Training in progress, epoch 4

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e07a097c1cc6571151d76b157094935dfc76ab4d1d534322f679ae1724ff78b8
+oid sha256:c897248b05a595815815af90cc11a34420404e8373577ba38254cec3b3d8db70
 size 343230128

tmp-checkpoint-5573/config.json

@@ -0,0 +1,36 @@
+{
+  "_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": "noncrack",
+    "1": "sealedcrack",
+    "2": "sealedpatch",
+    "3": "thincrack"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "noncrack": 0,
+    "sealedcrack": 1,
+    "sealedpatch": 2,
+    "thincrack": 3
+  },
+  "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-5573/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c897248b05a595815815af90cc11a34420404e8373577ba38254cec3b3d8db70
+size 343230128

tmp-checkpoint-5573/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc38ead823ba555cd314999d2a7dea3eeb68b769d09390f1c3200dfc2379395b
+size 686581178

tmp-checkpoint-5573/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-5573/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e552cb9eaa9f79b2e530d02d6ff7e293885a342dfcfe67bf8980446d6d9ceb12
+size 14244

tmp-checkpoint-5573/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a246d4eca6974bce96c9590300157e06acb504239072d0037798893c91ca485
+size 1064

tmp-checkpoint-5573/trainer_state.json

@@ -0,0 +1,3463 @@
+{
+  "best_metric": 0.1309136003255844,
+  "best_model_checkpoint": "vit-base-patch16-224-in21k-crack-detectorVITmain50epochs\\checkpoint-5573",
+  "epoch": 4.999327203408836,
+  "eval_steps": 500,
+  "global_step": 5573,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.976660682226213e-08,
+      "loss": 1.3944,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7953321364452426e-07,
+      "loss": 1.3875,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.692998204667864e-07,
+      "loss": 1.3999,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.590664272890485e-07,
+      "loss": 1.3915,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4883303411131064e-07,
+      "loss": 1.3856,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.385996409335728e-07,
+      "loss": 1.3912,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.283662477558349e-07,
+      "loss": 1.3902,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.18132854578097e-07,
+      "loss": 1.3851,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.078994614003591e-07,
+      "loss": 1.3755,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.976660682226213e-07,
+      "loss": 1.3793,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874326750448833e-07,
+      "loss": 1.3737,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0771992818671456e-06,
+      "loss": 1.3736,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1669658886894075e-06,
+      "loss": 1.3678,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2567324955116697e-06,
+      "loss": 1.3637,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3464991023339318e-06,
+      "loss": 1.357,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.436265709156194e-06,
+      "loss": 1.3583,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5260323159784561e-06,
+      "loss": 1.361,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6157989228007182e-06,
+      "loss": 1.3517,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7055655296229805e-06,
+      "loss": 1.3543,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7953321364452425e-06,
+      "loss": 1.3497,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8850987432675046e-06,
+      "loss": 1.3436,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9748653500897667e-06,
+      "loss": 1.3304,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.064631956912029e-06,
+      "loss": 1.3262,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1543985637342912e-06,
+      "loss": 1.3209,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.244165170556553e-06,
+      "loss": 1.3054,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.333931777378815e-06,
+      "loss": 1.3028,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.423698384201077e-06,
+      "loss": 1.2939,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5134649910233395e-06,
+      "loss": 1.2894,
+      "step": 280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6032315978456015e-06,
+      "loss": 1.2838,
+      "step": 290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6929982046678636e-06,
+      "loss": 1.2625,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7827648114901257e-06,
+      "loss": 1.2727,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.872531418312388e-06,
+      "loss": 1.2398,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9622980251346502e-06,
+      "loss": 1.2267,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0520646319569123e-06,
+      "loss": 1.2359,
+      "step": 340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1418312387791743e-06,
+      "loss": 1.218,
+      "step": 350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2315978456014364e-06,
+      "loss": 1.1877,
+      "step": 360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.321364452423698e-06,
+      "loss": 1.1764,
+      "step": 370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.411131059245961e-06,
+      "loss": 1.151,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.500897666068223e-06,
+      "loss": 1.1334,
+      "step": 390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.590664272890485e-06,
+      "loss": 1.1272,
+      "step": 400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.680430879712747e-06,
+      "loss": 1.1077,
+      "step": 410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.770197486535009e-06,
+      "loss": 1.1021,
+      "step": 420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.859964093357271e-06,
+      "loss": 1.0583,
+      "step": 430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.949730700179533e-06,
+      "loss": 1.0482,
+      "step": 440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.039497307001795e-06,
+      "loss": 1.0214,
+      "step": 450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.129263913824058e-06,
+      "loss": 1.0114,
+      "step": 460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.21903052064632e-06,
+      "loss": 1.0237,
+      "step": 470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3087971274685824e-06,
+      "loss": 0.969,
+      "step": 480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3985637342908445e-06,
+      "loss": 0.966,
+      "step": 490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.488330341113106e-06,
+      "loss": 0.9518,
+      "step": 500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.578096947935368e-06,
+      "loss": 0.9545,
+      "step": 510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.66786355475763e-06,
+      "loss": 0.9473,
+      "step": 520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.757630161579892e-06,
+      "loss": 0.9026,
+      "step": 530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.847396768402154e-06,
+      "loss": 0.8535,
+      "step": 540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.937163375224417e-06,
+      "loss": 0.8598,
+      "step": 550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.026929982046679e-06,
+      "loss": 0.8154,
+      "step": 560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116696588868941e-06,
+      "loss": 0.8561,
+      "step": 570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.206463195691203e-06,
+      "loss": 0.7924,
+      "step": 580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.296229802513465e-06,
+      "loss": 0.7947,
+      "step": 590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.385996409335727e-06,
+      "loss": 0.7999,
+      "step": 600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.475763016157989e-06,
+      "loss": 0.7771,
+      "step": 610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.565529622980251e-06,
+      "loss": 0.7826,
+      "step": 620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.655296229802514e-06,
+      "loss": 0.7321,
+      "step": 630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.745062836624776e-06,
+      "loss": 0.7689,
+      "step": 640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.834829443447038e-06,
+      "loss": 0.7546,
+      "step": 650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9245960502693004e-06,
+      "loss": 0.7336,
+      "step": 660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0143626570915625e-06,
+      "loss": 0.6889,
+      "step": 670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.1041292639138246e-06,
+      "loss": 0.6996,
+      "step": 680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.193895870736087e-06,
+      "loss": 0.6994,
+      "step": 690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283662477558349e-06,
+      "loss": 0.74,
+      "step": 700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.373429084380611e-06,
+      "loss": 0.711,
+      "step": 710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.463195691202873e-06,
+      "loss": 0.6765,
+      "step": 720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.552962298025135e-06,
+      "loss": 0.6717,
+      "step": 730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.642728904847396e-06,
+      "loss": 0.6776,
+      "step": 740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.732495511669659e-06,
+      "loss": 0.62,
+      "step": 750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.822262118491922e-06,
+      "loss": 0.6392,
+      "step": 760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.912028725314183e-06,
+      "loss": 0.62,
+      "step": 770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.001795332136446e-06,
+      "loss": 0.6103,
+      "step": 780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.091561938958707e-06,
+      "loss": 0.6504,
+      "step": 790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.18132854578097e-06,
+      "loss": 0.6126,
+      "step": 800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.271095152603231e-06,
+      "loss": 0.5707,
+      "step": 810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.360861759425494e-06,
+      "loss": 0.587,
+      "step": 820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.4506283662477555e-06,
+      "loss": 0.5684,
+      "step": 830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.540394973070018e-06,
+      "loss": 0.5586,
+      "step": 840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.630161579892281e-06,
+      "loss": 0.6074,
+      "step": 850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.719928186714543e-06,
+      "loss": 0.5638,
+      "step": 860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.809694793536805e-06,
+      "loss": 0.5952,
+      "step": 870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.899461400359067e-06,
+      "loss": 0.5647,
+      "step": 880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.98922800718133e-06,
+      "loss": 0.6049,
+      "step": 890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.07899461400359e-06,
+      "loss": 0.5525,
+      "step": 900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.168761220825854e-06,
+      "loss": 0.522,
+      "step": 910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.258527827648117e-06,
+      "loss": 0.5489,
+      "step": 920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.348294434470378e-06,
+      "loss": 0.5394,
+      "step": 930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.43806104129264e-06,
+      "loss": 0.5707,
+      "step": 940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.527827648114902e-06,
+      "loss": 0.5322,
+      "step": 950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.617594254937165e-06,
+      "loss": 0.5637,
+      "step": 960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.707360861759426e-06,
+      "loss": 0.529,
+      "step": 970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.797127468581689e-06,
+      "loss": 0.51,
+      "step": 980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.88689407540395e-06,
+      "loss": 0.4844,
+      "step": 990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.976660682226211e-06,
+      "loss": 0.5349,
+      "step": 1000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.066427289048474e-06,
+      "loss": 0.5377,
+      "step": 1010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.156193895870736e-06,
+      "loss": 0.4985,
+      "step": 1020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.245960502692998e-06,
+      "loss": 0.5533,
+      "step": 1030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.33572710951526e-06,
+      "loss": 0.4955,
+      "step": 1040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.425493716337523e-06,
+      "loss": 0.541,
+      "step": 1050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.515260323159784e-06,
+      "loss": 0.4757,
+      "step": 1060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.605026929982047e-06,
+      "loss": 0.5478,
+      "step": 1070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.694793536804308e-06,
+      "loss": 0.5281,
+      "step": 1080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.784560143626571e-06,
+      "loss": 0.5187,
+      "step": 1090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.874326750448834e-06,
+      "loss": 0.4591,
+      "step": 1100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.964093357271095e-06,
+      "loss": 0.5344,
+      "step": 1110
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": {
+        "accuracy": 0.8744113029827315
+      },
+      "eval_f1": {
+        "f1": 0.8645201833830978
+      },
+      "eval_loss": 0.40481075644493103,
+      "eval_precision": {
+        "precision": 0.8668175457367389
+      },
+      "eval_recall": {
+        "recall": 0.8631211355155497
+      },
+      "eval_runtime": 243.6123,
+      "eval_samples_per_second": 73.215,
+      "eval_steps_per_second": 4.577,
+      "step": 1114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0053859964093358e-05,
+      "loss": 0.5092,
+      "step": 1120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0143626570915619e-05,
+      "loss": 0.4834,
+      "step": 1130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0233393177737882e-05,
+      "loss": 0.5403,
+      "step": 1140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0323159784560143e-05,
+      "loss": 0.5109,
+      "step": 1150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0412926391382406e-05,
+      "loss": 0.5002,
+      "step": 1160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0502692998204669e-05,
+      "loss": 0.5052,
+      "step": 1170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.059245960502693e-05,
+      "loss": 0.4642,
+      "step": 1180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0682226211849193e-05,
+      "loss": 0.4479,
+      "step": 1190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0771992818671454e-05,
+      "loss": 0.4944,
+      "step": 1200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0861759425493717e-05,
+      "loss": 0.4583,
+      "step": 1210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0951526032315979e-05,
+      "loss": 0.4779,
+      "step": 1220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1041292639138241e-05,
+      "loss": 0.5038,
+      "step": 1230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1131059245960503e-05,
+      "loss": 0.4831,
+      "step": 1240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1220825852782766e-05,
+      "loss": 0.4999,
+      "step": 1250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1310592459605028e-05,
+      "loss": 0.4611,
+      "step": 1260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.140035906642729e-05,
+      "loss": 0.4775,
+      "step": 1270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1490125673249553e-05,
+      "loss": 0.4388,
+      "step": 1280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1579892280071814e-05,
+      "loss": 0.4374,
+      "step": 1290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1669658886894077e-05,
+      "loss": 0.4608,
+      "step": 1300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1759425493716338e-05,
+      "loss": 0.4973,
+      "step": 1310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1849192100538601e-05,
+      "loss": 0.4387,
+      "step": 1320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1938958707360862e-05,
+      "loss": 0.4617,
+      "step": 1330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2028725314183125e-05,
+      "loss": 0.4631,
+      "step": 1340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2118491921005388e-05,
+      "loss": 0.4869,
+      "step": 1350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2208258527827649e-05,
+      "loss": 0.4699,
+      "step": 1360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2298025134649912e-05,
+      "loss": 0.482,
+      "step": 1370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2387791741472173e-05,
+      "loss": 0.4067,
+      "step": 1380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2477558348294434e-05,
+      "loss": 0.4218,
+      "step": 1390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2567324955116697e-05,
+      "loss": 0.4643,
+      "step": 1400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2657091561938959e-05,
+      "loss": 0.4317,
+      "step": 1410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2746858168761221e-05,
+      "loss": 0.4303,
+      "step": 1420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836624775583483e-05,
+      "loss": 0.4247,
+      "step": 1430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2926391382405746e-05,
+      "loss": 0.4553,
+      "step": 1440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3016157989228009e-05,
+      "loss": 0.4527,
+      "step": 1450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.310592459605027e-05,
+      "loss": 0.4216,
+      "step": 1460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3195691202872531e-05,
+      "loss": 0.4481,
+      "step": 1470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3285457809694792e-05,
+      "loss": 0.3926,
+      "step": 1480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3375224416517057e-05,
+      "loss": 0.4644,
+      "step": 1490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3464991023339318e-05,
+      "loss": 0.4156,
+      "step": 1500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.355475763016158e-05,
+      "loss": 0.4165,
+      "step": 1510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3644524236983844e-05,
+      "loss": 0.427,
+      "step": 1520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3734290843806105e-05,
+      "loss": 0.4229,
+      "step": 1530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3824057450628366e-05,
+      "loss": 0.3961,
+      "step": 1540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3913824057450627e-05,
+      "loss": 0.4438,
+      "step": 1550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4003590664272892e-05,
+      "loss": 0.4037,
+      "step": 1560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4093357271095153e-05,
+      "loss": 0.4341,
+      "step": 1570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4183123877917415e-05,
+      "loss": 0.3715,
+      "step": 1580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4272890484739679e-05,
+      "loss": 0.4292,
+      "step": 1590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.436265709156194e-05,
+      "loss": 0.4539,
+      "step": 1600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4452423698384202e-05,
+      "loss": 0.409,
+      "step": 1610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4542190305206463e-05,
+      "loss": 0.4473,
+      "step": 1620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4631956912028727e-05,
+      "loss": 0.4202,
+      "step": 1630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4721723518850989e-05,
+      "loss": 0.3701,
+      "step": 1640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.481149012567325e-05,
+      "loss": 0.4451,
+      "step": 1650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4901256732495511e-05,
+      "loss": 0.4216,
+      "step": 1660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4991023339317776e-05,
+      "loss": 0.4116,
+      "step": 1670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5080789946140037e-05,
+      "loss": 0.4006,
+      "step": 1680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5170556552962298e-05,
+      "loss": 0.4303,
+      "step": 1690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5260323159784563e-05,
+      "loss": 0.441,
+      "step": 1700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5350089766606824e-05,
+      "loss": 0.4248,
+      "step": 1710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5439856373429085e-05,
+      "loss": 0.3771,
+      "step": 1720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5529622980251346e-05,
+      "loss": 0.4158,
+      "step": 1730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.561938958707361e-05,
+      "loss": 0.3815,
+      "step": 1740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5709156193895872e-05,
+      "loss": 0.4204,
+      "step": 1750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5798922800718133e-05,
+      "loss": 0.4886,
+      "step": 1760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5888689407540398e-05,
+      "loss": 0.4382,
+      "step": 1770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.597845601436266e-05,
+      "loss": 0.4502,
+      "step": 1780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.606822262118492e-05,
+      "loss": 0.4155,
+      "step": 1790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.615798922800718e-05,
+      "loss": 0.3801,
+      "step": 1800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6247755834829446e-05,
+      "loss": 0.3968,
+      "step": 1810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6337522441651707e-05,
+      "loss": 0.365,
+      "step": 1820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.642728904847397e-05,
+      "loss": 0.3638,
+      "step": 1830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6517055655296233e-05,
+      "loss": 0.3935,
+      "step": 1840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6606822262118494e-05,
+      "loss": 0.4092,
+      "step": 1850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6696588868940756e-05,
+      "loss": 0.3617,
+      "step": 1860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6786355475763017e-05,
+      "loss": 0.389,
+      "step": 1870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.687612208258528e-05,
+      "loss": 0.3357,
+      "step": 1880
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6965888689407543e-05,
+      "loss": 0.3925,
+      "step": 1890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7055655296229804e-05,
+      "loss": 0.4186,
+      "step": 1900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7145421903052065e-05,
+      "loss": 0.3913,
+      "step": 1910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.723518850987433e-05,
+      "loss": 0.3679,
+      "step": 1920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.732495511669659e-05,
+      "loss": 0.4476,
+      "step": 1930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7414721723518852e-05,
+      "loss": 0.4065,
+      "step": 1940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7504488330341113e-05,
+      "loss": 0.4079,
+      "step": 1950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7594254937163378e-05,
+      "loss": 0.3831,
+      "step": 1960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.768402154398564e-05,
+      "loss": 0.3985,
+      "step": 1970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.77737881508079e-05,
+      "loss": 0.3705,
+      "step": 1980
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.786355475763016e-05,
+      "loss": 0.3742,
+      "step": 1990
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7953321364452423e-05,
+      "loss": 0.3517,
+      "step": 2000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8043087971274687e-05,
+      "loss": 0.3871,
+      "step": 2010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.813285457809695e-05,
+      "loss": 0.3416,
+      "step": 2020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.822262118491921e-05,
+      "loss": 0.3856,
+      "step": 2030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.831238779174147e-05,
+      "loss": 0.4245,
+      "step": 2040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8402154398563732e-05,
+      "loss": 0.3645,
+      "step": 2050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8491921005385997e-05,
+      "loss": 0.3505,
+      "step": 2060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8581687612208258e-05,
+      "loss": 0.3746,
+      "step": 2070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.867145421903052e-05,
+      "loss": 0.352,
+      "step": 2080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8761220825852784e-05,
+      "loss": 0.4295,
+      "step": 2090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8850987432675045e-05,
+      "loss": 0.387,
+      "step": 2100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8940754039497306e-05,
+      "loss": 0.4056,
+      "step": 2110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9030520646319568e-05,
+      "loss": 0.3913,
+      "step": 2120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9120287253141832e-05,
+      "loss": 0.3789,
+      "step": 2130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9210053859964093e-05,
+      "loss": 0.3688,
+      "step": 2140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9299820466786355e-05,
+      "loss": 0.3499,
+      "step": 2150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9389587073608616e-05,
+      "loss": 0.3153,
+      "step": 2160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.947935368043088e-05,
+      "loss": 0.3558,
+      "step": 2170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9569120287253142e-05,
+      "loss": 0.3001,
+      "step": 2180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9658886894075403e-05,
+      "loss": 0.3109,
+      "step": 2190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9748653500897668e-05,
+      "loss": 0.3377,
+      "step": 2200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.983842010771993e-05,
+      "loss": 0.3888,
+      "step": 2210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.992818671454219e-05,
+      "loss": 0.3536,
+      "step": 2220
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": {
+        "accuracy": 0.9219555954249832
+      },
+      "eval_f1": {
+        "f1": 0.9164276462698699
+      },
+      "eval_loss": 0.2440667599439621,
+      "eval_precision": {
+        "precision": 0.9196643633917082
+      },
+      "eval_recall": {
+        "recall": 0.913748377704386
+      },
+      "eval_runtime": 164.5475,
+      "eval_samples_per_second": 108.394,
+      "eval_steps_per_second": 6.776,
+      "step": 2229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.001795332136445e-05,
+      "loss": 0.3716,
+      "step": 2230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0107719928186716e-05,
+      "loss": 0.3336,
+      "step": 2240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0197486535008977e-05,
+      "loss": 0.3309,
+      "step": 2250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0287253141831238e-05,
+      "loss": 0.4064,
+      "step": 2260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0377019748653503e-05,
+      "loss": 0.33,
+      "step": 2270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0466786355475764e-05,
+      "loss": 0.3911,
+      "step": 2280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0556552962298025e-05,
+      "loss": 0.3405,
+      "step": 2290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0646319569120286e-05,
+      "loss": 0.317,
+      "step": 2300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.073608617594255e-05,
+      "loss": 0.3305,
+      "step": 2310
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0825852782764812e-05,
+      "loss": 0.3201,
+      "step": 2320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0915619389587073e-05,
+      "loss": 0.3107,
+      "step": 2330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.1005385996409338e-05,
+      "loss": 0.351,
+      "step": 2340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.10951526032316e-05,
+      "loss": 0.3181,
+      "step": 2350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.118491921005386e-05,
+      "loss": 0.3494,
+      "step": 2360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1274685816876122e-05,
+      "loss": 0.3718,
+      "step": 2370
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1364452423698386e-05,
+      "loss": 0.34,
+      "step": 2380
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1454219030520648e-05,
+      "loss": 0.306,
+      "step": 2390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.154398563734291e-05,
+      "loss": 0.3147,
+      "step": 2400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.163375224416517e-05,
+      "loss": 0.2842,
+      "step": 2410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1723518850987435e-05,
+      "loss": 0.3461,
+      "step": 2420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1813285457809696e-05,
+      "loss": 0.3161,
+      "step": 2430
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1903052064631957e-05,
+      "loss": 0.3152,
+      "step": 2440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.199281867145422e-05,
+      "loss": 0.3104,
+      "step": 2450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2082585278276483e-05,
+      "loss": 0.3485,
+      "step": 2460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2172351885098744e-05,
+      "loss": 0.3682,
+      "step": 2470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2262118491921005e-05,
+      "loss": 0.2831,
+      "step": 2480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.235188509874327e-05,
+      "loss": 0.3097,
+      "step": 2490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.244165170556553e-05,
+      "loss": 0.3433,
+      "step": 2500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2531418312387792e-05,
+      "loss": 0.3319,
+      "step": 2510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2621184919210057e-05,
+      "loss": 0.3236,
+      "step": 2520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2710951526032318e-05,
+      "loss": 0.3531,
+      "step": 2530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.280071813285458e-05,
+      "loss": 0.2737,
+      "step": 2540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.289048473967684e-05,
+      "loss": 0.3774,
+      "step": 2550
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2980251346499105e-05,
+      "loss": 0.2936,
+      "step": 2560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3070017953321366e-05,
+      "loss": 0.3015,
+      "step": 2570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3159784560143628e-05,
+      "loss": 0.2933,
+      "step": 2580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3249551166965892e-05,
+      "loss": 0.3348,
+      "step": 2590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3339317773788153e-05,
+      "loss": 0.3249,
+      "step": 2600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3429084380610415e-05,
+      "loss": 0.3619,
+      "step": 2610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3518850987432676e-05,
+      "loss": 0.3428,
+      "step": 2620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.360861759425494e-05,
+      "loss": 0.3425,
+      "step": 2630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3698384201077202e-05,
+      "loss": 0.3027,
+      "step": 2640
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3788150807899463e-05,
+      "loss": 0.3723,
+      "step": 2650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3877917414721724e-05,
+      "loss": 0.3813,
+      "step": 2660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.396768402154399e-05,
+      "loss": 0.3423,
+      "step": 2670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.405745062836625e-05,
+      "loss": 0.355,
+      "step": 2680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.414721723518851e-05,
+      "loss": 0.3047,
+      "step": 2690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4236983842010776e-05,
+      "loss": 0.3049,
+      "step": 2700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4326750448833037e-05,
+      "loss": 0.2876,
+      "step": 2710
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4416517055655298e-05,
+      "loss": 0.3166,
+      "step": 2720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.450628366247756e-05,
+      "loss": 0.3468,
+      "step": 2730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.4596050269299824e-05,
+      "loss": 0.3173,
+      "step": 2740
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4685816876122085e-05,
+      "loss": 0.3189,
+      "step": 2750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4775583482944346e-05,
+      "loss": 0.3321,
+      "step": 2760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4865350089766608e-05,
+      "loss": 0.3314,
+      "step": 2770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.495511669658887e-05,
+      "loss": 0.3338,
+      "step": 2780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5044883303411134e-05,
+      "loss": 0.3212,
+      "step": 2790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5134649910233395e-05,
+      "loss": 0.3814,
+      "step": 2800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5224416517055656e-05,
+      "loss": 0.3548,
+      "step": 2810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5314183123877917e-05,
+      "loss": 0.3111,
+      "step": 2820
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5403949730700182e-05,
+      "loss": 0.2813,
+      "step": 2830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5493716337522443e-05,
+      "loss": 0.2877,
+      "step": 2840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5583482944344704e-05,
+      "loss": 0.3169,
+      "step": 2850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5673249551166965e-05,
+      "loss": 0.2842,
+      "step": 2860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5763016157989227e-05,
+      "loss": 0.3174,
+      "step": 2870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.585278276481149e-05,
+      "loss": 0.3094,
+      "step": 2880
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5942549371633752e-05,
+      "loss": 0.3405,
+      "step": 2890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6032315978456017e-05,
+      "loss": 0.3013,
+      "step": 2900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6122082585278278e-05,
+      "loss": 0.322,
+      "step": 2910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.621184919210054e-05,
+      "loss": 0.2959,
+      "step": 2920
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.63016157989228e-05,
+      "loss": 0.3343,
+      "step": 2930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6391382405745062e-05,
+      "loss": 0.3079,
+      "step": 2940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.6481149012567323e-05,
+      "loss": 0.2994,
+      "step": 2950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6570915619389584e-05,
+      "loss": 0.3091,
+      "step": 2960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6660682226211852e-05,
+      "loss": 0.3317,
+      "step": 2970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.6750448833034114e-05,
+      "loss": 0.2687,
+      "step": 2980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.6840215439856375e-05,
+      "loss": 0.2859,
+      "step": 2990
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.6929982046678636e-05,
+      "loss": 0.2997,
+      "step": 3000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.7019748653500897e-05,
+      "loss": 0.2603,
+      "step": 3010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.710951526032316e-05,
+      "loss": 0.3452,
+      "step": 3020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.719928186714542e-05,
+      "loss": 0.3475,
+      "step": 3030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7289048473967688e-05,
+      "loss": 0.317,
+      "step": 3040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.737881508078995e-05,
+      "loss": 0.3266,
+      "step": 3050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.746858168761221e-05,
+      "loss": 0.291,
+      "step": 3060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.755834829443447e-05,
+      "loss": 0.3323,
+      "step": 3070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7648114901256732e-05,
+      "loss": 0.32,
+      "step": 3080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7737881508078994e-05,
+      "loss": 0.2792,
+      "step": 3090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7827648114901255e-05,
+      "loss": 0.2858,
+      "step": 3100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7917414721723523e-05,
+      "loss": 0.2447,
+      "step": 3110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.8007181328545784e-05,
+      "loss": 0.3039,
+      "step": 3120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8096947935368045e-05,
+      "loss": 0.323,
+      "step": 3130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8186714542190307e-05,
+      "loss": 0.3041,
+      "step": 3140
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8276481149012568e-05,
+      "loss": 0.2496,
+      "step": 3150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.836624775583483e-05,
+      "loss": 0.3205,
+      "step": 3160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.845601436265709e-05,
+      "loss": 0.3079,
+      "step": 3170
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8545780969479358e-05,
+      "loss": 0.3112,
+      "step": 3180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.863554757630162e-05,
+      "loss": 0.2674,
+      "step": 3190
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.872531418312388e-05,
+      "loss": 0.3094,
+      "step": 3200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8815080789946142e-05,
+      "loss": 0.2895,
+      "step": 3210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.8904847396768403e-05,
+      "loss": 0.3456,
+      "step": 3220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.8994614003590664e-05,
+      "loss": 0.2602,
+      "step": 3230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9084380610412926e-05,
+      "loss": 0.3198,
+      "step": 3240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9174147217235194e-05,
+      "loss": 0.3463,
+      "step": 3250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9263913824057455e-05,
+      "loss": 0.3456,
+      "step": 3260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9353680430879716e-05,
+      "loss": 0.2803,
+      "step": 3270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9443447037701977e-05,
+      "loss": 0.302,
+      "step": 3280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.953321364452424e-05,
+      "loss": 0.2851,
+      "step": 3290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.96229802513465e-05,
+      "loss": 0.2752,
+      "step": 3300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.971274685816876e-05,
+      "loss": 0.2655,
+      "step": 3310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9802513464991022e-05,
+      "loss": 0.2716,
+      "step": 3320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.989228007181329e-05,
+      "loss": 0.2591,
+      "step": 3330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.998204667863555e-05,
+      "loss": 0.2794,
+      "step": 3340
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": {
+        "accuracy": 0.9377663153173357
+      },
+      "eval_f1": {
+        "f1": 0.9340905709341878
+      },
+      "eval_loss": 0.18664708733558655,
+      "eval_precision": {
+        "precision": 0.9374045110718889
+      },
+      "eval_recall": {
+        "recall": 0.9314497447723507
+      },
+      "eval_runtime": 164.7744,
+      "eval_samples_per_second": 108.245,
+      "eval_steps_per_second": 6.767,
+      "step": 3344
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0071813285457812e-05,
+      "loss": 0.3214,
+      "step": 3350
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0161579892280074e-05,
+      "loss": 0.3234,
+      "step": 3360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0251346499102335e-05,
+      "loss": 0.2471,
+      "step": 3370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0341113105924596e-05,
+      "loss": 0.2891,
+      "step": 3380
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0430879712746857e-05,
+      "loss": 0.2794,
+      "step": 3390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0520646319569125e-05,
+      "loss": 0.2889,
+      "step": 3400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.061041292639138e-05,
+      "loss": 0.3119,
+      "step": 3410
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.070017953321365e-05,
+      "loss": 0.2721,
+      "step": 3420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.0789946140035906e-05,
+      "loss": 0.319,
+      "step": 3430
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.087971274685817e-05,
+      "loss": 0.2623,
+      "step": 3440
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.096947935368043e-05,
+      "loss": 0.2689,
+      "step": 3450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.105924596050269e-05,
+      "loss": 0.2319,
+      "step": 3460
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.114901256732496e-05,
+      "loss": 0.2781,
+      "step": 3470
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.123877917414722e-05,
+      "loss": 0.3058,
+      "step": 3480
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.132854578096948e-05,
+      "loss": 0.2508,
+      "step": 3490
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.1418312387791744e-05,
+      "loss": 0.2737,
+      "step": 3500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1508078994614e-05,
+      "loss": 0.2592,
+      "step": 3510
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.159784560143627e-05,
+      "loss": 0.2965,
+      "step": 3520
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1687612208258525e-05,
+      "loss": 0.2821,
+      "step": 3530
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1777378815080796e-05,
+      "loss": 0.2622,
+      "step": 3540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1867145421903054e-05,
+      "loss": 0.3081,
+      "step": 3550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.195691202872532e-05,
+      "loss": 0.3187,
+      "step": 3560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.2046678635547576e-05,
+      "loss": 0.2525,
+      "step": 3570
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.213644524236984e-05,
+      "loss": 0.2896,
+      "step": 3580
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.22262118491921e-05,
+      "loss": 0.2649,
+      "step": 3590
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.231597845601436e-05,
+      "loss": 0.3238,
+      "step": 3600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.240574506283663e-05,
+      "loss": 0.298,
+      "step": 3610
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.249551166965889e-05,
+      "loss": 0.3019,
+      "step": 3620
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.258527827648115e-05,
+      "loss": 0.2835,
+      "step": 3630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.2675044883303415e-05,
+      "loss": 0.2717,
+      "step": 3640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.276481149012567e-05,
+      "loss": 0.2582,
+      "step": 3650
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.285457809694794e-05,
+      "loss": 0.2473,
+      "step": 3660
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2944344703770195e-05,
+      "loss": 0.3248,
+      "step": 3670
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3034111310592466e-05,
+      "loss": 0.2731,
+      "step": 3680
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3123877917414724e-05,
+      "loss": 0.2927,
+      "step": 3690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.321364452423699e-05,
+      "loss": 0.2665,
+      "step": 3700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.330341113105925e-05,
+      "loss": 0.2269,
+      "step": 3710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339317773788151e-05,
+      "loss": 0.279,
+      "step": 3720
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.348294434470377e-05,
+      "loss": 0.2727,
+      "step": 3730
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3572710951526034e-05,
+      "loss": 0.2918,
+      "step": 3740
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.36624775583483e-05,
+      "loss": 0.2857,
+      "step": 3750
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.375224416517056e-05,
+      "loss": 0.2641,
+      "step": 3760
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.384201077199282e-05,
+      "loss": 0.2134,
+      "step": 3770
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3931777378815085e-05,
+      "loss": 0.2589,
+      "step": 3780
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.402154398563734e-05,
+      "loss": 0.2597,
+      "step": 3790
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.411131059245961e-05,
+      "loss": 0.2538,
+      "step": 3800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.4201077199281866e-05,
+      "loss": 0.271,
+      "step": 3810
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.429084380610413e-05,
+      "loss": 0.2326,
+      "step": 3820
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4380610412926395e-05,
+      "loss": 0.26,
+      "step": 3830
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.447037701974866e-05,
+      "loss": 0.2703,
+      "step": 3840
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.456014362657092e-05,
+      "loss": 0.2996,
+      "step": 3850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.464991023339318e-05,
+      "loss": 0.2231,
+      "step": 3860
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.473967684021544e-05,
+      "loss": 0.287,
+      "step": 3870
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.4829443447037704e-05,
+      "loss": 0.2877,
+      "step": 3880
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.491921005385996e-05,
+      "loss": 0.2941,
+      "step": 3890
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.500897666068223e-05,
+      "loss": 0.283,
+      "step": 3900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.509874326750449e-05,
+      "loss": 0.2899,
+      "step": 3910
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.5188509874326756e-05,
+      "loss": 0.2669,
+      "step": 3920
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5278276481149014e-05,
+      "loss": 0.2673,
+      "step": 3930
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.536804308797128e-05,
+      "loss": 0.2427,
+      "step": 3940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.5457809694793536e-05,
+      "loss": 0.3008,
+      "step": 3950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.55475763016158e-05,
+      "loss": 0.2333,
+      "step": 3960
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5637342908438065e-05,
+      "loss": 0.2415,
+      "step": 3970
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.572710951526032e-05,
+      "loss": 0.2642,
+      "step": 3980
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.581687612208259e-05,
+      "loss": 0.2814,
+      "step": 3990
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5906642728904846e-05,
+      "loss": 0.2607,
+      "step": 4000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.599640933572711e-05,
+      "loss": 0.2062,
+      "step": 4010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6086175942549375e-05,
+      "loss": 0.2388,
+      "step": 4020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.617594254937163e-05,
+      "loss": 0.2364,
+      "step": 4030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.62657091561939e-05,
+      "loss": 0.2803,
+      "step": 4040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.635547576301616e-05,
+      "loss": 0.236,
+      "step": 4050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.644524236983842e-05,
+      "loss": 0.2795,
+      "step": 4060
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6535008976660684e-05,
+      "loss": 0.2728,
+      "step": 4070
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.662477558348294e-05,
+      "loss": 0.2897,
+      "step": 4080
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.671454219030521e-05,
+      "loss": 0.257,
+      "step": 4090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.6804308797127465e-05,
+      "loss": 0.2913,
+      "step": 4100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6894075403949736e-05,
+      "loss": 0.212,
+      "step": 4110
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.6983842010771994e-05,
+      "loss": 0.2672,
+      "step": 4120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.707360861759426e-05,
+      "loss": 0.2852,
+      "step": 4130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7163375224416516e-05,
+      "loss": 0.2238,
+      "step": 4140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.725314183123878e-05,
+      "loss": 0.2344,
+      "step": 4150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.734290843806104e-05,
+      "loss": 0.2695,
+      "step": 4160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.74326750448833e-05,
+      "loss": 0.2729,
+      "step": 4170
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.752244165170557e-05,
+      "loss": 0.2692,
+      "step": 4180
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.761220825852783e-05,
+      "loss": 0.2845,
+      "step": 4190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.770197486535009e-05,
+      "loss": 0.244,
+      "step": 4200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7791741472172355e-05,
+      "loss": 0.2621,
+      "step": 4210
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.788150807899461e-05,
+      "loss": 0.2581,
+      "step": 4220
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.797127468581688e-05,
+      "loss": 0.2201,
+      "step": 4230
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.8061041292639135e-05,
+      "loss": 0.2731,
+      "step": 4240
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.815080789946141e-05,
+      "loss": 0.3085,
+      "step": 4250
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.8240574506283664e-05,
+      "loss": 0.2729,
+      "step": 4260
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.833034111310593e-05,
+      "loss": 0.2282,
+      "step": 4270
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.842010771992819e-05,
+      "loss": 0.1955,
+      "step": 4280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.850987432675045e-05,
+      "loss": 0.2213,
+      "step": 4290
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.859964093357271e-05,
+      "loss": 0.2719,
+      "step": 4300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.8689407540394974e-05,
+      "loss": 0.2045,
+      "step": 4310
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.877917414721723e-05,
+      "loss": 0.2468,
+      "step": 4320
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.88689407540395e-05,
+      "loss": 0.259,
+      "step": 4330
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.895870736086176e-05,
+      "loss": 0.2462,
+      "step": 4340
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.9048473967684026e-05,
+      "loss": 0.2278,
+      "step": 4350
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.9138240574506283e-05,
+      "loss": 0.2701,
+      "step": 4360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.922800718132855e-05,
+      "loss": 0.269,
+      "step": 4370
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.9317773788150806e-05,
+      "loss": 0.2938,
+      "step": 4380
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.940754039497307e-05,
+      "loss": 0.2669,
+      "step": 4390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9497307001795335e-05,
+      "loss": 0.2557,
+      "step": 4400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.95870736086176e-05,
+      "loss": 0.2629,
+      "step": 4410
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.967684021543986e-05,
+      "loss": 0.2168,
+      "step": 4420
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.976660682226212e-05,
+      "loss": 0.2734,
+      "step": 4430
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.985637342908438e-05,
+      "loss": 0.2119,
+      "step": 4440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.9946140035906645e-05,
+      "loss": 0.2177,
+      "step": 4450
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": {
+        "accuracy": 0.9484189280107648
+      },
+      "eval_f1": {
+        "f1": 0.9455383407746016
+      },
+      "eval_loss": 0.1543792188167572,
+      "eval_precision": {
+        "precision": 0.9499526075822722
+      },
+      "eval_recall": {
+        "recall": 0.9425902511351058
+      },
+      "eval_runtime": 165.5499,
+      "eval_samples_per_second": 107.738,
+      "eval_steps_per_second": 6.735,
+      "step": 4459
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.00359066427289e-05,
+      "loss": 0.252,
+      "step": 4460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0125673249551174e-05,
+      "loss": 0.1988,
+      "step": 4470
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.021543985637343e-05,
+      "loss": 0.2268,
+      "step": 4480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0305206463195696e-05,
+      "loss": 0.2013,
+      "step": 4490
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.0394973070017954e-05,
+      "loss": 0.2409,
+      "step": 4500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.048473967684022e-05,
+      "loss": 0.2122,
+      "step": 4510
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.0574506283662476e-05,
+      "loss": 0.2246,
+      "step": 4520
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.066427289048474e-05,
+      "loss": 0.251,
+      "step": 4530
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.0754039497307006e-05,
+      "loss": 0.2115,
+      "step": 4540
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.084380610412927e-05,
+      "loss": 0.2536,
+      "step": 4550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.093357271095153e-05,
+      "loss": 0.2526,
+      "step": 4560
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.102333931777379e-05,
+      "loss": 0.2337,
+      "step": 4570
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.111310592459605e-05,
+      "loss": 0.2455,
+      "step": 4580
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.1202872531418315e-05,
+      "loss": 0.2178,
+      "step": 4590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.129263913824057e-05,
+      "loss": 0.2402,
+      "step": 4600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.1382405745062844e-05,
+      "loss": 0.2868,
+      "step": 4610
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.14721723518851e-05,
+      "loss": 0.1957,
+      "step": 4620
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.156193895870737e-05,
+      "loss": 0.2701,
+      "step": 4630
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.1651705565529625e-05,
+      "loss": 0.2711,
+      "step": 4640
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.174147217235189e-05,
+      "loss": 0.2992,
+      "step": 4650
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.183123877917415e-05,
+      "loss": 0.2284,
+      "step": 4660
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.192100538599641e-05,
+      "loss": 0.2522,
+      "step": 4670
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.2010771992818676e-05,
+      "loss": 0.2039,
+      "step": 4680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.2100538599640934e-05,
+      "loss": 0.2517,
+      "step": 4690
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.21903052064632e-05,
+      "loss": 0.2475,
+      "step": 4700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.228007181328546e-05,
+      "loss": 0.2224,
+      "step": 4710
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.236983842010772e-05,
+      "loss": 0.2761,
+      "step": 4720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2459605026929986e-05,
+      "loss": 0.2125,
+      "step": 4730
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2549371633752244e-05,
+      "loss": 0.2067,
+      "step": 4740
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.263913824057451e-05,
+      "loss": 0.2237,
+      "step": 4750
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.272890484739677e-05,
+      "loss": 0.2223,
+      "step": 4760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.281867145421903e-05,
+      "loss": 0.1756,
+      "step": 4770
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2908438061041295e-05,
+      "loss": 0.2134,
+      "step": 4780
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.299820466786356e-05,
+      "loss": 0.2791,
+      "step": 4790
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.308797127468582e-05,
+      "loss": 0.2446,
+      "step": 4800
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.317773788150808e-05,
+      "loss": 0.2578,
+      "step": 4810
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.326750448833034e-05,
+      "loss": 0.2233,
+      "step": 4820
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.3357271095152605e-05,
+      "loss": 0.2427,
+      "step": 4830
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.344703770197487e-05,
+      "loss": 0.2652,
+      "step": 4840
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.353680430879713e-05,
+      "loss": 0.252,
+      "step": 4850
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.362657091561939e-05,
+      "loss": 0.2174,
+      "step": 4860
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.371633752244165e-05,
+      "loss": 0.2391,
+      "step": 4870
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.3806104129263914e-05,
+      "loss": 0.2156,
+      "step": 4880
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.389587073608618e-05,
+      "loss": 0.2291,
+      "step": 4890
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.398563734290844e-05,
+      "loss": 0.2048,
+      "step": 4900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.40754039497307e-05,
+      "loss": 0.2036,
+      "step": 4910
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.4165170556552966e-05,
+      "loss": 0.2258,
+      "step": 4920
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4254937163375224e-05,
+      "loss": 0.2229,
+      "step": 4930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.434470377019749e-05,
+      "loss": 0.2562,
+      "step": 4940
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.4434470377019746e-05,
+      "loss": 0.2375,
+      "step": 4950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.452423698384201e-05,
+      "loss": 0.2262,
+      "step": 4960
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4614003590664275e-05,
+      "loss": 0.2465,
+      "step": 4970
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.470377019748654e-05,
+      "loss": 0.2164,
+      "step": 4980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.47935368043088e-05,
+      "loss": 0.194,
+      "step": 4990
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.488330341113106e-05,
+      "loss": 0.2824,
+      "step": 5000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.497307001795332e-05,
+      "loss": 0.2037,
+      "step": 5010
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.5062836624775585e-05,
+      "loss": 0.2506,
+      "step": 5020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.515260323159784e-05,
+      "loss": 0.1948,
+      "step": 5030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.5242369838420114e-05,
+      "loss": 0.2214,
+      "step": 5040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.533213644524237e-05,
+      "loss": 0.247,
+      "step": 5050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5421903052064636e-05,
+      "loss": 0.2344,
+      "step": 5060
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.5511669658886894e-05,
+      "loss": 0.1798,
+      "step": 5070
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.560143626570916e-05,
+      "loss": 0.1861,
+      "step": 5080
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.5691202872531417e-05,
+      "loss": 0.3008,
+      "step": 5090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.578096947935368e-05,
+      "loss": 0.2465,
+      "step": 5100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.5870736086175946e-05,
+      "loss": 0.2362,
+      "step": 5110
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.596050269299821e-05,
+      "loss": 0.2791,
+      "step": 5120
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.605026929982047e-05,
+      "loss": 0.2231,
+      "step": 5130
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.614003590664273e-05,
+      "loss": 0.2162,
+      "step": 5140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.622980251346499e-05,
+      "loss": 0.2517,
+      "step": 5150
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.6319569120287255e-05,
+      "loss": 0.238,
+      "step": 5160
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.640933572710951e-05,
+      "loss": 0.246,
+      "step": 5170
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.6499102333931784e-05,
+      "loss": 0.2644,
+      "step": 5180
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.658886894075404e-05,
+      "loss": 0.2377,
+      "step": 5190
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.667863554757631e-05,
+      "loss": 0.2456,
+      "step": 5200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.6768402154398565e-05,
+      "loss": 0.2361,
+      "step": 5210
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.685816876122083e-05,
+      "loss": 0.213,
+      "step": 5220
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.694793536804309e-05,
+      "loss": 0.198,
+      "step": 5230
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.703770197486535e-05,
+      "loss": 0.1916,
+      "step": 5240
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7127468581687616e-05,
+      "loss": 0.2777,
+      "step": 5250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.721723518850988e-05,
+      "loss": 0.223,
+      "step": 5260
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.730700179533214e-05,
+      "loss": 0.2366,
+      "step": 5270
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7396768402154403e-05,
+      "loss": 0.2375,
+      "step": 5280
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.748653500897666e-05,
+      "loss": 0.1849,
+      "step": 5290
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.7576301615798926e-05,
+      "loss": 0.1591,
+      "step": 5300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7666068222621184e-05,
+      "loss": 0.1891,
+      "step": 5310
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.775583482944345e-05,
+      "loss": 0.2269,
+      "step": 5320
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.784560143626571e-05,
+      "loss": 0.2275,
+      "step": 5330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.793536804308798e-05,
+      "loss": 0.1934,
+      "step": 5340
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.8025134649910235e-05,
+      "loss": 0.1878,
+      "step": 5350
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.81149012567325e-05,
+      "loss": 0.2362,
+      "step": 5360
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.820466786355476e-05,
+      "loss": 0.1989,
+      "step": 5370
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.829443447037702e-05,
+      "loss": 0.2118,
+      "step": 5380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.838420107719928e-05,
+      "loss": 0.2637,
+      "step": 5390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.847396768402155e-05,
+      "loss": 0.2251,
+      "step": 5400
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.856373429084381e-05,
+      "loss": 0.2176,
+      "step": 5410
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.8653500897666074e-05,
+      "loss": 0.2177,
+      "step": 5420
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.874326750448833e-05,
+      "loss": 0.1898,
+      "step": 5430
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.8833034111310596e-05,
+      "loss": 0.2589,
+      "step": 5440
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.8922800718132854e-05,
+      "loss": 0.2425,
+      "step": 5450
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.901256732495512e-05,
+      "loss": 0.2131,
+      "step": 5460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.9102333931777383e-05,
+      "loss": 0.2607,
+      "step": 5470
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.919210053859965e-05,
+      "loss": 0.2929,
+      "step": 5480
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.9281867145421906e-05,
+      "loss": 0.207,
+      "step": 5490
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.937163375224417e-05,
+      "loss": 0.3048,
+      "step": 5500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.946140035906643e-05,
+      "loss": 0.2231,
+      "step": 5510
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.955116696588869e-05,
+      "loss": 0.1971,
+      "step": 5520
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.964093357271095e-05,
+      "loss": 0.2188,
+      "step": 5530
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.9730700179533215e-05,
+      "loss": 0.162,
+      "step": 5540
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.982046678635548e-05,
+      "loss": 0.1836,
+      "step": 5550
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.991023339317774e-05,
+      "loss": 0.2438,
+      "step": 5560
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5e-05,
+      "loss": 0.2397,
+      "step": 5570
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": {
+        "accuracy": 0.9569410181655079
+      },
+      "eval_f1": {
+        "f1": 0.9546849040800426
+      },
+      "eval_loss": 0.1309136003255844,
+      "eval_precision": {
+        "precision": 0.9547567778919024
+      },
+      "eval_recall": {
+        "recall": 0.9546610577932907
+      },
+      "eval_runtime": 167.576,
+      "eval_samples_per_second": 106.435,
+      "eval_steps_per_second": 6.654,
+      "step": 5573
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 55700,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 500,
+  "total_flos": 2.764188817764434e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

tmp-checkpoint-5573/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3394b871720811033add862f1e1383ca3790e6ab55972b19ed900a5d6b59c978
+size 4792