Upload folder using huggingface_hub

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 8.0,
+    "total_flos": 3.8629591832685773e+18,
+    "train_loss": 0.2533179052472592,
+    "train_runtime": 4957.8804,
+    "train_samples_per_second": 10.069,
+    "train_steps_per_second": 0.629
+}

checkpoint-1200/config.json

@@ -0,0 +1,40 @@
+{
+  "_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": "Brazil",
+    "1": "Canada",
+    "2": "Finland",
+    "3": "Japan",
+    "4": "United-Kingdom",
+    "5": "United_States"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "Brazil": "0",
+    "Canada": "1",
+    "Finland": "2",
+    "Japan": "3",
+    "United-Kingdom": "4",
+    "United_States": "5"
+  },
+  "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": "4.31.0"
+}

checkpoint-1200/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b53cf19e0fe68219c3cde14ec10a9e31bbfe37f2bddc8174b04e0adce4f1221e
+size 686537861

checkpoint-1200/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": "ViTFeatureExtractor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 224,
+    "width": 224
+  }
+}

checkpoint-1200/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e4cf7738c400c63d307d24c20ed9c77e1a7e56df0099b73936659a2e13d9670
+size 343278253

checkpoint-1200/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15eae6ef31a72124390c6c3b9dc58b9cb83b5171321842da3e78c939836919a1
+size 13617

checkpoint-1200/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a81ad569a08b812a8c2c69f31db173295038be7ba83b72561fb876fd57769fc
+size 627

checkpoint-1200/trainer_state.json

@@ -0,0 +1,1006 @@
+{
+  "best_metric": 0.6381103992462158,
+  "best_model_checkpoint": "./vit-base-beans/checkpoint-1200",
+  "epoch": 3.076923076923077,
+  "global_step": 1200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019935897435897437,
+      "loss": 1.7678,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 1.7014,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001980769230769231,
+      "loss": 1.4957,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 1.5043,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.44556677890011226,
+      "eval_loss": 1.493240237236023,
+      "eval_runtime": 59.3866,
+      "eval_samples_per_second": 60.081,
+      "eval_steps_per_second": 7.51,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019679487179487178,
+      "loss": 1.3931,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 1.3029,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001955128205128205,
+      "loss": 1.1753,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 1.3894,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.5297418630751964,
+      "eval_loss": 1.2430940866470337,
+      "eval_runtime": 38.1635,
+      "eval_samples_per_second": 93.492,
+      "eval_steps_per_second": 11.687,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019423076923076924,
+      "loss": 1.118,
+      "step": 90
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 1.1395,
+      "step": 100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019294871794871797,
+      "loss": 1.0946,
+      "step": 110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 1.208,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.6464646464646465,
+      "eval_loss": 0.9881709218025208,
+      "eval_runtime": 34.6718,
+      "eval_samples_per_second": 102.908,
+      "eval_steps_per_second": 12.863,
+      "step": 120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 0.9775,
+      "step": 130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 1.0216,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019038461538461538,
+      "loss": 1.1942,
+      "step": 150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.9847,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.6599326599326599,
+      "eval_loss": 0.950343132019043,
+      "eval_runtime": 40.3536,
+      "eval_samples_per_second": 88.418,
+      "eval_steps_per_second": 11.052,
+      "step": 160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001891025641025641,
+      "loss": 0.9707,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 1.1941,
+      "step": 180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018782051282051283,
+      "loss": 1.0309,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 1.102,
+      "step": 200
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 0.6374859708193041,
+      "eval_loss": 0.9868837594985962,
+      "eval_runtime": 32.1858,
+      "eval_samples_per_second": 110.856,
+      "eval_steps_per_second": 13.857,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018653846153846154,
+      "loss": 0.9899,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8926,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018525641025641027,
+      "loss": 0.9552,
+      "step": 230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.9109,
+      "step": 240
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.6380471380471381,
+      "eval_loss": 0.9887320399284363,
+      "eval_runtime": 32.5312,
+      "eval_samples_per_second": 109.679,
+      "eval_steps_per_second": 13.71,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018397435897435897,
+      "loss": 0.832,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.9323,
+      "step": 260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001826923076923077,
+      "loss": 1.0061,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.9627,
+      "step": 280
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.6156004489337823,
+      "eval_loss": 1.057395100593567,
+      "eval_runtime": 34.1152,
+      "eval_samples_per_second": 104.587,
+      "eval_steps_per_second": 13.073,
+      "step": 280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018141025641025643,
+      "loss": 0.9613,
+      "step": 290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.8986,
+      "step": 300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018012820512820513,
+      "loss": 0.9344,
+      "step": 310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.7031,
+      "step": 320
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.7087542087542088,
+      "eval_loss": 0.813530683517456,
+      "eval_runtime": 39.1342,
+      "eval_samples_per_second": 91.173,
+      "eval_steps_per_second": 11.397,
+      "step": 320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017884615384615386,
+      "loss": 0.7596,
+      "step": 330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.7207,
+      "step": 340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017756410256410257,
+      "loss": 0.8904,
+      "step": 350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.7605,
+      "step": 360
+    },
+    {
+      "epoch": 0.92,
+      "eval_accuracy": 0.7272727272727273,
+      "eval_loss": 0.7620912790298462,
+      "eval_runtime": 33.0054,
+      "eval_samples_per_second": 108.103,
+      "eval_steps_per_second": 13.513,
+      "step": 360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001762820512820513,
+      "loss": 0.8093,
+      "step": 370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.8198,
+      "step": 380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000175,
+      "loss": 0.9515,
+      "step": 390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.8467,
+      "step": 400
+    },
+    {
+      "epoch": 1.03,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7974632978439331,
+      "eval_runtime": 33.3512,
+      "eval_samples_per_second": 106.983,
+      "eval_steps_per_second": 13.373,
+      "step": 400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017371794871794873,
+      "loss": 0.5829,
+      "step": 410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.5576,
+      "step": 420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017243589743589746,
+      "loss": 0.6035,
+      "step": 430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.6653,
+      "step": 440
+    },
+    {
+      "epoch": 1.13,
+      "eval_accuracy": 0.7317620650953984,
+      "eval_loss": 0.7574185132980347,
+      "eval_runtime": 32.4855,
+      "eval_samples_per_second": 109.834,
+      "eval_steps_per_second": 13.729,
+      "step": 440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017115384615384616,
+      "loss": 0.5067,
+      "step": 450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.4973,
+      "step": 460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00016987179487179486,
+      "loss": 0.5721,
+      "step": 470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.5467,
+      "step": 480
+    },
+    {
+      "epoch": 1.23,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7471520900726318,
+      "eval_runtime": 32.6752,
+      "eval_samples_per_second": 109.196,
+      "eval_steps_per_second": 13.65,
+      "step": 480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001685897435897436,
+      "loss": 0.5494,
+      "step": 490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.6158,
+      "step": 500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016730769230769232,
+      "loss": 0.5412,
+      "step": 510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.388,
+      "step": 520
+    },
+    {
+      "epoch": 1.33,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 0.6508304476737976,
+      "eval_runtime": 36.0929,
+      "eval_samples_per_second": 98.856,
+      "eval_steps_per_second": 12.357,
+      "step": 520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016602564102564105,
+      "loss": 0.4511,
+      "step": 530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.566,
+      "step": 540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016474358974358976,
+      "loss": 0.4476,
+      "step": 550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.4699,
+      "step": 560
+    },
+    {
+      "epoch": 1.44,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.773759663105011,
+      "eval_runtime": 31.7272,
+      "eval_samples_per_second": 112.459,
+      "eval_steps_per_second": 14.057,
+      "step": 560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 0.402,
+      "step": 570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.425,
+      "step": 580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001621794871794872,
+      "loss": 0.5171,
+      "step": 590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.5344,
+      "step": 600
+    },
+    {
+      "epoch": 1.54,
+      "eval_accuracy": 0.7328843995510662,
+      "eval_loss": 0.7624912858009338,
+      "eval_runtime": 32.007,
+      "eval_samples_per_second": 111.476,
+      "eval_steps_per_second": 13.934,
+      "step": 600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016089743589743592,
+      "loss": 0.4913,
+      "step": 610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.5042,
+      "step": 620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015961538461538462,
+      "loss": 0.6051,
+      "step": 630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.5753,
+      "step": 640
+    },
+    {
+      "epoch": 1.64,
+      "eval_accuracy": 0.7598204264870931,
+      "eval_loss": 0.6743783950805664,
+      "eval_runtime": 30.705,
+      "eval_samples_per_second": 116.203,
+      "eval_steps_per_second": 14.525,
+      "step": 640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 0.524,
+      "step": 650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.5007,
+      "step": 660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015705128205128205,
+      "loss": 0.57,
+      "step": 670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.5533,
+      "step": 680
+    },
+    {
+      "epoch": 1.74,
+      "eval_accuracy": 0.734006734006734,
+      "eval_loss": 0.7230738997459412,
+      "eval_runtime": 33.3034,
+      "eval_samples_per_second": 107.136,
+      "eval_steps_per_second": 13.392,
+      "step": 680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015576923076923078,
+      "loss": 0.4399,
+      "step": 690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.453,
+      "step": 700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015448717948717951,
+      "loss": 0.4504,
+      "step": 710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.4244,
+      "step": 720
+    },
+    {
+      "epoch": 1.85,
+      "eval_accuracy": 0.7687991021324355,
+      "eval_loss": 0.6673141717910767,
+      "eval_runtime": 32.6948,
+      "eval_samples_per_second": 109.131,
+      "eval_steps_per_second": 13.641,
+      "step": 720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00015320512820512822,
+      "loss": 0.46,
+      "step": 730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.4132,
+      "step": 740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015192307692307692,
+      "loss": 0.4849,
+      "step": 750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.5423,
+      "step": 760
+    },
+    {
+      "epoch": 1.95,
+      "eval_accuracy": 0.7413019079685746,
+      "eval_loss": 0.7185856699943542,
+      "eval_runtime": 35.9539,
+      "eval_samples_per_second": 99.238,
+      "eval_steps_per_second": 12.405,
+      "step": 760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015064102564102565,
+      "loss": 0.3997,
+      "step": 770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4744,
+      "step": 780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00014935897435897438,
+      "loss": 0.4717,
+      "step": 790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.3384,
+      "step": 800
+    },
+    {
+      "epoch": 2.05,
+      "eval_accuracy": 0.7351290684624018,
+      "eval_loss": 0.7269920706748962,
+      "eval_runtime": 33.5897,
+      "eval_samples_per_second": 106.223,
+      "eval_steps_per_second": 13.278,
+      "step": 800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00014807692307692308,
+      "loss": 0.2602,
+      "step": 810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.2345,
+      "step": 820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014679487179487178,
+      "loss": 0.2135,
+      "step": 830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.2797,
+      "step": 840
+    },
+    {
+      "epoch": 2.15,
+      "eval_accuracy": 0.7497194163860831,
+      "eval_loss": 0.7742622494697571,
+      "eval_runtime": 31.2996,
+      "eval_samples_per_second": 113.995,
+      "eval_steps_per_second": 14.249,
+      "step": 840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014551282051282051,
+      "loss": 0.1862,
+      "step": 850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.2334,
+      "step": 860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014423076923076924,
+      "loss": 0.3361,
+      "step": 870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.2939,
+      "step": 880
+    },
+    {
+      "epoch": 2.26,
+      "eval_accuracy": 0.7732884399551067,
+      "eval_loss": 0.6985616087913513,
+      "eval_runtime": 32.9047,
+      "eval_samples_per_second": 108.435,
+      "eval_steps_per_second": 13.554,
+      "step": 880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014294871794871795,
+      "loss": 0.2846,
+      "step": 890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.2555,
+      "step": 900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 0.2629,
+      "step": 910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.2288,
+      "step": 920
+    },
+    {
+      "epoch": 2.36,
+      "eval_accuracy": 0.7480359147025814,
+      "eval_loss": 0.761025607585907,
+      "eval_runtime": 34.1426,
+      "eval_samples_per_second": 104.503,
+      "eval_steps_per_second": 13.063,
+      "step": 920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00014038461538461538,
+      "loss": 0.2115,
+      "step": 930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.2036,
+      "step": 940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001391025641025641,
+      "loss": 0.2913,
+      "step": 950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.2204,
+      "step": 960
+    },
+    {
+      "epoch": 2.46,
+      "eval_accuracy": 0.7570145903479237,
+      "eval_loss": 0.7840890884399414,
+      "eval_runtime": 35.7983,
+      "eval_samples_per_second": 99.67,
+      "eval_steps_per_second": 12.459,
+      "step": 960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013782051282051284,
+      "loss": 0.2759,
+      "step": 970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.3945,
+      "step": 980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00013653846153846154,
+      "loss": 0.3639,
+      "step": 990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.4397,
+      "step": 1000
+    },
+    {
+      "epoch": 2.56,
+      "eval_accuracy": 0.7789001122334456,
+      "eval_loss": 0.6565700173377991,
+      "eval_runtime": 31.6553,
+      "eval_samples_per_second": 112.714,
+      "eval_steps_per_second": 14.089,
+      "step": 1000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00013525641025641027,
+      "loss": 0.2707,
+      "step": 1010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.3451,
+      "step": 1020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013397435897435897,
+      "loss": 0.3277,
+      "step": 1030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.2219,
+      "step": 1040
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.6900522112846375,
+      "eval_runtime": 31.7845,
+      "eval_samples_per_second": 112.256,
+      "eval_steps_per_second": 14.032,
+      "step": 1040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001326923076923077,
+      "loss": 0.3174,
+      "step": 1050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.189,
+      "step": 1060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001314102564102564,
+      "loss": 0.2619,
+      "step": 1070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.2297,
+      "step": 1080
+    },
+    {
+      "epoch": 2.77,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.7095093131065369,
+      "eval_runtime": 38.4561,
+      "eval_samples_per_second": 92.781,
+      "eval_steps_per_second": 11.598,
+      "step": 1080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00013012820512820514,
+      "loss": 0.2201,
+      "step": 1090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.2261,
+      "step": 1100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00012884615384615387,
+      "loss": 0.2095,
+      "step": 1110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.209,
+      "step": 1120
+    },
+    {
+      "epoch": 2.87,
+      "eval_accuracy": 0.7738496071829405,
+      "eval_loss": 0.7127913236618042,
+      "eval_runtime": 32.9997,
+      "eval_samples_per_second": 108.122,
+      "eval_steps_per_second": 13.515,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00012756410256410257,
+      "loss": 0.3365,
+      "step": 1130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.2639,
+      "step": 1140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00012628205128205127,
+      "loss": 0.1892,
+      "step": 1150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.1853,
+      "step": 1160
+    },
+    {
+      "epoch": 2.97,
+      "eval_accuracy": 0.7710437710437711,
+      "eval_loss": 0.6986069083213806,
+      "eval_runtime": 32.2598,
+      "eval_samples_per_second": 110.602,
+      "eval_steps_per_second": 13.825,
+      "step": 1160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000125,
+      "loss": 0.1807,
+      "step": 1170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.1554,
+      "step": 1180
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00012371794871794873,
+      "loss": 0.1845,
+      "step": 1190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.1322,
+      "step": 1200
+    },
+    {
+      "epoch": 3.08,
+      "eval_accuracy": 0.7962962962962963,
+      "eval_loss": 0.6381103992462158,
+      "eval_runtime": 35.1257,
+      "eval_samples_per_second": 101.578,
+      "eval_steps_per_second": 12.697,
+      "step": 1200
+    }
+  ],
+  "max_steps": 3120,
+  "num_train_epochs": 8,
+  "total_flos": 1.4858071822502093e+18,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1200/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38c25b2b280e8f8dde2511a004bf1b097d7641bf1aae99dd53711356509cd78e
+size 3899

checkpoint-3120/config.json

@@ -0,0 +1,40 @@
+{
+  "_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": "Brazil",
+    "1": "Canada",
+    "2": "Finland",
+    "3": "Japan",
+    "4": "United-Kingdom",
+    "5": "United_States"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "Brazil": "0",
+    "Canada": "1",
+    "Finland": "2",
+    "Japan": "3",
+    "United-Kingdom": "4",
+    "United_States": "5"
+  },
+  "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": "4.31.0"
+}

checkpoint-3120/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4a8f329c33f0b869453a941fbff52267fa7967b9aeeff87f529f98ab308f8f8
+size 686537861

checkpoint-3120/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": "ViTFeatureExtractor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 224,
+    "width": 224
+  }
+}

checkpoint-3120/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f7265735d9427d338f6934c4ac7bc62a2c11b9c0842576be5b8f6032cfa8508
+size 343278253

checkpoint-3120/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5603187b2e7109f06291eba7b2b6fdeed2976f8fc03c3ec1a91df1a163133bf0
+size 13617

checkpoint-3120/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac683e6fa1db5f0ebc121b7681a83c3a8a325b59f341f4ccfd5c5bedb8cb70f3
+size 627

checkpoint-3120/trainer_state.json

@@ -0,0 +1,2590 @@
+{
+  "best_metric": 0.6381103992462158,
+  "best_model_checkpoint": "./vit-base-beans/checkpoint-1200",
+  "epoch": 8.0,
+  "global_step": 3120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019935897435897437,
+      "loss": 1.7678,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 1.7014,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001980769230769231,
+      "loss": 1.4957,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 1.5043,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.44556677890011226,
+      "eval_loss": 1.493240237236023,
+      "eval_runtime": 59.3866,
+      "eval_samples_per_second": 60.081,
+      "eval_steps_per_second": 7.51,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019679487179487178,
+      "loss": 1.3931,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 1.3029,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001955128205128205,
+      "loss": 1.1753,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 1.3894,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.5297418630751964,
+      "eval_loss": 1.2430940866470337,
+      "eval_runtime": 38.1635,
+      "eval_samples_per_second": 93.492,
+      "eval_steps_per_second": 11.687,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019423076923076924,
+      "loss": 1.118,
+      "step": 90
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 1.1395,
+      "step": 100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019294871794871797,
+      "loss": 1.0946,
+      "step": 110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 1.208,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.6464646464646465,
+      "eval_loss": 0.9881709218025208,
+      "eval_runtime": 34.6718,
+      "eval_samples_per_second": 102.908,
+      "eval_steps_per_second": 12.863,
+      "step": 120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 0.9775,
+      "step": 130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 1.0216,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019038461538461538,
+      "loss": 1.1942,
+      "step": 150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.9847,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.6599326599326599,
+      "eval_loss": 0.950343132019043,
+      "eval_runtime": 40.3536,
+      "eval_samples_per_second": 88.418,
+      "eval_steps_per_second": 11.052,
+      "step": 160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001891025641025641,
+      "loss": 0.9707,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 1.1941,
+      "step": 180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018782051282051283,
+      "loss": 1.0309,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 1.102,
+      "step": 200
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 0.6374859708193041,
+      "eval_loss": 0.9868837594985962,
+      "eval_runtime": 32.1858,
+      "eval_samples_per_second": 110.856,
+      "eval_steps_per_second": 13.857,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018653846153846154,
+      "loss": 0.9899,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8926,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018525641025641027,
+      "loss": 0.9552,
+      "step": 230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.9109,
+      "step": 240
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.6380471380471381,
+      "eval_loss": 0.9887320399284363,
+      "eval_runtime": 32.5312,
+      "eval_samples_per_second": 109.679,
+      "eval_steps_per_second": 13.71,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018397435897435897,
+      "loss": 0.832,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.9323,
+      "step": 260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001826923076923077,
+      "loss": 1.0061,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.9627,
+      "step": 280
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.6156004489337823,
+      "eval_loss": 1.057395100593567,
+      "eval_runtime": 34.1152,
+      "eval_samples_per_second": 104.587,
+      "eval_steps_per_second": 13.073,
+      "step": 280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018141025641025643,
+      "loss": 0.9613,
+      "step": 290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.8986,
+      "step": 300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018012820512820513,
+      "loss": 0.9344,
+      "step": 310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.7031,
+      "step": 320
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.7087542087542088,
+      "eval_loss": 0.813530683517456,
+      "eval_runtime": 39.1342,
+      "eval_samples_per_second": 91.173,
+      "eval_steps_per_second": 11.397,
+      "step": 320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017884615384615386,
+      "loss": 0.7596,
+      "step": 330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.7207,
+      "step": 340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017756410256410257,
+      "loss": 0.8904,
+      "step": 350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.7605,
+      "step": 360
+    },
+    {
+      "epoch": 0.92,
+      "eval_accuracy": 0.7272727272727273,
+      "eval_loss": 0.7620912790298462,
+      "eval_runtime": 33.0054,
+      "eval_samples_per_second": 108.103,
+      "eval_steps_per_second": 13.513,
+      "step": 360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001762820512820513,
+      "loss": 0.8093,
+      "step": 370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.8198,
+      "step": 380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000175,
+      "loss": 0.9515,
+      "step": 390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.8467,
+      "step": 400
+    },
+    {
+      "epoch": 1.03,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7974632978439331,
+      "eval_runtime": 33.3512,
+      "eval_samples_per_second": 106.983,
+      "eval_steps_per_second": 13.373,
+      "step": 400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017371794871794873,
+      "loss": 0.5829,
+      "step": 410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.5576,
+      "step": 420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017243589743589746,
+      "loss": 0.6035,
+      "step": 430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.6653,
+      "step": 440
+    },
+    {
+      "epoch": 1.13,
+      "eval_accuracy": 0.7317620650953984,
+      "eval_loss": 0.7574185132980347,
+      "eval_runtime": 32.4855,
+      "eval_samples_per_second": 109.834,
+      "eval_steps_per_second": 13.729,
+      "step": 440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017115384615384616,
+      "loss": 0.5067,
+      "step": 450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.4973,
+      "step": 460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00016987179487179486,
+      "loss": 0.5721,
+      "step": 470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.5467,
+      "step": 480
+    },
+    {
+      "epoch": 1.23,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7471520900726318,
+      "eval_runtime": 32.6752,
+      "eval_samples_per_second": 109.196,
+      "eval_steps_per_second": 13.65,
+      "step": 480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001685897435897436,
+      "loss": 0.5494,
+      "step": 490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.6158,
+      "step": 500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016730769230769232,
+      "loss": 0.5412,
+      "step": 510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.388,
+      "step": 520
+    },
+    {
+      "epoch": 1.33,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 0.6508304476737976,
+      "eval_runtime": 36.0929,
+      "eval_samples_per_second": 98.856,
+      "eval_steps_per_second": 12.357,
+      "step": 520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016602564102564105,
+      "loss": 0.4511,
+      "step": 530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.566,
+      "step": 540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016474358974358976,
+      "loss": 0.4476,
+      "step": 550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.4699,
+      "step": 560
+    },
+    {
+      "epoch": 1.44,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.773759663105011,
+      "eval_runtime": 31.7272,
+      "eval_samples_per_second": 112.459,
+      "eval_steps_per_second": 14.057,
+      "step": 560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 0.402,
+      "step": 570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.425,
+      "step": 580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001621794871794872,
+      "loss": 0.5171,
+      "step": 590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.5344,
+      "step": 600
+    },
+    {
+      "epoch": 1.54,
+      "eval_accuracy": 0.7328843995510662,
+      "eval_loss": 0.7624912858009338,
+      "eval_runtime": 32.007,
+      "eval_samples_per_second": 111.476,
+      "eval_steps_per_second": 13.934,
+      "step": 600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016089743589743592,
+      "loss": 0.4913,
+      "step": 610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.5042,
+      "step": 620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015961538461538462,
+      "loss": 0.6051,
+      "step": 630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.5753,
+      "step": 640
+    },
+    {
+      "epoch": 1.64,
+      "eval_accuracy": 0.7598204264870931,
+      "eval_loss": 0.6743783950805664,
+      "eval_runtime": 30.705,
+      "eval_samples_per_second": 116.203,
+      "eval_steps_per_second": 14.525,
+      "step": 640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 0.524,
+      "step": 650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.5007,
+      "step": 660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015705128205128205,
+      "loss": 0.57,
+      "step": 670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.5533,
+      "step": 680
+    },
+    {
+      "epoch": 1.74,
+      "eval_accuracy": 0.734006734006734,
+      "eval_loss": 0.7230738997459412,
+      "eval_runtime": 33.3034,
+      "eval_samples_per_second": 107.136,
+      "eval_steps_per_second": 13.392,
+      "step": 680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015576923076923078,
+      "loss": 0.4399,
+      "step": 690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.453,
+      "step": 700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015448717948717951,
+      "loss": 0.4504,
+      "step": 710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.4244,
+      "step": 720
+    },
+    {
+      "epoch": 1.85,
+      "eval_accuracy": 0.7687991021324355,
+      "eval_loss": 0.6673141717910767,
+      "eval_runtime": 32.6948,
+      "eval_samples_per_second": 109.131,
+      "eval_steps_per_second": 13.641,
+      "step": 720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00015320512820512822,
+      "loss": 0.46,
+      "step": 730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.4132,
+      "step": 740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015192307692307692,
+      "loss": 0.4849,
+      "step": 750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.5423,
+      "step": 760
+    },
+    {
+      "epoch": 1.95,
+      "eval_accuracy": 0.7413019079685746,
+      "eval_loss": 0.7185856699943542,
+      "eval_runtime": 35.9539,
+      "eval_samples_per_second": 99.238,
+      "eval_steps_per_second": 12.405,
+      "step": 760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015064102564102565,
+      "loss": 0.3997,
+      "step": 770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4744,
+      "step": 780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00014935897435897438,
+      "loss": 0.4717,
+      "step": 790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.3384,
+      "step": 800
+    },
+    {
+      "epoch": 2.05,
+      "eval_accuracy": 0.7351290684624018,
+      "eval_loss": 0.7269920706748962,
+      "eval_runtime": 33.5897,
+      "eval_samples_per_second": 106.223,
+      "eval_steps_per_second": 13.278,
+      "step": 800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00014807692307692308,
+      "loss": 0.2602,
+      "step": 810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.2345,
+      "step": 820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014679487179487178,
+      "loss": 0.2135,
+      "step": 830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.2797,
+      "step": 840
+    },
+    {
+      "epoch": 2.15,
+      "eval_accuracy": 0.7497194163860831,
+      "eval_loss": 0.7742622494697571,
+      "eval_runtime": 31.2996,
+      "eval_samples_per_second": 113.995,
+      "eval_steps_per_second": 14.249,
+      "step": 840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014551282051282051,
+      "loss": 0.1862,
+      "step": 850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.2334,
+      "step": 860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014423076923076924,
+      "loss": 0.3361,
+      "step": 870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.2939,
+      "step": 880
+    },
+    {
+      "epoch": 2.26,
+      "eval_accuracy": 0.7732884399551067,
+      "eval_loss": 0.6985616087913513,
+      "eval_runtime": 32.9047,
+      "eval_samples_per_second": 108.435,
+      "eval_steps_per_second": 13.554,
+      "step": 880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014294871794871795,
+      "loss": 0.2846,
+      "step": 890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.2555,
+      "step": 900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 0.2629,
+      "step": 910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.2288,
+      "step": 920
+    },
+    {
+      "epoch": 2.36,
+      "eval_accuracy": 0.7480359147025814,
+      "eval_loss": 0.761025607585907,
+      "eval_runtime": 34.1426,
+      "eval_samples_per_second": 104.503,
+      "eval_steps_per_second": 13.063,
+      "step": 920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00014038461538461538,
+      "loss": 0.2115,
+      "step": 930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.2036,
+      "step": 940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001391025641025641,
+      "loss": 0.2913,
+      "step": 950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.2204,
+      "step": 960
+    },
+    {
+      "epoch": 2.46,
+      "eval_accuracy": 0.7570145903479237,
+      "eval_loss": 0.7840890884399414,
+      "eval_runtime": 35.7983,
+      "eval_samples_per_second": 99.67,
+      "eval_steps_per_second": 12.459,
+      "step": 960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013782051282051284,
+      "loss": 0.2759,
+      "step": 970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.3945,
+      "step": 980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00013653846153846154,
+      "loss": 0.3639,
+      "step": 990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.4397,
+      "step": 1000
+    },
+    {
+      "epoch": 2.56,
+      "eval_accuracy": 0.7789001122334456,
+      "eval_loss": 0.6565700173377991,
+      "eval_runtime": 31.6553,
+      "eval_samples_per_second": 112.714,
+      "eval_steps_per_second": 14.089,
+      "step": 1000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00013525641025641027,
+      "loss": 0.2707,
+      "step": 1010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.3451,
+      "step": 1020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013397435897435897,
+      "loss": 0.3277,
+      "step": 1030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.2219,
+      "step": 1040
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.6900522112846375,
+      "eval_runtime": 31.7845,
+      "eval_samples_per_second": 112.256,
+      "eval_steps_per_second": 14.032,
+      "step": 1040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001326923076923077,
+      "loss": 0.3174,
+      "step": 1050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.189,
+      "step": 1060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001314102564102564,
+      "loss": 0.2619,
+      "step": 1070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.2297,
+      "step": 1080
+    },
+    {
+      "epoch": 2.77,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.7095093131065369,
+      "eval_runtime": 38.4561,
+      "eval_samples_per_second": 92.781,
+      "eval_steps_per_second": 11.598,
+      "step": 1080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00013012820512820514,
+      "loss": 0.2201,
+      "step": 1090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.2261,
+      "step": 1100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00012884615384615387,
+      "loss": 0.2095,
+      "step": 1110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.209,
+      "step": 1120
+    },
+    {
+      "epoch": 2.87,
+      "eval_accuracy": 0.7738496071829405,
+      "eval_loss": 0.7127913236618042,
+      "eval_runtime": 32.9997,
+      "eval_samples_per_second": 108.122,
+      "eval_steps_per_second": 13.515,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00012756410256410257,
+      "loss": 0.3365,
+      "step": 1130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.2639,
+      "step": 1140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00012628205128205127,
+      "loss": 0.1892,
+      "step": 1150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.1853,
+      "step": 1160
+    },
+    {
+      "epoch": 2.97,
+      "eval_accuracy": 0.7710437710437711,
+      "eval_loss": 0.6986069083213806,
+      "eval_runtime": 32.2598,
+      "eval_samples_per_second": 110.602,
+      "eval_steps_per_second": 13.825,
+      "step": 1160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000125,
+      "loss": 0.1807,
+      "step": 1170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.1554,
+      "step": 1180
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00012371794871794873,
+      "loss": 0.1845,
+      "step": 1190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.1322,
+      "step": 1200
+    },
+    {
+      "epoch": 3.08,
+      "eval_accuracy": 0.7962962962962963,
+      "eval_loss": 0.6381103992462158,
+      "eval_runtime": 35.1257,
+      "eval_samples_per_second": 101.578,
+      "eval_steps_per_second": 12.697,
+      "step": 1200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00012243589743589746,
+      "loss": 0.0598,
+      "step": 1210
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.0853,
+      "step": 1220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00012115384615384615,
+      "loss": 0.1616,
+      "step": 1230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.2603,
+      "step": 1240
+    },
+    {
+      "epoch": 3.18,
+      "eval_accuracy": 0.7682379349046016,
+      "eval_loss": 0.7860389351844788,
+      "eval_runtime": 35.648,
+      "eval_samples_per_second": 100.09,
+      "eval_steps_per_second": 12.511,
+      "step": 1240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00011987179487179487,
+      "loss": 0.1104,
+      "step": 1250
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.0634,
+      "step": 1260
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001185897435897436,
+      "loss": 0.0717,
+      "step": 1270
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.1031,
+      "step": 1280
+    },
+    {
+      "epoch": 3.28,
+      "eval_accuracy": 0.7895622895622896,
+      "eval_loss": 0.7322177886962891,
+      "eval_runtime": 30.6139,
+      "eval_samples_per_second": 116.549,
+      "eval_steps_per_second": 14.569,
+      "step": 1280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00011730769230769231,
+      "loss": 0.054,
+      "step": 1290
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.0883,
+      "step": 1300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00011602564102564104,
+      "loss": 0.1443,
+      "step": 1310
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.0763,
+      "step": 1320
+    },
+    {
+      "epoch": 3.38,
+      "eval_accuracy": 0.7839506172839507,
+      "eval_loss": 0.77773118019104,
+      "eval_runtime": 31.4828,
+      "eval_samples_per_second": 113.332,
+      "eval_steps_per_second": 14.166,
+      "step": 1320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00011474358974358975,
+      "loss": 0.1328,
+      "step": 1330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.0751,
+      "step": 1340
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00011346153846153846,
+      "loss": 0.1193,
+      "step": 1350
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.1437,
+      "step": 1360
+    },
+    {
+      "epoch": 3.49,
+      "eval_accuracy": 0.7665544332210998,
+      "eval_loss": 0.9293356537818909,
+      "eval_runtime": 31.1681,
+      "eval_samples_per_second": 114.476,
+      "eval_steps_per_second": 14.31,
+      "step": 1360
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00011217948717948718,
+      "loss": 0.1805,
+      "step": 1370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.0931,
+      "step": 1380
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00011089743589743591,
+      "loss": 0.1227,
+      "step": 1390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.0818,
+      "step": 1400
+    },
+    {
+      "epoch": 3.59,
+      "eval_accuracy": 0.7940516273849607,
+      "eval_loss": 0.796922504901886,
+      "eval_runtime": 31.9044,
+      "eval_samples_per_second": 111.834,
+      "eval_steps_per_second": 13.979,
+      "step": 1400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00010961538461538463,
+      "loss": 0.1193,
+      "step": 1410
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.1237,
+      "step": 1420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00010833333333333333,
+      "loss": 0.1074,
+      "step": 1430
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.1355,
+      "step": 1440
+    },
+    {
+      "epoch": 3.69,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 0.8145824670791626,
+      "eval_runtime": 37.6519,
+      "eval_samples_per_second": 94.763,
+      "eval_steps_per_second": 11.845,
+      "step": 1440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00010705128205128206,
+      "loss": 0.0844,
+      "step": 1450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 0.1067,
+      "step": 1460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00010576923076923077,
+      "loss": 0.0342,
+      "step": 1470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.0802,
+      "step": 1480
+    },
+    {
+      "epoch": 3.79,
+      "eval_accuracy": 0.8075196408529742,
+      "eval_loss": 0.6976819038391113,
+      "eval_runtime": 32.2393,
+      "eval_samples_per_second": 110.672,
+      "eval_steps_per_second": 13.834,
+      "step": 1480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0001044871794871795,
+      "loss": 0.1143,
+      "step": 1490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.1332,
+      "step": 1500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00010320512820512822,
+      "loss": 0.069,
+      "step": 1510
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.032,
+      "step": 1520
+    },
+    {
+      "epoch": 3.9,
+      "eval_accuracy": 0.7929292929292929,
+      "eval_loss": 0.8349816203117371,
+      "eval_runtime": 32.9637,
+      "eval_samples_per_second": 108.24,
+      "eval_steps_per_second": 13.53,
+      "step": 1520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00010192307692307692,
+      "loss": 0.0722,
+      "step": 1530
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.1383,
+      "step": 1540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00010064102564102564,
+      "loss": 0.0861,
+      "step": 1550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1560
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7968574635241302,
+      "eval_loss": 0.775885820388794,
+      "eval_runtime": 33.0589,
+      "eval_samples_per_second": 107.928,
+      "eval_steps_per_second": 13.491,
+      "step": 1560
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.935897435897437e-05,
+      "loss": 0.0486,
+      "step": 1570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.0369,
+      "step": 1580
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.807692307692307e-05,
+      "loss": 0.0209,
+      "step": 1590
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.0391,
+      "step": 1600
+    },
+    {
+      "epoch": 4.1,
+      "eval_accuracy": 0.8013468013468014,
+      "eval_loss": 0.8052350282669067,
+      "eval_runtime": 32.0537,
+      "eval_samples_per_second": 111.313,
+      "eval_steps_per_second": 13.914,
+      "step": 1600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.67948717948718e-05,
+      "loss": 0.0354,
+      "step": 1610
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.0738,
+      "step": 1620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.551282051282052e-05,
+      "loss": 0.0275,
+      "step": 1630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.0406,
+      "step": 1640
+    },
+    {
+      "epoch": 4.21,
+      "eval_accuracy": 0.7828282828282829,
+      "eval_loss": 0.9251748919487,
+      "eval_runtime": 31.6082,
+      "eval_samples_per_second": 112.882,
+      "eval_steps_per_second": 14.11,
+      "step": 1640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.423076923076924e-05,
+      "loss": 0.1377,
+      "step": 1650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.0526,
+      "step": 1660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.294871794871795e-05,
+      "loss": 0.152,
+      "step": 1670
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.0488,
+      "step": 1680
+    },
+    {
+      "epoch": 4.31,
+      "eval_accuracy": 0.7901234567901234,
+      "eval_loss": 0.8203706741333008,
+      "eval_runtime": 31.8606,
+      "eval_samples_per_second": 111.988,
+      "eval_steps_per_second": 13.998,
+      "step": 1680
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.0372,
+      "step": 1690
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.0968,
+      "step": 1700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.038461538461538e-05,
+      "loss": 0.0147,
+      "step": 1710
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.016,
+      "step": 1720
+    },
+    {
+      "epoch": 4.41,
+      "eval_accuracy": 0.7794612794612794,
+      "eval_loss": 0.9169319868087769,
+      "eval_runtime": 33.724,
+      "eval_samples_per_second": 105.8,
+      "eval_steps_per_second": 13.225,
+      "step": 1720
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.910256410256411e-05,
+      "loss": 0.0393,
+      "step": 1730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.0258,
+      "step": 1740
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.782051282051283e-05,
+      "loss": 0.009,
+      "step": 1750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.0994,
+      "step": 1760
+    },
+    {
+      "epoch": 4.51,
+      "eval_accuracy": 0.8120089786756454,
+      "eval_loss": 0.7930631637573242,
+      "eval_runtime": 34.0477,
+      "eval_samples_per_second": 104.794,
+      "eval_steps_per_second": 13.099,
+      "step": 1760
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.653846153846155e-05,
+      "loss": 0.0418,
+      "step": 1770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.0793,
+      "step": 1780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.525641025641026e-05,
+      "loss": 0.0122,
+      "step": 1790
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.07,
+      "step": 1800
+    },
+    {
+      "epoch": 4.62,
+      "eval_accuracy": 0.8103254769921436,
+      "eval_loss": 0.8245030045509338,
+      "eval_runtime": 32.6583,
+      "eval_samples_per_second": 109.252,
+      "eval_steps_per_second": 13.657,
+      "step": 1800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.397435897435898e-05,
+      "loss": 0.0656,
+      "step": 1810
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.0364,
+      "step": 1820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.26923076923077e-05,
+      "loss": 0.0639,
+      "step": 1830
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 0.0088,
+      "step": 1840
+    },
+    {
+      "epoch": 4.72,
+      "eval_accuracy": 0.7985409652076318,
+      "eval_loss": 0.907173752784729,
+      "eval_runtime": 40.0812,
+      "eval_samples_per_second": 89.019,
+      "eval_steps_per_second": 11.127,
+      "step": 1840
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.141025641025641e-05,
+      "loss": 0.0557,
+      "step": 1850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.0548,
+      "step": 1860
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.012820512820514e-05,
+      "loss": 0.0055,
+      "step": 1870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.0085,
+      "step": 1880
+    },
+    {
+      "epoch": 4.82,
+      "eval_accuracy": 0.8114478114478114,
+      "eval_loss": 0.7862226963043213,
+      "eval_runtime": 33.9761,
+      "eval_samples_per_second": 105.015,
+      "eval_steps_per_second": 13.127,
+      "step": 1880
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.884615384615384e-05,
+      "loss": 0.0423,
+      "step": 1890
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 0.0581,
+      "step": 1900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.756410256410257e-05,
+      "loss": 0.0043,
+      "step": 1910
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.083,
+      "step": 1920
+    },
+    {
+      "epoch": 4.92,
+      "eval_accuracy": 0.8198653198653199,
+      "eval_loss": 0.7796981334686279,
+      "eval_runtime": 32.3864,
+      "eval_samples_per_second": 110.17,
+      "eval_steps_per_second": 13.771,
+      "step": 1920
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.628205128205128e-05,
+      "loss": 0.1308,
+      "step": 1930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.0063,
+      "step": 1940
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.0132,
+      "step": 1950
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.0055,
+      "step": 1960
+    },
+    {
+      "epoch": 5.03,
+      "eval_accuracy": 0.8069584736251403,
+      "eval_loss": 0.8723996877670288,
+      "eval_runtime": 33.6185,
+      "eval_samples_per_second": 106.132,
+      "eval_steps_per_second": 13.267,
+      "step": 1960
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.371794871794872e-05,
+      "loss": 0.0366,
+      "step": 1970
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.06,
+      "step": 1980
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.243589743589744e-05,
+      "loss": 0.0065,
+      "step": 1990
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.0223,
+      "step": 2000
+    },
+    {
+      "epoch": 5.13,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.7518730163574219,
+      "eval_runtime": 43.1922,
+      "eval_samples_per_second": 82.607,
+      "eval_steps_per_second": 10.326,
+      "step": 2000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.115384615384616e-05,
+      "loss": 0.0215,
+      "step": 2010
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 0.0494,
+      "step": 2020
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.987179487179487e-05,
+      "loss": 0.0441,
+      "step": 2030
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.0042,
+      "step": 2040
+    },
+    {
+      "epoch": 5.23,
+      "eval_accuracy": 0.8232323232323232,
+      "eval_loss": 0.7583897113800049,
+      "eval_runtime": 31.893,
+      "eval_samples_per_second": 111.874,
+      "eval_steps_per_second": 13.984,
+      "step": 2040
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.858974358974359e-05,
+      "loss": 0.0131,
+      "step": 2050
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 0.0037,
+      "step": 2060
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.730769230769232e-05,
+      "loss": 0.02,
+      "step": 2070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.0178,
+      "step": 2080
+    },
+    {
+      "epoch": 5.33,
+      "eval_accuracy": 0.8080808080808081,
+      "eval_loss": 0.8523600697517395,
+      "eval_runtime": 32.5102,
+      "eval_samples_per_second": 109.75,
+      "eval_steps_per_second": 13.719,
+      "step": 2080
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.602564102564102e-05,
+      "loss": 0.0035,
+      "step": 2090
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.0137,
+      "step": 2100
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.474358974358975e-05,
+      "loss": 0.0033,
+      "step": 2110
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.0172,
+      "step": 2120
+    },
+    {
+      "epoch": 5.44,
+      "eval_accuracy": 0.8215488215488216,
+      "eval_loss": 0.7729219794273376,
+      "eval_runtime": 31.7305,
+      "eval_samples_per_second": 112.447,
+      "eval_steps_per_second": 14.056,
+      "step": 2120
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.346153846153847e-05,
+      "loss": 0.0252,
+      "step": 2130
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.0034,
+      "step": 2140
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.217948717948718e-05,
+      "loss": 0.0045,
+      "step": 2150
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.0044,
+      "step": 2160
+    },
+    {
+      "epoch": 5.54,
+      "eval_accuracy": 0.808641975308642,
+      "eval_loss": 0.8701183795928955,
+      "eval_runtime": 31.0299,
+      "eval_samples_per_second": 114.986,
+      "eval_steps_per_second": 14.373,
+      "step": 2160
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.089743589743589e-05,
+      "loss": 0.0373,
+      "step": 2170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.0029,
+      "step": 2180
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.9615384615384616e-05,
+      "loss": 0.0032,
+      "step": 2190
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.003,
+      "step": 2200
+    },
+    {
+      "epoch": 5.64,
+      "eval_accuracy": 0.8237934904601572,
+      "eval_loss": 0.8060529828071594,
+      "eval_runtime": 32.4421,
+      "eval_samples_per_second": 109.98,
+      "eval_steps_per_second": 13.748,
+      "step": 2200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.0034,
+      "step": 2210
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.0025,
+      "step": 2220
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.705128205128205e-05,
+      "loss": 0.0029,
+      "step": 2230
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.0033,
+      "step": 2240
+    },
+    {
+      "epoch": 5.74,
+      "eval_accuracy": 0.8226711560044894,
+      "eval_loss": 0.813855767250061,
+      "eval_runtime": 32.378,
+      "eval_samples_per_second": 110.198,
+      "eval_steps_per_second": 13.775,
+      "step": 2240
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.576923076923077e-05,
+      "loss": 0.0129,
+      "step": 2250
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 0.0079,
+      "step": 2260
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.448717948717948e-05,
+      "loss": 0.0117,
+      "step": 2270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.0023,
+      "step": 2280
+    },
+    {
+      "epoch": 5.85,
+      "eval_accuracy": 0.8164983164983165,
+      "eval_loss": 0.8478395342826843,
+      "eval_runtime": 35.321,
+      "eval_samples_per_second": 101.016,
+      "eval_steps_per_second": 12.627,
+      "step": 2280
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.3205128205128205e-05,
+      "loss": 0.0075,
+      "step": 2290
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 0.0023,
+      "step": 2300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.192307692307693e-05,
+      "loss": 0.0058,
+      "step": 2310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.003,
+      "step": 2320
+    },
+    {
+      "epoch": 5.95,
+      "eval_accuracy": 0.8120089786756454,
+      "eval_loss": 0.8443180918693542,
+      "eval_runtime": 32.2001,
+      "eval_samples_per_second": 110.807,
+      "eval_steps_per_second": 13.851,
+      "step": 2320
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.0641025641025644e-05,
+      "loss": 0.0512,
+      "step": 2330
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5e-05,
+      "loss": 0.0024,
+      "step": 2340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.935897435897436e-05,
+      "loss": 0.0022,
+      "step": 2350
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.031,
+      "step": 2360
+    },
+    {
+      "epoch": 6.05,
+      "eval_accuracy": 0.8047138047138047,
+      "eval_loss": 0.9272196888923645,
+      "eval_runtime": 33.2145,
+      "eval_samples_per_second": 107.423,
+      "eval_steps_per_second": 13.428,
+      "step": 2360
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 0.0456,
+      "step": 2370
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 0.0024,
+      "step": 2380
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.67948717948718e-05,
+      "loss": 0.0023,
+      "step": 2390
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.0021,
+      "step": 2400
+    },
+    {
+      "epoch": 6.15,
+      "eval_accuracy": 0.8204264870931538,
+      "eval_loss": 0.8369048237800598,
+      "eval_runtime": 36.1395,
+      "eval_samples_per_second": 98.728,
+      "eval_steps_per_second": 12.341,
+      "step": 2400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.5512820512820516e-05,
+      "loss": 0.0021,
+      "step": 2410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.0026,
+      "step": 2420
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.423076923076923e-05,
+      "loss": 0.0021,
+      "step": 2430
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.0019,
+      "step": 2440
+    },
+    {
+      "epoch": 6.26,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.828059196472168,
+      "eval_runtime": 32.1149,
+      "eval_samples_per_second": 111.101,
+      "eval_steps_per_second": 13.888,
+      "step": 2440
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.294871794871795e-05,
+      "loss": 0.0024,
+      "step": 2450
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.0019,
+      "step": 2460
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.002,
+      "step": 2470
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.0019,
+      "step": 2480
+    },
+    {
+      "epoch": 6.36,
+      "eval_accuracy": 0.8198653198653199,
+      "eval_loss": 0.834019124507904,
+      "eval_runtime": 35.8439,
+      "eval_samples_per_second": 99.543,
+      "eval_steps_per_second": 12.443,
+      "step": 2480
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.038461538461539e-05,
+      "loss": 0.002,
+      "step": 2490
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 0.0019,
+      "step": 2500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9102564102564105e-05,
+      "loss": 0.0018,
+      "step": 2510
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.0018,
+      "step": 2520
+    },
+    {
+      "epoch": 6.46,
+      "eval_accuracy": 0.8204264870931538,
+      "eval_loss": 0.8339148759841919,
+      "eval_runtime": 34.0288,
+      "eval_samples_per_second": 104.853,
+      "eval_steps_per_second": 13.107,
+      "step": 2520
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.782051282051282e-05,
+      "loss": 0.0018,
+      "step": 2530
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 0.0018,
+      "step": 2540
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.653846153846154e-05,
+      "loss": 0.0018,
+      "step": 2550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 0.002,
+      "step": 2560
+    },
+    {
+      "epoch": 6.56,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.8293877840042114,
+      "eval_runtime": 33.9846,
+      "eval_samples_per_second": 104.989,
+      "eval_steps_per_second": 13.124,
+      "step": 2560
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.525641025641026e-05,
+      "loss": 0.0017,
+      "step": 2570
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 0.0017,
+      "step": 2580
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.397435897435898e-05,
+      "loss": 0.0018,
+      "step": 2590
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0017,
+      "step": 2600
+    },
+    {
+      "epoch": 6.67,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.8266403079032898,
+      "eval_runtime": 36.0162,
+      "eval_samples_per_second": 99.067,
+      "eval_steps_per_second": 12.383,
+      "step": 2600
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.269230769230769e-05,
+      "loss": 0.002,
+      "step": 2610
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.0017,
+      "step": 2620
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.141025641025641e-05,
+      "loss": 0.0017,
+      "step": 2630
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.0017,
+      "step": 2640
+    },
+    {
+      "epoch": 6.77,
+      "eval_accuracy": 0.8226711560044894,
+      "eval_loss": 0.8177886009216309,
+      "eval_runtime": 36.3329,
+      "eval_samples_per_second": 98.203,
+      "eval_steps_per_second": 12.275,
+      "step": 2640
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.012820512820513e-05,
+      "loss": 0.0016,
+      "step": 2650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 0.0016,
+      "step": 2660
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 0.0016,
+      "step": 2670
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 0.0017,
+      "step": 2680
+    },
+    {
+      "epoch": 6.87,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8174560070037842,
+      "eval_runtime": 32.3063,
+      "eval_samples_per_second": 110.443,
+      "eval_steps_per_second": 13.805,
+      "step": 2680
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.756410256410257e-05,
+      "loss": 0.0017,
+      "step": 2690
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.0017,
+      "step": 2700
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6282051282051285e-05,
+      "loss": 0.0017,
+      "step": 2710
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 0.0017,
+      "step": 2720
+    },
+    {
+      "epoch": 6.97,
+      "eval_accuracy": 0.8237934904601572,
+      "eval_loss": 0.8166114687919617,
+      "eval_runtime": 33.2936,
+      "eval_samples_per_second": 107.168,
+      "eval_steps_per_second": 13.396,
+      "step": 2720
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 2730
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 0.0016,
+      "step": 2740
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.3717948717948718e-05,
+      "loss": 0.0016,
+      "step": 2750
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.0015,
+      "step": 2760
+    },
+    {
+      "epoch": 7.08,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8176218271255493,
+      "eval_runtime": 32.3011,
+      "eval_samples_per_second": 110.46,
+      "eval_steps_per_second": 13.808,
+      "step": 2760
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.2435897435897437e-05,
+      "loss": 0.0015,
+      "step": 2770
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 0.0015,
+      "step": 2780
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.1153846153846154e-05,
+      "loss": 0.0015,
+      "step": 2790
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.0015,
+      "step": 2800
+    },
+    {
+      "epoch": 7.18,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8185549378395081,
+      "eval_runtime": 32.7126,
+      "eval_samples_per_second": 109.071,
+      "eval_steps_per_second": 13.634,
+      "step": 2800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.987179487179487e-05,
+      "loss": 0.0015,
+      "step": 2810
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.0015,
+      "step": 2820
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.858974358974359e-05,
+      "loss": 0.0015,
+      "step": 2830
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.0015,
+      "step": 2840
+    },
+    {
+      "epoch": 7.28,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.819684624671936,
+      "eval_runtime": 35.8925,
+      "eval_samples_per_second": 99.408,
+      "eval_steps_per_second": 12.426,
+      "step": 2840
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.0016,
+      "step": 2850
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0015,
+      "step": 2860
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.0015,
+      "step": 2870
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.0015,
+      "step": 2880
+    },
+    {
+      "epoch": 7.38,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8204275965690613,
+      "eval_runtime": 31.8471,
+      "eval_samples_per_second": 112.035,
+      "eval_steps_per_second": 14.004,
+      "step": 2880
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.0015,
+      "step": 2890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 0.0015,
+      "step": 2900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3461538461538462e-05,
+      "loss": 0.0015,
+      "step": 2910
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 0.0015,
+      "step": 2920
+    },
+    {
+      "epoch": 7.49,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8213893175125122,
+      "eval_runtime": 30.8496,
+      "eval_samples_per_second": 115.658,
+      "eval_steps_per_second": 14.457,
+      "step": 2920
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.0015,
+      "step": 2930
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 0.0015,
+      "step": 2940
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.0015,
+      "step": 2950
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.0015,
+      "step": 2960
+    },
+    {
+      "epoch": 7.59,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8220140933990479,
+      "eval_runtime": 31.6074,
+      "eval_samples_per_second": 112.885,
+      "eval_steps_per_second": 14.111,
+      "step": 2960
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.0016,
+      "step": 2970
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.0014,
+      "step": 2980
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0015,
+      "step": 2990
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.0015,
+      "step": 3000
+    },
+    {
+      "epoch": 7.69,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8220102787017822,
+      "eval_runtime": 30.2325,
+      "eval_samples_per_second": 118.019,
+      "eval_steps_per_second": 14.752,
+      "step": 3000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.051282051282052e-06,
+      "loss": 0.0014,
+      "step": 3010
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 0.0014,
+      "step": 3020
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 0.0015,
+      "step": 3030
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.0015,
+      "step": 3040
+    },
+    {
+      "epoch": 7.79,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8217305541038513,
+      "eval_runtime": 38.9983,
+      "eval_samples_per_second": 91.491,
+      "eval_steps_per_second": 11.436,
+      "step": 3040
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 0.0014,
+      "step": 3050
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.0015,
+      "step": 3060
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.205128205128205e-06,
+      "loss": 0.0014,
+      "step": 3070
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.0014,
+      "step": 3080
+    },
+    {
+      "epoch": 7.9,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8224795460700989,
+      "eval_runtime": 31.9073,
+      "eval_samples_per_second": 111.824,
+      "eval_steps_per_second": 13.978,
+      "step": 3080
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.0014,
+      "step": 3090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.0015,
+      "step": 3100
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 0.0014,
+      "step": 3110
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0,
+      "loss": 0.0014,
+      "step": 3120
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.8260381593714927,
+      "eval_loss": 0.8224756717681885,
+      "eval_runtime": 30.7985,
+      "eval_samples_per_second": 115.85,
+      "eval_steps_per_second": 14.481,
+      "step": 3120
+    }
+  ],
+  "max_steps": 3120,
+  "num_train_epochs": 8,
+  "total_flos": 3.8629591832685773e+18,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3120/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38c25b2b280e8f8dde2511a004bf1b097d7641bf1aae99dd53711356509cd78e
+size 3899

config.json

@@ -0,0 +1,40 @@
+{
+  "_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": "Brazil",
+    "1": "Canada",
+    "2": "Finland",
+    "3": "Japan",
+    "4": "United-Kingdom",
+    "5": "United_States"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "Brazil": "0",
+    "Canada": "1",
+    "Finland": "2",
+    "Japan": "3",
+    "United-Kingdom": "4",
+    "United_States": "5"
+  },
+  "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": "4.31.0"
+}

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": "ViTFeatureExtractor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 224,
+    "width": 224
+  }
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e4cf7738c400c63d307d24c20ed9c77e1a7e56df0099b73936659a2e13d9670
+size 343278253

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 8.0,
+    "total_flos": 3.8629591832685773e+18,
+    "train_loss": 0.2533179052472592,
+    "train_runtime": 4957.8804,
+    "train_samples_per_second": 10.069,
+    "train_steps_per_second": 0.629
+}

trainer_state.json

@@ -0,0 +1,2599 @@
+{
+  "best_metric": 0.6381103992462158,
+  "best_model_checkpoint": "./vit-base-beans/checkpoint-1200",
+  "epoch": 8.0,
+  "global_step": 3120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019935897435897437,
+      "loss": 1.7678,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 1.7014,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001980769230769231,
+      "loss": 1.4957,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 1.5043,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.44556677890011226,
+      "eval_loss": 1.493240237236023,
+      "eval_runtime": 59.3866,
+      "eval_samples_per_second": 60.081,
+      "eval_steps_per_second": 7.51,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019679487179487178,
+      "loss": 1.3931,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 1.3029,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001955128205128205,
+      "loss": 1.1753,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 1.3894,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.5297418630751964,
+      "eval_loss": 1.2430940866470337,
+      "eval_runtime": 38.1635,
+      "eval_samples_per_second": 93.492,
+      "eval_steps_per_second": 11.687,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019423076923076924,
+      "loss": 1.118,
+      "step": 90
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 1.1395,
+      "step": 100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019294871794871797,
+      "loss": 1.0946,
+      "step": 110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 1.208,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.6464646464646465,
+      "eval_loss": 0.9881709218025208,
+      "eval_runtime": 34.6718,
+      "eval_samples_per_second": 102.908,
+      "eval_steps_per_second": 12.863,
+      "step": 120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 0.9775,
+      "step": 130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 1.0216,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019038461538461538,
+      "loss": 1.1942,
+      "step": 150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.9847,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.6599326599326599,
+      "eval_loss": 0.950343132019043,
+      "eval_runtime": 40.3536,
+      "eval_samples_per_second": 88.418,
+      "eval_steps_per_second": 11.052,
+      "step": 160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001891025641025641,
+      "loss": 0.9707,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 1.1941,
+      "step": 180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018782051282051283,
+      "loss": 1.0309,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 1.102,
+      "step": 200
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 0.6374859708193041,
+      "eval_loss": 0.9868837594985962,
+      "eval_runtime": 32.1858,
+      "eval_samples_per_second": 110.856,
+      "eval_steps_per_second": 13.857,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018653846153846154,
+      "loss": 0.9899,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8926,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018525641025641027,
+      "loss": 0.9552,
+      "step": 230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.9109,
+      "step": 240
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.6380471380471381,
+      "eval_loss": 0.9887320399284363,
+      "eval_runtime": 32.5312,
+      "eval_samples_per_second": 109.679,
+      "eval_steps_per_second": 13.71,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018397435897435897,
+      "loss": 0.832,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.9323,
+      "step": 260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001826923076923077,
+      "loss": 1.0061,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.9627,
+      "step": 280
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.6156004489337823,
+      "eval_loss": 1.057395100593567,
+      "eval_runtime": 34.1152,
+      "eval_samples_per_second": 104.587,
+      "eval_steps_per_second": 13.073,
+      "step": 280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018141025641025643,
+      "loss": 0.9613,
+      "step": 290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.8986,
+      "step": 300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018012820512820513,
+      "loss": 0.9344,
+      "step": 310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.7031,
+      "step": 320
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.7087542087542088,
+      "eval_loss": 0.813530683517456,
+      "eval_runtime": 39.1342,
+      "eval_samples_per_second": 91.173,
+      "eval_steps_per_second": 11.397,
+      "step": 320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017884615384615386,
+      "loss": 0.7596,
+      "step": 330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.7207,
+      "step": 340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017756410256410257,
+      "loss": 0.8904,
+      "step": 350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.7605,
+      "step": 360
+    },
+    {
+      "epoch": 0.92,
+      "eval_accuracy": 0.7272727272727273,
+      "eval_loss": 0.7620912790298462,
+      "eval_runtime": 33.0054,
+      "eval_samples_per_second": 108.103,
+      "eval_steps_per_second": 13.513,
+      "step": 360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001762820512820513,
+      "loss": 0.8093,
+      "step": 370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.8198,
+      "step": 380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000175,
+      "loss": 0.9515,
+      "step": 390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.8467,
+      "step": 400
+    },
+    {
+      "epoch": 1.03,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7974632978439331,
+      "eval_runtime": 33.3512,
+      "eval_samples_per_second": 106.983,
+      "eval_steps_per_second": 13.373,
+      "step": 400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017371794871794873,
+      "loss": 0.5829,
+      "step": 410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.5576,
+      "step": 420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017243589743589746,
+      "loss": 0.6035,
+      "step": 430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.6653,
+      "step": 440
+    },
+    {
+      "epoch": 1.13,
+      "eval_accuracy": 0.7317620650953984,
+      "eval_loss": 0.7574185132980347,
+      "eval_runtime": 32.4855,
+      "eval_samples_per_second": 109.834,
+      "eval_steps_per_second": 13.729,
+      "step": 440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017115384615384616,
+      "loss": 0.5067,
+      "step": 450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.4973,
+      "step": 460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00016987179487179486,
+      "loss": 0.5721,
+      "step": 470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.5467,
+      "step": 480
+    },
+    {
+      "epoch": 1.23,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.7471520900726318,
+      "eval_runtime": 32.6752,
+      "eval_samples_per_second": 109.196,
+      "eval_steps_per_second": 13.65,
+      "step": 480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001685897435897436,
+      "loss": 0.5494,
+      "step": 490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.6158,
+      "step": 500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016730769230769232,
+      "loss": 0.5412,
+      "step": 510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.388,
+      "step": 520
+    },
+    {
+      "epoch": 1.33,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 0.6508304476737976,
+      "eval_runtime": 36.0929,
+      "eval_samples_per_second": 98.856,
+      "eval_steps_per_second": 12.357,
+      "step": 520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016602564102564105,
+      "loss": 0.4511,
+      "step": 530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.566,
+      "step": 540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016474358974358976,
+      "loss": 0.4476,
+      "step": 550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.4699,
+      "step": 560
+    },
+    {
+      "epoch": 1.44,
+      "eval_accuracy": 0.7244668911335578,
+      "eval_loss": 0.773759663105011,
+      "eval_runtime": 31.7272,
+      "eval_samples_per_second": 112.459,
+      "eval_steps_per_second": 14.057,
+      "step": 560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 0.402,
+      "step": 570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.425,
+      "step": 580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001621794871794872,
+      "loss": 0.5171,
+      "step": 590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.5344,
+      "step": 600
+    },
+    {
+      "epoch": 1.54,
+      "eval_accuracy": 0.7328843995510662,
+      "eval_loss": 0.7624912858009338,
+      "eval_runtime": 32.007,
+      "eval_samples_per_second": 111.476,
+      "eval_steps_per_second": 13.934,
+      "step": 600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016089743589743592,
+      "loss": 0.4913,
+      "step": 610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.5042,
+      "step": 620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015961538461538462,
+      "loss": 0.6051,
+      "step": 630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.5753,
+      "step": 640
+    },
+    {
+      "epoch": 1.64,
+      "eval_accuracy": 0.7598204264870931,
+      "eval_loss": 0.6743783950805664,
+      "eval_runtime": 30.705,
+      "eval_samples_per_second": 116.203,
+      "eval_steps_per_second": 14.525,
+      "step": 640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 0.524,
+      "step": 650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.5007,
+      "step": 660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015705128205128205,
+      "loss": 0.57,
+      "step": 670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.5533,
+      "step": 680
+    },
+    {
+      "epoch": 1.74,
+      "eval_accuracy": 0.734006734006734,
+      "eval_loss": 0.7230738997459412,
+      "eval_runtime": 33.3034,
+      "eval_samples_per_second": 107.136,
+      "eval_steps_per_second": 13.392,
+      "step": 680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015576923076923078,
+      "loss": 0.4399,
+      "step": 690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.453,
+      "step": 700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00015448717948717951,
+      "loss": 0.4504,
+      "step": 710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.4244,
+      "step": 720
+    },
+    {
+      "epoch": 1.85,
+      "eval_accuracy": 0.7687991021324355,
+      "eval_loss": 0.6673141717910767,
+      "eval_runtime": 32.6948,
+      "eval_samples_per_second": 109.131,
+      "eval_steps_per_second": 13.641,
+      "step": 720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00015320512820512822,
+      "loss": 0.46,
+      "step": 730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.4132,
+      "step": 740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015192307692307692,
+      "loss": 0.4849,
+      "step": 750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.5423,
+      "step": 760
+    },
+    {
+      "epoch": 1.95,
+      "eval_accuracy": 0.7413019079685746,
+      "eval_loss": 0.7185856699943542,
+      "eval_runtime": 35.9539,
+      "eval_samples_per_second": 99.238,
+      "eval_steps_per_second": 12.405,
+      "step": 760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015064102564102565,
+      "loss": 0.3997,
+      "step": 770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4744,
+      "step": 780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00014935897435897438,
+      "loss": 0.4717,
+      "step": 790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.3384,
+      "step": 800
+    },
+    {
+      "epoch": 2.05,
+      "eval_accuracy": 0.7351290684624018,
+      "eval_loss": 0.7269920706748962,
+      "eval_runtime": 33.5897,
+      "eval_samples_per_second": 106.223,
+      "eval_steps_per_second": 13.278,
+      "step": 800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00014807692307692308,
+      "loss": 0.2602,
+      "step": 810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.2345,
+      "step": 820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014679487179487178,
+      "loss": 0.2135,
+      "step": 830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.2797,
+      "step": 840
+    },
+    {
+      "epoch": 2.15,
+      "eval_accuracy": 0.7497194163860831,
+      "eval_loss": 0.7742622494697571,
+      "eval_runtime": 31.2996,
+      "eval_samples_per_second": 113.995,
+      "eval_steps_per_second": 14.249,
+      "step": 840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014551282051282051,
+      "loss": 0.1862,
+      "step": 850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.2334,
+      "step": 860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014423076923076924,
+      "loss": 0.3361,
+      "step": 870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.2939,
+      "step": 880
+    },
+    {
+      "epoch": 2.26,
+      "eval_accuracy": 0.7732884399551067,
+      "eval_loss": 0.6985616087913513,
+      "eval_runtime": 32.9047,
+      "eval_samples_per_second": 108.435,
+      "eval_steps_per_second": 13.554,
+      "step": 880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014294871794871795,
+      "loss": 0.2846,
+      "step": 890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.2555,
+      "step": 900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 0.2629,
+      "step": 910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.2288,
+      "step": 920
+    },
+    {
+      "epoch": 2.36,
+      "eval_accuracy": 0.7480359147025814,
+      "eval_loss": 0.761025607585907,
+      "eval_runtime": 34.1426,
+      "eval_samples_per_second": 104.503,
+      "eval_steps_per_second": 13.063,
+      "step": 920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00014038461538461538,
+      "loss": 0.2115,
+      "step": 930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.2036,
+      "step": 940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001391025641025641,
+      "loss": 0.2913,
+      "step": 950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.2204,
+      "step": 960
+    },
+    {
+      "epoch": 2.46,
+      "eval_accuracy": 0.7570145903479237,
+      "eval_loss": 0.7840890884399414,
+      "eval_runtime": 35.7983,
+      "eval_samples_per_second": 99.67,
+      "eval_steps_per_second": 12.459,
+      "step": 960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013782051282051284,
+      "loss": 0.2759,
+      "step": 970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.3945,
+      "step": 980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00013653846153846154,
+      "loss": 0.3639,
+      "step": 990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.4397,
+      "step": 1000
+    },
+    {
+      "epoch": 2.56,
+      "eval_accuracy": 0.7789001122334456,
+      "eval_loss": 0.6565700173377991,
+      "eval_runtime": 31.6553,
+      "eval_samples_per_second": 112.714,
+      "eval_steps_per_second": 14.089,
+      "step": 1000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00013525641025641027,
+      "loss": 0.2707,
+      "step": 1010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.3451,
+      "step": 1020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013397435897435897,
+      "loss": 0.3277,
+      "step": 1030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.2219,
+      "step": 1040
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.6900522112846375,
+      "eval_runtime": 31.7845,
+      "eval_samples_per_second": 112.256,
+      "eval_steps_per_second": 14.032,
+      "step": 1040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001326923076923077,
+      "loss": 0.3174,
+      "step": 1050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.189,
+      "step": 1060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001314102564102564,
+      "loss": 0.2619,
+      "step": 1070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.2297,
+      "step": 1080
+    },
+    {
+      "epoch": 2.77,
+      "eval_accuracy": 0.7581369248035915,
+      "eval_loss": 0.7095093131065369,
+      "eval_runtime": 38.4561,
+      "eval_samples_per_second": 92.781,
+      "eval_steps_per_second": 11.598,
+      "step": 1080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00013012820512820514,
+      "loss": 0.2201,
+      "step": 1090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.2261,
+      "step": 1100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00012884615384615387,
+      "loss": 0.2095,
+      "step": 1110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.209,
+      "step": 1120
+    },
+    {
+      "epoch": 2.87,
+      "eval_accuracy": 0.7738496071829405,
+      "eval_loss": 0.7127913236618042,
+      "eval_runtime": 32.9997,
+      "eval_samples_per_second": 108.122,
+      "eval_steps_per_second": 13.515,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00012756410256410257,
+      "loss": 0.3365,
+      "step": 1130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.2639,
+      "step": 1140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00012628205128205127,
+      "loss": 0.1892,
+      "step": 1150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.1853,
+      "step": 1160
+    },
+    {
+      "epoch": 2.97,
+      "eval_accuracy": 0.7710437710437711,
+      "eval_loss": 0.6986069083213806,
+      "eval_runtime": 32.2598,
+      "eval_samples_per_second": 110.602,
+      "eval_steps_per_second": 13.825,
+      "step": 1160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000125,
+      "loss": 0.1807,
+      "step": 1170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.1554,
+      "step": 1180
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00012371794871794873,
+      "loss": 0.1845,
+      "step": 1190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.1322,
+      "step": 1200
+    },
+    {
+      "epoch": 3.08,
+      "eval_accuracy": 0.7962962962962963,
+      "eval_loss": 0.6381103992462158,
+      "eval_runtime": 35.1257,
+      "eval_samples_per_second": 101.578,
+      "eval_steps_per_second": 12.697,
+      "step": 1200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00012243589743589746,
+      "loss": 0.0598,
+      "step": 1210
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.0853,
+      "step": 1220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00012115384615384615,
+      "loss": 0.1616,
+      "step": 1230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.2603,
+      "step": 1240
+    },
+    {
+      "epoch": 3.18,
+      "eval_accuracy": 0.7682379349046016,
+      "eval_loss": 0.7860389351844788,
+      "eval_runtime": 35.648,
+      "eval_samples_per_second": 100.09,
+      "eval_steps_per_second": 12.511,
+      "step": 1240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00011987179487179487,
+      "loss": 0.1104,
+      "step": 1250
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.0634,
+      "step": 1260
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001185897435897436,
+      "loss": 0.0717,
+      "step": 1270
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.1031,
+      "step": 1280
+    },
+    {
+      "epoch": 3.28,
+      "eval_accuracy": 0.7895622895622896,
+      "eval_loss": 0.7322177886962891,
+      "eval_runtime": 30.6139,
+      "eval_samples_per_second": 116.549,
+      "eval_steps_per_second": 14.569,
+      "step": 1280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00011730769230769231,
+      "loss": 0.054,
+      "step": 1290
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.0883,
+      "step": 1300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00011602564102564104,
+      "loss": 0.1443,
+      "step": 1310
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.0763,
+      "step": 1320
+    },
+    {
+      "epoch": 3.38,
+      "eval_accuracy": 0.7839506172839507,
+      "eval_loss": 0.77773118019104,
+      "eval_runtime": 31.4828,
+      "eval_samples_per_second": 113.332,
+      "eval_steps_per_second": 14.166,
+      "step": 1320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00011474358974358975,
+      "loss": 0.1328,
+      "step": 1330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.0751,
+      "step": 1340
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00011346153846153846,
+      "loss": 0.1193,
+      "step": 1350
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.1437,
+      "step": 1360
+    },
+    {
+      "epoch": 3.49,
+      "eval_accuracy": 0.7665544332210998,
+      "eval_loss": 0.9293356537818909,
+      "eval_runtime": 31.1681,
+      "eval_samples_per_second": 114.476,
+      "eval_steps_per_second": 14.31,
+      "step": 1360
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00011217948717948718,
+      "loss": 0.1805,
+      "step": 1370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.0931,
+      "step": 1380
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00011089743589743591,
+      "loss": 0.1227,
+      "step": 1390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.0818,
+      "step": 1400
+    },
+    {
+      "epoch": 3.59,
+      "eval_accuracy": 0.7940516273849607,
+      "eval_loss": 0.796922504901886,
+      "eval_runtime": 31.9044,
+      "eval_samples_per_second": 111.834,
+      "eval_steps_per_second": 13.979,
+      "step": 1400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00010961538461538463,
+      "loss": 0.1193,
+      "step": 1410
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.1237,
+      "step": 1420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00010833333333333333,
+      "loss": 0.1074,
+      "step": 1430
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.1355,
+      "step": 1440
+    },
+    {
+      "epoch": 3.69,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 0.8145824670791626,
+      "eval_runtime": 37.6519,
+      "eval_samples_per_second": 94.763,
+      "eval_steps_per_second": 11.845,
+      "step": 1440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00010705128205128206,
+      "loss": 0.0844,
+      "step": 1450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 0.1067,
+      "step": 1460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00010576923076923077,
+      "loss": 0.0342,
+      "step": 1470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.0802,
+      "step": 1480
+    },
+    {
+      "epoch": 3.79,
+      "eval_accuracy": 0.8075196408529742,
+      "eval_loss": 0.6976819038391113,
+      "eval_runtime": 32.2393,
+      "eval_samples_per_second": 110.672,
+      "eval_steps_per_second": 13.834,
+      "step": 1480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0001044871794871795,
+      "loss": 0.1143,
+      "step": 1490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.1332,
+      "step": 1500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00010320512820512822,
+      "loss": 0.069,
+      "step": 1510
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.032,
+      "step": 1520
+    },
+    {
+      "epoch": 3.9,
+      "eval_accuracy": 0.7929292929292929,
+      "eval_loss": 0.8349816203117371,
+      "eval_runtime": 32.9637,
+      "eval_samples_per_second": 108.24,
+      "eval_steps_per_second": 13.53,
+      "step": 1520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00010192307692307692,
+      "loss": 0.0722,
+      "step": 1530
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.1383,
+      "step": 1540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00010064102564102564,
+      "loss": 0.0861,
+      "step": 1550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1560
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7968574635241302,
+      "eval_loss": 0.775885820388794,
+      "eval_runtime": 33.0589,
+      "eval_samples_per_second": 107.928,
+      "eval_steps_per_second": 13.491,
+      "step": 1560
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.935897435897437e-05,
+      "loss": 0.0486,
+      "step": 1570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.0369,
+      "step": 1580
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.807692307692307e-05,
+      "loss": 0.0209,
+      "step": 1590
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.0391,
+      "step": 1600
+    },
+    {
+      "epoch": 4.1,
+      "eval_accuracy": 0.8013468013468014,
+      "eval_loss": 0.8052350282669067,
+      "eval_runtime": 32.0537,
+      "eval_samples_per_second": 111.313,
+      "eval_steps_per_second": 13.914,
+      "step": 1600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.67948717948718e-05,
+      "loss": 0.0354,
+      "step": 1610
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.0738,
+      "step": 1620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.551282051282052e-05,
+      "loss": 0.0275,
+      "step": 1630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.0406,
+      "step": 1640
+    },
+    {
+      "epoch": 4.21,
+      "eval_accuracy": 0.7828282828282829,
+      "eval_loss": 0.9251748919487,
+      "eval_runtime": 31.6082,
+      "eval_samples_per_second": 112.882,
+      "eval_steps_per_second": 14.11,
+      "step": 1640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.423076923076924e-05,
+      "loss": 0.1377,
+      "step": 1650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.0526,
+      "step": 1660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.294871794871795e-05,
+      "loss": 0.152,
+      "step": 1670
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.0488,
+      "step": 1680
+    },
+    {
+      "epoch": 4.31,
+      "eval_accuracy": 0.7901234567901234,
+      "eval_loss": 0.8203706741333008,
+      "eval_runtime": 31.8606,
+      "eval_samples_per_second": 111.988,
+      "eval_steps_per_second": 13.998,
+      "step": 1680
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.0372,
+      "step": 1690
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.0968,
+      "step": 1700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.038461538461538e-05,
+      "loss": 0.0147,
+      "step": 1710
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.016,
+      "step": 1720
+    },
+    {
+      "epoch": 4.41,
+      "eval_accuracy": 0.7794612794612794,
+      "eval_loss": 0.9169319868087769,
+      "eval_runtime": 33.724,
+      "eval_samples_per_second": 105.8,
+      "eval_steps_per_second": 13.225,
+      "step": 1720
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.910256410256411e-05,
+      "loss": 0.0393,
+      "step": 1730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.0258,
+      "step": 1740
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.782051282051283e-05,
+      "loss": 0.009,
+      "step": 1750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.0994,
+      "step": 1760
+    },
+    {
+      "epoch": 4.51,
+      "eval_accuracy": 0.8120089786756454,
+      "eval_loss": 0.7930631637573242,
+      "eval_runtime": 34.0477,
+      "eval_samples_per_second": 104.794,
+      "eval_steps_per_second": 13.099,
+      "step": 1760
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.653846153846155e-05,
+      "loss": 0.0418,
+      "step": 1770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.0793,
+      "step": 1780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.525641025641026e-05,
+      "loss": 0.0122,
+      "step": 1790
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.07,
+      "step": 1800
+    },
+    {
+      "epoch": 4.62,
+      "eval_accuracy": 0.8103254769921436,
+      "eval_loss": 0.8245030045509338,
+      "eval_runtime": 32.6583,
+      "eval_samples_per_second": 109.252,
+      "eval_steps_per_second": 13.657,
+      "step": 1800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.397435897435898e-05,
+      "loss": 0.0656,
+      "step": 1810
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.0364,
+      "step": 1820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.26923076923077e-05,
+      "loss": 0.0639,
+      "step": 1830
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 0.0088,
+      "step": 1840
+    },
+    {
+      "epoch": 4.72,
+      "eval_accuracy": 0.7985409652076318,
+      "eval_loss": 0.907173752784729,
+      "eval_runtime": 40.0812,
+      "eval_samples_per_second": 89.019,
+      "eval_steps_per_second": 11.127,
+      "step": 1840
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.141025641025641e-05,
+      "loss": 0.0557,
+      "step": 1850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.0548,
+      "step": 1860
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.012820512820514e-05,
+      "loss": 0.0055,
+      "step": 1870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.0085,
+      "step": 1880
+    },
+    {
+      "epoch": 4.82,
+      "eval_accuracy": 0.8114478114478114,
+      "eval_loss": 0.7862226963043213,
+      "eval_runtime": 33.9761,
+      "eval_samples_per_second": 105.015,
+      "eval_steps_per_second": 13.127,
+      "step": 1880
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.884615384615384e-05,
+      "loss": 0.0423,
+      "step": 1890
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 0.0581,
+      "step": 1900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.756410256410257e-05,
+      "loss": 0.0043,
+      "step": 1910
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.083,
+      "step": 1920
+    },
+    {
+      "epoch": 4.92,
+      "eval_accuracy": 0.8198653198653199,
+      "eval_loss": 0.7796981334686279,
+      "eval_runtime": 32.3864,
+      "eval_samples_per_second": 110.17,
+      "eval_steps_per_second": 13.771,
+      "step": 1920
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.628205128205128e-05,
+      "loss": 0.1308,
+      "step": 1930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.0063,
+      "step": 1940
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.0132,
+      "step": 1950
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.0055,
+      "step": 1960
+    },
+    {
+      "epoch": 5.03,
+      "eval_accuracy": 0.8069584736251403,
+      "eval_loss": 0.8723996877670288,
+      "eval_runtime": 33.6185,
+      "eval_samples_per_second": 106.132,
+      "eval_steps_per_second": 13.267,
+      "step": 1960
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.371794871794872e-05,
+      "loss": 0.0366,
+      "step": 1970
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.06,
+      "step": 1980
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.243589743589744e-05,
+      "loss": 0.0065,
+      "step": 1990
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.0223,
+      "step": 2000
+    },
+    {
+      "epoch": 5.13,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.7518730163574219,
+      "eval_runtime": 43.1922,
+      "eval_samples_per_second": 82.607,
+      "eval_steps_per_second": 10.326,
+      "step": 2000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.115384615384616e-05,
+      "loss": 0.0215,
+      "step": 2010
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 0.0494,
+      "step": 2020
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.987179487179487e-05,
+      "loss": 0.0441,
+      "step": 2030
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.0042,
+      "step": 2040
+    },
+    {
+      "epoch": 5.23,
+      "eval_accuracy": 0.8232323232323232,
+      "eval_loss": 0.7583897113800049,
+      "eval_runtime": 31.893,
+      "eval_samples_per_second": 111.874,
+      "eval_steps_per_second": 13.984,
+      "step": 2040
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.858974358974359e-05,
+      "loss": 0.0131,
+      "step": 2050
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 0.0037,
+      "step": 2060
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.730769230769232e-05,
+      "loss": 0.02,
+      "step": 2070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.0178,
+      "step": 2080
+    },
+    {
+      "epoch": 5.33,
+      "eval_accuracy": 0.8080808080808081,
+      "eval_loss": 0.8523600697517395,
+      "eval_runtime": 32.5102,
+      "eval_samples_per_second": 109.75,
+      "eval_steps_per_second": 13.719,
+      "step": 2080
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.602564102564102e-05,
+      "loss": 0.0035,
+      "step": 2090
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.0137,
+      "step": 2100
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.474358974358975e-05,
+      "loss": 0.0033,
+      "step": 2110
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.0172,
+      "step": 2120
+    },
+    {
+      "epoch": 5.44,
+      "eval_accuracy": 0.8215488215488216,
+      "eval_loss": 0.7729219794273376,
+      "eval_runtime": 31.7305,
+      "eval_samples_per_second": 112.447,
+      "eval_steps_per_second": 14.056,
+      "step": 2120
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.346153846153847e-05,
+      "loss": 0.0252,
+      "step": 2130
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.0034,
+      "step": 2140
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.217948717948718e-05,
+      "loss": 0.0045,
+      "step": 2150
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.0044,
+      "step": 2160
+    },
+    {
+      "epoch": 5.54,
+      "eval_accuracy": 0.808641975308642,
+      "eval_loss": 0.8701183795928955,
+      "eval_runtime": 31.0299,
+      "eval_samples_per_second": 114.986,
+      "eval_steps_per_second": 14.373,
+      "step": 2160
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.089743589743589e-05,
+      "loss": 0.0373,
+      "step": 2170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.0029,
+      "step": 2180
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.9615384615384616e-05,
+      "loss": 0.0032,
+      "step": 2190
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.003,
+      "step": 2200
+    },
+    {
+      "epoch": 5.64,
+      "eval_accuracy": 0.8237934904601572,
+      "eval_loss": 0.8060529828071594,
+      "eval_runtime": 32.4421,
+      "eval_samples_per_second": 109.98,
+      "eval_steps_per_second": 13.748,
+      "step": 2200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.0034,
+      "step": 2210
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.0025,
+      "step": 2220
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.705128205128205e-05,
+      "loss": 0.0029,
+      "step": 2230
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.0033,
+      "step": 2240
+    },
+    {
+      "epoch": 5.74,
+      "eval_accuracy": 0.8226711560044894,
+      "eval_loss": 0.813855767250061,
+      "eval_runtime": 32.378,
+      "eval_samples_per_second": 110.198,
+      "eval_steps_per_second": 13.775,
+      "step": 2240
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.576923076923077e-05,
+      "loss": 0.0129,
+      "step": 2250
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 0.0079,
+      "step": 2260
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.448717948717948e-05,
+      "loss": 0.0117,
+      "step": 2270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.0023,
+      "step": 2280
+    },
+    {
+      "epoch": 5.85,
+      "eval_accuracy": 0.8164983164983165,
+      "eval_loss": 0.8478395342826843,
+      "eval_runtime": 35.321,
+      "eval_samples_per_second": 101.016,
+      "eval_steps_per_second": 12.627,
+      "step": 2280
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.3205128205128205e-05,
+      "loss": 0.0075,
+      "step": 2290
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 0.0023,
+      "step": 2300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.192307692307693e-05,
+      "loss": 0.0058,
+      "step": 2310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.003,
+      "step": 2320
+    },
+    {
+      "epoch": 5.95,
+      "eval_accuracy": 0.8120089786756454,
+      "eval_loss": 0.8443180918693542,
+      "eval_runtime": 32.2001,
+      "eval_samples_per_second": 110.807,
+      "eval_steps_per_second": 13.851,
+      "step": 2320
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.0641025641025644e-05,
+      "loss": 0.0512,
+      "step": 2330
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5e-05,
+      "loss": 0.0024,
+      "step": 2340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.935897435897436e-05,
+      "loss": 0.0022,
+      "step": 2350
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.031,
+      "step": 2360
+    },
+    {
+      "epoch": 6.05,
+      "eval_accuracy": 0.8047138047138047,
+      "eval_loss": 0.9272196888923645,
+      "eval_runtime": 33.2145,
+      "eval_samples_per_second": 107.423,
+      "eval_steps_per_second": 13.428,
+      "step": 2360
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 0.0456,
+      "step": 2370
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 0.0024,
+      "step": 2380
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.67948717948718e-05,
+      "loss": 0.0023,
+      "step": 2390
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.0021,
+      "step": 2400
+    },
+    {
+      "epoch": 6.15,
+      "eval_accuracy": 0.8204264870931538,
+      "eval_loss": 0.8369048237800598,
+      "eval_runtime": 36.1395,
+      "eval_samples_per_second": 98.728,
+      "eval_steps_per_second": 12.341,
+      "step": 2400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.5512820512820516e-05,
+      "loss": 0.0021,
+      "step": 2410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.0026,
+      "step": 2420
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.423076923076923e-05,
+      "loss": 0.0021,
+      "step": 2430
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.0019,
+      "step": 2440
+    },
+    {
+      "epoch": 6.26,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.828059196472168,
+      "eval_runtime": 32.1149,
+      "eval_samples_per_second": 111.101,
+      "eval_steps_per_second": 13.888,
+      "step": 2440
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.294871794871795e-05,
+      "loss": 0.0024,
+      "step": 2450
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.0019,
+      "step": 2460
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.002,
+      "step": 2470
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.0019,
+      "step": 2480
+    },
+    {
+      "epoch": 6.36,
+      "eval_accuracy": 0.8198653198653199,
+      "eval_loss": 0.834019124507904,
+      "eval_runtime": 35.8439,
+      "eval_samples_per_second": 99.543,
+      "eval_steps_per_second": 12.443,
+      "step": 2480
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.038461538461539e-05,
+      "loss": 0.002,
+      "step": 2490
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 0.0019,
+      "step": 2500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9102564102564105e-05,
+      "loss": 0.0018,
+      "step": 2510
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.0018,
+      "step": 2520
+    },
+    {
+      "epoch": 6.46,
+      "eval_accuracy": 0.8204264870931538,
+      "eval_loss": 0.8339148759841919,
+      "eval_runtime": 34.0288,
+      "eval_samples_per_second": 104.853,
+      "eval_steps_per_second": 13.107,
+      "step": 2520
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.782051282051282e-05,
+      "loss": 0.0018,
+      "step": 2530
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 0.0018,
+      "step": 2540
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.653846153846154e-05,
+      "loss": 0.0018,
+      "step": 2550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 0.002,
+      "step": 2560
+    },
+    {
+      "epoch": 6.56,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.8293877840042114,
+      "eval_runtime": 33.9846,
+      "eval_samples_per_second": 104.989,
+      "eval_steps_per_second": 13.124,
+      "step": 2560
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.525641025641026e-05,
+      "loss": 0.0017,
+      "step": 2570
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 0.0017,
+      "step": 2580
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.397435897435898e-05,
+      "loss": 0.0018,
+      "step": 2590
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0017,
+      "step": 2600
+    },
+    {
+      "epoch": 6.67,
+      "eval_accuracy": 0.8209876543209876,
+      "eval_loss": 0.8266403079032898,
+      "eval_runtime": 36.0162,
+      "eval_samples_per_second": 99.067,
+      "eval_steps_per_second": 12.383,
+      "step": 2600
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.269230769230769e-05,
+      "loss": 0.002,
+      "step": 2610
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.0017,
+      "step": 2620
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.141025641025641e-05,
+      "loss": 0.0017,
+      "step": 2630
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.0017,
+      "step": 2640
+    },
+    {
+      "epoch": 6.77,
+      "eval_accuracy": 0.8226711560044894,
+      "eval_loss": 0.8177886009216309,
+      "eval_runtime": 36.3329,
+      "eval_samples_per_second": 98.203,
+      "eval_steps_per_second": 12.275,
+      "step": 2640
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.012820512820513e-05,
+      "loss": 0.0016,
+      "step": 2650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 0.0016,
+      "step": 2660
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 0.0016,
+      "step": 2670
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 0.0017,
+      "step": 2680
+    },
+    {
+      "epoch": 6.87,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8174560070037842,
+      "eval_runtime": 32.3063,
+      "eval_samples_per_second": 110.443,
+      "eval_steps_per_second": 13.805,
+      "step": 2680
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.756410256410257e-05,
+      "loss": 0.0017,
+      "step": 2690
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.0017,
+      "step": 2700
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6282051282051285e-05,
+      "loss": 0.0017,
+      "step": 2710
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 0.0017,
+      "step": 2720
+    },
+    {
+      "epoch": 6.97,
+      "eval_accuracy": 0.8237934904601572,
+      "eval_loss": 0.8166114687919617,
+      "eval_runtime": 33.2936,
+      "eval_samples_per_second": 107.168,
+      "eval_steps_per_second": 13.396,
+      "step": 2720
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 2730
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 0.0016,
+      "step": 2740
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.3717948717948718e-05,
+      "loss": 0.0016,
+      "step": 2750
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.0015,
+      "step": 2760
+    },
+    {
+      "epoch": 7.08,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8176218271255493,
+      "eval_runtime": 32.3011,
+      "eval_samples_per_second": 110.46,
+      "eval_steps_per_second": 13.808,
+      "step": 2760
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.2435897435897437e-05,
+      "loss": 0.0015,
+      "step": 2770
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 0.0015,
+      "step": 2780
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.1153846153846154e-05,
+      "loss": 0.0015,
+      "step": 2790
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.0015,
+      "step": 2800
+    },
+    {
+      "epoch": 7.18,
+      "eval_accuracy": 0.824354657687991,
+      "eval_loss": 0.8185549378395081,
+      "eval_runtime": 32.7126,
+      "eval_samples_per_second": 109.071,
+      "eval_steps_per_second": 13.634,
+      "step": 2800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.987179487179487e-05,
+      "loss": 0.0015,
+      "step": 2810
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.0015,
+      "step": 2820
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.858974358974359e-05,
+      "loss": 0.0015,
+      "step": 2830
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.0015,
+      "step": 2840
+    },
+    {
+      "epoch": 7.28,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.819684624671936,
+      "eval_runtime": 35.8925,
+      "eval_samples_per_second": 99.408,
+      "eval_steps_per_second": 12.426,
+      "step": 2840
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.0016,
+      "step": 2850
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0015,
+      "step": 2860
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.0015,
+      "step": 2870
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.0015,
+      "step": 2880
+    },
+    {
+      "epoch": 7.38,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8204275965690613,
+      "eval_runtime": 31.8471,
+      "eval_samples_per_second": 112.035,
+      "eval_steps_per_second": 14.004,
+      "step": 2880
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.0015,
+      "step": 2890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 0.0015,
+      "step": 2900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3461538461538462e-05,
+      "loss": 0.0015,
+      "step": 2910
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 0.0015,
+      "step": 2920
+    },
+    {
+      "epoch": 7.49,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8213893175125122,
+      "eval_runtime": 30.8496,
+      "eval_samples_per_second": 115.658,
+      "eval_steps_per_second": 14.457,
+      "step": 2920
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.0015,
+      "step": 2930
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 0.0015,
+      "step": 2940
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.0015,
+      "step": 2950
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.0015,
+      "step": 2960
+    },
+    {
+      "epoch": 7.59,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.8220140933990479,
+      "eval_runtime": 31.6074,
+      "eval_samples_per_second": 112.885,
+      "eval_steps_per_second": 14.111,
+      "step": 2960
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.0016,
+      "step": 2970
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.0014,
+      "step": 2980
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0015,
+      "step": 2990
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.0015,
+      "step": 3000
+    },
+    {
+      "epoch": 7.69,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8220102787017822,
+      "eval_runtime": 30.2325,
+      "eval_samples_per_second": 118.019,
+      "eval_steps_per_second": 14.752,
+      "step": 3000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.051282051282052e-06,
+      "loss": 0.0014,
+      "step": 3010
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 0.0014,
+      "step": 3020
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 0.0015,
+      "step": 3030
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.0015,
+      "step": 3040
+    },
+    {
+      "epoch": 7.79,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8217305541038513,
+      "eval_runtime": 38.9983,
+      "eval_samples_per_second": 91.491,
+      "eval_steps_per_second": 11.436,
+      "step": 3040
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 0.0014,
+      "step": 3050
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.0015,
+      "step": 3060
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.205128205128205e-06,
+      "loss": 0.0014,
+      "step": 3070
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.0014,
+      "step": 3080
+    },
+    {
+      "epoch": 7.9,
+      "eval_accuracy": 0.8254769921436588,
+      "eval_loss": 0.8224795460700989,
+      "eval_runtime": 31.9073,
+      "eval_samples_per_second": 111.824,
+      "eval_steps_per_second": 13.978,
+      "step": 3080
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.0014,
+      "step": 3090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.0015,
+      "step": 3100
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 0.0014,
+      "step": 3110
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0,
+      "loss": 0.0014,
+      "step": 3120
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.8260381593714927,
+      "eval_loss": 0.8224756717681885,
+      "eval_runtime": 30.7985,
+      "eval_samples_per_second": 115.85,
+      "eval_steps_per_second": 14.481,
+      "step": 3120
+    },
+    {
+      "epoch": 8.0,
+      "step": 3120,
+      "total_flos": 3.8629591832685773e+18,
+      "train_loss": 0.2533179052472592,
+      "train_runtime": 4957.8804,
+      "train_samples_per_second": 10.069,
+      "train_steps_per_second": 0.629
+    }
+  ],
+  "max_steps": 3120,
+  "num_train_epochs": 8,
+  "total_flos": 3.8629591832685773e+18,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38c25b2b280e8f8dde2511a004bf1b097d7641bf1aae99dd53711356509cd78e
+size 3899