 12.00573476702509,
+      "grad_norm": 2.8716249465942383,
+      "learning_rate": 1.0553564317005179e-05,
+      "loss": 0.2319,
+      "step": 2260
+    },
+    {
+      "epoch": 12.00931899641577,
+      "grad_norm": 1.9340592622756958,
+      "learning_rate": 1.0354440461967344e-05,
+      "loss": 0.1552,
+      "step": 2270
+    },
+    {
+      "epoch": 12.012903225806452,
+      "grad_norm": 0.8818116784095764,
+      "learning_rate": 1.015531660692951e-05,
+      "loss": 0.0837,
+      "step": 2280
+    },
+    {
+      "epoch": 12.016487455197133,
+      "grad_norm": 0.1528209149837494,
+      "learning_rate": 9.956192751891677e-06,
+      "loss": 0.0903,
+      "step": 2290
+    },
+    {
+      "epoch": 12.020071684587814,
+      "grad_norm": 0.201262429356575,
+      "learning_rate": 9.757068896853844e-06,
+      "loss": 0.137,
+      "step": 2300
+    },
+    {
+      "epoch": 12.023655913978494,
+      "grad_norm": 23.355026245117188,
+      "learning_rate": 9.557945041816011e-06,
+      "loss": 0.2225,
+      "step": 2310
+    },
+    {
+      "epoch": 12.027240143369175,
+      "grad_norm": 22.52867889404297,
+      "learning_rate": 9.358821186778176e-06,
+      "loss": 0.1936,
+      "step": 2320
+    },
+    {
+      "epoch": 12.030824372759856,
+      "grad_norm": 5.563544273376465,
+      "learning_rate": 9.159697331740342e-06,
+      "loss": 0.3059,
+      "step": 2330
+    },
+    {
+      "epoch": 12.034408602150538,
+      "grad_norm": 26.484102249145508,
+      "learning_rate": 8.960573476702509e-06,
+      "loss": 0.2575,
+      "step": 2340
+    },
+    {
+      "epoch": 12.037992831541219,
+      "grad_norm": 0.5328412055969238,
+      "learning_rate": 8.761449621664676e-06,
+      "loss": 0.2211,
+      "step": 2350
+    },
+    {
+      "epoch": 12.0415770609319,
+      "grad_norm": 29.48992156982422,
+      "learning_rate": 8.562325766626843e-06,
+      "loss": 0.2132,
+      "step": 2360
+    },
+    {
+      "epoch": 12.04516129032258,
+      "grad_norm": 0.5142113566398621,
+      "learning_rate": 8.363201911589009e-06,
+      "loss": 0.1998,
+      "step": 2370
+    },
+    {
+      "epoch": 12.048745519713261,
+      "grad_norm": 22.66319465637207,
+      "learning_rate": 8.164078056551176e-06,
+      "loss": 0.2969,
+      "step": 2380
+    },
+    {
+      "epoch": 12.052329749103942,
+      "grad_norm": 27.633180618286133,
+      "learning_rate": 7.964954201513341e-06,
+      "loss": 0.1352,
+      "step": 2390
+    },
+    {
+      "epoch": 12.055913978494624,
+      "grad_norm": 8.562397956848145,
+      "learning_rate": 7.765830346475508e-06,
+      "loss": 0.1831,
+      "step": 2400
+    },
+    {
+      "epoch": 12.059498207885305,
+      "grad_norm": 8.322550773620605,
+      "learning_rate": 7.566706491437675e-06,
+      "loss": 0.138,
+      "step": 2410
+    },
+    {
+      "epoch": 12.063082437275986,
+      "grad_norm": 24.926652908325195,
+      "learning_rate": 7.367582636399841e-06,
+      "loss": 0.1709,
+      "step": 2420
+    },
+    {
+      "epoch": 12.066666666666666,
+      "grad_norm": 17.362070083618164,
+      "learning_rate": 7.168458781362007e-06,
+      "loss": 0.3158,
+      "step": 2430
+    },
+    {
+      "epoch": 12.067025089605735,
+      "eval_accuracy": 0.8589743589743589,
+      "eval_loss": 0.5392084717750549,
+      "eval_runtime": 50.6612,
+      "eval_samples_per_second": 3.079,
+      "eval_steps_per_second": 0.395,
+      "step": 2431
+    },
+    {
+      "epoch": 13.003225806451614,
+      "grad_norm": 17.38275146484375,
+      "learning_rate": 6.969334926324174e-06,
+      "loss": 0.0979,
+      "step": 2440
+    },
+    {
+      "epoch": 13.006810035842294,
+      "grad_norm": 0.16219447553157806,
+      "learning_rate": 6.77021107128634e-06,
+      "loss": 0.1422,
+      "step": 2450
+    },
+    {
+      "epoch": 13.010394265232975,
+      "grad_norm": 0.48416832089424133,
+      "learning_rate": 6.5710872162485075e-06,
+      "loss": 0.0831,
+      "step": 2460
+    },
+    {
+      "epoch": 13.013978494623656,
+      "grad_norm": 3.7526910305023193,
+      "learning_rate": 6.371963361210673e-06,
+      "loss": 0.2188,
+      "step": 2470
+    },
+    {
+      "epoch": 13.017562724014336,
+      "grad_norm": 0.4946393072605133,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 0.0438,
+      "step": 2480
+    },
+    {
+      "epoch": 13.021146953405017,
+      "grad_norm": 14.273146629333496,
+      "learning_rate": 5.973715651135006e-06,
+      "loss": 0.1903,
+      "step": 2490
+    },
+    {
+      "epoch": 13.0247311827957,
+      "grad_norm": 0.20919331908226013,
+      "learning_rate": 5.7745917960971725e-06,
+      "loss": 0.1862,
+      "step": 2500
+    },
+    {
+      "epoch": 13.02831541218638,
+      "grad_norm": 44.509796142578125,
+      "learning_rate": 5.575467941059339e-06,
+      "loss": 0.3252,
+      "step": 2510
+    },
+    {
+      "epoch": 13.031899641577061,
+      "grad_norm": 1.4209870100021362,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 0.1379,
+      "step": 2520
+    },
+    {
+      "epoch": 13.035483870967742,
+      "grad_norm": 32.848148345947266,
+      "learning_rate": 5.177220230983672e-06,
+      "loss": 0.1815,
+      "step": 2530
+    },
+    {
+      "epoch": 13.039068100358422,
+      "grad_norm": 8.186372756958008,
+      "learning_rate": 4.978096375945838e-06,
+      "loss": 0.1547,
+      "step": 2540
+    },
+    {
+      "epoch": 13.042652329749105,
+      "grad_norm": 2.0492045879364014,
+      "learning_rate": 4.7789725209080055e-06,
+      "loss": 0.1371,
+      "step": 2550
+    },
+    {
+      "epoch": 13.046236559139786,
+      "grad_norm": 0.1944122165441513,
+      "learning_rate": 4.579848665870171e-06,
+      "loss": 0.0394,
+      "step": 2560
+    },
+    {
+      "epoch": 13.049820788530466,
+      "grad_norm": 0.04899689927697182,
+      "learning_rate": 4.380724810832338e-06,
+      "loss": 0.0782,
+      "step": 2570
+    },
+    {
+      "epoch": 13.053405017921147,
+      "grad_norm": 2.100203275680542,
+      "learning_rate": 4.181600955794504e-06,
+      "loss": 0.2894,
+      "step": 2580
+    },
+    {
+      "epoch": 13.056989247311828,
+      "grad_norm": 8.302490234375,
+      "learning_rate": 3.9824771007566706e-06,
+      "loss": 0.4249,
+      "step": 2590
+    },
+    {
+      "epoch": 13.060573476702508,
+      "grad_norm": 0.5646193027496338,
+      "learning_rate": 3.7833532457188377e-06,
+      "loss": 0.1387,
+      "step": 2600
+    },
+    {
+      "epoch": 13.06415770609319,
+      "grad_norm": 19.096433639526367,
+      "learning_rate": 3.5842293906810035e-06,
+      "loss": 0.1726,
+      "step": 2610
+    },
+    {
+      "epoch": 13.067025089605735,
+      "eval_accuracy": 0.8333333333333334,
+      "eval_loss": 0.5430171489715576,
+      "eval_runtime": 52.1786,
+      "eval_samples_per_second": 2.99,
+      "eval_steps_per_second": 0.383,
+      "step": 2618
+    },
+    {
+      "epoch": 14.000716845878136,
+      "grad_norm": 0.11462392657995224,
+      "learning_rate": 3.38510553564317e-06,
+      "loss": 0.1894,
+      "step": 2620
+    },
+    {
+      "epoch": 14.004301075268817,
+      "grad_norm": 0.08781033009290695,
+      "learning_rate": 3.1859816806053364e-06,
+      "loss": 0.2443,
+      "step": 2630
+    },
+    {
+      "epoch": 14.007885304659498,
+      "grad_norm": 54.70309066772461,
+      "learning_rate": 2.986857825567503e-06,
+      "loss": 0.1809,
+      "step": 2640
+    },
+    {
+      "epoch": 14.01146953405018,
+      "grad_norm": 0.0693989247083664,
+      "learning_rate": 2.7877339705296694e-06,
+      "loss": 0.0067,
+      "step": 2650
+    },
+    {
+      "epoch": 14.01505376344086,
+      "grad_norm": 0.1357724964618683,
+      "learning_rate": 2.588610115491836e-06,
+      "loss": 0.0544,
+      "step": 2660
+    },
+    {
+      "epoch": 14.018637992831541,
+      "grad_norm": 16.85883331298828,
+      "learning_rate": 2.3894862604540028e-06,
+      "loss": 0.1216,
+      "step": 2670
+    },
+    {
+      "epoch": 14.022222222222222,
+      "grad_norm": 7.547858715057373,
+      "learning_rate": 2.190362405416169e-06,
+      "loss": 0.2463,
+      "step": 2680
+    },
+    {
+      "epoch": 14.025806451612903,
+      "grad_norm": 2.388237714767456,
+      "learning_rate": 1.9912385503783353e-06,
+      "loss": 0.1674,
+      "step": 2690
+    },
+    {
+      "epoch": 14.029390681003584,
+      "grad_norm": 0.4825267493724823,
+      "learning_rate": 1.7921146953405017e-06,
+      "loss": 0.1415,
+      "step": 2700
+    },
+    {
+      "epoch": 14.032974910394266,
+      "grad_norm": 15.646313667297363,
+      "learning_rate": 1.5929908403026682e-06,
+      "loss": 0.2143,
+      "step": 2710
+    },
+    {
+      "epoch": 14.036559139784947,
+      "grad_norm": 0.05193038657307625,
+      "learning_rate": 1.3938669852648347e-06,
+      "loss": 0.1399,
+      "step": 2720
+    },
+    {
+      "epoch": 14.040143369175627,
+      "grad_norm": 10.611710548400879,
+      "learning_rate": 1.1947431302270014e-06,
+      "loss": 0.1608,
+      "step": 2730
+    },
+    {
+      "epoch": 14.043727598566308,
+      "grad_norm": 28.249521255493164,
+      "learning_rate": 9.956192751891676e-07,
+      "loss": 0.3168,
+      "step": 2740
+    },
+    {
+      "epoch": 14.047311827956989,
+      "grad_norm": 0.169923797249794,
+      "learning_rate": 7.964954201513341e-07,
+      "loss": 0.2273,
+      "step": 2750
+    },
+    {
+      "epoch": 14.05089605734767,
+      "grad_norm": 0.6709271669387817,
+      "learning_rate": 5.973715651135007e-07,
+      "loss": 0.0479,
+      "step": 2760
+    },
+    {
+      "epoch": 14.054480286738352,
+      "grad_norm": 23.617549896240234,
+      "learning_rate": 3.9824771007566706e-07,
+      "loss": 0.1528,
+      "step": 2770
+    },
+    {
+      "epoch": 14.058064516129033,
+      "grad_norm": 0.37845826148986816,
+      "learning_rate": 1.9912385503783353e-07,
+      "loss": 0.2054,
+      "step": 2780
+    },
+    {
+      "epoch": 14.061648745519713,
+      "grad_norm": 0.18122656643390656,
+      "learning_rate": 0.0,
+      "loss": 0.2543,
+      "step": 2790
+    },
+    {
+      "epoch": 14.061648745519713,
+      "eval_accuracy": 0.8717948717948718,
+      "eval_loss": 0.49779054522514343,
+      "eval_runtime": 59.5535,
+      "eval_samples_per_second": 2.619,
+      "eval_steps_per_second": 0.336,
+      "step": 2790
+    },
+    {
+      "epoch": 14.061648745519713,
+      "step": 2790,
+      "total_flos": 2.7788486409809363e+19,
+      "train_loss": 0.30916261622234914,
+      "train_runtime": 10278.2383,
+      "train_samples_per_second": 2.172,
+      "train_steps_per_second": 0.271
+    },
+    {
+      "epoch": 14.061648745519713,
+      "eval_accuracy": 0.81875,
+      "eval_loss": 0.6550884246826172,
+      "eval_runtime": 302.2934,
+      "eval_samples_per_second": 2.646,
+      "eval_steps_per_second": 0.331,
+      "step": 2790
+    },
+    {
+      "epoch": 14.061648745519713,
+      "eval_accuracy": 0.81875,
+      "eval_loss": 0.6550884246826172,
+      "eval_runtime": 306.835,
+      "eval_samples_per_second": 2.607,
+      "eval_steps_per_second": 0.326,
+      "step": 2790
     }
   ],
   "logging_steps": 10,
-  "max_steps": 930,
+  "max_steps": 2790,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 9223372036854775807,
   "save_steps": 500,
@@ -794,7 +2141,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 9.225862220448793e+18,
+  "total_flos": 2.7788486409809363e+19,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null