diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5132 @@
+{
+  "best_metric": 0.18093784153461456,
+  "best_model_checkpoint": "./stable_diffusion_prompts/checkpoint-8500",
+  "epoch": 2.0,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9976470588235294e-05,
+      "loss": 2.8307,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.995294117647059e-05,
+      "loss": 2.3977,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9929411764705884e-05,
+      "loss": 2.0917,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9905882352941176e-05,
+      "loss": 1.8467,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.988235294117647e-05,
+      "loss": 1.6254,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9858823529411767e-05,
+      "loss": 1.4769,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9835294117647062e-05,
+      "loss": 1.3717,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9811764705882354e-05,
+      "loss": 1.2786,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.978823529411765e-05,
+      "loss": 1.2022,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9764705882352945e-05,
+      "loss": 1.1503,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9741176470588237e-05,
+      "loss": 1.0988,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.971764705882353e-05,
+      "loss": 1.0437,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9694117647058824e-05,
+      "loss": 1.0052,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.967058823529412e-05,
+      "loss": 0.9693,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.964705882352941e-05,
+      "loss": 0.9348,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9623529411764707e-05,
+      "loss": 0.9004,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.8651,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9576470588235297e-05,
+      "loss": 0.8442,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.955294117647059e-05,
+      "loss": 0.8065,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9529411764705885e-05,
+      "loss": 0.782,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.950588235294118e-05,
+      "loss": 0.7742,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9482352941176472e-05,
+      "loss": 0.7353,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9458823529411764e-05,
+      "loss": 0.732,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9435294117647063e-05,
+      "loss": 0.7053,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 0.6894,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9388235294117647e-05,
+      "loss": 0.6857,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9364705882352942e-05,
+      "loss": 0.6455,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9341176470588237e-05,
+      "loss": 0.6497,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9317647058823533e-05,
+      "loss": 0.6298,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9294117647058825e-05,
+      "loss": 0.6188,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.927058823529412e-05,
+      "loss": 0.601,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9247058823529415e-05,
+      "loss": 0.5927,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9223529411764707e-05,
+      "loss": 0.5796,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.5738,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9176470588235298e-05,
+      "loss": 0.5542,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.915294117647059e-05,
+      "loss": 0.5291,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9129411764705882e-05,
+      "loss": 0.5372,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9105882352941177e-05,
+      "loss": 0.5434,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9082352941176472e-05,
+      "loss": 0.514,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9058823529411764e-05,
+      "loss": 0.505,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.903529411764706e-05,
+      "loss": 0.4934,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9011764705882355e-05,
+      "loss": 0.4931,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.898823529411765e-05,
+      "loss": 0.4858,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8964705882352942e-05,
+      "loss": 0.4746,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8941176470588238e-05,
+      "loss": 0.4684,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8917647058823533e-05,
+      "loss": 0.4533,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8894117647058825e-05,
+      "loss": 0.4563,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8870588235294117e-05,
+      "loss": 0.4342,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8847058823529412e-05,
+      "loss": 0.4398,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.428,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.88e-05,
+      "loss": 0.4192,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8776470588235295e-05,
+      "loss": 0.4147,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.875294117647059e-05,
+      "loss": 0.4178,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8729411764705886e-05,
+      "loss": 0.3974,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8705882352941178e-05,
+      "loss": 0.3988,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8682352941176473e-05,
+      "loss": 0.3943,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8658823529411768e-05,
+      "loss": 0.3902,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.863529411764706e-05,
+      "loss": 0.3768,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8611764705882352e-05,
+      "loss": 0.3779,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8588235294117647e-05,
+      "loss": 0.3682,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8564705882352943e-05,
+      "loss": 0.3683,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8541176470588235e-05,
+      "loss": 0.3671,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.851764705882353e-05,
+      "loss": 0.3637,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8494117647058825e-05,
+      "loss": 0.3594,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.847058823529412e-05,
+      "loss": 0.3524,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8447058823529413e-05,
+      "loss": 0.3497,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8423529411764708e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.3399,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8376470588235295e-05,
+      "loss": 0.3273,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8352941176470587e-05,
+      "loss": 0.3333,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8329411764705886e-05,
+      "loss": 0.3225,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8305882352941178e-05,
+      "loss": 0.3306,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.828235294117647e-05,
+      "loss": 0.3343,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8258823529411765e-05,
+      "loss": 0.3274,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.823529411764706e-05,
+      "loss": 0.3257,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8211764705882356e-05,
+      "loss": 0.3196,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8188235294117648e-05,
+      "loss": 0.3226,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8164705882352943e-05,
+      "loss": 0.3139,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.814117647058824e-05,
+      "loss": 0.3239,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.811764705882353e-05,
+      "loss": 0.3103,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8094117647058826e-05,
+      "loss": 0.3037,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.807058823529412e-05,
+      "loss": 0.3058,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8047058823529413e-05,
+      "loss": 0.3095,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8023529411764705e-05,
+      "loss": 0.2998,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8e-05,
+      "loss": 0.3016,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7976470588235296e-05,
+      "loss": 0.3057,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.795294117647059e-05,
+      "loss": 0.2959,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7929411764705883e-05,
+      "loss": 0.2957,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790588235294118e-05,
+      "loss": 0.2934,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7882352941176474e-05,
+      "loss": 0.2903,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7858823529411766e-05,
+      "loss": 0.292,
+      "step": 910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783529411764706e-05,
+      "loss": 0.2843,
+      "step": 920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7811764705882356e-05,
+      "loss": 0.2788,
+      "step": 930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778823529411765e-05,
+      "loss": 0.2796,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776470588235294e-05,
+      "loss": 0.2815,
+      "step": 950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7741176470588236e-05,
+      "loss": 0.2871,
+      "step": 960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.771764705882353e-05,
+      "loss": 0.2858,
+      "step": 970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7694117647058826e-05,
+      "loss": 0.2804,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7670588235294118e-05,
+      "loss": 0.284,
+      "step": 990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.2767,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.762352941176471e-05,
+      "loss": 0.2788,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76e-05,
+      "loss": 0.2797,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7576470588235296e-05,
+      "loss": 0.2735,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.755294117647059e-05,
+      "loss": 0.2693,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529411764705884e-05,
+      "loss": 0.2694,
+      "step": 1050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505882352941175e-05,
+      "loss": 0.2749,
+      "step": 1060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482352941176474e-05,
+      "loss": 0.2644,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7458823529411766e-05,
+      "loss": 0.2582,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435294117647058e-05,
+      "loss": 0.2596,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411764705882353e-05,
+      "loss": 0.2686,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738823529411765e-05,
+      "loss": 0.2665,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364705882352944e-05,
+      "loss": 0.2598,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7341176470588236e-05,
+      "loss": 0.2675,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.731764705882353e-05,
+      "loss": 0.2628,
+      "step": 1140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7294117647058827e-05,
+      "loss": 0.2591,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.727058823529412e-05,
+      "loss": 0.2595,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.724705882352941e-05,
+      "loss": 0.2543,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.722352941176471e-05,
+      "loss": 0.2577,
+      "step": 1180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.72e-05,
+      "loss": 0.2572,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7176470588235293e-05,
+      "loss": 0.2578,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.715294117647059e-05,
+      "loss": 0.258,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7129411764705884e-05,
+      "loss": 0.2518,
+      "step": 1220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.710588235294118e-05,
+      "loss": 0.2498,
+      "step": 1230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.708235294117647e-05,
+      "loss": 0.2588,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 0.2524,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7035294117647062e-05,
+      "loss": 0.2524,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7011764705882354e-05,
+      "loss": 0.2528,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.698823529411765e-05,
+      "loss": 0.2509,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6964705882352945e-05,
+      "loss": 0.2489,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6941176470588237e-05,
+      "loss": 0.2434,
+      "step": 1300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.691764705882353e-05,
+      "loss": 0.2485,
+      "step": 1310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6894117647058824e-05,
+      "loss": 0.2492,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.687058823529412e-05,
+      "loss": 0.25,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6847058823529414e-05,
+      "loss": 0.2472,
+      "step": 1340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6823529411764706e-05,
+      "loss": 0.2448,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.2428,
+      "step": 1360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6776470588235297e-05,
+      "loss": 0.2461,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.675294117647059e-05,
+      "loss": 0.2436,
+      "step": 1380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6729411764705884e-05,
+      "loss": 0.2428,
+      "step": 1390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.670588235294118e-05,
+      "loss": 0.2432,
+      "step": 1400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6682352941176472e-05,
+      "loss": 0.2459,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6658823529411764e-05,
+      "loss": 0.2431,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.663529411764706e-05,
+      "loss": 0.2404,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6611764705882354e-05,
+      "loss": 0.2385,
+      "step": 1440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.658823529411765e-05,
+      "loss": 0.24,
+      "step": 1450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.656470588235294e-05,
+      "loss": 0.2364,
+      "step": 1460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6541176470588237e-05,
+      "loss": 0.2402,
+      "step": 1470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6517647058823532e-05,
+      "loss": 0.24,
+      "step": 1480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6494117647058824e-05,
+      "loss": 0.2369,
+      "step": 1490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.237,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6447058823529415e-05,
+      "loss": 0.2373,
+      "step": 1510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6423529411764707e-05,
+      "loss": 0.2378,
+      "step": 1520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.64e-05,
+      "loss": 0.2393,
+      "step": 1530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6376470588235298e-05,
+      "loss": 0.2384,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.635294117647059e-05,
+      "loss": 0.2351,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6329411764705885e-05,
+      "loss": 0.2345,
+      "step": 1560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6305882352941177e-05,
+      "loss": 0.2355,
+      "step": 1570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6282352941176472e-05,
+      "loss": 0.2344,
+      "step": 1580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6258823529411767e-05,
+      "loss": 0.2329,
+      "step": 1590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623529411764706e-05,
+      "loss": 0.2338,
+      "step": 1600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6211764705882355e-05,
+      "loss": 0.2389,
+      "step": 1610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618823529411765e-05,
+      "loss": 0.2352,
+      "step": 1620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6164705882352942e-05,
+      "loss": 0.2371,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6141176470588234e-05,
+      "loss": 0.2351,
+      "step": 1640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6117647058823533e-05,
+      "loss": 0.2309,
+      "step": 1650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6094117647058825e-05,
+      "loss": 0.2322,
+      "step": 1660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607058823529412e-05,
+      "loss": 0.232,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6047058823529412e-05,
+      "loss": 0.2403,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6023529411764707e-05,
+      "loss": 0.2315,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2304,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5976470588235295e-05,
+      "loss": 0.2261,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.595294117647059e-05,
+      "loss": 0.2316,
+      "step": 1720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5929411764705885e-05,
+      "loss": 0.2261,
+      "step": 1730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5905882352941177e-05,
+      "loss": 0.2247,
+      "step": 1740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5882352941176473e-05,
+      "loss": 0.2272,
+      "step": 1750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5858823529411768e-05,
+      "loss": 0.2273,
+      "step": 1760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.583529411764706e-05,
+      "loss": 0.2284,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5811764705882352e-05,
+      "loss": 0.2247,
+      "step": 1780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5788235294117647e-05,
+      "loss": 0.2301,
+      "step": 1790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5764705882352943e-05,
+      "loss": 0.2276,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5741176470588238e-05,
+      "loss": 0.2225,
+      "step": 1810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.571764705882353e-05,
+      "loss": 0.2292,
+      "step": 1820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5694117647058825e-05,
+      "loss": 0.2253,
+      "step": 1830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.567058823529412e-05,
+      "loss": 0.2226,
+      "step": 1840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5647058823529412e-05,
+      "loss": 0.2181,
+      "step": 1850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5623529411764708e-05,
+      "loss": 0.2248,
+      "step": 1860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.2307,
+      "step": 1870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5576470588235295e-05,
+      "loss": 0.2296,
+      "step": 1880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5552941176470587e-05,
+      "loss": 0.2231,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5529411764705882e-05,
+      "loss": 0.2251,
+      "step": 1900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5505882352941178e-05,
+      "loss": 0.2245,
+      "step": 1910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5482352941176473e-05,
+      "loss": 0.2213,
+      "step": 1920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5458823529411765e-05,
+      "loss": 0.2239,
+      "step": 1930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.543529411764706e-05,
+      "loss": 0.2216,
+      "step": 1940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5411764705882356e-05,
+      "loss": 0.2225,
+      "step": 1950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5388235294117648e-05,
+      "loss": 0.2186,
+      "step": 1960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5364705882352943e-05,
+      "loss": 0.2206,
+      "step": 1970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5341176470588238e-05,
+      "loss": 0.2209,
+      "step": 1980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.531764705882353e-05,
+      "loss": 0.2185,
+      "step": 1990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.2201,
+      "step": 2000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.527058823529412e-05,
+      "loss": 0.2146,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5247058823529413e-05,
+      "loss": 0.2175,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5223529411764708e-05,
+      "loss": 0.2209,
+      "step": 2030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.2172,
+      "step": 2040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5176470588235295e-05,
+      "loss": 0.2166,
+      "step": 2050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.515294117647059e-05,
+      "loss": 0.2204,
+      "step": 2060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5129411764705883e-05,
+      "loss": 0.218,
+      "step": 2070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5105882352941176e-05,
+      "loss": 0.2196,
+      "step": 2080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5082352941176472e-05,
+      "loss": 0.2192,
+      "step": 2090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5058823529411765e-05,
+      "loss": 0.2169,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5035294117647059e-05,
+      "loss": 0.2205,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5011764705882354e-05,
+      "loss": 0.2154,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4988235294117648e-05,
+      "loss": 0.22,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4964705882352943e-05,
+      "loss": 0.2159,
+      "step": 2140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4941176470588237e-05,
+      "loss": 0.2184,
+      "step": 2150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.491764705882353e-05,
+      "loss": 0.2166,
+      "step": 2160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4894117647058826e-05,
+      "loss": 0.2174,
+      "step": 2170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.487058823529412e-05,
+      "loss": 0.2158,
+      "step": 2180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4847058823529412e-05,
+      "loss": 0.2201,
+      "step": 2190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4823529411764707e-05,
+      "loss": 0.2139,
+      "step": 2200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.48e-05,
+      "loss": 0.2147,
+      "step": 2210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4776470588235294e-05,
+      "loss": 0.2186,
+      "step": 2220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.475294117647059e-05,
+      "loss": 0.2141,
+      "step": 2230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4729411764705883e-05,
+      "loss": 0.2159,
+      "step": 2240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4705882352941179e-05,
+      "loss": 0.217,
+      "step": 2250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4682352941176472e-05,
+      "loss": 0.2172,
+      "step": 2260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4658823529411766e-05,
+      "loss": 0.2178,
+      "step": 2270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4635294117647061e-05,
+      "loss": 0.2127,
+      "step": 2280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4611764705882355e-05,
+      "loss": 0.2161,
+      "step": 2290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4588235294117647e-05,
+      "loss": 0.2137,
+      "step": 2300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4564705882352944e-05,
+      "loss": 0.214,
+      "step": 2310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4541176470588236e-05,
+      "loss": 0.2157,
+      "step": 2320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.451764705882353e-05,
+      "loss": 0.2167,
+      "step": 2330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4494117647058825e-05,
+      "loss": 0.2172,
+      "step": 2340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4470588235294118e-05,
+      "loss": 0.2095,
+      "step": 2350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4447058823529414e-05,
+      "loss": 0.2177,
+      "step": 2360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4423529411764707e-05,
+      "loss": 0.2151,
+      "step": 2370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.2154,
+      "step": 2380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4376470588235296e-05,
+      "loss": 0.2084,
+      "step": 2390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.435294117647059e-05,
+      "loss": 0.2155,
+      "step": 2400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4329411764705882e-05,
+      "loss": 0.21,
+      "step": 2410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4305882352941179e-05,
+      "loss": 0.213,
+      "step": 2420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4282352941176471e-05,
+      "loss": 0.2142,
+      "step": 2430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4258823529411765e-05,
+      "loss": 0.2143,
+      "step": 2440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.423529411764706e-05,
+      "loss": 0.214,
+      "step": 2450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4211764705882354e-05,
+      "loss": 0.2143,
+      "step": 2460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4188235294117649e-05,
+      "loss": 0.215,
+      "step": 2470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4164705882352943e-05,
+      "loss": 0.2088,
+      "step": 2480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4141176470588236e-05,
+      "loss": 0.2096,
+      "step": 2490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.2101,
+      "step": 2500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4094117647058825e-05,
+      "loss": 0.2129,
+      "step": 2510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4070588235294119e-05,
+      "loss": 0.2095,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4047058823529414e-05,
+      "loss": 0.2111,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4023529411764706e-05,
+      "loss": 0.2126,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4e-05,
+      "loss": 0.213,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3976470588235295e-05,
+      "loss": 0.2147,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3952941176470589e-05,
+      "loss": 0.2078,
+      "step": 2570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3929411764705882e-05,
+      "loss": 0.2089,
+      "step": 2580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3905882352941178e-05,
+      "loss": 0.2083,
+      "step": 2590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3882352941176471e-05,
+      "loss": 0.2116,
+      "step": 2600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3858823529411767e-05,
+      "loss": 0.2106,
+      "step": 2610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.383529411764706e-05,
+      "loss": 0.2113,
+      "step": 2620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3811764705882354e-05,
+      "loss": 0.2108,
+      "step": 2630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.378823529411765e-05,
+      "loss": 0.2098,
+      "step": 2640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3764705882352943e-05,
+      "loss": 0.2097,
+      "step": 2650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3741176470588235e-05,
+      "loss": 0.2077,
+      "step": 2660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.371764705882353e-05,
+      "loss": 0.2138,
+      "step": 2670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3694117647058824e-05,
+      "loss": 0.2111,
+      "step": 2680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3670588235294118e-05,
+      "loss": 0.2107,
+      "step": 2690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3647058823529413e-05,
+      "loss": 0.2109,
+      "step": 2700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3623529411764707e-05,
+      "loss": 0.2091,
+      "step": 2710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.209,
+      "step": 2720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3576470588235296e-05,
+      "loss": 0.2085,
+      "step": 2730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.355294117647059e-05,
+      "loss": 0.2075,
+      "step": 2740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3529411764705885e-05,
+      "loss": 0.2036,
+      "step": 2750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3505882352941178e-05,
+      "loss": 0.208,
+      "step": 2760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.348235294117647e-05,
+      "loss": 0.2086,
+      "step": 2770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3458823529411767e-05,
+      "loss": 0.2078,
+      "step": 2780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3435294117647059e-05,
+      "loss": 0.2086,
+      "step": 2790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3411764705882353e-05,
+      "loss": 0.2088,
+      "step": 2800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3388235294117648e-05,
+      "loss": 0.2095,
+      "step": 2810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3364705882352942e-05,
+      "loss": 0.2061,
+      "step": 2820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3341176470588237e-05,
+      "loss": 0.2123,
+      "step": 2830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.331764705882353e-05,
+      "loss": 0.2089,
+      "step": 2840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3294117647058824e-05,
+      "loss": 0.2088,
+      "step": 2850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.327058823529412e-05,
+      "loss": 0.2097,
+      "step": 2860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3247058823529413e-05,
+      "loss": 0.2075,
+      "step": 2870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3223529411764705e-05,
+      "loss": 0.2049,
+      "step": 2880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.208,
+      "step": 2890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3176470588235294e-05,
+      "loss": 0.2073,
+      "step": 2900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3152941176470588e-05,
+      "loss": 0.2098,
+      "step": 2910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3129411764705883e-05,
+      "loss": 0.2037,
+      "step": 2920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3105882352941177e-05,
+      "loss": 0.2075,
+      "step": 2930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3082352941176472e-05,
+      "loss": 0.2066,
+      "step": 2940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3058823529411766e-05,
+      "loss": 0.2056,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.303529411764706e-05,
+      "loss": 0.2056,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3011764705882355e-05,
+      "loss": 0.205,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2988235294117649e-05,
+      "loss": 0.2089,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2964705882352942e-05,
+      "loss": 0.2029,
+      "step": 2990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.209,
+      "step": 3000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.291764705882353e-05,
+      "loss": 0.2023,
+      "step": 3010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2894117647058823e-05,
+      "loss": 0.2032,
+      "step": 3020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2870588235294118e-05,
+      "loss": 0.2053,
+      "step": 3030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2847058823529412e-05,
+      "loss": 0.2056,
+      "step": 3040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2823529411764707e-05,
+      "loss": 0.2064,
+      "step": 3050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.2062,
+      "step": 3060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2776470588235295e-05,
+      "loss": 0.2051,
+      "step": 3070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.275294117647059e-05,
+      "loss": 0.2092,
+      "step": 3080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2729411764705884e-05,
+      "loss": 0.2036,
+      "step": 3090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2705882352941177e-05,
+      "loss": 0.2053,
+      "step": 3100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2682352941176473e-05,
+      "loss": 0.2029,
+      "step": 3110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2658823529411766e-05,
+      "loss": 0.2049,
+      "step": 3120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2635294117647058e-05,
+      "loss": 0.2043,
+      "step": 3130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2611764705882354e-05,
+      "loss": 0.2108,
+      "step": 3140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2588235294117647e-05,
+      "loss": 0.2038,
+      "step": 3150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2564705882352943e-05,
+      "loss": 0.205,
+      "step": 3160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2541176470588236e-05,
+      "loss": 0.202,
+      "step": 3170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.251764705882353e-05,
+      "loss": 0.2057,
+      "step": 3180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2494117647058825e-05,
+      "loss": 0.2076,
+      "step": 3190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2470588235294119e-05,
+      "loss": 0.2058,
+      "step": 3200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2447058823529413e-05,
+      "loss": 0.2042,
+      "step": 3210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2423529411764708e-05,
+      "loss": 0.2027,
+      "step": 3220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.202,
+      "step": 3230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2376470588235294e-05,
+      "loss": 0.2052,
+      "step": 3240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.235294117647059e-05,
+      "loss": 0.2024,
+      "step": 3250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2329411764705882e-05,
+      "loss": 0.2025,
+      "step": 3260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2305882352941176e-05,
+      "loss": 0.2045,
+      "step": 3270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2282352941176471e-05,
+      "loss": 0.2069,
+      "step": 3280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2258823529411765e-05,
+      "loss": 0.2022,
+      "step": 3290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223529411764706e-05,
+      "loss": 0.2043,
+      "step": 3300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2211764705882354e-05,
+      "loss": 0.2034,
+      "step": 3310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2188235294117648e-05,
+      "loss": 0.2034,
+      "step": 3320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2164705882352943e-05,
+      "loss": 0.2005,
+      "step": 3330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2141176470588237e-05,
+      "loss": 0.2048,
+      "step": 3340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.211764705882353e-05,
+      "loss": 0.203,
+      "step": 3350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2094117647058826e-05,
+      "loss": 0.1987,
+      "step": 3360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2070588235294118e-05,
+      "loss": 0.2007,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2047058823529411e-05,
+      "loss": 0.2023,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2023529411764707e-05,
+      "loss": 0.2033,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2e-05,
+      "loss": 0.203,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1976470588235296e-05,
+      "loss": 0.2008,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.195294117647059e-05,
+      "loss": 0.2034,
+      "step": 3420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1929411764705883e-05,
+      "loss": 0.2032,
+      "step": 3430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1905882352941178e-05,
+      "loss": 0.2001,
+      "step": 3440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1882352941176472e-05,
+      "loss": 0.2021,
+      "step": 3450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1858823529411766e-05,
+      "loss": 0.1996,
+      "step": 3460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1835294117647061e-05,
+      "loss": 0.2004,
+      "step": 3470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1811764705882353e-05,
+      "loss": 0.2028,
+      "step": 3480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1788235294117647e-05,
+      "loss": 0.2021,
+      "step": 3490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.1999,
+      "step": 3500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1741176470588235e-05,
+      "loss": 0.2009,
+      "step": 3510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.171764705882353e-05,
+      "loss": 0.2009,
+      "step": 3520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1694117647058824e-05,
+      "loss": 0.205,
+      "step": 3530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1670588235294118e-05,
+      "loss": 0.2038,
+      "step": 3540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1647058823529413e-05,
+      "loss": 0.2026,
+      "step": 3550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1623529411764707e-05,
+      "loss": 0.199,
+      "step": 3560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.16e-05,
+      "loss": 0.2014,
+      "step": 3570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1576470588235296e-05,
+      "loss": 0.2022,
+      "step": 3580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.155294117647059e-05,
+      "loss": 0.2031,
+      "step": 3590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1529411764705882e-05,
+      "loss": 0.1994,
+      "step": 3600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1505882352941177e-05,
+      "loss": 0.1987,
+      "step": 3610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148235294117647e-05,
+      "loss": 0.2021,
+      "step": 3620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1458823529411766e-05,
+      "loss": 0.202,
+      "step": 3630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.143529411764706e-05,
+      "loss": 0.2036,
+      "step": 3640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1411764705882353e-05,
+      "loss": 0.2004,
+      "step": 3650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1388235294117649e-05,
+      "loss": 0.1974,
+      "step": 3660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1364705882352942e-05,
+      "loss": 0.2004,
+      "step": 3670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1341176470588236e-05,
+      "loss": 0.1999,
+      "step": 3680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1317647058823531e-05,
+      "loss": 0.2016,
+      "step": 3690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1294117647058825e-05,
+      "loss": 0.2027,
+      "step": 3700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1270588235294117e-05,
+      "loss": 0.1962,
+      "step": 3710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1247058823529414e-05,
+      "loss": 0.2023,
+      "step": 3720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1223529411764706e-05,
+      "loss": 0.1987,
+      "step": 3730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.2011,
+      "step": 3740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.2001,
+      "step": 3750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1152941176470588e-05,
+      "loss": 0.2024,
+      "step": 3760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1129411764705884e-05,
+      "loss": 0.1977,
+      "step": 3770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1105882352941177e-05,
+      "loss": 0.1991,
+      "step": 3780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1082352941176471e-05,
+      "loss": 0.2028,
+      "step": 3790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1058823529411766e-05,
+      "loss": 0.1989,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.103529411764706e-05,
+      "loss": 0.2008,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1011764705882354e-05,
+      "loss": 0.2034,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0988235294117649e-05,
+      "loss": 0.1994,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0964705882352941e-05,
+      "loss": 0.2041,
+      "step": 3840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0941176470588238e-05,
+      "loss": 0.2019,
+      "step": 3850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.091764705882353e-05,
+      "loss": 0.2003,
+      "step": 3860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0894117647058824e-05,
+      "loss": 0.2002,
+      "step": 3870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0870588235294119e-05,
+      "loss": 0.2006,
+      "step": 3880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0847058823529413e-05,
+      "loss": 0.1983,
+      "step": 3890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0823529411764706e-05,
+      "loss": 0.1984,
+      "step": 3900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.1978,
+      "step": 3910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0776470588235295e-05,
+      "loss": 0.1972,
+      "step": 3920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0752941176470589e-05,
+      "loss": 0.2001,
+      "step": 3930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0729411764705884e-05,
+      "loss": 0.1976,
+      "step": 3940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0705882352941178e-05,
+      "loss": 0.2017,
+      "step": 3950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.068235294117647e-05,
+      "loss": 0.1975,
+      "step": 3960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0658823529411765e-05,
+      "loss": 0.2006,
+      "step": 3970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0635294117647059e-05,
+      "loss": 0.197,
+      "step": 3980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0611764705882354e-05,
+      "loss": 0.2016,
+      "step": 3990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.1978,
+      "step": 4000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0564705882352941e-05,
+      "loss": 0.1956,
+      "step": 4010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0541176470588237e-05,
+      "loss": 0.1967,
+      "step": 4020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.051764705882353e-05,
+      "loss": 0.1953,
+      "step": 4030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0494117647058824e-05,
+      "loss": 0.1995,
+      "step": 4040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.047058823529412e-05,
+      "loss": 0.1993,
+      "step": 4050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0447058823529413e-05,
+      "loss": 0.1969,
+      "step": 4060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0423529411764705e-05,
+      "loss": 0.2,
+      "step": 4070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.04e-05,
+      "loss": 0.195,
+      "step": 4080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0376470588235294e-05,
+      "loss": 0.1998,
+      "step": 4090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.035294117647059e-05,
+      "loss": 0.1985,
+      "step": 4100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0329411764705883e-05,
+      "loss": 0.1981,
+      "step": 4110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0305882352941177e-05,
+      "loss": 0.2001,
+      "step": 4120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0282352941176472e-05,
+      "loss": 0.2018,
+      "step": 4130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0258823529411766e-05,
+      "loss": 0.197,
+      "step": 4140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.023529411764706e-05,
+      "loss": 0.1998,
+      "step": 4150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0211764705882355e-05,
+      "loss": 0.1983,
+      "step": 4160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0188235294117648e-05,
+      "loss": 0.1993,
+      "step": 4170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.016470588235294e-05,
+      "loss": 0.1989,
+      "step": 4180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0141176470588237e-05,
+      "loss": 0.1958,
+      "step": 4190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.011764705882353e-05,
+      "loss": 0.1973,
+      "step": 4200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0094117647058825e-05,
+      "loss": 0.1969,
+      "step": 4210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0070588235294118e-05,
+      "loss": 0.1945,
+      "step": 4220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0047058823529412e-05,
+      "loss": 0.1992,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0023529411764707e-05,
+      "loss": 0.1965,
+      "step": 4240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.1955,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.18427707254886627,
+      "eval_runtime": 116.059,
+      "eval_samples_per_second": 146.477,
+      "eval_steps_per_second": 9.159,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.976470588235294e-06,
+      "loss": 0.1999,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.952941176470588e-06,
+      "loss": 0.1962,
+      "step": 4270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.929411764705883e-06,
+      "loss": 0.1973,
+      "step": 4280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.905882352941177e-06,
+      "loss": 0.1961,
+      "step": 4290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.882352941176472e-06,
+      "loss": 0.1962,
+      "step": 4300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.858823529411764e-06,
+      "loss": 0.1984,
+      "step": 4310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.83529411764706e-06,
+      "loss": 0.1949,
+      "step": 4320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.811764705882353e-06,
+      "loss": 0.1975,
+      "step": 4330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.788235294117649e-06,
+      "loss": 0.2107,
+      "step": 4340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.764705882352942e-06,
+      "loss": 0.1995,
+      "step": 4350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.741176470588236e-06,
+      "loss": 0.1956,
+      "step": 4360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.717647058823531e-06,
+      "loss": 0.1978,
+      "step": 4370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.694117647058823e-06,
+      "loss": 0.1983,
+      "step": 4380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.670588235294119e-06,
+      "loss": 0.2008,
+      "step": 4390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.647058823529412e-06,
+      "loss": 0.1971,
+      "step": 4400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.623529411764708e-06,
+      "loss": 0.1972,
+      "step": 4410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.1977,
+      "step": 4420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.576470588235295e-06,
+      "loss": 0.2014,
+      "step": 4430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.552941176470589e-06,
+      "loss": 0.1947,
+      "step": 4440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.529411764705882e-06,
+      "loss": 0.197,
+      "step": 4450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.505882352941178e-06,
+      "loss": 0.1979,
+      "step": 4460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.482352941176471e-06,
+      "loss": 0.1963,
+      "step": 4470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.458823529411767e-06,
+      "loss": 0.1984,
+      "step": 4480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.435294117647058e-06,
+      "loss": 0.1944,
+      "step": 4490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.1946,
+      "step": 4500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.388235294117647e-06,
+      "loss": 0.1942,
+      "step": 4510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.364705882352943e-06,
+      "loss": 0.197,
+      "step": 4520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.341176470588236e-06,
+      "loss": 0.1972,
+      "step": 4530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.31764705882353e-06,
+      "loss": 0.2001,
+      "step": 4540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.294117647058824e-06,
+      "loss": 0.1983,
+      "step": 4550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.270588235294117e-06,
+      "loss": 0.1924,
+      "step": 4560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.247058823529413e-06,
+      "loss": 0.1932,
+      "step": 4570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.223529411764706e-06,
+      "loss": 0.198,
+      "step": 4580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.1962,
+      "step": 4590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.176470588235294e-06,
+      "loss": 0.1969,
+      "step": 4600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.152941176470589e-06,
+      "loss": 0.2002,
+      "step": 4610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.129411764705883e-06,
+      "loss": 0.1964,
+      "step": 4620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.105882352941178e-06,
+      "loss": 0.1951,
+      "step": 4630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.082352941176472e-06,
+      "loss": 0.1943,
+      "step": 4640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.058823529411765e-06,
+      "loss": 0.1967,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.03529411764706e-06,
+      "loss": 0.1969,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.011764705882353e-06,
+      "loss": 0.1956,
+      "step": 4670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.988235294117648e-06,
+      "loss": 0.1947,
+      "step": 4680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.964705882352942e-06,
+      "loss": 0.1965,
+      "step": 4690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.941176470588237e-06,
+      "loss": 0.199,
+      "step": 4700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.91764705882353e-06,
+      "loss": 0.1995,
+      "step": 4710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.894117647058824e-06,
+      "loss": 0.195,
+      "step": 4720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.870588235294118e-06,
+      "loss": 0.194,
+      "step": 4730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.847058823529413e-06,
+      "loss": 0.1961,
+      "step": 4740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.1987,
+      "step": 4750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.8e-06,
+      "loss": 0.1958,
+      "step": 4760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.776470588235296e-06,
+      "loss": 0.1927,
+      "step": 4770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.752941176470588e-06,
+      "loss": 0.2,
+      "step": 4780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.729411764705883e-06,
+      "loss": 0.2001,
+      "step": 4790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.705882352941177e-06,
+      "loss": 0.1953,
+      "step": 4800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.682352941176472e-06,
+      "loss": 0.1971,
+      "step": 4810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.658823529411766e-06,
+      "loss": 0.1957,
+      "step": 4820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.63529411764706e-06,
+      "loss": 0.1963,
+      "step": 4830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.611764705882355e-06,
+      "loss": 0.1965,
+      "step": 4840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.588235294117647e-06,
+      "loss": 0.1971,
+      "step": 4850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.564705882352942e-06,
+      "loss": 0.1965,
+      "step": 4860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.541176470588236e-06,
+      "loss": 0.195,
+      "step": 4870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.517647058823531e-06,
+      "loss": 0.1974,
+      "step": 4880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.494117647058825e-06,
+      "loss": 0.1953,
+      "step": 4890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.470588235294118e-06,
+      "loss": 0.1958,
+      "step": 4900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.447058823529412e-06,
+      "loss": 0.1943,
+      "step": 4910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.423529411764707e-06,
+      "loss": 0.1981,
+      "step": 4920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.1934,
+      "step": 4930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.376470588235295e-06,
+      "loss": 0.1947,
+      "step": 4940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.35294117647059e-06,
+      "loss": 0.1902,
+      "step": 4950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.329411764705882e-06,
+      "loss": 0.195,
+      "step": 4960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.305882352941177e-06,
+      "loss": 0.1946,
+      "step": 4970
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.28235294117647e-06,
+      "loss": 0.1973,
+      "step": 4980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.258823529411766e-06,
+      "loss": 0.1957,
+      "step": 4990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.1966,
+      "step": 5000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.211764705882353e-06,
+      "loss": 0.1948,
+      "step": 5010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.188235294117649e-06,
+      "loss": 0.1966,
+      "step": 5020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.164705882352942e-06,
+      "loss": 0.1949,
+      "step": 5030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.141176470588236e-06,
+      "loss": 0.1924,
+      "step": 5040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.11764705882353e-06,
+      "loss": 0.1957,
+      "step": 5050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.094117647058825e-06,
+      "loss": 0.1982,
+      "step": 5060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.070588235294117e-06,
+      "loss": 0.1948,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.047058823529412e-06,
+      "loss": 0.1961,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.023529411764706e-06,
+      "loss": 0.1924,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.1935,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.976470588235295e-06,
+      "loss": 0.1936,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.952941176470589e-06,
+      "loss": 0.1946,
+      "step": 5120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.929411764705884e-06,
+      "loss": 0.1948,
+      "step": 5130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.905882352941176e-06,
+      "loss": 0.1926,
+      "step": 5140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.882352941176471e-06,
+      "loss": 0.1943,
+      "step": 5150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.858823529411765e-06,
+      "loss": 0.1932,
+      "step": 5160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.83529411764706e-06,
+      "loss": 0.1954,
+      "step": 5170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.811764705882354e-06,
+      "loss": 0.196,
+      "step": 5180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.788235294117648e-06,
+      "loss": 0.1958,
+      "step": 5190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.764705882352941e-06,
+      "loss": 0.1941,
+      "step": 5200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.741176470588237e-06,
+      "loss": 0.1926,
+      "step": 5210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.71764705882353e-06,
+      "loss": 0.1922,
+      "step": 5220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.694117647058824e-06,
+      "loss": 0.1955,
+      "step": 5230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.670588235294119e-06,
+      "loss": 0.195,
+      "step": 5240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.1926,
+      "step": 5250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.6235294117647064e-06,
+      "loss": 0.1954,
+      "step": 5260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.1934,
+      "step": 5270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.576470588235295e-06,
+      "loss": 0.1966,
+      "step": 5280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.552941176470588e-06,
+      "loss": 0.1899,
+      "step": 5290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.529411764705883e-06,
+      "loss": 0.1946,
+      "step": 5300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.505882352941177e-06,
+      "loss": 0.1962,
+      "step": 5310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.482352941176472e-06,
+      "loss": 0.1945,
+      "step": 5320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.458823529411765e-06,
+      "loss": 0.1967,
+      "step": 5330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.43529411764706e-06,
+      "loss": 0.196,
+      "step": 5340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.4117647058823535e-06,
+      "loss": 0.1952,
+      "step": 5350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.388235294117647e-06,
+      "loss": 0.1921,
+      "step": 5360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.364705882352942e-06,
+      "loss": 0.1947,
+      "step": 5370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.341176470588236e-06,
+      "loss": 0.1937,
+      "step": 5380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.317647058823531e-06,
+      "loss": 0.1909,
+      "step": 5390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.294117647058823e-06,
+      "loss": 0.1942,
+      "step": 5400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.270588235294118e-06,
+      "loss": 0.1948,
+      "step": 5410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.247058823529412e-06,
+      "loss": 0.1928,
+      "step": 5420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.223529411764707e-06,
+      "loss": 0.1931,
+      "step": 5430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.1985,
+      "step": 5440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.176470588235295e-06,
+      "loss": 0.1965,
+      "step": 5450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.1529411764705895e-06,
+      "loss": 0.1951,
+      "step": 5460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.129411764705882e-06,
+      "loss": 0.1964,
+      "step": 5470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.105882352941177e-06,
+      "loss": 0.1975,
+      "step": 5480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.082352941176471e-06,
+      "loss": 0.1912,
+      "step": 5490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.1932,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0352941176470594e-06,
+      "loss": 0.1947,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.011764705882353e-06,
+      "loss": 0.1936,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.9882352941176476e-06,
+      "loss": 0.1934,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.964705882352941e-06,
+      "loss": 0.1948,
+      "step": 5540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.941176470588236e-06,
+      "loss": 0.1946,
+      "step": 5550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.91764705882353e-06,
+      "loss": 0.1928,
+      "step": 5560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.894117647058825e-06,
+      "loss": 0.1914,
+      "step": 5570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.8705882352941175e-06,
+      "loss": 0.193,
+      "step": 5580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.847058823529412e-06,
+      "loss": 0.1936,
+      "step": 5590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8235294117647065e-06,
+      "loss": 0.1938,
+      "step": 5600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.1938,
+      "step": 5610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.776470588235295e-06,
+      "loss": 0.1982,
+      "step": 5620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.752941176470589e-06,
+      "loss": 0.1952,
+      "step": 5630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.729411764705884e-06,
+      "loss": 0.1902,
+      "step": 5640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.705882352941176e-06,
+      "loss": 0.194,
+      "step": 5650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.682352941176471e-06,
+      "loss": 0.1934,
+      "step": 5660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.658823529411765e-06,
+      "loss": 0.1939,
+      "step": 5670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.63529411764706e-06,
+      "loss": 0.1925,
+      "step": 5680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.611764705882353e-06,
+      "loss": 0.1934,
+      "step": 5690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.588235294117647e-06,
+      "loss": 0.193,
+      "step": 5700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.564705882352942e-06,
+      "loss": 0.1953,
+      "step": 5710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.541176470588236e-06,
+      "loss": 0.1966,
+      "step": 5720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.51764705882353e-06,
+      "loss": 0.1961,
+      "step": 5730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.494117647058824e-06,
+      "loss": 0.193,
+      "step": 5740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.1974,
+      "step": 5750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4470588235294116e-06,
+      "loss": 0.1952,
+      "step": 5760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.423529411764706e-06,
+      "loss": 0.1933,
+      "step": 5770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.1941,
+      "step": 5780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.376470588235295e-06,
+      "loss": 0.1949,
+      "step": 5790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.352941176470589e-06,
+      "loss": 0.1923,
+      "step": 5800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.329411764705883e-06,
+      "loss": 0.1987,
+      "step": 5810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.305882352941177e-06,
+      "loss": 0.1929,
+      "step": 5820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.282352941176471e-06,
+      "loss": 0.1922,
+      "step": 5830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.258823529411765e-06,
+      "loss": 0.1952,
+      "step": 5840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2352941176470595e-06,
+      "loss": 0.1941,
+      "step": 5850
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.211764705882354e-06,
+      "loss": 0.1917,
+      "step": 5860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.188235294117647e-06,
+      "loss": 0.1976,
+      "step": 5870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.164705882352941e-06,
+      "loss": 0.1907,
+      "step": 5880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.141176470588236e-06,
+      "loss": 0.1961,
+      "step": 5890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.11764705882353e-06,
+      "loss": 0.1921,
+      "step": 5900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.094117647058824e-06,
+      "loss": 0.1939,
+      "step": 5910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.070588235294118e-06,
+      "loss": 0.1919,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.047058823529413e-06,
+      "loss": 0.193,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.023529411764706e-06,
+      "loss": 0.1936,
+      "step": 5940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6e-06,
+      "loss": 0.1926,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.976470588235295e-06,
+      "loss": 0.1922,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.952941176470589e-06,
+      "loss": 0.1948,
+      "step": 5970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.929411764705883e-06,
+      "loss": 0.191,
+      "step": 5980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9058823529411764e-06,
+      "loss": 0.1941,
+      "step": 5990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.1941,
+      "step": 6000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.858823529411765e-06,
+      "loss": 0.1939,
+      "step": 6010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.835294117647059e-06,
+      "loss": 0.196,
+      "step": 6020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8117647058823536e-06,
+      "loss": 0.1962,
+      "step": 6030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.788235294117648e-06,
+      "loss": 0.1932,
+      "step": 6040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.764705882352941e-06,
+      "loss": 0.1943,
+      "step": 6050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.741176470588235e-06,
+      "loss": 0.1918,
+      "step": 6060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.71764705882353e-06,
+      "loss": 0.1913,
+      "step": 6070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.694117647058824e-06,
+      "loss": 0.1923,
+      "step": 6080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.670588235294118e-06,
+      "loss": 0.1951,
+      "step": 6090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6470588235294125e-06,
+      "loss": 0.1914,
+      "step": 6100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.623529411764707e-06,
+      "loss": 0.1935,
+      "step": 6110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.1902,
+      "step": 6120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.576470588235294e-06,
+      "loss": 0.1921,
+      "step": 6130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.552941176470589e-06,
+      "loss": 0.1917,
+      "step": 6140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.529411764705883e-06,
+      "loss": 0.1886,
+      "step": 6150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.505882352941177e-06,
+      "loss": 0.1917,
+      "step": 6160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4823529411764705e-06,
+      "loss": 0.1942,
+      "step": 6170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.458823529411765e-06,
+      "loss": 0.1918,
+      "step": 6180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4352941176470595e-06,
+      "loss": 0.1921,
+      "step": 6190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.411764705882353e-06,
+      "loss": 0.192,
+      "step": 6200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.388235294117648e-06,
+      "loss": 0.192,
+      "step": 6210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.364705882352942e-06,
+      "loss": 0.1936,
+      "step": 6220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.341176470588235e-06,
+      "loss": 0.192,
+      "step": 6230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.317647058823529e-06,
+      "loss": 0.193,
+      "step": 6240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.1909,
+      "step": 6250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.270588235294118e-06,
+      "loss": 0.1902,
+      "step": 6260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.247058823529412e-06,
+      "loss": 0.1901,
+      "step": 6270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.2235294117647065e-06,
+      "loss": 0.1935,
+      "step": 6280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.2e-06,
+      "loss": 0.1933,
+      "step": 6290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.176470588235295e-06,
+      "loss": 0.1925,
+      "step": 6300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.152941176470588e-06,
+      "loss": 0.193,
+      "step": 6310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.129411764705883e-06,
+      "loss": 0.1908,
+      "step": 6320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.105882352941177e-06,
+      "loss": 0.1908,
+      "step": 6330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.08235294117647e-06,
+      "loss": 0.1901,
+      "step": 6340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.058823529411765e-06,
+      "loss": 0.1915,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.035294117647059e-06,
+      "loss": 0.1919,
+      "step": 6360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.011764705882354e-06,
+      "loss": 0.1929,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.988235294117647e-06,
+      "loss": 0.1894,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.964705882352942e-06,
+      "loss": 0.1934,
+      "step": 6390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.941176470588236e-06,
+      "loss": 0.1957,
+      "step": 6400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.91764705882353e-06,
+      "loss": 0.1918,
+      "step": 6410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.894117647058824e-06,
+      "loss": 0.1897,
+      "step": 6420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.870588235294118e-06,
+      "loss": 0.1931,
+      "step": 6430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.847058823529412e-06,
+      "loss": 0.1959,
+      "step": 6440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.823529411764706e-06,
+      "loss": 0.1918,
+      "step": 6450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.1925,
+      "step": 6460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.776470588235294e-06,
+      "loss": 0.1887,
+      "step": 6470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.752941176470589e-06,
+      "loss": 0.1874,
+      "step": 6480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.729411764705883e-06,
+      "loss": 0.1906,
+      "step": 6490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.1947,
+      "step": 6500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.682352941176471e-06,
+      "loss": 0.1928,
+      "step": 6510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.658823529411765e-06,
+      "loss": 0.1915,
+      "step": 6520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.635294117647059e-06,
+      "loss": 0.1908,
+      "step": 6530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.611764705882353e-06,
+      "loss": 0.1946,
+      "step": 6540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.588235294117647e-06,
+      "loss": 0.1939,
+      "step": 6550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.564705882352941e-06,
+      "loss": 0.1908,
+      "step": 6560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.541176470588236e-06,
+      "loss": 0.1898,
+      "step": 6570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.51764705882353e-06,
+      "loss": 0.1911,
+      "step": 6580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.494117647058824e-06,
+      "loss": 0.1941,
+      "step": 6590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4705882352941184e-06,
+      "loss": 0.1907,
+      "step": 6600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.447058823529412e-06,
+      "loss": 0.1935,
+      "step": 6610
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.423529411764707e-06,
+      "loss": 0.1948,
+      "step": 6620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4e-06,
+      "loss": 0.194,
+      "step": 6630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.376470588235294e-06,
+      "loss": 0.1884,
+      "step": 6640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.352941176470588e-06,
+      "loss": 0.1928,
+      "step": 6650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.329411764705883e-06,
+      "loss": 0.1934,
+      "step": 6660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.305882352941177e-06,
+      "loss": 0.194,
+      "step": 6670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.282352941176471e-06,
+      "loss": 0.1913,
+      "step": 6680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2588235294117655e-06,
+      "loss": 0.193,
+      "step": 6690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.235294117647059e-06,
+      "loss": 0.1929,
+      "step": 6700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211764705882354e-06,
+      "loss": 0.1917,
+      "step": 6710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.188235294117647e-06,
+      "loss": 0.1925,
+      "step": 6720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.164705882352941e-06,
+      "loss": 0.1961,
+      "step": 6730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.141176470588235e-06,
+      "loss": 0.1922,
+      "step": 6740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.205,
+      "step": 6750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.094117647058824e-06,
+      "loss": 0.19,
+      "step": 6760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.070588235294118e-06,
+      "loss": 0.1976,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0470588235294125e-06,
+      "loss": 0.1925,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.023529411764706e-06,
+      "loss": 0.1901,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.1911,
+      "step": 6800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976470588235294e-06,
+      "loss": 0.1911,
+      "step": 6810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.952941176470588e-06,
+      "loss": 0.1894,
+      "step": 6820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9294117647058824e-06,
+      "loss": 0.1884,
+      "step": 6830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.905882352941177e-06,
+      "loss": 0.1931,
+      "step": 6840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.882352941176471e-06,
+      "loss": 0.1891,
+      "step": 6850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.858823529411765e-06,
+      "loss": 0.1922,
+      "step": 6860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8352941176470596e-06,
+      "loss": 0.1908,
+      "step": 6870
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8117647058823532e-06,
+      "loss": 0.1892,
+      "step": 6880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7882352941176477e-06,
+      "loss": 0.1925,
+      "step": 6890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7647058823529414e-06,
+      "loss": 0.1918,
+      "step": 6900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.741176470588236e-06,
+      "loss": 0.1896,
+      "step": 6910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.71764705882353e-06,
+      "loss": 0.193,
+      "step": 6920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6941176470588236e-06,
+      "loss": 0.194,
+      "step": 6930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.670588235294118e-06,
+      "loss": 0.1906,
+      "step": 6940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6470588235294117e-06,
+      "loss": 0.1892,
+      "step": 6950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.623529411764706e-06,
+      "loss": 0.1943,
+      "step": 6960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.1902,
+      "step": 6970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5764705882352948e-06,
+      "loss": 0.1873,
+      "step": 6980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5529411764705884e-06,
+      "loss": 0.1895,
+      "step": 6990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.1887,
+      "step": 7000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5058823529411765e-06,
+      "loss": 0.1938,
+      "step": 7010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4823529411764706e-06,
+      "loss": 0.1883,
+      "step": 7020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.458823529411765e-06,
+      "loss": 0.1922,
+      "step": 7030
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4352941176470587e-06,
+      "loss": 0.1917,
+      "step": 7040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4117647058823532e-06,
+      "loss": 0.1903,
+      "step": 7050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3882352941176473e-06,
+      "loss": 0.1939,
+      "step": 7060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.364705882352942e-06,
+      "loss": 0.1922,
+      "step": 7070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3411764705882354e-06,
+      "loss": 0.1947,
+      "step": 7080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.31764705882353e-06,
+      "loss": 0.1929,
+      "step": 7090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2941176470588236e-06,
+      "loss": 0.1915,
+      "step": 7100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.270588235294118e-06,
+      "loss": 0.1932,
+      "step": 7110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.247058823529412e-06,
+      "loss": 0.1891,
+      "step": 7120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2235294117647058e-06,
+      "loss": 0.191,
+      "step": 7130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.1907,
+      "step": 7140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1764705882352943e-06,
+      "loss": 0.19,
+      "step": 7150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1529411764705884e-06,
+      "loss": 0.1945,
+      "step": 7160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1294117647058825e-06,
+      "loss": 0.1913,
+      "step": 7170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.105882352941177e-06,
+      "loss": 0.1912,
+      "step": 7180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0823529411764706e-06,
+      "loss": 0.1923,
+      "step": 7190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.058823529411765e-06,
+      "loss": 0.192,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.035294117647059e-06,
+      "loss": 0.1892,
+      "step": 7210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.011764705882353e-06,
+      "loss": 0.1905,
+      "step": 7220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9882352941176473e-06,
+      "loss": 0.192,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9647058823529414e-06,
+      "loss": 0.1892,
+      "step": 7240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.191,
+      "step": 7250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9176470588235295e-06,
+      "loss": 0.1912,
+      "step": 7260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.894117647058824e-06,
+      "loss": 0.1903,
+      "step": 7270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8705882352941177e-06,
+      "loss": 0.1895,
+      "step": 7280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.847058823529412e-06,
+      "loss": 0.1922,
+      "step": 7290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8235294117647062e-06,
+      "loss": 0.1937,
+      "step": 7300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.189,
+      "step": 7310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7764705882352944e-06,
+      "loss": 0.1923,
+      "step": 7320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7529411764705884e-06,
+      "loss": 0.1914,
+      "step": 7330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7294117647058825e-06,
+      "loss": 0.1901,
+      "step": 7340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7058823529411766e-06,
+      "loss": 0.1906,
+      "step": 7350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.682352941176471e-06,
+      "loss": 0.1911,
+      "step": 7360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6588235294117647e-06,
+      "loss": 0.19,
+      "step": 7370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.635294117647059e-06,
+      "loss": 0.192,
+      "step": 7380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6117647058823533e-06,
+      "loss": 0.1895,
+      "step": 7390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5882352941176473e-06,
+      "loss": 0.1928,
+      "step": 7400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5647058823529414e-06,
+      "loss": 0.19,
+      "step": 7410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.541176470588235e-06,
+      "loss": 0.1912,
+      "step": 7420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5176470588235295e-06,
+      "loss": 0.1915,
+      "step": 7430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4941176470588236e-06,
+      "loss": 0.1888,
+      "step": 7440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.470588235294118e-06,
+      "loss": 0.1929,
+      "step": 7450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.447058823529412e-06,
+      "loss": 0.1915,
+      "step": 7460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.423529411764706e-06,
+      "loss": 0.1894,
+      "step": 7470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.1926,
+      "step": 7480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3764705882352944e-06,
+      "loss": 0.1892,
+      "step": 7490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.1916,
+      "step": 7500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3294117647058825e-06,
+      "loss": 0.1891,
+      "step": 7510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3058823529411766e-06,
+      "loss": 0.1965,
+      "step": 7520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2823529411764707e-06,
+      "loss": 0.1927,
+      "step": 7530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.258823529411765e-06,
+      "loss": 0.1892,
+      "step": 7540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2352941176470592e-06,
+      "loss": 0.1922,
+      "step": 7550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2117647058823533e-06,
+      "loss": 0.1915,
+      "step": 7560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.188235294117647e-06,
+      "loss": 0.1896,
+      "step": 7570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1647058823529414e-06,
+      "loss": 0.1894,
+      "step": 7580
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1411764705882355e-06,
+      "loss": 0.1901,
+      "step": 7590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1176470588235296e-06,
+      "loss": 0.1891,
+      "step": 7600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0941176470588236e-06,
+      "loss": 0.1899,
+      "step": 7610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0705882352941177e-06,
+      "loss": 0.1925,
+      "step": 7620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.047058823529412e-06,
+      "loss": 0.1902,
+      "step": 7630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0235294117647063e-06,
+      "loss": 0.1926,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.1917,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.976470588235294e-06,
+      "loss": 0.1912,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9529411764705885e-06,
+      "loss": 0.1926,
+      "step": 7670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9294117647058825e-06,
+      "loss": 0.1914,
+      "step": 7680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9058823529411766e-06,
+      "loss": 0.1868,
+      "step": 7690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8823529411764707e-06,
+      "loss": 0.192,
+      "step": 7700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.858823529411765e-06,
+      "loss": 0.1905,
+      "step": 7710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.835294117647059e-06,
+      "loss": 0.1889,
+      "step": 7720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.811764705882353e-06,
+      "loss": 0.1888,
+      "step": 7730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7882352941176474e-06,
+      "loss": 0.1906,
+      "step": 7740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 0.1931,
+      "step": 7750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7411764705882353e-06,
+      "loss": 0.1888,
+      "step": 7760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7176470588235294e-06,
+      "loss": 0.1896,
+      "step": 7770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6941176470588237e-06,
+      "loss": 0.1901,
+      "step": 7780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6705882352941177e-06,
+      "loss": 0.1882,
+      "step": 7790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6470588235294118e-06,
+      "loss": 0.1905,
+      "step": 7800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.623529411764706e-06,
+      "loss": 0.1911,
+      "step": 7810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.1883,
+      "step": 7820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5764705882352942e-06,
+      "loss": 0.191,
+      "step": 7830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5529411764705885e-06,
+      "loss": 0.1909,
+      "step": 7840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5294117647058826e-06,
+      "loss": 0.1906,
+      "step": 7850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5058823529411764e-06,
+      "loss": 0.1907,
+      "step": 7860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4823529411764707e-06,
+      "loss": 0.1884,
+      "step": 7870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4588235294117648e-06,
+      "loss": 0.1892,
+      "step": 7880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4352941176470588e-06,
+      "loss": 0.1903,
+      "step": 7890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4117647058823531e-06,
+      "loss": 0.1919,
+      "step": 7900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3882352941176472e-06,
+      "loss": 0.1942,
+      "step": 7910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3647058823529413e-06,
+      "loss": 0.1903,
+      "step": 7920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3411764705882355e-06,
+      "loss": 0.1903,
+      "step": 7930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3176470588235296e-06,
+      "loss": 0.1914,
+      "step": 7940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2941176470588237e-06,
+      "loss": 0.1902,
+      "step": 7950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2705882352941175e-06,
+      "loss": 0.1909,
+      "step": 7960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2470588235294118e-06,
+      "loss": 0.1902,
+      "step": 7970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.223529411764706e-06,
+      "loss": 0.1901,
+      "step": 7980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.1922,
+      "step": 7990
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.1879,
+      "step": 8000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1529411764705883e-06,
+      "loss": 0.1897,
+      "step": 8010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1294117647058826e-06,
+      "loss": 0.192,
+      "step": 8020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1058823529411766e-06,
+      "loss": 0.195,
+      "step": 8030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0823529411764707e-06,
+      "loss": 0.1921,
+      "step": 8040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0588235294117648e-06,
+      "loss": 0.1904,
+      "step": 8050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0352941176470589e-06,
+      "loss": 0.1904,
+      "step": 8060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0117647058823531e-06,
+      "loss": 0.1923,
+      "step": 8070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.88235294117647e-07,
+      "loss": 0.1885,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.647058823529413e-07,
+      "loss": 0.1895,
+      "step": 8090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.411764705882353e-07,
+      "loss": 0.1899,
+      "step": 8100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.176470588235295e-07,
+      "loss": 0.1909,
+      "step": 8110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.941176470588237e-07,
+      "loss": 0.1939,
+      "step": 8120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.705882352941177e-07,
+      "loss": 0.1877,
+      "step": 8130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.470588235294118e-07,
+      "loss": 0.1917,
+      "step": 8140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.235294117647059e-07,
+      "loss": 0.1912,
+      "step": 8150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.1939,
+      "step": 8160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.764705882352942e-07,
+      "loss": 0.1919,
+      "step": 8170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.529411764705882e-07,
+      "loss": 0.1917,
+      "step": 8180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.294117647058824e-07,
+      "loss": 0.1879,
+      "step": 8190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.058823529411766e-07,
+      "loss": 0.1924,
+      "step": 8200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.823529411764706e-07,
+      "loss": 0.1898,
+      "step": 8210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.588235294117648e-07,
+      "loss": 0.1878,
+      "step": 8220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.352941176470588e-07,
+      "loss": 0.1893,
+      "step": 8230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.11764705882353e-07,
+      "loss": 0.1905,
+      "step": 8240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.1916,
+      "step": 8250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.647058823529413e-07,
+      "loss": 0.1915,
+      "step": 8260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.411764705882354e-07,
+      "loss": 0.1867,
+      "step": 8270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.176470588235294e-07,
+      "loss": 0.1904,
+      "step": 8280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.941176470588235e-07,
+      "loss": 0.1904,
+      "step": 8290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7058823529411767e-07,
+      "loss": 0.1927,
+      "step": 8300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4705882352941184e-07,
+      "loss": 0.1899,
+      "step": 8310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.235294117647059e-07,
+      "loss": 0.1888,
+      "step": 8320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.1887,
+      "step": 8330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.764705882352941e-07,
+      "loss": 0.1892,
+      "step": 8340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.529411764705883e-07,
+      "loss": 0.1865,
+      "step": 8350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.294117647058824e-07,
+      "loss": 0.1884,
+      "step": 8360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.058823529411765e-07,
+      "loss": 0.1908,
+      "step": 8370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8235294117647064e-07,
+      "loss": 0.1991,
+      "step": 8380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.588235294117647e-07,
+      "loss": 0.1895,
+      "step": 8390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3529411764705883e-07,
+      "loss": 0.1904,
+      "step": 8400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1176470588235296e-07,
+      "loss": 0.19,
+      "step": 8410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8823529411764705e-07,
+      "loss": 0.1905,
+      "step": 8420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.647058823529412e-07,
+      "loss": 0.1908,
+      "step": 8430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4117647058823532e-07,
+      "loss": 0.1886,
+      "step": 8440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1764705882352942e-07,
+      "loss": 0.1868,
+      "step": 8450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.411764705882353e-08,
+      "loss": 0.1886,
+      "step": 8460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.058823529411766e-08,
+      "loss": 0.192,
+      "step": 8470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.705882352941176e-08,
+      "loss": 0.1884,
+      "step": 8480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.352941176470588e-08,
+      "loss": 0.1881,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1948,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.18093784153461456,
+      "eval_runtime": 116.1321,
+      "eval_samples_per_second": 146.385,
+      "eval_steps_per_second": 9.153,
+      "step": 8500
+    }
+  ],
+  "max_steps": 8500,
+  "num_train_epochs": 2,
+  "total_flos": 4442044760064000.0,
+  "trial_name": null,
+  "trial_params": null
+}