Training in progress, epoch 3

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0626740999d70436a63bb393abae1f5a7de83f31e6824b6bc182b233c02cf30d
+oid sha256:612b46cd150d8cd39c0d39eb5420e5e0fcc68049ad183fe3ad15ce3041c1bc9a
 size 343233204

tmp-checkpoint-5340/config.json

@@ -0,0 +1,38 @@
+{
+  "_name_or_path": "google/vit-base-patch16-224-in21k",
+  "architectures": [
+    "ViTForImageClassification"
+  ],
+  "attention_probs_dropout_prob": 0.0,
+  "encoder_stride": 16,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.0,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "crack",
+    "1": "for review",
+    "2": "material deterioration",
+    "3": "repair",
+    "4": "water leakage"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "crack": 0,
+    "for review": 1,
+    "material deterioration": 2,
+    "repair": 3,
+    "water leakage": 4
+  },
+  "layer_norm_eps": 1e-12,
+  "model_type": "vit",
+  "num_attention_heads": 12,
+  "num_channels": 3,
+  "num_hidden_layers": 12,
+  "patch_size": 16,
+  "problem_type": "single_label_classification",
+  "qkv_bias": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.37.2"
+}

tmp-checkpoint-5340/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:612b46cd150d8cd39c0d39eb5420e5e0fcc68049ad183fe3ad15ce3041c1bc9a
+size 343233204

tmp-checkpoint-5340/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0afea4d52aea8faaf689386090206414d76638cdc280f8069154386fb7edda10
+size 686587322

tmp-checkpoint-5340/preprocessor_config.json

@@ -0,0 +1,22 @@
+{
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "ViTImageProcessor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 224,
+    "width": 224
+  }
+}

tmp-checkpoint-5340/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:498dfd6df7d34a110349149457dc7beb4402714e058155340398617601d3434e
+size 14244

tmp-checkpoint-5340/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8152fea158481e7c271483446f23cffd4276bd1959b34aa8a5d466a1478dc456
+size 1064

tmp-checkpoint-5340/trainer_state.json

@@ -0,0 +1,3285 @@
+{
+  "best_metric": 0.22899627685546875,
+  "best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT15\\checkpoint-5340",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 5340,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8726591760299626e-07,
+      "loss": 1.6071,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7453183520599253e-07,
+      "loss": 1.6248,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.617977528089887e-07,
+      "loss": 1.6194,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.490636704119851e-07,
+      "loss": 1.6239,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.363295880149814e-07,
+      "loss": 1.6197,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1235955056179775e-06,
+      "loss": 1.6005,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.310861423220974e-06,
+      "loss": 1.6125,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4981273408239701e-06,
+      "loss": 1.6071,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6853932584269663e-06,
+      "loss": 1.5943,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8726591760299627e-06,
+      "loss": 1.6106,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0599250936329587e-06,
+      "loss": 1.597,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.247191011235955e-06,
+      "loss": 1.5831,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4344569288389516e-06,
+      "loss": 1.5911,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.621722846441948e-06,
+      "loss": 1.5827,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.808988764044944e-06,
+      "loss": 1.5803,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9962546816479402e-06,
+      "loss": 1.5632,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1835205992509364e-06,
+      "loss": 1.568,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3707865168539327e-06,
+      "loss": 1.5746,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.558052434456929e-06,
+      "loss": 1.5403,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7453183520599255e-06,
+      "loss": 1.5729,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.932584269662922e-06,
+      "loss": 1.535,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1198501872659175e-06,
+      "loss": 1.5219,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.307116104868914e-06,
+      "loss": 1.5319,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.49438202247191e-06,
+      "loss": 1.5031,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6816479400749066e-06,
+      "loss": 1.4887,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868913857677903e-06,
+      "loss": 1.4869,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.056179775280899e-06,
+      "loss": 1.4823,
+      "step": 270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.243445692883896e-06,
+      "loss": 1.4644,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.430711610486891e-06,
+      "loss": 1.4454,
+      "step": 290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.617977528089888e-06,
+      "loss": 1.4372,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.805243445692885e-06,
+      "loss": 1.3914,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9925093632958805e-06,
+      "loss": 1.4091,
+      "step": 320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.179775280898876e-06,
+      "loss": 1.3276,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.367041198501873e-06,
+      "loss": 1.3891,
+      "step": 340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.554307116104869e-06,
+      "loss": 1.3041,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.741573033707865e-06,
+      "loss": 1.3156,
+      "step": 360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.928838951310862e-06,
+      "loss": 1.2533,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.116104868913858e-06,
+      "loss": 1.2701,
+      "step": 380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.303370786516854e-06,
+      "loss": 1.2659,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.490636704119851e-06,
+      "loss": 1.2466,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.677902621722846e-06,
+      "loss": 1.2126,
+      "step": 410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.865168539325843e-06,
+      "loss": 1.1697,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.05243445692884e-06,
+      "loss": 1.2085,
+      "step": 430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.239700374531835e-06,
+      "loss": 1.1072,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.426966292134832e-06,
+      "loss": 1.0911,
+      "step": 450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.614232209737828e-06,
+      "loss": 1.0768,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.801498127340826e-06,
+      "loss": 1.0506,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.98876404494382e-06,
+      "loss": 0.9857,
+      "step": 480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.176029962546817e-06,
+      "loss": 0.94,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.363295880149813e-06,
+      "loss": 0.9744,
+      "step": 500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.550561797752809e-06,
+      "loss": 0.9008,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.737827715355806e-06,
+      "loss": 0.9664,
+      "step": 520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.925093632958802e-06,
+      "loss": 0.8886,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0112359550561798e-05,
+      "loss": 0.8804,
+      "step": 540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0299625468164795e-05,
+      "loss": 0.8209,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.0486891385767791e-05,
+      "loss": 0.7627,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0674157303370787e-05,
+      "loss": 0.8034,
+      "step": 570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0861423220973783e-05,
+      "loss": 0.8372,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.104868913857678e-05,
+      "loss": 0.716,
+      "step": 590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1235955056179776e-05,
+      "loss": 0.8199,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1423220973782772e-05,
+      "loss": 0.6924,
+      "step": 610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.161048689138577e-05,
+      "loss": 0.7584,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1797752808988765e-05,
+      "loss": 0.7057,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1985018726591761e-05,
+      "loss": 0.6971,
+      "step": 640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2172284644194758e-05,
+      "loss": 0.5725,
+      "step": 650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2359550561797752e-05,
+      "loss": 0.646,
+      "step": 660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.254681647940075e-05,
+      "loss": 0.6461,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2734082397003746e-05,
+      "loss": 0.6068,
+      "step": 680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2921348314606743e-05,
+      "loss": 0.6562,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3108614232209737e-05,
+      "loss": 0.4778,
+      "step": 700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3295880149812733e-05,
+      "loss": 0.5758,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.348314606741573e-05,
+      "loss": 0.6472,
+      "step": 720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3670411985018728e-05,
+      "loss": 0.6138,
+      "step": 730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3857677902621724e-05,
+      "loss": 0.6494,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4044943820224721e-05,
+      "loss": 0.5153,
+      "step": 750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4232209737827715e-05,
+      "loss": 0.6178,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4419475655430711e-05,
+      "loss": 0.5731,
+      "step": 770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4606741573033709e-05,
+      "loss": 0.4859,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4794007490636705e-05,
+      "loss": 0.4832,
+      "step": 790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4981273408239702e-05,
+      "loss": 0.5471,
+      "step": 800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5168539325842698e-05,
+      "loss": 0.4091,
+      "step": 810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5355805243445692e-05,
+      "loss": 0.5861,
+      "step": 820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.554307116104869e-05,
+      "loss": 0.4113,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5730337078651687e-05,
+      "loss": 0.5322,
+      "step": 840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.591760299625468e-05,
+      "loss": 0.5145,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.610486891385768e-05,
+      "loss": 0.5974,
+      "step": 860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6292134831460676e-05,
+      "loss": 0.642,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.647940074906367e-05,
+      "loss": 0.6079,
+      "step": 880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.6456,
+      "step": 890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6853932584269665e-05,
+      "loss": 0.5845,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.704119850187266e-05,
+      "loss": 0.39,
+      "step": 910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7228464419475657e-05,
+      "loss": 0.4569,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7415730337078654e-05,
+      "loss": 0.3825,
+      "step": 930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.760299625468165e-05,
+      "loss": 0.539,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7790262172284646e-05,
+      "loss": 0.3787,
+      "step": 950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.797752808988764e-05,
+      "loss": 0.4516,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8164794007490637e-05,
+      "loss": 0.4029,
+      "step": 970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8352059925093635e-05,
+      "loss": 0.3875,
+      "step": 980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8539325842696632e-05,
+      "loss": 0.2843,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726591760299626e-05,
+      "loss": 0.3438,
+      "step": 1000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.891385767790262e-05,
+      "loss": 0.4593,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9101123595505618e-05,
+      "loss": 0.4925,
+      "step": 1020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9288389513108615e-05,
+      "loss": 0.4471,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9475655430711613e-05,
+      "loss": 0.4646,
+      "step": 1040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9662921348314607e-05,
+      "loss": 0.4893,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9850187265917604e-05,
+      "loss": 0.4647,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.00374531835206e-05,
+      "loss": 0.5963,
+      "step": 1070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0224719101123596e-05,
+      "loss": 0.5337,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0411985018726593e-05,
+      "loss": 0.5736,
+      "step": 1090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.059925093632959e-05,
+      "loss": 0.4304,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0786516853932585e-05,
+      "loss": 0.4311,
+      "step": 1110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0973782771535582e-05,
+      "loss": 0.3767,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.1161048689138577e-05,
+      "loss": 0.3995,
+      "step": 1130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.1348314606741574e-05,
+      "loss": 0.2452,
+      "step": 1140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.153558052434457e-05,
+      "loss": 0.4263,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.1722846441947566e-05,
+      "loss": 0.3038,
+      "step": 1160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.1910112359550563e-05,
+      "loss": 0.4827,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.209737827715356e-05,
+      "loss": 0.3673,
+      "step": 1180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.2284644194756555e-05,
+      "loss": 0.3788,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.2471910112359552e-05,
+      "loss": 0.4111,
+      "step": 1200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2659176029962546e-05,
+      "loss": 0.4552,
+      "step": 1210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2846441947565544e-05,
+      "loss": 0.4149,
+      "step": 1220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.303370786516854e-05,
+      "loss": 0.265,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.322097378277154e-05,
+      "loss": 0.2473,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3408239700374533e-05,
+      "loss": 0.4116,
+      "step": 1250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.359550561797753e-05,
+      "loss": 0.4774,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3782771535580524e-05,
+      "loss": 0.3525,
+      "step": 1270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3970037453183522e-05,
+      "loss": 0.4028,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.415730337078652e-05,
+      "loss": 0.5239,
+      "step": 1290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4344569288389517e-05,
+      "loss": 0.4398,
+      "step": 1300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.453183520599251e-05,
+      "loss": 0.413,
+      "step": 1310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4719101123595505e-05,
+      "loss": 0.2971,
+      "step": 1320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4906367041198502e-05,
+      "loss": 0.3945,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.50936329588015e-05,
+      "loss": 0.3862,
+      "step": 1340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5280898876404497e-05,
+      "loss": 0.2727,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.546816479400749e-05,
+      "loss": 0.4686,
+      "step": 1360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565543071161049e-05,
+      "loss": 0.4213,
+      "step": 1370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5842696629213486e-05,
+      "loss": 0.3585,
+      "step": 1380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6029962546816484e-05,
+      "loss": 0.3737,
+      "step": 1390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6217228464419475e-05,
+      "loss": 0.4583,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6404494382022472e-05,
+      "loss": 0.241,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.6591760299625466e-05,
+      "loss": 0.491,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.6779026217228464e-05,
+      "loss": 0.4018,
+      "step": 1430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.696629213483146e-05,
+      "loss": 0.4581,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.715355805243446e-05,
+      "loss": 0.4194,
+      "step": 1450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7340823970037456e-05,
+      "loss": 0.4169,
+      "step": 1460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.752808988764045e-05,
+      "loss": 0.4541,
+      "step": 1470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7715355805243448e-05,
+      "loss": 0.255,
+      "step": 1480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7902621722846445e-05,
+      "loss": 0.2751,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8089887640449443e-05,
+      "loss": 0.5636,
+      "step": 1500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8277153558052437e-05,
+      "loss": 0.2853,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.846441947565543e-05,
+      "loss": 0.4982,
+      "step": 1520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8651685393258425e-05,
+      "loss": 0.1949,
+      "step": 1530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8838951310861422e-05,
+      "loss": 0.2742,
+      "step": 1540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.902621722846442e-05,
+      "loss": 0.4109,
+      "step": 1550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9213483146067417e-05,
+      "loss": 0.3774,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.940074906367041e-05,
+      "loss": 0.281,
+      "step": 1570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.958801498127341e-05,
+      "loss": 0.4742,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9775280898876406e-05,
+      "loss": 0.4149,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9962546816479404e-05,
+      "loss": 0.3106,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.01498127340824e-05,
+      "loss": 0.4442,
+      "step": 1610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0337078651685396e-05,
+      "loss": 0.1809,
+      "step": 1620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052434456928839e-05,
+      "loss": 0.4556,
+      "step": 1630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0711610486891384e-05,
+      "loss": 0.3343,
+      "step": 1640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.089887640449438e-05,
+      "loss": 0.5095,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.108614232209738e-05,
+      "loss": 0.2521,
+      "step": 1660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1273408239700376e-05,
+      "loss": 0.2387,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1460674157303374e-05,
+      "loss": 0.2964,
+      "step": 1680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.164794007490637e-05,
+      "loss": 0.2401,
+      "step": 1690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.183520599250936e-05,
+      "loss": 0.6202,
+      "step": 1700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.202247191011236e-05,
+      "loss": 0.3587,
+      "step": 1710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.220973782771536e-05,
+      "loss": 0.5592,
+      "step": 1720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2397003745318354e-05,
+      "loss": 0.3333,
+      "step": 1730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.258426966292135e-05,
+      "loss": 0.3082,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.277153558052435e-05,
+      "loss": 0.5302,
+      "step": 1750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.295880149812734e-05,
+      "loss": 0.3574,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.314606741573034e-05,
+      "loss": 0.4965,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3548,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": {
+        "accuracy": 0.9118225217635496
+      },
+      "eval_f1": {
+        "f1": 0.912598515170384
+      },
+      "eval_loss": 0.2847656011581421,
+      "eval_precision": {
+        "precision": 0.913326374297146
+      },
+      "eval_recall": {
+        "recall": 0.9157022464716918
+      },
+      "eval_runtime": 123.7534,
+      "eval_samples_per_second": 28.775,
+      "eval_steps_per_second": 14.392,
+      "step": 1780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.352059925093633e-05,
+      "loss": 0.21,
+      "step": 1790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.370786516853933e-05,
+      "loss": 0.1671,
+      "step": 1800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.389513108614232e-05,
+      "loss": 0.2274,
+      "step": 1810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.408239700374532e-05,
+      "loss": 0.3181,
+      "step": 1820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.4269662921348316e-05,
+      "loss": 0.3126,
+      "step": 1830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.445692883895131e-05,
+      "loss": 0.1818,
+      "step": 1840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.464419475655431e-05,
+      "loss": 0.4245,
+      "step": 1850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.483146067415731e-05,
+      "loss": 0.1466,
+      "step": 1860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.5018726591760305e-05,
+      "loss": 0.4405,
+      "step": 1870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.52059925093633e-05,
+      "loss": 0.1791,
+      "step": 1880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5393258426966294e-05,
+      "loss": 0.2027,
+      "step": 1890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.558052434456929e-05,
+      "loss": 0.5071,
+      "step": 1900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.576779026217228e-05,
+      "loss": 0.3631,
+      "step": 1910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.595505617977528e-05,
+      "loss": 0.3622,
+      "step": 1920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.614232209737828e-05,
+      "loss": 0.2463,
+      "step": 1930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6329588014981274e-05,
+      "loss": 0.3801,
+      "step": 1940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.651685393258427e-05,
+      "loss": 0.3729,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.670411985018727e-05,
+      "loss": 0.2867,
+      "step": 1960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.689138576779027e-05,
+      "loss": 0.6275,
+      "step": 1970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.7078651685393264e-05,
+      "loss": 0.3692,
+      "step": 1980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.726591760299626e-05,
+      "loss": 0.2219,
+      "step": 1990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.745318352059925e-05,
+      "loss": 0.5232,
+      "step": 2000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.764044943820225e-05,
+      "loss": 0.243,
+      "step": 2010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.782771535580524e-05,
+      "loss": 0.3025,
+      "step": 2020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.801498127340824e-05,
+      "loss": 0.3635,
+      "step": 2030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8202247191011236e-05,
+      "loss": 0.2452,
+      "step": 2040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.838951310861423e-05,
+      "loss": 0.3114,
+      "step": 2050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.857677902621723e-05,
+      "loss": 0.3641,
+      "step": 2060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.876404494382023e-05,
+      "loss": 0.1288,
+      "step": 2070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8951310861423226e-05,
+      "loss": 0.3868,
+      "step": 2080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.913857677902622e-05,
+      "loss": 0.1675,
+      "step": 2090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9325842696629214e-05,
+      "loss": 0.5064,
+      "step": 2100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.951310861423221e-05,
+      "loss": 0.3433,
+      "step": 2110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.970037453183521e-05,
+      "loss": 0.4016,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.98876404494382e-05,
+      "loss": 0.5303,
+      "step": 2130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.00749063670412e-05,
+      "loss": 0.1433,
+      "step": 2140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.0262172284644194e-05,
+      "loss": 0.3427,
+      "step": 2150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.044943820224719e-05,
+      "loss": 0.3481,
+      "step": 2160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.063670411985019e-05,
+      "loss": 0.3738,
+      "step": 2170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.082397003745319e-05,
+      "loss": 0.3195,
+      "step": 2180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1011235955056184e-05,
+      "loss": 0.1761,
+      "step": 2190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.119850187265918e-05,
+      "loss": 0.3804,
+      "step": 2200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.138576779026217e-05,
+      "loss": 0.2417,
+      "step": 2210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.157303370786517e-05,
+      "loss": 0.5015,
+      "step": 2220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.176029962546817e-05,
+      "loss": 0.5068,
+      "step": 2230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.1947565543071165e-05,
+      "loss": 0.5047,
+      "step": 2240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2134831460674156e-05,
+      "loss": 0.399,
+      "step": 2250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.232209737827715e-05,
+      "loss": 0.2417,
+      "step": 2260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.250936329588015e-05,
+      "loss": 0.2412,
+      "step": 2270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.269662921348315e-05,
+      "loss": 0.2515,
+      "step": 2280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.2883895131086146e-05,
+      "loss": 0.3938,
+      "step": 2290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.307116104868914e-05,
+      "loss": 0.6135,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3258426966292134e-05,
+      "loss": 0.4047,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.344569288389513e-05,
+      "loss": 0.3186,
+      "step": 2320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.363295880149813e-05,
+      "loss": 0.2476,
+      "step": 2330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3820224719101126e-05,
+      "loss": 0.4001,
+      "step": 2340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.4007490636704124e-05,
+      "loss": 0.3103,
+      "step": 2350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.419475655430712e-05,
+      "loss": 0.2314,
+      "step": 2360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.438202247191011e-05,
+      "loss": 0.1733,
+      "step": 2370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.456928838951311e-05,
+      "loss": 0.2724,
+      "step": 2380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.475655430711611e-05,
+      "loss": 0.216,
+      "step": 2390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.4943820224719104e-05,
+      "loss": 0.1945,
+      "step": 2400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.51310861423221e-05,
+      "loss": 0.2007,
+      "step": 2410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.531835205992509e-05,
+      "loss": 0.2947,
+      "step": 2420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.550561797752809e-05,
+      "loss": 0.4243,
+      "step": 2430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.569288389513109e-05,
+      "loss": 0.2945,
+      "step": 2440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5880149812734085e-05,
+      "loss": 0.2609,
+      "step": 2450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606741573033708e-05,
+      "loss": 0.4327,
+      "step": 2460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.625468164794008e-05,
+      "loss": 0.5988,
+      "step": 2470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.644194756554308e-05,
+      "loss": 0.1898,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.662921348314607e-05,
+      "loss": 0.279,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6816479400749066e-05,
+      "loss": 0.2209,
+      "step": 2500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.700374531835206e-05,
+      "loss": 0.3227,
+      "step": 2510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.719101123595506e-05,
+      "loss": 0.2589,
+      "step": 2520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.737827715355805e-05,
+      "loss": 0.406,
+      "step": 2530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.756554307116105e-05,
+      "loss": 0.2744,
+      "step": 2540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7752808988764046e-05,
+      "loss": 0.3877,
+      "step": 2550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7940074906367044e-05,
+      "loss": 0.3208,
+      "step": 2560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.812734082397004e-05,
+      "loss": 0.1757,
+      "step": 2570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.831460674157304e-05,
+      "loss": 0.3472,
+      "step": 2580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.8501872659176036e-05,
+      "loss": 0.2122,
+      "step": 2590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.8689138576779034e-05,
+      "loss": 0.3728,
+      "step": 2600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.8876404494382024e-05,
+      "loss": 0.2358,
+      "step": 2610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.906367041198502e-05,
+      "loss": 0.2261,
+      "step": 2620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.925093632958801e-05,
+      "loss": 0.2761,
+      "step": 2630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.943820224719101e-05,
+      "loss": 0.2807,
+      "step": 2640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.962546816479401e-05,
+      "loss": 0.1926,
+      "step": 2650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.9812734082397005e-05,
+      "loss": 0.3198,
+      "step": 2660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-05,
+      "loss": 0.2685,
+      "step": 2670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.997919267582189e-05,
+      "loss": 0.2652,
+      "step": 2680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.995838535164378e-05,
+      "loss": 0.5231,
+      "step": 2690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.993757802746567e-05,
+      "loss": 0.3469,
+      "step": 2700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.991677070328756e-05,
+      "loss": 0.2902,
+      "step": 2710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9895963379109454e-05,
+      "loss": 0.2788,
+      "step": 2720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.987515605493134e-05,
+      "loss": 0.2888,
+      "step": 2730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.9854348730753226e-05,
+      "loss": 0.2221,
+      "step": 2740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.983354140657512e-05,
+      "loss": 0.283,
+      "step": 2750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9812734082397005e-05,
+      "loss": 0.253,
+      "step": 2760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.979192675821889e-05,
+      "loss": 0.1786,
+      "step": 2770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9771119434040784e-05,
+      "loss": 0.3892,
+      "step": 2780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.975031210986267e-05,
+      "loss": 0.1414,
+      "step": 2790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.972950478568456e-05,
+      "loss": 0.0931,
+      "step": 2800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9708697461506456e-05,
+      "loss": 0.0699,
+      "step": 2810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.968789013732834e-05,
+      "loss": 0.3096,
+      "step": 2820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.966708281315023e-05,
+      "loss": 0.1819,
+      "step": 2830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.964627548897212e-05,
+      "loss": 0.4443,
+      "step": 2840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.962546816479401e-05,
+      "loss": 0.4339,
+      "step": 2850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.96046608406159e-05,
+      "loss": 0.2916,
+      "step": 2860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9583853516437787e-05,
+      "loss": 0.1922,
+      "step": 2870
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.956304619225968e-05,
+      "loss": 0.3508,
+      "step": 2880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.954223886808157e-05,
+      "loss": 0.2477,
+      "step": 2890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.952143154390346e-05,
+      "loss": 0.2358,
+      "step": 2900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.9500624219725345e-05,
+      "loss": 0.407,
+      "step": 2910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.947981689554724e-05,
+      "loss": 0.3142,
+      "step": 2920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9459009571369124e-05,
+      "loss": 0.1523,
+      "step": 2930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.943820224719101e-05,
+      "loss": 0.3927,
+      "step": 2940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.94173949230129e-05,
+      "loss": 0.4299,
+      "step": 2950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9396587598834796e-05,
+      "loss": 0.2918,
+      "step": 2960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.937578027465668e-05,
+      "loss": 0.4283,
+      "step": 2970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9354972950478575e-05,
+      "loss": 0.156,
+      "step": 2980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.933416562630046e-05,
+      "loss": 0.1685,
+      "step": 2990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.931335830212235e-05,
+      "loss": 0.502,
+      "step": 3000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.929255097794424e-05,
+      "loss": 0.2719,
+      "step": 3010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9271743653766126e-05,
+      "loss": 0.2534,
+      "step": 3020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.925093632958801e-05,
+      "loss": 0.2632,
+      "step": 3030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9230129005409905e-05,
+      "loss": 0.5752,
+      "step": 3040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.92093216812318e-05,
+      "loss": 0.2694,
+      "step": 3050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9188514357053684e-05,
+      "loss": 0.2839,
+      "step": 3060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.916770703287558e-05,
+      "loss": 0.1957,
+      "step": 3070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9146899708697464e-05,
+      "loss": 0.2485,
+      "step": 3080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.912609238451935e-05,
+      "loss": 0.4064,
+      "step": 3090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.910528506034124e-05,
+      "loss": 0.3101,
+      "step": 3100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.908447773616313e-05,
+      "loss": 0.1904,
+      "step": 3110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.906367041198502e-05,
+      "loss": 0.1904,
+      "step": 3120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9042863087806915e-05,
+      "loss": 0.1766,
+      "step": 3130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.90220557636288e-05,
+      "loss": 0.2133,
+      "step": 3140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.900124843945069e-05,
+      "loss": 0.2306,
+      "step": 3150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.898044111527258e-05,
+      "loss": 0.185,
+      "step": 3160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.8959633791094466e-05,
+      "loss": 0.2374,
+      "step": 3170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.893882646691635e-05,
+      "loss": 0.2548,
+      "step": 3180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.8918019142738245e-05,
+      "loss": 0.3173,
+      "step": 3190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.889721181856013e-05,
+      "loss": 0.4484,
+      "step": 3200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.8876404494382024e-05,
+      "loss": 0.4081,
+      "step": 3210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.885559717020392e-05,
+      "loss": 0.359,
+      "step": 3220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.88347898460258e-05,
+      "loss": 0.1906,
+      "step": 3230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.8813982521847696e-05,
+      "loss": 0.3136,
+      "step": 3240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.879317519766958e-05,
+      "loss": 0.2744,
+      "step": 3250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.877236787349147e-05,
+      "loss": 0.2581,
+      "step": 3260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.875156054931336e-05,
+      "loss": 0.3342,
+      "step": 3270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.873075322513525e-05,
+      "loss": 0.4871,
+      "step": 3280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.870994590095714e-05,
+      "loss": 0.2652,
+      "step": 3290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.8689138576779034e-05,
+      "loss": 0.203,
+      "step": 3300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.866833125260092e-05,
+      "loss": 0.1886,
+      "step": 3310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8647523928422806e-05,
+      "loss": 0.2189,
+      "step": 3320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.86267166042447e-05,
+      "loss": 0.1404,
+      "step": 3330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8605909280066585e-05,
+      "loss": 0.4393,
+      "step": 3340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.858510195588847e-05,
+      "loss": 0.2466,
+      "step": 3350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8564294631710364e-05,
+      "loss": 0.2869,
+      "step": 3360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.854348730753226e-05,
+      "loss": 0.2776,
+      "step": 3370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.852267998335414e-05,
+      "loss": 0.5125,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.8501872659176036e-05,
+      "loss": 0.2889,
+      "step": 3390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.848106533499792e-05,
+      "loss": 0.3192,
+      "step": 3400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.846025801081981e-05,
+      "loss": 0.2064,
+      "step": 3410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.84394506866417e-05,
+      "loss": 0.3253,
+      "step": 3420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.841864336246359e-05,
+      "loss": 0.0968,
+      "step": 3430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.8397836038285474e-05,
+      "loss": 0.1386,
+      "step": 3440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8377028714107367e-05,
+      "loss": 0.3869,
+      "step": 3450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.835622138992926e-05,
+      "loss": 0.2375,
+      "step": 3460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.8335414065751146e-05,
+      "loss": 0.3227,
+      "step": 3470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.831460674157304e-05,
+      "loss": 0.3983,
+      "step": 3480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8293799417394925e-05,
+      "loss": 0.3662,
+      "step": 3490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.827299209321681e-05,
+      "loss": 0.3329,
+      "step": 3500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.8252184769038704e-05,
+      "loss": 0.3773,
+      "step": 3510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.823137744486059e-05,
+      "loss": 0.1765,
+      "step": 3520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.821057012068248e-05,
+      "loss": 0.4403,
+      "step": 3530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8189762796504376e-05,
+      "loss": 0.2278,
+      "step": 3540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.816895547232626e-05,
+      "loss": 0.1792,
+      "step": 3550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.1718,
+      "step": 3560
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": {
+        "accuracy": 0.9005897219882055
+      },
+      "eval_f1": {
+        "f1": 0.9021520907258462
+      },
+      "eval_loss": 0.34348464012145996,
+      "eval_precision": {
+        "precision": 0.9071588887385811
+      },
+      "eval_recall": {
+        "recall": 0.9088734326741875
+      },
+      "eval_runtime": 93.9787,
+      "eval_samples_per_second": 37.892,
+      "eval_steps_per_second": 18.951,
+      "step": 3560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.812734082397004e-05,
+      "loss": 0.4564,
+      "step": 3570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.810653349979193e-05,
+      "loss": 0.2619,
+      "step": 3580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.808572617561382e-05,
+      "loss": 0.1274,
+      "step": 3590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8064918851435706e-05,
+      "loss": 0.2,
+      "step": 3600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.804411152725759e-05,
+      "loss": 0.1242,
+      "step": 3610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8023304203079485e-05,
+      "loss": 0.1887,
+      "step": 3620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.800249687890138e-05,
+      "loss": 0.2478,
+      "step": 3630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.7981689554723265e-05,
+      "loss": 0.0844,
+      "step": 3640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796088223054516e-05,
+      "loss": 0.3901,
+      "step": 3650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7940074906367044e-05,
+      "loss": 0.2306,
+      "step": 3660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.791926758218893e-05,
+      "loss": 0.281,
+      "step": 3670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.789846025801082e-05,
+      "loss": 0.2839,
+      "step": 3680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.787765293383271e-05,
+      "loss": 0.1552,
+      "step": 3690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.78568456096546e-05,
+      "loss": 0.2969,
+      "step": 3700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7836038285476495e-05,
+      "loss": 0.3101,
+      "step": 3710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.781523096129838e-05,
+      "loss": 0.117,
+      "step": 3720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.779442363712027e-05,
+      "loss": 0.191,
+      "step": 3730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.777361631294216e-05,
+      "loss": 0.1365,
+      "step": 3740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.7752808988764046e-05,
+      "loss": 0.1637,
+      "step": 3750
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.773200166458593e-05,
+      "loss": 0.3449,
+      "step": 3760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.7711194340407825e-05,
+      "loss": 0.5397,
+      "step": 3770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.769038701622972e-05,
+      "loss": 0.4197,
+      "step": 3780
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.7669579692051604e-05,
+      "loss": 0.1912,
+      "step": 3790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.76487723678735e-05,
+      "loss": 0.1763,
+      "step": 3800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.7627965043695383e-05,
+      "loss": 0.1706,
+      "step": 3810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.760715771951727e-05,
+      "loss": 0.4229,
+      "step": 3820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.758635039533916e-05,
+      "loss": 0.2296,
+      "step": 3830
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.756554307116105e-05,
+      "loss": 0.2251,
+      "step": 3840
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.7544735746982935e-05,
+      "loss": 0.0553,
+      "step": 3850
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.752392842280483e-05,
+      "loss": 0.2203,
+      "step": 3860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.750312109862672e-05,
+      "loss": 0.0682,
+      "step": 3870
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.748231377444861e-05,
+      "loss": 0.2284,
+      "step": 3880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.74615064502705e-05,
+      "loss": 0.4361,
+      "step": 3890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.7440699126092386e-05,
+      "loss": 0.1676,
+      "step": 3900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.741989180191427e-05,
+      "loss": 0.0995,
+      "step": 3910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.7399084477736165e-05,
+      "loss": 0.2302,
+      "step": 3920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.737827715355805e-05,
+      "loss": 0.1351,
+      "step": 3930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.7357469829379944e-05,
+      "loss": 0.2093,
+      "step": 3940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.733666250520184e-05,
+      "loss": 0.0935,
+      "step": 3950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.731585518102372e-05,
+      "loss": 0.2482,
+      "step": 3960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.7295047856845616e-05,
+      "loss": 0.142,
+      "step": 3970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.72742405326675e-05,
+      "loss": 0.3728,
+      "step": 3980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.725343320848939e-05,
+      "loss": 0.3502,
+      "step": 3990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.723262588431128e-05,
+      "loss": 0.2604,
+      "step": 4000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.721181856013317e-05,
+      "loss": 0.0286,
+      "step": 4010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.719101123595506e-05,
+      "loss": 0.2325,
+      "step": 4020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.7170203911776953e-05,
+      "loss": 0.3871,
+      "step": 4030
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.714939658759884e-05,
+      "loss": 0.3771,
+      "step": 4040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7128589263420726e-05,
+      "loss": 0.179,
+      "step": 4050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.710778193924262e-05,
+      "loss": 0.1487,
+      "step": 4060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7086974615064505e-05,
+      "loss": 0.0621,
+      "step": 4070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.706616729088639e-05,
+      "loss": 0.3428,
+      "step": 4080
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.7045359966708284e-05,
+      "loss": 0.4521,
+      "step": 4090
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.702455264253017e-05,
+      "loss": 0.3151,
+      "step": 4100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.700374531835206e-05,
+      "loss": 0.4056,
+      "step": 4110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6982937994173956e-05,
+      "loss": 0.289,
+      "step": 4120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.696213066999584e-05,
+      "loss": 0.2563,
+      "step": 4130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.694132334581773e-05,
+      "loss": 0.1494,
+      "step": 4140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.692051602163962e-05,
+      "loss": 0.2708,
+      "step": 4150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.689970869746151e-05,
+      "loss": 0.1988,
+      "step": 4160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6878901373283394e-05,
+      "loss": 0.0919,
+      "step": 4170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6858094049105286e-05,
+      "loss": 0.1862,
+      "step": 4180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.683728672492718e-05,
+      "loss": 0.0622,
+      "step": 4190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6816479400749066e-05,
+      "loss": 0.1565,
+      "step": 4200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.679567207657096e-05,
+      "loss": 0.1914,
+      "step": 4210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6774864752392845e-05,
+      "loss": 0.1417,
+      "step": 4220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.675405742821473e-05,
+      "loss": 0.2142,
+      "step": 4230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6733250104036624e-05,
+      "loss": 0.2943,
+      "step": 4240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.671244277985851e-05,
+      "loss": 0.0698,
+      "step": 4250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6691635455680396e-05,
+      "loss": 0.172,
+      "step": 4260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.667082813150229e-05,
+      "loss": 0.3883,
+      "step": 4270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.665002080732418e-05,
+      "loss": 0.2394,
+      "step": 4280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.662921348314607e-05,
+      "loss": 0.0763,
+      "step": 4290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.660840615896796e-05,
+      "loss": 0.2724,
+      "step": 4300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.658759883478985e-05,
+      "loss": 0.1771,
+      "step": 4310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.656679151061174e-05,
+      "loss": 0.1131,
+      "step": 4320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.6545984186433626e-05,
+      "loss": 0.2075,
+      "step": 4330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.652517686225551e-05,
+      "loss": 0.1966,
+      "step": 4340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.6504369538077405e-05,
+      "loss": 0.1479,
+      "step": 4350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.64835622138993e-05,
+      "loss": 0.1561,
+      "step": 4360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.6462754889721184e-05,
+      "loss": 0.2182,
+      "step": 4370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.644194756554308e-05,
+      "loss": 0.2924,
+      "step": 4380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.6421140241364964e-05,
+      "loss": 0.1758,
+      "step": 4390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.640033291718685e-05,
+      "loss": 0.2548,
+      "step": 4400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.637952559300874e-05,
+      "loss": 0.322,
+      "step": 4410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.635871826883063e-05,
+      "loss": 0.1558,
+      "step": 4420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.633791094465252e-05,
+      "loss": 0.3469,
+      "step": 4430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.6317103620474415e-05,
+      "loss": 0.0522,
+      "step": 4440
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.2128,
+      "step": 4450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.627548897211819e-05,
+      "loss": 0.1909,
+      "step": 4460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.625468164794008e-05,
+      "loss": 0.3989,
+      "step": 4470
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.6233874323761966e-05,
+      "loss": 0.0935,
+      "step": 4480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.621306699958385e-05,
+      "loss": 0.2124,
+      "step": 4490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.6192259675405745e-05,
+      "loss": 0.3213,
+      "step": 4500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.617145235122763e-05,
+      "loss": 0.0931,
+      "step": 4510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.6150645027049524e-05,
+      "loss": 0.1241,
+      "step": 4520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.612983770287142e-05,
+      "loss": 0.1819,
+      "step": 4530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.61090303786933e-05,
+      "loss": 0.1545,
+      "step": 4540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.608822305451519e-05,
+      "loss": 0.3003,
+      "step": 4550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.606741573033708e-05,
+      "loss": 0.2651,
+      "step": 4560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.604660840615897e-05,
+      "loss": 0.269,
+      "step": 4570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.6025801081980855e-05,
+      "loss": 0.3127,
+      "step": 4580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.600499375780275e-05,
+      "loss": 0.0244,
+      "step": 4590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.598418643362464e-05,
+      "loss": 0.1026,
+      "step": 4600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.596337910944653e-05,
+      "loss": 0.3314,
+      "step": 4610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.594257178526842e-05,
+      "loss": 0.3053,
+      "step": 4620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5921764461090306e-05,
+      "loss": 0.2489,
+      "step": 4630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.590095713691219e-05,
+      "loss": 0.0668,
+      "step": 4640
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5880149812734085e-05,
+      "loss": 0.1628,
+      "step": 4650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.585934248855597e-05,
+      "loss": 0.064,
+      "step": 4660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.583853516437786e-05,
+      "loss": 0.1781,
+      "step": 4670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.581772784019975e-05,
+      "loss": 0.1673,
+      "step": 4680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.579692051602164e-05,
+      "loss": 0.0803,
+      "step": 4690
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5776113191843536e-05,
+      "loss": 0.2954,
+      "step": 4700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.575530586766542e-05,
+      "loss": 0.3284,
+      "step": 4710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.573449854348731e-05,
+      "loss": 0.199,
+      "step": 4720
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.57136912193092e-05,
+      "loss": 0.0978,
+      "step": 4730
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.569288389513109e-05,
+      "loss": 0.2259,
+      "step": 4740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5672076570952974e-05,
+      "loss": 0.1172,
+      "step": 4750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5651269246774867e-05,
+      "loss": 0.386,
+      "step": 4760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.563046192259676e-05,
+      "loss": 0.2982,
+      "step": 4770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5609654598418646e-05,
+      "loss": 0.1203,
+      "step": 4780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.558884727424054e-05,
+      "loss": 0.0712,
+      "step": 4790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5568039950062425e-05,
+      "loss": 0.0417,
+      "step": 4800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.554723262588431e-05,
+      "loss": 0.2212,
+      "step": 4810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5526425301706204e-05,
+      "loss": 0.2469,
+      "step": 4820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.550561797752809e-05,
+      "loss": 0.1038,
+      "step": 4830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.548481065334998e-05,
+      "loss": 0.2489,
+      "step": 4840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5464003329171876e-05,
+      "loss": 0.1279,
+      "step": 4850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544319600499376e-05,
+      "loss": 0.447,
+      "step": 4860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.542238868081565e-05,
+      "loss": 0.117,
+      "step": 4870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.540158135663754e-05,
+      "loss": 0.1692,
+      "step": 4880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.538077403245943e-05,
+      "loss": 0.3346,
+      "step": 4890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.535996670828131e-05,
+      "loss": 0.2527,
+      "step": 4900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5339159384103206e-05,
+      "loss": 0.1485,
+      "step": 4910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.531835205992509e-05,
+      "loss": 0.2969,
+      "step": 4920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5297544735746985e-05,
+      "loss": 0.1685,
+      "step": 4930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.527673741156888e-05,
+      "loss": 0.1816,
+      "step": 4940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5255930087390764e-05,
+      "loss": 0.1887,
+      "step": 4950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.523512276321265e-05,
+      "loss": 0.3821,
+      "step": 4960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5214315439034544e-05,
+      "loss": 0.0828,
+      "step": 4970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.519350811485643e-05,
+      "loss": 0.167,
+      "step": 4980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5172700790678316e-05,
+      "loss": 0.1685,
+      "step": 4990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.515189346650021e-05,
+      "loss": 0.2478,
+      "step": 5000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.51310861423221e-05,
+      "loss": 0.2478,
+      "step": 5010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.511027881814399e-05,
+      "loss": 0.1545,
+      "step": 5020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.508947149396588e-05,
+      "loss": 0.1951,
+      "step": 5030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.506866416978777e-05,
+      "loss": 0.2085,
+      "step": 5040
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.504785684560966e-05,
+      "loss": 0.1854,
+      "step": 5050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5027049521431546e-05,
+      "loss": 0.3196,
+      "step": 5060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.500624219725343e-05,
+      "loss": 0.1474,
+      "step": 5070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.4985434873075325e-05,
+      "loss": 0.2622,
+      "step": 5080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.496462754889721e-05,
+      "loss": 0.2549,
+      "step": 5090
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.4943820224719104e-05,
+      "loss": 0.0239,
+      "step": 5100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.4923012900541e-05,
+      "loss": 0.0897,
+      "step": 5110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.490220557636288e-05,
+      "loss": 0.1313,
+      "step": 5120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.488139825218477e-05,
+      "loss": 0.2186,
+      "step": 5130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.486059092800666e-05,
+      "loss": 0.12,
+      "step": 5140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.483978360382855e-05,
+      "loss": 0.5581,
+      "step": 5150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.4818976279650435e-05,
+      "loss": 0.2464,
+      "step": 5160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.479816895547233e-05,
+      "loss": 0.2357,
+      "step": 5170
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.477736163129422e-05,
+      "loss": 0.2009,
+      "step": 5180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.475655430711611e-05,
+      "loss": 0.1367,
+      "step": 5190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4735746982938e-05,
+      "loss": 0.1218,
+      "step": 5200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4714939658759886e-05,
+      "loss": 0.256,
+      "step": 5210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.469413233458177e-05,
+      "loss": 0.1986,
+      "step": 5220
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4673325010403665e-05,
+      "loss": 0.2234,
+      "step": 5230
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.465251768622555e-05,
+      "loss": 0.4343,
+      "step": 5240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.4631710362047444e-05,
+      "loss": 0.1247,
+      "step": 5250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.461090303786934e-05,
+      "loss": 0.1964,
+      "step": 5260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.459009571369122e-05,
+      "loss": 0.2728,
+      "step": 5270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.456928838951311e-05,
+      "loss": 0.135,
+      "step": 5280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4548481065335e-05,
+      "loss": 0.3644,
+      "step": 5290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.452767374115689e-05,
+      "loss": 0.2721,
+      "step": 5300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.4506866416978775e-05,
+      "loss": 0.2715,
+      "step": 5310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.448605909280067e-05,
+      "loss": 0.2618,
+      "step": 5320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.4465251768622554e-05,
+      "loss": 0.2132,
+      "step": 5330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.1956,
+      "step": 5340
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": {
+        "accuracy": 0.9337264813254704
+      },
+      "eval_f1": {
+        "f1": 0.9345043308561282
+      },
+      "eval_loss": 0.22899627685546875,
+      "eval_precision": {
+        "precision": 0.9371641968965463
+      },
+      "eval_recall": {
+        "recall": 0.9353444695340449
+      },
+      "eval_runtime": 94.2073,
+      "eval_samples_per_second": 37.8,
+      "eval_steps_per_second": 18.905,
+      "step": 5340
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 26700,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 500,
+  "total_flos": 3.310546808093737e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

tmp-checkpoint-5340/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc08deef7462e0688eea4cd78f5a41e4fed672d5f9bbda16fe9dab2683952f5b
+size 4728