diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,25296 @@
+{
+  "best_metric": 1.1405349969863892,
+  "best_model_checkpoint": "pretraining_outputs_full_eval_intrinsic_easy_nopronpt/checkpoint-390000",
+  "epoch": 1.7935128639710167,
+  "global_step": 400000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.85e-07,
+      "loss": 2.9676,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.85e-07,
+      "loss": 2.2829,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.85e-07,
+      "loss": 2.0164,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.185e-06,
+      "loss": 1.8594,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4850000000000002e-06,
+      "loss": 1.8078,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.785e-06,
+      "loss": 1.7464,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0850000000000004e-06,
+      "loss": 1.708,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.385e-06,
+      "loss": 1.6553,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.685e-06,
+      "loss": 1.6595,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.982e-06,
+      "loss": 1.5951,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.282e-06,
+      "loss": 1.6185,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.582e-06,
+      "loss": 1.6282,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8819999999999994e-06,
+      "loss": 1.6045,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.182e-06,
+      "loss": 1.5522,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4820000000000005e-06,
+      "loss": 1.5741,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782e-06,
+      "loss": 1.5447,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.082e-06,
+      "loss": 1.5024,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.382e-06,
+      "loss": 1.5379,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.682000000000001e-06,
+      "loss": 1.5337,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.982e-06,
+      "loss": 1.5011,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282e-06,
+      "loss": 1.5182,
+      "step": 2100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.582000000000001e-06,
+      "loss": 1.4732,
+      "step": 2200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.882e-06,
+      "loss": 1.48,
+      "step": 2300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.182e-06,
+      "loss": 1.508,
+      "step": 2400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4820000000000005e-06,
+      "loss": 1.4707,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.9636703729629517,
+      "eval_runtime": 135.3491,
+      "eval_samples_per_second": 14.777,
+      "eval_steps_per_second": 0.924,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.782000000000001e-06,
+      "loss": 1.4525,
+      "step": 2600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.082e-06,
+      "loss": 1.4686,
+      "step": 2700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379e-06,
+      "loss": 1.4489,
+      "step": 2800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.679000000000001e-06,
+      "loss": 1.4372,
+      "step": 2900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.979e-06,
+      "loss": 1.4392,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.279e-06,
+      "loss": 1.4603,
+      "step": 3100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.579e-06,
+      "loss": 1.4308,
+      "step": 3200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.879e-06,
+      "loss": 1.4471,
+      "step": 3300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0179e-05,
+      "loss": 1.4567,
+      "step": 3400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0479e-05,
+      "loss": 1.415,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0779000000000001e-05,
+      "loss": 1.4615,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1079e-05,
+      "loss": 1.4438,
+      "step": 3700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1379e-05,
+      "loss": 1.4292,
+      "step": 3800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1679e-05,
+      "loss": 1.4371,
+      "step": 3900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1979e-05,
+      "loss": 1.4299,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2279e-05,
+      "loss": 1.415,
+      "step": 4100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2579000000000001e-05,
+      "loss": 1.4043,
+      "step": 4200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2879000000000002e-05,
+      "loss": 1.4097,
+      "step": 4300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3179e-05,
+      "loss": 1.406,
+      "step": 4400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3479e-05,
+      "loss": 1.3928,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3779e-05,
+      "loss": 1.3455,
+      "step": 4600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4079e-05,
+      "loss": 1.4007,
+      "step": 4700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4379e-05,
+      "loss": 1.3532,
+      "step": 4800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4679000000000001e-05,
+      "loss": 1.388,
+      "step": 4900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4979000000000002e-05,
+      "loss": 1.3594,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 1.837215781211853,
+      "eval_runtime": 136.1059,
+      "eval_samples_per_second": 14.694,
+      "eval_steps_per_second": 0.918,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5279e-05,
+      "loss": 1.3522,
+      "step": 5100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5579e-05,
+      "loss": 1.3171,
+      "step": 5200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5879e-05,
+      "loss": 1.3313,
+      "step": 5300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6179000000000002e-05,
+      "loss": 1.3396,
+      "step": 5400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6479e-05,
+      "loss": 1.3525,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6779e-05,
+      "loss": 1.318,
+      "step": 5600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7079e-05,
+      "loss": 1.3697,
+      "step": 5700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7379e-05,
+      "loss": 1.3748,
+      "step": 5800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7679e-05,
+      "loss": 1.3267,
+      "step": 5900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7979000000000002e-05,
+      "loss": 1.3655,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8279e-05,
+      "loss": 1.3704,
+      "step": 6100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8579e-05,
+      "loss": 1.3201,
+      "step": 6200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8879e-05,
+      "loss": 1.313,
+      "step": 6300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9179e-05,
+      "loss": 1.3585,
+      "step": 6400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9479e-05,
+      "loss": 1.3698,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9779e-05,
+      "loss": 1.2822,
+      "step": 6600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0079000000000002e-05,
+      "loss": 1.3266,
+      "step": 6700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0379000000000003e-05,
+      "loss": 1.3346,
+      "step": 6800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0679e-05,
+      "loss": 1.3228,
+      "step": 6900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0979e-05,
+      "loss": 1.3023,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1279e-05,
+      "loss": 1.3868,
+      "step": 7100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1579e-05,
+      "loss": 1.3614,
+      "step": 7200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1879e-05,
+      "loss": 1.3135,
+      "step": 7300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2179e-05,
+      "loss": 1.2921,
+      "step": 7400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2479e-05,
+      "loss": 1.3557,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 1.8817667961120605,
+      "eval_runtime": 133.6804,
+      "eval_samples_per_second": 14.961,
+      "eval_steps_per_second": 0.935,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2779e-05,
+      "loss": 1.3126,
+      "step": 7600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3079e-05,
+      "loss": 1.3126,
+      "step": 7700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3379e-05,
+      "loss": 1.3013,
+      "step": 7800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3679e-05,
+      "loss": 1.3007,
+      "step": 7900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3976000000000002e-05,
+      "loss": 1.3019,
+      "step": 8000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4276000000000003e-05,
+      "loss": 1.322,
+      "step": 8100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4576000000000003e-05,
+      "loss": 1.296,
+      "step": 8200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4876000000000004e-05,
+      "loss": 1.2825,
+      "step": 8300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5175999999999997e-05,
+      "loss": 1.3016,
+      "step": 8400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5475999999999998e-05,
+      "loss": 1.3127,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5776e-05,
+      "loss": 1.3082,
+      "step": 8600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6076e-05,
+      "loss": 1.334,
+      "step": 8700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6376e-05,
+      "loss": 1.3268,
+      "step": 8800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6673e-05,
+      "loss": 1.2714,
+      "step": 8900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6973e-05,
+      "loss": 1.2581,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7273e-05,
+      "loss": 1.3306,
+      "step": 9100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7573000000000002e-05,
+      "loss": 1.3202,
+      "step": 9200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7873000000000002e-05,
+      "loss": 1.2766,
+      "step": 9300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8173000000000003e-05,
+      "loss": 1.3235,
+      "step": 9400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8473000000000003e-05,
+      "loss": 1.3148,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8773e-05,
+      "loss": 1.3185,
+      "step": 9600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9073e-05,
+      "loss": 1.306,
+      "step": 9700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9373e-05,
+      "loss": 1.3494,
+      "step": 9800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9673e-05,
+      "loss": 1.3514,
+      "step": 9900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9973e-05,
+      "loss": 1.2738,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 1.7686232328414917,
+      "eval_runtime": 133.186,
+      "eval_samples_per_second": 15.017,
+      "eval_steps_per_second": 0.939,
+      "step": 10000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9993000000000002e-05,
+      "loss": 1.2775,
+      "step": 10100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9985307692307693e-05,
+      "loss": 1.2648,
+      "step": 10200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9977615384615385e-05,
+      "loss": 1.2317,
+      "step": 10300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.996992307692308e-05,
+      "loss": 1.2885,
+      "step": 10400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.996223076923077e-05,
+      "loss": 1.291,
+      "step": 10500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9954538461538465e-05,
+      "loss": 1.3181,
+      "step": 10600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9946846153846152e-05,
+      "loss": 1.2913,
+      "step": 10700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9939153846153847e-05,
+      "loss": 1.2766,
+      "step": 10800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.993146153846154e-05,
+      "loss": 1.2741,
+      "step": 10900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9923769230769232e-05,
+      "loss": 1.3108,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9916076923076923e-05,
+      "loss": 1.2748,
+      "step": 11100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9908384615384614e-05,
+      "loss": 1.3238,
+      "step": 11200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990069230769231e-05,
+      "loss": 1.3178,
+      "step": 11300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893e-05,
+      "loss": 1.2615,
+      "step": 11400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9885307692307694e-05,
+      "loss": 1.2758,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9877615384615385e-05,
+      "loss": 1.2741,
+      "step": 11600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869923076923077e-05,
+      "loss": 1.3008,
+      "step": 11700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986223076923077e-05,
+      "loss": 1.2459,
+      "step": 11800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9854538461538462e-05,
+      "loss": 1.272,
+      "step": 11900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9846846153846153e-05,
+      "loss": 1.2414,
+      "step": 12000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9839153846153848e-05,
+      "loss": 1.2706,
+      "step": 12100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.983146153846154e-05,
+      "loss": 1.2383,
+      "step": 12200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9823769230769233e-05,
+      "loss": 1.2149,
+      "step": 12300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9816076923076924e-05,
+      "loss": 1.266,
+      "step": 12400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9808384615384615e-05,
+      "loss": 1.2325,
+      "step": 12500
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.777679443359375,
+      "eval_runtime": 132.497,
+      "eval_samples_per_second": 15.095,
+      "eval_steps_per_second": 0.943,
+      "step": 12500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980069230769231e-05,
+      "loss": 1.2255,
+      "step": 12600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9793e-05,
+      "loss": 1.2838,
+      "step": 12700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9785307692307695e-05,
+      "loss": 1.2265,
+      "step": 12800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9777615384615383e-05,
+      "loss": 1.2662,
+      "step": 12900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9769923076923078e-05,
+      "loss": 1.2306,
+      "step": 13000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9762230769230772e-05,
+      "loss": 1.2725,
+      "step": 13100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9754538461538463e-05,
+      "loss": 1.2307,
+      "step": 13200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9746846153846154e-05,
+      "loss": 1.264,
+      "step": 13300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9739153846153845e-05,
+      "loss": 1.2152,
+      "step": 13400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.973146153846154e-05,
+      "loss": 1.2681,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.972376923076923e-05,
+      "loss": 1.2385,
+      "step": 13600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9716076923076925e-05,
+      "loss": 1.2763,
+      "step": 13700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9708384615384616e-05,
+      "loss": 1.2277,
+      "step": 13800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9700692307692307e-05,
+      "loss": 1.2973,
+      "step": 13900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9693000000000002e-05,
+      "loss": 1.2376,
+      "step": 14000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9685307692307693e-05,
+      "loss": 1.2514,
+      "step": 14100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9677615384615384e-05,
+      "loss": 1.2785,
+      "step": 14200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.966992307692308e-05,
+      "loss": 1.2425,
+      "step": 14300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.966223076923077e-05,
+      "loss": 1.2581,
+      "step": 14400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9654538461538464e-05,
+      "loss": 1.2026,
+      "step": 14500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.964684615384615e-05,
+      "loss": 1.2581,
+      "step": 14600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9639153846153846e-05,
+      "loss": 1.2763,
+      "step": 14700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.963146153846154e-05,
+      "loss": 1.2471,
+      "step": 14800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9623769230769232e-05,
+      "loss": 1.2585,
+      "step": 14900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9616076923076926e-05,
+      "loss": 1.2159,
+      "step": 15000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 1.765824794769287,
+      "eval_runtime": 132.2613,
+      "eval_samples_per_second": 15.122,
+      "eval_steps_per_second": 0.945,
+      "step": 15000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9608384615384614e-05,
+      "loss": 1.2836,
+      "step": 15100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.960069230769231e-05,
+      "loss": 1.2389,
+      "step": 15200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9593000000000003e-05,
+      "loss": 1.1892,
+      "step": 15300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9585307692307694e-05,
+      "loss": 1.2571,
+      "step": 15400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9577615384615385e-05,
+      "loss": 1.2238,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9569923076923076e-05,
+      "loss": 1.2476,
+      "step": 15600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.956223076923077e-05,
+      "loss": 1.2342,
+      "step": 15700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9554538461538465e-05,
+      "loss": 1.2241,
+      "step": 15800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9546846153846153e-05,
+      "loss": 1.2854,
+      "step": 15900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9539153846153847e-05,
+      "loss": 1.1961,
+      "step": 16000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9531461538461538e-05,
+      "loss": 1.2139,
+      "step": 16100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9523769230769233e-05,
+      "loss": 1.1906,
+      "step": 16200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9516076923076924e-05,
+      "loss": 1.2203,
+      "step": 16300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9508384615384615e-05,
+      "loss": 1.2217,
+      "step": 16400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.950069230769231e-05,
+      "loss": 1.229,
+      "step": 16500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9493e-05,
+      "loss": 1.259,
+      "step": 16600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9485307692307695e-05,
+      "loss": 1.2056,
+      "step": 16700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9477615384615383e-05,
+      "loss": 1.2103,
+      "step": 16800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9469923076923077e-05,
+      "loss": 1.2095,
+      "step": 16900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.946223076923077e-05,
+      "loss": 1.2089,
+      "step": 17000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9454538461538463e-05,
+      "loss": 1.1992,
+      "step": 17100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9446846153846157e-05,
+      "loss": 1.2035,
+      "step": 17200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9439153846153845e-05,
+      "loss": 1.2001,
+      "step": 17300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.943146153846154e-05,
+      "loss": 1.259,
+      "step": 17400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9423769230769234e-05,
+      "loss": 1.2324,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 1.8244938850402832,
+      "eval_runtime": 131.4308,
+      "eval_samples_per_second": 15.217,
+      "eval_steps_per_second": 0.951,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9416076923076925e-05,
+      "loss": 1.2114,
+      "step": 17600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9408384615384616e-05,
+      "loss": 1.2328,
+      "step": 17700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9400692307692307e-05,
+      "loss": 1.2405,
+      "step": 17800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9393e-05,
+      "loss": 1.2286,
+      "step": 17900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9385307692307696e-05,
+      "loss": 1.2027,
+      "step": 18000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9377615384615383e-05,
+      "loss": 1.1934,
+      "step": 18100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9369923076923078e-05,
+      "loss": 1.1917,
+      "step": 18200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.936223076923077e-05,
+      "loss": 1.2031,
+      "step": 18300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9354538461538463e-05,
+      "loss": 1.2112,
+      "step": 18400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9346846153846155e-05,
+      "loss": 1.2184,
+      "step": 18500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9339153846153846e-05,
+      "loss": 1.2152,
+      "step": 18600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.933146153846154e-05,
+      "loss": 1.193,
+      "step": 18700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.932376923076923e-05,
+      "loss": 1.1974,
+      "step": 18800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9316076923076926e-05,
+      "loss": 1.2214,
+      "step": 18900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.930846153846154e-05,
+      "loss": 1.2201,
+      "step": 19000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9300769230769228e-05,
+      "loss": 1.1957,
+      "step": 19100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9293076923076923e-05,
+      "loss": 1.2007,
+      "step": 19200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9285384615384617e-05,
+      "loss": 1.1798,
+      "step": 19300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9277692307692308e-05,
+      "loss": 1.2324,
+      "step": 19400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9270000000000003e-05,
+      "loss": 1.2094,
+      "step": 19500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.926230769230769e-05,
+      "loss": 1.2166,
+      "step": 19600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9254615384615385e-05,
+      "loss": 1.2204,
+      "step": 19700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.924692307692308e-05,
+      "loss": 1.1814,
+      "step": 19800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.923923076923077e-05,
+      "loss": 1.2115,
+      "step": 19900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.923153846153846e-05,
+      "loss": 1.149,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.779923439025879,
+      "eval_runtime": 130.9794,
+      "eval_samples_per_second": 15.27,
+      "eval_steps_per_second": 0.954,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9223846153846153e-05,
+      "loss": 1.2278,
+      "step": 20100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9216153846153847e-05,
+      "loss": 1.2197,
+      "step": 20200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9208538461538462e-05,
+      "loss": 1.2295,
+      "step": 20300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9200846153846156e-05,
+      "loss": 1.1635,
+      "step": 20400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9193153846153847e-05,
+      "loss": 1.2152,
+      "step": 20500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.918546153846154e-05,
+      "loss": 1.1861,
+      "step": 20600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.917776923076923e-05,
+      "loss": 1.232,
+      "step": 20700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9170076923076924e-05,
+      "loss": 1.2149,
+      "step": 20800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.916238461538462e-05,
+      "loss": 1.2204,
+      "step": 20900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9154692307692306e-05,
+      "loss": 1.1699,
+      "step": 21000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9147e-05,
+      "loss": 1.1646,
+      "step": 21100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9139307692307692e-05,
+      "loss": 1.202,
+      "step": 21200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9131615384615386e-05,
+      "loss": 1.152,
+      "step": 21300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.912392307692308e-05,
+      "loss": 1.2322,
+      "step": 21400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.911623076923077e-05,
+      "loss": 1.1943,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9108538461538463e-05,
+      "loss": 1.1711,
+      "step": 21600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9100846153846154e-05,
+      "loss": 1.179,
+      "step": 21700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.909315384615385e-05,
+      "loss": 1.2446,
+      "step": 21800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.908546153846154e-05,
+      "loss": 1.1959,
+      "step": 21900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.907776923076923e-05,
+      "loss": 1.1935,
+      "step": 22000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9070153846153845e-05,
+      "loss": 1.1793,
+      "step": 22100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.906246153846154e-05,
+      "loss": 1.2054,
+      "step": 22200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.905476923076923e-05,
+      "loss": 1.2324,
+      "step": 22300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9047076923076925e-05,
+      "loss": 1.2169,
+      "step": 22400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9039384615384616e-05,
+      "loss": 1.1853,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 1.7166931629180908,
+      "eval_runtime": 130.9188,
+      "eval_samples_per_second": 15.277,
+      "eval_steps_per_second": 0.955,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9031692307692308e-05,
+      "loss": 1.1718,
+      "step": 22600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9024000000000002e-05,
+      "loss": 1.141,
+      "step": 22700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9016307692307693e-05,
+      "loss": 1.2313,
+      "step": 22800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9008615384615384e-05,
+      "loss": 1.1851,
+      "step": 22900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.900092307692308e-05,
+      "loss": 1.1837,
+      "step": 23000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.899323076923077e-05,
+      "loss": 1.2211,
+      "step": 23100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8985538461538464e-05,
+      "loss": 1.1629,
+      "step": 23200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8977846153846155e-05,
+      "loss": 1.217,
+      "step": 23300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8970153846153846e-05,
+      "loss": 1.222,
+      "step": 23400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8962461538461537e-05,
+      "loss": 1.2448,
+      "step": 23500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8954769230769232e-05,
+      "loss": 1.1594,
+      "step": 23600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8947076923076926e-05,
+      "loss": 1.1468,
+      "step": 23700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8939384615384614e-05,
+      "loss": 1.2215,
+      "step": 23800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.893169230769231e-05,
+      "loss": 1.1473,
+      "step": 23900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8924e-05,
+      "loss": 1.1991,
+      "step": 24000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8916307692307694e-05,
+      "loss": 1.1629,
+      "step": 24100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8908615384615385e-05,
+      "loss": 1.2094,
+      "step": 24200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8900923076923076e-05,
+      "loss": 1.1517,
+      "step": 24300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.889323076923077e-05,
+      "loss": 1.1687,
+      "step": 24400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8885538461538462e-05,
+      "loss": 1.2191,
+      "step": 24500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8877846153846156e-05,
+      "loss": 1.1585,
+      "step": 24600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8870153846153847e-05,
+      "loss": 1.2094,
+      "step": 24700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.886246153846154e-05,
+      "loss": 1.1635,
+      "step": 24800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8854769230769233e-05,
+      "loss": 1.2067,
+      "step": 24900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8847076923076924e-05,
+      "loss": 1.2011,
+      "step": 25000
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 1.7367329597473145,
+      "eval_runtime": 132.5691,
+      "eval_samples_per_second": 15.086,
+      "eval_steps_per_second": 0.943,
+      "step": 25000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8839384615384615e-05,
+      "loss": 1.1901,
+      "step": 25100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.883169230769231e-05,
+      "loss": 1.1971,
+      "step": 25200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8824e-05,
+      "loss": 1.1739,
+      "step": 25300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8816307692307695e-05,
+      "loss": 1.191,
+      "step": 25400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8808615384615386e-05,
+      "loss": 1.1638,
+      "step": 25500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8800923076923077e-05,
+      "loss": 1.1845,
+      "step": 25600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8793230769230768e-05,
+      "loss": 1.2117,
+      "step": 25700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8785538461538463e-05,
+      "loss": 1.1761,
+      "step": 25800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8777846153846157e-05,
+      "loss": 1.1205,
+      "step": 25900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8770153846153845e-05,
+      "loss": 1.1141,
+      "step": 26000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.876246153846154e-05,
+      "loss": 1.1909,
+      "step": 26100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.875476923076923e-05,
+      "loss": 1.2082,
+      "step": 26200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8747076923076925e-05,
+      "loss": 1.172,
+      "step": 26300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8739384615384616e-05,
+      "loss": 1.1583,
+      "step": 26400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8731692307692307e-05,
+      "loss": 1.2064,
+      "step": 26500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8724e-05,
+      "loss": 1.1594,
+      "step": 26600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8716307692307693e-05,
+      "loss": 1.1844,
+      "step": 26700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8708615384615387e-05,
+      "loss": 1.1335,
+      "step": 26800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8700923076923078e-05,
+      "loss": 1.1706,
+      "step": 26900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.869323076923077e-05,
+      "loss": 1.1953,
+      "step": 27000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8685538461538464e-05,
+      "loss": 1.186,
+      "step": 27100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8677846153846155e-05,
+      "loss": 1.1748,
+      "step": 27200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8670153846153846e-05,
+      "loss": 1.2005,
+      "step": 27300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.866246153846154e-05,
+      "loss": 1.1952,
+      "step": 27400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.865476923076923e-05,
+      "loss": 1.1794,
+      "step": 27500
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 1.7217485904693604,
+      "eval_runtime": 131.3468,
+      "eval_samples_per_second": 15.227,
+      "eval_steps_per_second": 0.952,
+      "step": 27500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8647076923076926e-05,
+      "loss": 1.1524,
+      "step": 27600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8639384615384613e-05,
+      "loss": 1.1471,
+      "step": 27700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8631692307692308e-05,
+      "loss": 1.1629,
+      "step": 27800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8624000000000002e-05,
+      "loss": 1.1718,
+      "step": 27900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8616307692307693e-05,
+      "loss": 1.1438,
+      "step": 28000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8608615384615388e-05,
+      "loss": 1.1868,
+      "step": 28100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8600923076923076e-05,
+      "loss": 1.2246,
+      "step": 28200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.859323076923077e-05,
+      "loss": 1.1205,
+      "step": 28300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.858553846153846e-05,
+      "loss": 1.1759,
+      "step": 28400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8577846153846156e-05,
+      "loss": 1.163,
+      "step": 28500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8570153846153847e-05,
+      "loss": 1.1668,
+      "step": 28600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.856253846153846e-05,
+      "loss": 1.1739,
+      "step": 28700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8554846153846153e-05,
+      "loss": 1.1554,
+      "step": 28800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8547153846153847e-05,
+      "loss": 1.1529,
+      "step": 28900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8539461538461538e-05,
+      "loss": 1.1811,
+      "step": 29000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8531769230769233e-05,
+      "loss": 1.1667,
+      "step": 29100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8524076923076924e-05,
+      "loss": 1.1565,
+      "step": 29200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8516384615384615e-05,
+      "loss": 1.1916,
+      "step": 29300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.850869230769231e-05,
+      "loss": 1.1469,
+      "step": 29400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8501e-05,
+      "loss": 1.1621,
+      "step": 29500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.849330769230769e-05,
+      "loss": 1.1835,
+      "step": 29600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8485615384615386e-05,
+      "loss": 1.1705,
+      "step": 29700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8477923076923077e-05,
+      "loss": 1.162,
+      "step": 29800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.847023076923077e-05,
+      "loss": 1.1888,
+      "step": 29900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8462538461538463e-05,
+      "loss": 1.1769,
+      "step": 30000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 1.6898554563522339,
+      "eval_runtime": 130.1248,
+      "eval_samples_per_second": 15.37,
+      "eval_steps_per_second": 0.961,
+      "step": 30000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8454846153846154e-05,
+      "loss": 1.2021,
+      "step": 30100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.844723076923077e-05,
+      "loss": 1.1885,
+      "step": 30200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8439538461538463e-05,
+      "loss": 1.1512,
+      "step": 30300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8431846153846154e-05,
+      "loss": 1.1621,
+      "step": 30400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.842415384615385e-05,
+      "loss": 1.1653,
+      "step": 30500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8416461538461536e-05,
+      "loss": 1.0979,
+      "step": 30600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.840876923076923e-05,
+      "loss": 1.1135,
+      "step": 30700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8401076923076925e-05,
+      "loss": 1.1335,
+      "step": 30800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8393384615384616e-05,
+      "loss": 1.1839,
+      "step": 30900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.838569230769231e-05,
+      "loss": 1.1848,
+      "step": 31000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8378e-05,
+      "loss": 1.179,
+      "step": 31100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8370307692307693e-05,
+      "loss": 1.1592,
+      "step": 31200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8362615384615387e-05,
+      "loss": 1.1597,
+      "step": 31300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.835492307692308e-05,
+      "loss": 1.1617,
+      "step": 31400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.834723076923077e-05,
+      "loss": 1.1746,
+      "step": 31500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.833953846153846e-05,
+      "loss": 1.191,
+      "step": 31600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8331846153846155e-05,
+      "loss": 1.1232,
+      "step": 31700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.832415384615385e-05,
+      "loss": 1.1497,
+      "step": 31800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.831646153846154e-05,
+      "loss": 1.1787,
+      "step": 31900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.830876923076923e-05,
+      "loss": 1.1908,
+      "step": 32000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8301076923076923e-05,
+      "loss": 1.1959,
+      "step": 32100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8293384615384617e-05,
+      "loss": 1.1618,
+      "step": 32200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8285692307692308e-05,
+      "loss": 1.1541,
+      "step": 32300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8278e-05,
+      "loss": 1.1517,
+      "step": 32400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8270307692307694e-05,
+      "loss": 1.171,
+      "step": 32500
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 1.7089424133300781,
+      "eval_runtime": 130.5805,
+      "eval_samples_per_second": 15.316,
+      "eval_steps_per_second": 0.957,
+      "step": 32500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8262615384615385e-05,
+      "loss": 1.149,
+      "step": 32600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.825492307692308e-05,
+      "loss": 1.181,
+      "step": 32700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8247230769230767e-05,
+      "loss": 1.1446,
+      "step": 32800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.823953846153846e-05,
+      "loss": 1.1544,
+      "step": 32900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8231846153846156e-05,
+      "loss": 1.1102,
+      "step": 33000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8224153846153847e-05,
+      "loss": 1.1114,
+      "step": 33100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.821646153846154e-05,
+      "loss": 1.1536,
+      "step": 33200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.820876923076923e-05,
+      "loss": 1.1991,
+      "step": 33300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8201076923076924e-05,
+      "loss": 1.1558,
+      "step": 33400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8193384615384618e-05,
+      "loss": 1.1355,
+      "step": 33500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.818569230769231e-05,
+      "loss": 1.1426,
+      "step": 33600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8178e-05,
+      "loss": 1.1778,
+      "step": 33700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.817030769230769e-05,
+      "loss": 1.1805,
+      "step": 33800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8162615384615386e-05,
+      "loss": 1.1544,
+      "step": 33900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.815492307692308e-05,
+      "loss": 1.1456,
+      "step": 34000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8147230769230768e-05,
+      "loss": 1.1588,
+      "step": 34100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8139538461538462e-05,
+      "loss": 1.1878,
+      "step": 34200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8131846153846153e-05,
+      "loss": 1.1829,
+      "step": 34300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8124153846153848e-05,
+      "loss": 1.1453,
+      "step": 34400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.811646153846154e-05,
+      "loss": 1.1481,
+      "step": 34500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.810876923076923e-05,
+      "loss": 1.1498,
+      "step": 34600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8101076923076925e-05,
+      "loss": 1.1647,
+      "step": 34700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8093384615384616e-05,
+      "loss": 1.1051,
+      "step": 34800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.808569230769231e-05,
+      "loss": 1.169,
+      "step": 34900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8077999999999998e-05,
+      "loss": 1.1402,
+      "step": 35000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 1.6927934885025024,
+      "eval_runtime": 130.8746,
+      "eval_samples_per_second": 15.282,
+      "eval_steps_per_second": 0.955,
+      "step": 35000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8070307692307692e-05,
+      "loss": 1.1177,
+      "step": 35100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8062692307692307e-05,
+      "loss": 1.1229,
+      "step": 35200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8055e-05,
+      "loss": 1.1593,
+      "step": 35300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8047307692307693e-05,
+      "loss": 1.1548,
+      "step": 35400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8039615384615387e-05,
+      "loss": 1.1568,
+      "step": 35500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8031923076923075e-05,
+      "loss": 1.187,
+      "step": 35600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.802423076923077e-05,
+      "loss": 1.1594,
+      "step": 35700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8016538461538464e-05,
+      "loss": 1.1674,
+      "step": 35800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8008846153846155e-05,
+      "loss": 1.1301,
+      "step": 35900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8001153846153846e-05,
+      "loss": 1.1714,
+      "step": 36000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7993461538461537e-05,
+      "loss": 1.187,
+      "step": 36100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.798576923076923e-05,
+      "loss": 1.1288,
+      "step": 36200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7978076923076926e-05,
+      "loss": 1.1592,
+      "step": 36300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7970384615384617e-05,
+      "loss": 1.1701,
+      "step": 36400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7962692307692308e-05,
+      "loss": 1.1334,
+      "step": 36500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7955e-05,
+      "loss": 1.2074,
+      "step": 36600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7947307692307694e-05,
+      "loss": 1.1783,
+      "step": 36700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7939615384615388e-05,
+      "loss": 1.1415,
+      "step": 36800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7931923076923076e-05,
+      "loss": 1.1398,
+      "step": 36900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.792423076923077e-05,
+      "loss": 1.1628,
+      "step": 37000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.791653846153846e-05,
+      "loss": 1.157,
+      "step": 37100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7908846153846156e-05,
+      "loss": 1.1618,
+      "step": 37200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7901153846153847e-05,
+      "loss": 1.1012,
+      "step": 37300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7893461538461538e-05,
+      "loss": 1.1554,
+      "step": 37400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7885769230769232e-05,
+      "loss": 1.1692,
+      "step": 37500
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.6816636323928833,
+      "eval_runtime": 131.8181,
+      "eval_samples_per_second": 15.172,
+      "eval_steps_per_second": 0.948,
+      "step": 37500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7878076923076923e-05,
+      "loss": 1.1467,
+      "step": 37600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7870384615384618e-05,
+      "loss": 1.1539,
+      "step": 37700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7862692307692306e-05,
+      "loss": 1.1348,
+      "step": 37800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7855e-05,
+      "loss": 1.0675,
+      "step": 37900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7847307692307695e-05,
+      "loss": 1.1467,
+      "step": 38000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7839615384615386e-05,
+      "loss": 1.1466,
+      "step": 38100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7831923076923077e-05,
+      "loss": 1.1818,
+      "step": 38200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7824230769230768e-05,
+      "loss": 1.1545,
+      "step": 38300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7816538461538462e-05,
+      "loss": 1.1591,
+      "step": 38400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7808846153846157e-05,
+      "loss": 1.1536,
+      "step": 38500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7801153846153848e-05,
+      "loss": 1.1264,
+      "step": 38600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.779346153846154e-05,
+      "loss": 1.1126,
+      "step": 38700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.778576923076923e-05,
+      "loss": 1.1788,
+      "step": 38800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7778076923076924e-05,
+      "loss": 1.145,
+      "step": 38900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.777038461538462e-05,
+      "loss": 1.1082,
+      "step": 39000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7762692307692307e-05,
+      "loss": 1.1794,
+      "step": 39100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.775507692307692e-05,
+      "loss": 1.1118,
+      "step": 39200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7747384615384616e-05,
+      "loss": 1.1369,
+      "step": 39300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7739692307692307e-05,
+      "loss": 1.1952,
+      "step": 39400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7732e-05,
+      "loss": 1.1512,
+      "step": 39500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7724384615384616e-05,
+      "loss": 1.143,
+      "step": 39600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.771669230769231e-05,
+      "loss": 1.1691,
+      "step": 39700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7709e-05,
+      "loss": 1.1933,
+      "step": 39800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7701307692307693e-05,
+      "loss": 1.1513,
+      "step": 39900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7693615384615384e-05,
+      "loss": 1.167,
+      "step": 40000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.666869044303894,
+      "eval_runtime": 130.3813,
+      "eval_samples_per_second": 15.34,
+      "eval_steps_per_second": 0.959,
+      "step": 40000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.768592307692308e-05,
+      "loss": 1.124,
+      "step": 40100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.767823076923077e-05,
+      "loss": 1.1154,
+      "step": 40200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.767053846153846e-05,
+      "loss": 1.1336,
+      "step": 40300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7662846153846155e-05,
+      "loss": 1.1911,
+      "step": 40400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7655153846153846e-05,
+      "loss": 1.0898,
+      "step": 40500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.764746153846154e-05,
+      "loss": 1.1008,
+      "step": 40600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7639769230769232e-05,
+      "loss": 1.1778,
+      "step": 40700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7632076923076923e-05,
+      "loss": 1.152,
+      "step": 40800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7624384615384617e-05,
+      "loss": 1.1463,
+      "step": 40900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.761669230769231e-05,
+      "loss": 1.1726,
+      "step": 41000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7609e-05,
+      "loss": 1.141,
+      "step": 41100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7601307692307694e-05,
+      "loss": 1.1191,
+      "step": 41200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7593615384615385e-05,
+      "loss": 1.0983,
+      "step": 41300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.758592307692308e-05,
+      "loss": 1.1467,
+      "step": 41400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.757823076923077e-05,
+      "loss": 1.1471,
+      "step": 41500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.757053846153846e-05,
+      "loss": 1.0858,
+      "step": 41600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7562846153846153e-05,
+      "loss": 1.1448,
+      "step": 41700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7555153846153847e-05,
+      "loss": 1.0814,
+      "step": 41800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.754746153846154e-05,
+      "loss": 1.147,
+      "step": 41900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.753976923076923e-05,
+      "loss": 1.1083,
+      "step": 42000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7532076923076924e-05,
+      "loss": 1.157,
+      "step": 42100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7524384615384615e-05,
+      "loss": 1.1416,
+      "step": 42200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.751669230769231e-05,
+      "loss": 1.1543,
+      "step": 42300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7509e-05,
+      "loss": 1.1642,
+      "step": 42400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.750130769230769e-05,
+      "loss": 1.1575,
+      "step": 42500
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.6338238716125488,
+      "eval_runtime": 131.9287,
+      "eval_samples_per_second": 15.16,
+      "eval_steps_per_second": 0.947,
+      "step": 42500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7493615384615386e-05,
+      "loss": 1.0936,
+      "step": 42600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7485923076923077e-05,
+      "loss": 1.1118,
+      "step": 42700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.747823076923077e-05,
+      "loss": 1.0946,
+      "step": 42800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7470538461538463e-05,
+      "loss": 1.1033,
+      "step": 42900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7462846153846154e-05,
+      "loss": 1.1015,
+      "step": 43000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.745523076923077e-05,
+      "loss": 1.1056,
+      "step": 43100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7447538461538463e-05,
+      "loss": 1.1354,
+      "step": 43200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7439846153846154e-05,
+      "loss": 1.1701,
+      "step": 43300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.743215384615385e-05,
+      "loss": 1.1702,
+      "step": 43400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.742446153846154e-05,
+      "loss": 1.0999,
+      "step": 43500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.741676923076923e-05,
+      "loss": 1.0987,
+      "step": 43600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7409076923076925e-05,
+      "loss": 1.104,
+      "step": 43700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7401384615384616e-05,
+      "loss": 1.1948,
+      "step": 43800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7393692307692307e-05,
+      "loss": 1.1225,
+      "step": 43900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7386000000000002e-05,
+      "loss": 1.135,
+      "step": 44000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7378307692307693e-05,
+      "loss": 1.1453,
+      "step": 44100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7370615384615387e-05,
+      "loss": 1.1382,
+      "step": 44200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7362923076923075e-05,
+      "loss": 1.0959,
+      "step": 44300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.735523076923077e-05,
+      "loss": 1.1326,
+      "step": 44400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7347538461538464e-05,
+      "loss": 1.0941,
+      "step": 44500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7339846153846155e-05,
+      "loss": 1.1374,
+      "step": 44600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.733215384615385e-05,
+      "loss": 1.1109,
+      "step": 44700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7324461538461537e-05,
+      "loss": 1.1298,
+      "step": 44800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.731676923076923e-05,
+      "loss": 1.1385,
+      "step": 44900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7309076923076923e-05,
+      "loss": 1.1173,
+      "step": 45000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.636410117149353,
+      "eval_runtime": 130.2597,
+      "eval_samples_per_second": 15.354,
+      "eval_steps_per_second": 0.96,
+      "step": 45000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7301384615384617e-05,
+      "loss": 1.0904,
+      "step": 45100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7293692307692308e-05,
+      "loss": 1.1356,
+      "step": 45200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7286e-05,
+      "loss": 1.1085,
+      "step": 45300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7278384615384614e-05,
+      "loss": 1.1325,
+      "step": 45400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.727069230769231e-05,
+      "loss": 1.0876,
+      "step": 45500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7263e-05,
+      "loss": 1.1306,
+      "step": 45600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7255307692307694e-05,
+      "loss": 1.1558,
+      "step": 45700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7247615384615385e-05,
+      "loss": 1.136,
+      "step": 45800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7239923076923076e-05,
+      "loss": 1.0917,
+      "step": 45900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.723223076923077e-05,
+      "loss": 1.106,
+      "step": 46000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7224538461538462e-05,
+      "loss": 1.1676,
+      "step": 46100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7216846153846153e-05,
+      "loss": 1.1716,
+      "step": 46200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7209153846153847e-05,
+      "loss": 1.0806,
+      "step": 46300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.720146153846154e-05,
+      "loss": 1.1369,
+      "step": 46400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7193769230769233e-05,
+      "loss": 1.1441,
+      "step": 46500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7186076923076924e-05,
+      "loss": 1.1056,
+      "step": 46600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7178384615384615e-05,
+      "loss": 1.1116,
+      "step": 46700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.717069230769231e-05,
+      "loss": 1.2066,
+      "step": 46800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7163e-05,
+      "loss": 1.1302,
+      "step": 46900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7155307692307695e-05,
+      "loss": 1.1222,
+      "step": 47000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7147615384615383e-05,
+      "loss": 1.1468,
+      "step": 47100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7139923076923077e-05,
+      "loss": 1.0751,
+      "step": 47200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7132230769230772e-05,
+      "loss": 1.1509,
+      "step": 47300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7124538461538463e-05,
+      "loss": 1.1283,
+      "step": 47400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7116846153846154e-05,
+      "loss": 1.1442,
+      "step": 47500
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.6622968912124634,
+      "eval_runtime": 130.1999,
+      "eval_samples_per_second": 15.361,
+      "eval_steps_per_second": 0.96,
+      "step": 47500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7109153846153845e-05,
+      "loss": 1.1764,
+      "step": 47600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.710146153846154e-05,
+      "loss": 1.0903,
+      "step": 47700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7093769230769234e-05,
+      "loss": 1.1128,
+      "step": 47800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7086076923076925e-05,
+      "loss": 1.1371,
+      "step": 47900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7078384615384616e-05,
+      "loss": 1.107,
+      "step": 48000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7070692307692307e-05,
+      "loss": 1.1003,
+      "step": 48100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7063e-05,
+      "loss": 1.161,
+      "step": 48200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7055307692307693e-05,
+      "loss": 1.1402,
+      "step": 48300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.704776923076923e-05,
+      "loss": 1.1321,
+      "step": 48400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7040076923076926e-05,
+      "loss": 1.1587,
+      "step": 48500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7032384615384617e-05,
+      "loss": 1.1748,
+      "step": 48600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7024692307692308e-05,
+      "loss": 1.1237,
+      "step": 48700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7017e-05,
+      "loss": 1.1565,
+      "step": 48800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7009307692307693e-05,
+      "loss": 1.1317,
+      "step": 48900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7001615384615388e-05,
+      "loss": 1.1085,
+      "step": 49000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6993923076923076e-05,
+      "loss": 1.1167,
+      "step": 49100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.698623076923077e-05,
+      "loss": 1.1675,
+      "step": 49200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.697853846153846e-05,
+      "loss": 1.1726,
+      "step": 49300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6970846153846156e-05,
+      "loss": 1.1171,
+      "step": 49400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.696315384615385e-05,
+      "loss": 1.1177,
+      "step": 49500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6955461538461538e-05,
+      "loss": 1.0996,
+      "step": 49600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6947769230769232e-05,
+      "loss": 1.1287,
+      "step": 49700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6940076923076923e-05,
+      "loss": 1.1164,
+      "step": 49800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6932384615384618e-05,
+      "loss": 1.1166,
+      "step": 49900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6924692307692305e-05,
+      "loss": 1.1148,
+      "step": 50000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.6451276540756226,
+      "eval_runtime": 130.6707,
+      "eval_samples_per_second": 15.306,
+      "eval_steps_per_second": 0.957,
+      "step": 50000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6917e-05,
+      "loss": 1.1177,
+      "step": 50100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6909307692307694e-05,
+      "loss": 1.1317,
+      "step": 50200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6901615384615385e-05,
+      "loss": 1.0997,
+      "step": 50300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6893923076923077e-05,
+      "loss": 1.1004,
+      "step": 50400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6886230769230768e-05,
+      "loss": 1.1051,
+      "step": 50500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6878538461538462e-05,
+      "loss": 1.0663,
+      "step": 50600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6870846153846157e-05,
+      "loss": 1.1021,
+      "step": 50700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6863153846153848e-05,
+      "loss": 1.1266,
+      "step": 50800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.685546153846154e-05,
+      "loss": 1.0954,
+      "step": 50900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.684776923076923e-05,
+      "loss": 1.1401,
+      "step": 51000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6840076923076924e-05,
+      "loss": 1.1354,
+      "step": 51100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.683238461538462e-05,
+      "loss": 1.1561,
+      "step": 51200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6824692307692306e-05,
+      "loss": 1.1029,
+      "step": 51300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6817e-05,
+      "loss": 1.1044,
+      "step": 51400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6809307692307692e-05,
+      "loss": 1.1474,
+      "step": 51500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6801615384615386e-05,
+      "loss": 1.1148,
+      "step": 51600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6793923076923078e-05,
+      "loss": 1.1211,
+      "step": 51700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.678623076923077e-05,
+      "loss": 1.0887,
+      "step": 51800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6778538461538463e-05,
+      "loss": 1.108,
+      "step": 51900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6770846153846154e-05,
+      "loss": 1.1096,
+      "step": 52000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.676315384615385e-05,
+      "loss": 1.1365,
+      "step": 52100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6755461538461536e-05,
+      "loss": 1.1332,
+      "step": 52200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.674776923076923e-05,
+      "loss": 1.1081,
+      "step": 52300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6740076923076925e-05,
+      "loss": 1.0648,
+      "step": 52400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.673246153846154e-05,
+      "loss": 1.0995,
+      "step": 52500
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.6107012033462524,
+      "eval_runtime": 130.8204,
+      "eval_samples_per_second": 15.288,
+      "eval_steps_per_second": 0.956,
+      "step": 52500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.672476923076923e-05,
+      "loss": 1.1135,
+      "step": 52600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6717076923076926e-05,
+      "loss": 1.1292,
+      "step": 52700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6709384615384613e-05,
+      "loss": 1.117,
+      "step": 52800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6701692307692308e-05,
+      "loss": 1.1043,
+      "step": 52900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6694076923076923e-05,
+      "loss": 1.077,
+      "step": 53000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6686384615384617e-05,
+      "loss": 1.1052,
+      "step": 53100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6678692307692308e-05,
+      "loss": 1.1297,
+      "step": 53200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6671e-05,
+      "loss": 1.1183,
+      "step": 53300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6663307692307694e-05,
+      "loss": 1.0677,
+      "step": 53400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6655615384615385e-05,
+      "loss": 1.1373,
+      "step": 53500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.664792307692308e-05,
+      "loss": 1.1407,
+      "step": 53600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.664023076923077e-05,
+      "loss": 1.1458,
+      "step": 53700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.663253846153846e-05,
+      "loss": 1.1556,
+      "step": 53800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6624846153846153e-05,
+      "loss": 1.1213,
+      "step": 53900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6617153846153847e-05,
+      "loss": 1.1054,
+      "step": 54000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.660946153846154e-05,
+      "loss": 1.1266,
+      "step": 54100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.660176923076923e-05,
+      "loss": 1.0873,
+      "step": 54200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6594076923076924e-05,
+      "loss": 1.086,
+      "step": 54300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6586384615384615e-05,
+      "loss": 1.1026,
+      "step": 54400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.657869230769231e-05,
+      "loss": 1.0979,
+      "step": 54500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6571000000000004e-05,
+      "loss": 1.1339,
+      "step": 54600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.656330769230769e-05,
+      "loss": 1.1594,
+      "step": 54700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6555615384615386e-05,
+      "loss": 1.1285,
+      "step": 54800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6547923076923077e-05,
+      "loss": 1.0981,
+      "step": 54900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.654023076923077e-05,
+      "loss": 1.1103,
+      "step": 55000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.5922576189041138,
+      "eval_runtime": 130.728,
+      "eval_samples_per_second": 15.299,
+      "eval_steps_per_second": 0.956,
+      "step": 55000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6532538461538462e-05,
+      "loss": 1.125,
+      "step": 55100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6524846153846153e-05,
+      "loss": 1.0981,
+      "step": 55200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6517153846153848e-05,
+      "loss": 1.124,
+      "step": 55300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.650946153846154e-05,
+      "loss": 1.0997,
+      "step": 55400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.650176923076923e-05,
+      "loss": 1.144,
+      "step": 55500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6494076923076925e-05,
+      "loss": 1.1179,
+      "step": 55600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6486384615384616e-05,
+      "loss": 1.102,
+      "step": 55700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.647869230769231e-05,
+      "loss": 1.1287,
+      "step": 55800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6471e-05,
+      "loss": 1.0978,
+      "step": 55900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6463307692307692e-05,
+      "loss": 1.1475,
+      "step": 56000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6455615384615383e-05,
+      "loss": 1.0875,
+      "step": 56100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6447923076923078e-05,
+      "loss": 1.0838,
+      "step": 56200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6440230769230772e-05,
+      "loss": 1.1167,
+      "step": 56300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.643253846153846e-05,
+      "loss": 1.1329,
+      "step": 56400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6424846153846154e-05,
+      "loss": 1.0987,
+      "step": 56500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6417153846153845e-05,
+      "loss": 1.1039,
+      "step": 56600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.640946153846154e-05,
+      "loss": 1.0955,
+      "step": 56700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.640176923076923e-05,
+      "loss": 1.0824,
+      "step": 56800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6394076923076922e-05,
+      "loss": 1.1085,
+      "step": 56900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6386461538461537e-05,
+      "loss": 1.0952,
+      "step": 57000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.637876923076923e-05,
+      "loss": 1.1411,
+      "step": 57100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6371076923076923e-05,
+      "loss": 1.1403,
+      "step": 57200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6363384615384617e-05,
+      "loss": 1.1245,
+      "step": 57300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6355692307692308e-05,
+      "loss": 1.1414,
+      "step": 57400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6348e-05,
+      "loss": 1.1045,
+      "step": 57500
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.6080684661865234,
+      "eval_runtime": 131.9112,
+      "eval_samples_per_second": 15.162,
+      "eval_steps_per_second": 0.948,
+      "step": 57500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6340307692307694e-05,
+      "loss": 1.1186,
+      "step": 57600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6332615384615385e-05,
+      "loss": 1.0996,
+      "step": 57700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.632492307692308e-05,
+      "loss": 1.0463,
+      "step": 57800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.631723076923077e-05,
+      "loss": 1.0772,
+      "step": 57900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.630953846153846e-05,
+      "loss": 1.1165,
+      "step": 58000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6301846153846156e-05,
+      "loss": 1.0845,
+      "step": 58100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6294153846153847e-05,
+      "loss": 1.0989,
+      "step": 58200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6286461538461538e-05,
+      "loss": 1.1044,
+      "step": 58300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6278769230769232e-05,
+      "loss": 1.1138,
+      "step": 58400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6271076923076923e-05,
+      "loss": 1.1091,
+      "step": 58500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6263384615384618e-05,
+      "loss": 1.1104,
+      "step": 58600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6255692307692306e-05,
+      "loss": 1.1633,
+      "step": 58700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6248e-05,
+      "loss": 1.1216,
+      "step": 58800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6240307692307695e-05,
+      "loss": 1.1115,
+      "step": 58900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6232615384615386e-05,
+      "loss": 1.1413,
+      "step": 59000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.622492307692308e-05,
+      "loss": 1.0937,
+      "step": 59100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6217230769230768e-05,
+      "loss": 1.0965,
+      "step": 59200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6209538461538462e-05,
+      "loss": 1.0732,
+      "step": 59300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6201846153846153e-05,
+      "loss": 1.0935,
+      "step": 59400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6194153846153848e-05,
+      "loss": 1.1135,
+      "step": 59500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.618646153846154e-05,
+      "loss": 1.1009,
+      "step": 59600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.617876923076923e-05,
+      "loss": 1.1144,
+      "step": 59700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6171076923076924e-05,
+      "loss": 1.1109,
+      "step": 59800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6163384615384616e-05,
+      "loss": 1.0797,
+      "step": 59900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6155692307692307e-05,
+      "loss": 1.0964,
+      "step": 60000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.6355774402618408,
+      "eval_runtime": 130.4082,
+      "eval_samples_per_second": 15.336,
+      "eval_steps_per_second": 0.959,
+      "step": 60000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6148e-05,
+      "loss": 1.0916,
+      "step": 60100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6140384615384616e-05,
+      "loss": 1.0991,
+      "step": 60200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6132692307692307e-05,
+      "loss": 1.1386,
+      "step": 60300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6125e-05,
+      "loss": 1.1198,
+      "step": 60400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6117307692307693e-05,
+      "loss": 1.1145,
+      "step": 60500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6109615384615384e-05,
+      "loss": 1.1087,
+      "step": 60600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6101923076923078e-05,
+      "loss": 1.0991,
+      "step": 60700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.609423076923077e-05,
+      "loss": 1.1439,
+      "step": 60800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6086538461538464e-05,
+      "loss": 1.1152,
+      "step": 60900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6078846153846155e-05,
+      "loss": 1.1196,
+      "step": 61000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6071153846153846e-05,
+      "loss": 1.1199,
+      "step": 61100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.606346153846154e-05,
+      "loss": 1.0671,
+      "step": 61200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.605576923076923e-05,
+      "loss": 1.1475,
+      "step": 61300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6048076923076926e-05,
+      "loss": 1.1237,
+      "step": 61400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6040384615384613e-05,
+      "loss": 1.1251,
+      "step": 61500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6032692307692308e-05,
+      "loss": 1.1202,
+      "step": 61600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6025000000000002e-05,
+      "loss": 1.0884,
+      "step": 61700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6017307692307693e-05,
+      "loss": 1.1034,
+      "step": 61800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6009615384615385e-05,
+      "loss": 1.0906,
+      "step": 61900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6001923076923076e-05,
+      "loss": 1.1129,
+      "step": 62000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.599423076923077e-05,
+      "loss": 1.0835,
+      "step": 62100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5986538461538465e-05,
+      "loss": 1.1235,
+      "step": 62200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5978846153846156e-05,
+      "loss": 1.1113,
+      "step": 62300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5971153846153847e-05,
+      "loss": 1.128,
+      "step": 62400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5963461538461538e-05,
+      "loss": 1.1014,
+      "step": 62500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.592538833618164,
+      "eval_runtime": 130.825,
+      "eval_samples_per_second": 15.288,
+      "eval_steps_per_second": 0.955,
+      "step": 62500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5955769230769232e-05,
+      "loss": 1.0906,
+      "step": 62600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5948076923076923e-05,
+      "loss": 1.0951,
+      "step": 62700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5940384615384614e-05,
+      "loss": 1.0765,
+      "step": 62800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5932769230769233e-05,
+      "loss": 1.0841,
+      "step": 62900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5925076923076924e-05,
+      "loss": 1.1431,
+      "step": 63000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5917384615384615e-05,
+      "loss": 1.0886,
+      "step": 63100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.590969230769231e-05,
+      "loss": 1.1147,
+      "step": 63200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5902e-05,
+      "loss": 1.0749,
+      "step": 63300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.589430769230769e-05,
+      "loss": 1.1353,
+      "step": 63400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5886615384615386e-05,
+      "loss": 1.0788,
+      "step": 63500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5878923076923077e-05,
+      "loss": 1.1074,
+      "step": 63600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.587123076923077e-05,
+      "loss": 1.1008,
+      "step": 63700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.586353846153846e-05,
+      "loss": 1.1116,
+      "step": 63800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5855846153846154e-05,
+      "loss": 1.0897,
+      "step": 63900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5848153846153848e-05,
+      "loss": 1.1502,
+      "step": 64000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.584046153846154e-05,
+      "loss": 1.1106,
+      "step": 64100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5832769230769234e-05,
+      "loss": 1.0798,
+      "step": 64200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.582507692307692e-05,
+      "loss": 1.1259,
+      "step": 64300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5817384615384616e-05,
+      "loss": 1.0993,
+      "step": 64400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.580969230769231e-05,
+      "loss": 1.1089,
+      "step": 64500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5802e-05,
+      "loss": 1.0732,
+      "step": 64600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5794384615384616e-05,
+      "loss": 1.0891,
+      "step": 64700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5786692307692307e-05,
+      "loss": 1.0926,
+      "step": 64800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5779e-05,
+      "loss": 1.0659,
+      "step": 64900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5771307692307693e-05,
+      "loss": 1.0702,
+      "step": 65000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.5834455490112305,
+      "eval_runtime": 131.4164,
+      "eval_samples_per_second": 15.219,
+      "eval_steps_per_second": 0.951,
+      "step": 65000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5763615384615387e-05,
+      "loss": 1.0826,
+      "step": 65100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.575592307692308e-05,
+      "loss": 1.0499,
+      "step": 65200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.574823076923077e-05,
+      "loss": 1.0996,
+      "step": 65300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.574053846153846e-05,
+      "loss": 1.0963,
+      "step": 65400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5732846153846155e-05,
+      "loss": 1.1083,
+      "step": 65500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.572515384615385e-05,
+      "loss": 1.0771,
+      "step": 65600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5717461538461537e-05,
+      "loss": 1.0973,
+      "step": 65700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.570976923076923e-05,
+      "loss": 1.0877,
+      "step": 65800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5702076923076923e-05,
+      "loss": 1.0814,
+      "step": 65900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5694384615384617e-05,
+      "loss": 1.055,
+      "step": 66000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.568669230769231e-05,
+      "loss": 1.0955,
+      "step": 66100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5679e-05,
+      "loss": 1.1238,
+      "step": 66200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5671307692307694e-05,
+      "loss": 1.0789,
+      "step": 66300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5663615384615385e-05,
+      "loss": 1.1134,
+      "step": 66400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.565592307692308e-05,
+      "loss": 1.0918,
+      "step": 66500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5648230769230767e-05,
+      "loss": 1.0988,
+      "step": 66600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.564053846153846e-05,
+      "loss": 1.0999,
+      "step": 66700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5632846153846156e-05,
+      "loss": 1.1293,
+      "step": 66800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5625153846153847e-05,
+      "loss": 1.0958,
+      "step": 66900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5617461538461538e-05,
+      "loss": 1.1095,
+      "step": 67000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.560976923076923e-05,
+      "loss": 1.1307,
+      "step": 67100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5602076923076924e-05,
+      "loss": 1.0444,
+      "step": 67200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5594384615384618e-05,
+      "loss": 1.0667,
+      "step": 67300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.558669230769231e-05,
+      "loss": 1.1003,
+      "step": 67400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5579e-05,
+      "loss": 1.0955,
+      "step": 67500
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.6075574159622192,
+      "eval_runtime": 130.2607,
+      "eval_samples_per_second": 15.354,
+      "eval_steps_per_second": 0.96,
+      "step": 67500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.557130769230769e-05,
+      "loss": 1.1237,
+      "step": 67600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5563615384615386e-05,
+      "loss": 1.0991,
+      "step": 67700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.555592307692308e-05,
+      "loss": 1.1124,
+      "step": 67800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5548230769230768e-05,
+      "loss": 1.0887,
+      "step": 67900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5540538461538462e-05,
+      "loss": 1.0718,
+      "step": 68000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5532846153846153e-05,
+      "loss": 1.134,
+      "step": 68100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.552523076923077e-05,
+      "loss": 1.083,
+      "step": 68200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5517538461538463e-05,
+      "loss": 1.0713,
+      "step": 68300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5509846153846157e-05,
+      "loss": 1.0765,
+      "step": 68400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5502153846153845e-05,
+      "loss": 1.0675,
+      "step": 68500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.549446153846154e-05,
+      "loss": 1.1374,
+      "step": 68600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.548676923076923e-05,
+      "loss": 1.0974,
+      "step": 68700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5479076923076925e-05,
+      "loss": 1.0833,
+      "step": 68800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5471384615384616e-05,
+      "loss": 1.0975,
+      "step": 68900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5463692307692307e-05,
+      "loss": 1.1018,
+      "step": 69000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5456e-05,
+      "loss": 1.0893,
+      "step": 69100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5448307692307693e-05,
+      "loss": 1.0971,
+      "step": 69200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5440615384615387e-05,
+      "loss": 1.109,
+      "step": 69300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5432923076923078e-05,
+      "loss": 1.0443,
+      "step": 69400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.542523076923077e-05,
+      "loss": 1.1025,
+      "step": 69500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5417538461538464e-05,
+      "loss": 1.1093,
+      "step": 69600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5409846153846155e-05,
+      "loss": 1.0987,
+      "step": 69700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5402153846153846e-05,
+      "loss": 1.0932,
+      "step": 69800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5394461538461537e-05,
+      "loss": 1.0748,
+      "step": 69900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.538676923076923e-05,
+      "loss": 1.0867,
+      "step": 70000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.5765336751937866,
+      "eval_runtime": 130.1659,
+      "eval_samples_per_second": 15.365,
+      "eval_steps_per_second": 0.96,
+      "step": 70000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5379076923076926e-05,
+      "loss": 1.0854,
+      "step": 70100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5371384615384614e-05,
+      "loss": 1.0838,
+      "step": 70200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5363692307692308e-05,
+      "loss": 1.059,
+      "step": 70300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5356e-05,
+      "loss": 1.0849,
+      "step": 70400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5348307692307694e-05,
+      "loss": 1.0807,
+      "step": 70500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5340615384615388e-05,
+      "loss": 1.0784,
+      "step": 70600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5332923076923076e-05,
+      "loss": 1.0905,
+      "step": 70700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.532523076923077e-05,
+      "loss": 1.0449,
+      "step": 70800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531753846153846e-05,
+      "loss": 1.0844,
+      "step": 70900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5309846153846156e-05,
+      "loss": 1.0909,
+      "step": 71000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5302153846153847e-05,
+      "loss": 1.1218,
+      "step": 71100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5294461538461538e-05,
+      "loss": 1.0805,
+      "step": 71200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5286769230769232e-05,
+      "loss": 1.0766,
+      "step": 71300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5279076923076923e-05,
+      "loss": 1.0636,
+      "step": 71400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5271384615384615e-05,
+      "loss": 1.1179,
+      "step": 71500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.526369230769231e-05,
+      "loss": 1.0872,
+      "step": 71600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5256e-05,
+      "loss": 1.0963,
+      "step": 71700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5248384615384615e-05,
+      "loss": 1.0911,
+      "step": 71800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.524069230769231e-05,
+      "loss": 1.0864,
+      "step": 71900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5233e-05,
+      "loss": 1.0634,
+      "step": 72000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.522530769230769e-05,
+      "loss": 1.103,
+      "step": 72100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5217615384615386e-05,
+      "loss": 1.0612,
+      "step": 72200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5209923076923077e-05,
+      "loss": 1.086,
+      "step": 72300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.520223076923077e-05,
+      "loss": 1.0317,
+      "step": 72400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5194538461538463e-05,
+      "loss": 1.0831,
+      "step": 72500
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.5194767713546753,
+      "eval_runtime": 130.7933,
+      "eval_samples_per_second": 15.291,
+      "eval_steps_per_second": 0.956,
+      "step": 72500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5186846153846154e-05,
+      "loss": 1.1067,
+      "step": 72600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5179153846153848e-05,
+      "loss": 1.0913,
+      "step": 72700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.517146153846154e-05,
+      "loss": 1.0885,
+      "step": 72800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5163769230769234e-05,
+      "loss": 1.0706,
+      "step": 72900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.515607692307692e-05,
+      "loss": 1.0774,
+      "step": 73000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5148384615384616e-05,
+      "loss": 1.0536,
+      "step": 73100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5140692307692307e-05,
+      "loss": 1.0659,
+      "step": 73200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5133e-05,
+      "loss": 1.0637,
+      "step": 73300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5125307692307693e-05,
+      "loss": 1.0853,
+      "step": 73400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5117615384615384e-05,
+      "loss": 1.0526,
+      "step": 73500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5109923076923078e-05,
+      "loss": 1.0829,
+      "step": 73600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.510223076923077e-05,
+      "loss": 1.1177,
+      "step": 73700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5094538461538464e-05,
+      "loss": 1.1019,
+      "step": 73800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5086846153846155e-05,
+      "loss": 1.1163,
+      "step": 73900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5079153846153846e-05,
+      "loss": 1.0905,
+      "step": 74000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507146153846154e-05,
+      "loss": 1.1056,
+      "step": 74100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506376923076923e-05,
+      "loss": 1.0645,
+      "step": 74200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5056076923076922e-05,
+      "loss": 1.0983,
+      "step": 74300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5048384615384617e-05,
+      "loss": 1.1101,
+      "step": 74400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5040692307692308e-05,
+      "loss": 1.1152,
+      "step": 74500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5033000000000002e-05,
+      "loss": 1.0757,
+      "step": 74600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.502530769230769e-05,
+      "loss": 1.0934,
+      "step": 74700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.5017615384615385e-05,
+      "loss": 1.0946,
+      "step": 74800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.500992307692308e-05,
+      "loss": 1.1071,
+      "step": 74900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.500223076923077e-05,
+      "loss": 1.0869,
+      "step": 75000
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.5480155944824219,
+      "eval_runtime": 130.6026,
+      "eval_samples_per_second": 15.314,
+      "eval_steps_per_second": 0.957,
+      "step": 75000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4994538461538465e-05,
+      "loss": 1.111,
+      "step": 75100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4986846153846152e-05,
+      "loss": 1.097,
+      "step": 75200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4979153846153847e-05,
+      "loss": 1.0569,
+      "step": 75300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4971461538461538e-05,
+      "loss": 1.0797,
+      "step": 75400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4963769230769232e-05,
+      "loss": 1.0574,
+      "step": 75500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4956076923076923e-05,
+      "loss": 1.0255,
+      "step": 75600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4948384615384614e-05,
+      "loss": 1.0274,
+      "step": 75700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.494069230769231e-05,
+      "loss": 1.0517,
+      "step": 75800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4933e-05,
+      "loss": 1.0775,
+      "step": 75900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4925307692307694e-05,
+      "loss": 1.049,
+      "step": 76000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4917615384615386e-05,
+      "loss": 1.0512,
+      "step": 76100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4909923076923077e-05,
+      "loss": 1.0576,
+      "step": 76200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.490223076923077e-05,
+      "loss": 1.0661,
+      "step": 76300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4894538461538462e-05,
+      "loss": 1.0783,
+      "step": 76400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4886846153846153e-05,
+      "loss": 1.0823,
+      "step": 76500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4879153846153848e-05,
+      "loss": 1.0227,
+      "step": 76600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.487146153846154e-05,
+      "loss": 1.0977,
+      "step": 76700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4863769230769233e-05,
+      "loss": 1.0778,
+      "step": 76800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.485607692307692e-05,
+      "loss": 1.067,
+      "step": 76900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4848384615384615e-05,
+      "loss": 1.0703,
+      "step": 77000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.484069230769231e-05,
+      "loss": 1.0598,
+      "step": 77100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4833e-05,
+      "loss": 1.1039,
+      "step": 77200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4825307692307695e-05,
+      "loss": 1.0667,
+      "step": 77300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4817615384615383e-05,
+      "loss": 1.0747,
+      "step": 77400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4809999999999998e-05,
+      "loss": 1.0652,
+      "step": 77500
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.5700453519821167,
+      "eval_runtime": 130.2756,
+      "eval_samples_per_second": 15.352,
+      "eval_steps_per_second": 0.96,
+      "step": 77500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4802307692307692e-05,
+      "loss": 1.1306,
+      "step": 77600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4794615384615387e-05,
+      "loss": 1.0605,
+      "step": 77700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4786923076923078e-05,
+      "loss": 1.0771,
+      "step": 77800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.477923076923077e-05,
+      "loss": 1.055,
+      "step": 77900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.477153846153846e-05,
+      "loss": 1.0351,
+      "step": 78000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4763846153846155e-05,
+      "loss": 1.09,
+      "step": 78100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.475615384615385e-05,
+      "loss": 1.0842,
+      "step": 78200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.474846153846154e-05,
+      "loss": 1.0457,
+      "step": 78300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.474076923076923e-05,
+      "loss": 1.0852,
+      "step": 78400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4733076923076922e-05,
+      "loss": 1.1265,
+      "step": 78500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4725384615384617e-05,
+      "loss": 1.0988,
+      "step": 78600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4717692307692308e-05,
+      "loss": 1.0694,
+      "step": 78700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.471e-05,
+      "loss": 1.0923,
+      "step": 78800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4702307692307693e-05,
+      "loss": 1.0827,
+      "step": 78900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4694615384615384e-05,
+      "loss": 1.1097,
+      "step": 79000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.468692307692308e-05,
+      "loss": 1.0411,
+      "step": 79100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.467923076923077e-05,
+      "loss": 1.0316,
+      "step": 79200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.467153846153846e-05,
+      "loss": 1.0778,
+      "step": 79300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4663846153846156e-05,
+      "loss": 1.0385,
+      "step": 79400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4656153846153847e-05,
+      "loss": 1.0688,
+      "step": 79500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.464846153846154e-05,
+      "loss": 1.1079,
+      "step": 79600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.464076923076923e-05,
+      "loss": 1.0375,
+      "step": 79700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4633076923076923e-05,
+      "loss": 1.0714,
+      "step": 79800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4625384615384618e-05,
+      "loss": 1.0434,
+      "step": 79900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4617769230769233e-05,
+      "loss": 1.0679,
+      "step": 80000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.5356096029281616,
+      "eval_runtime": 130.108,
+      "eval_samples_per_second": 15.372,
+      "eval_steps_per_second": 0.961,
+      "step": 80000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4610076923076924e-05,
+      "loss": 1.0691,
+      "step": 80100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4602384615384618e-05,
+      "loss": 1.1132,
+      "step": 80200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4594692307692306e-05,
+      "loss": 1.092,
+      "step": 80300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4587e-05,
+      "loss": 1.0504,
+      "step": 80400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4579307692307695e-05,
+      "loss": 1.0718,
+      "step": 80500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4571615384615386e-05,
+      "loss": 1.0837,
+      "step": 80600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4563923076923077e-05,
+      "loss": 1.0713,
+      "step": 80700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4556230769230768e-05,
+      "loss": 1.0387,
+      "step": 80800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4548538461538462e-05,
+      "loss": 1.0764,
+      "step": 80900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4540846153846157e-05,
+      "loss": 1.0703,
+      "step": 81000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4533153846153848e-05,
+      "loss": 1.1426,
+      "step": 81100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.452546153846154e-05,
+      "loss": 1.066,
+      "step": 81200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.451776923076923e-05,
+      "loss": 1.0512,
+      "step": 81300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4510076923076925e-05,
+      "loss": 1.0785,
+      "step": 81400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.450238461538462e-05,
+      "loss": 1.0644,
+      "step": 81500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4494692307692307e-05,
+      "loss": 1.0736,
+      "step": 81600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4487e-05,
+      "loss": 1.0304,
+      "step": 81700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4479307692307692e-05,
+      "loss": 1.0517,
+      "step": 81800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4471615384615387e-05,
+      "loss": 1.0873,
+      "step": 81900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4463923076923074e-05,
+      "loss": 1.0651,
+      "step": 82000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.445623076923077e-05,
+      "loss": 1.0996,
+      "step": 82100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4448538461538463e-05,
+      "loss": 1.1139,
+      "step": 82200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4440846153846154e-05,
+      "loss": 1.0254,
+      "step": 82300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.443315384615385e-05,
+      "loss": 1.0673,
+      "step": 82400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4425461538461537e-05,
+      "loss": 1.0799,
+      "step": 82500
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.5397475957870483,
+      "eval_runtime": 131.6323,
+      "eval_samples_per_second": 15.194,
+      "eval_steps_per_second": 0.95,
+      "step": 82500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.441776923076923e-05,
+      "loss": 1.0758,
+      "step": 82600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4410076923076926e-05,
+      "loss": 1.0535,
+      "step": 82700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4402384615384617e-05,
+      "loss": 1.0797,
+      "step": 82800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4394692307692308e-05,
+      "loss": 1.0468,
+      "step": 82900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4387e-05,
+      "loss": 1.0578,
+      "step": 83000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4379384615384614e-05,
+      "loss": 1.067,
+      "step": 83100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4371692307692308e-05,
+      "loss": 1.0865,
+      "step": 83200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4364000000000003e-05,
+      "loss": 1.0594,
+      "step": 83300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4356307692307694e-05,
+      "loss": 1.082,
+      "step": 83400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4348615384615385e-05,
+      "loss": 1.091,
+      "step": 83500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4340923076923076e-05,
+      "loss": 1.1018,
+      "step": 83600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.433323076923077e-05,
+      "loss": 1.1077,
+      "step": 83700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4325538461538465e-05,
+      "loss": 1.0558,
+      "step": 83800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4317846153846152e-05,
+      "loss": 1.0183,
+      "step": 83900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4310153846153847e-05,
+      "loss": 1.0664,
+      "step": 84000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4302461538461538e-05,
+      "loss": 1.092,
+      "step": 84100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4294769230769232e-05,
+      "loss": 1.0443,
+      "step": 84200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4287076923076923e-05,
+      "loss": 1.058,
+      "step": 84300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4279384615384615e-05,
+      "loss": 1.0425,
+      "step": 84400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.427169230769231e-05,
+      "loss": 1.0411,
+      "step": 84500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4264e-05,
+      "loss": 1.0612,
+      "step": 84600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4256307692307695e-05,
+      "loss": 1.0698,
+      "step": 84700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4248615384615386e-05,
+      "loss": 1.0765,
+      "step": 84800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4240923076923077e-05,
+      "loss": 1.0556,
+      "step": 84900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.423323076923077e-05,
+      "loss": 1.0697,
+      "step": 85000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.547716498374939,
+      "eval_runtime": 130.8651,
+      "eval_samples_per_second": 15.283,
+      "eval_steps_per_second": 0.955,
+      "step": 85000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4225538461538462e-05,
+      "loss": 1.0601,
+      "step": 85100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4217846153846153e-05,
+      "loss": 1.0573,
+      "step": 85200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4210153846153844e-05,
+      "loss": 1.0509,
+      "step": 85300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.420246153846154e-05,
+      "loss": 1.0788,
+      "step": 85400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4194769230769233e-05,
+      "loss": 1.0338,
+      "step": 85500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4187076923076924e-05,
+      "loss": 1.1074,
+      "step": 85600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.417946153846154e-05,
+      "loss": 1.1148,
+      "step": 85700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.417176923076923e-05,
+      "loss": 1.0875,
+      "step": 85800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.416407692307692e-05,
+      "loss": 1.0676,
+      "step": 85900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4156384615384616e-05,
+      "loss": 1.1043,
+      "step": 86000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.414869230769231e-05,
+      "loss": 1.0762,
+      "step": 86100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4140999999999998e-05,
+      "loss": 1.0562,
+      "step": 86200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4133307692307693e-05,
+      "loss": 1.0835,
+      "step": 86300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4125615384615384e-05,
+      "loss": 1.0483,
+      "step": 86400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4117923076923078e-05,
+      "loss": 1.094,
+      "step": 86500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4110230769230773e-05,
+      "loss": 1.0468,
+      "step": 86600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.410253846153846e-05,
+      "loss": 1.1263,
+      "step": 86700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4094846153846155e-05,
+      "loss": 1.0748,
+      "step": 86800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4087153846153846e-05,
+      "loss": 1.0867,
+      "step": 86900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.407946153846154e-05,
+      "loss": 1.0435,
+      "step": 87000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.407176923076923e-05,
+      "loss": 1.0414,
+      "step": 87100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4064076923076922e-05,
+      "loss": 1.0812,
+      "step": 87200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4056384615384617e-05,
+      "loss": 1.0664,
+      "step": 87300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4048692307692308e-05,
+      "loss": 1.0468,
+      "step": 87400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4041000000000002e-05,
+      "loss": 1.0822,
+      "step": 87500
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.5006853342056274,
+      "eval_runtime": 131.593,
+      "eval_samples_per_second": 15.198,
+      "eval_steps_per_second": 0.95,
+      "step": 87500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4033307692307693e-05,
+      "loss": 1.0844,
+      "step": 87600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4025615384615385e-05,
+      "loss": 1.0459,
+      "step": 87700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.401792307692308e-05,
+      "loss": 1.0833,
+      "step": 87800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.401023076923077e-05,
+      "loss": 1.1121,
+      "step": 87900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.400253846153846e-05,
+      "loss": 1.0666,
+      "step": 88000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3994846153846156e-05,
+      "loss": 1.0666,
+      "step": 88100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3987153846153847e-05,
+      "loss": 1.0679,
+      "step": 88200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.397946153846154e-05,
+      "loss": 1.089,
+      "step": 88300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.397176923076923e-05,
+      "loss": 1.041,
+      "step": 88400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3964076923076923e-05,
+      "loss": 1.0825,
+      "step": 88500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3956384615384614e-05,
+      "loss": 1.0668,
+      "step": 88600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.394869230769231e-05,
+      "loss": 1.0791,
+      "step": 88700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3941000000000003e-05,
+      "loss": 1.0767,
+      "step": 88800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.393330769230769e-05,
+      "loss": 1.0845,
+      "step": 88900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3925615384615386e-05,
+      "loss": 1.0834,
+      "step": 89000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3917923076923077e-05,
+      "loss": 1.1095,
+      "step": 89100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.391023076923077e-05,
+      "loss": 1.0422,
+      "step": 89200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3902538461538462e-05,
+      "loss": 1.0737,
+      "step": 89300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3894846153846153e-05,
+      "loss": 1.0839,
+      "step": 89400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3887153846153848e-05,
+      "loss": 1.0444,
+      "step": 89500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.387946153846154e-05,
+      "loss": 1.0795,
+      "step": 89600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.387176923076923e-05,
+      "loss": 1.0703,
+      "step": 89700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3864076923076924e-05,
+      "loss": 1.0334,
+      "step": 89800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.385646153846154e-05,
+      "loss": 1.0797,
+      "step": 89900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.384876923076923e-05,
+      "loss": 1.0666,
+      "step": 90000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.5210641622543335,
+      "eval_runtime": 131.0046,
+      "eval_samples_per_second": 15.267,
+      "eval_steps_per_second": 0.954,
+      "step": 90000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3841153846153845e-05,
+      "loss": 1.0827,
+      "step": 90100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.383346153846154e-05,
+      "loss": 1.0864,
+      "step": 90200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.382576923076923e-05,
+      "loss": 1.0915,
+      "step": 90300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3818076923076925e-05,
+      "loss": 1.0804,
+      "step": 90400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3810384615384616e-05,
+      "loss": 1.0402,
+      "step": 90500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3802692307692307e-05,
+      "loss": 1.0711,
+      "step": 90600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3795000000000002e-05,
+      "loss": 1.0514,
+      "step": 90700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3787307692307693e-05,
+      "loss": 1.0649,
+      "step": 90800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3779615384615384e-05,
+      "loss": 1.0687,
+      "step": 90900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.377192307692308e-05,
+      "loss": 1.0434,
+      "step": 91000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.376423076923077e-05,
+      "loss": 1.0649,
+      "step": 91100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3756538461538464e-05,
+      "loss": 1.0095,
+      "step": 91200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.374884615384615e-05,
+      "loss": 1.0059,
+      "step": 91300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3741153846153846e-05,
+      "loss": 1.0642,
+      "step": 91400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.373346153846154e-05,
+      "loss": 1.0627,
+      "step": 91500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.372576923076923e-05,
+      "loss": 1.0941,
+      "step": 91600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3718076923076926e-05,
+      "loss": 1.0316,
+      "step": 91700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3710384615384614e-05,
+      "loss": 1.111,
+      "step": 91800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3702692307692308e-05,
+      "loss": 1.1001,
+      "step": 91900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3695000000000003e-05,
+      "loss": 1.0193,
+      "step": 92000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3687307692307694e-05,
+      "loss": 1.0485,
+      "step": 92100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3679615384615385e-05,
+      "loss": 1.0721,
+      "step": 92200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3671923076923076e-05,
+      "loss": 1.0393,
+      "step": 92300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.366430769230769e-05,
+      "loss": 1.0147,
+      "step": 92400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3656615384615385e-05,
+      "loss": 1.0469,
+      "step": 92500
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.5373603105545044,
+      "eval_runtime": 131.1214,
+      "eval_samples_per_second": 15.253,
+      "eval_steps_per_second": 0.953,
+      "step": 92500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.364892307692308e-05,
+      "loss": 1.0659,
+      "step": 92600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.364123076923077e-05,
+      "loss": 1.0542,
+      "step": 92700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3633538461538462e-05,
+      "loss": 1.0143,
+      "step": 92800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3625846153846153e-05,
+      "loss": 1.0684,
+      "step": 92900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3618153846153847e-05,
+      "loss": 1.0164,
+      "step": 93000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.361046153846154e-05,
+      "loss": 1.034,
+      "step": 93100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.360276923076923e-05,
+      "loss": 1.0634,
+      "step": 93200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3595076923076924e-05,
+      "loss": 1.0504,
+      "step": 93300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3587384615384615e-05,
+      "loss": 0.9939,
+      "step": 93400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.357969230769231e-05,
+      "loss": 1.0502,
+      "step": 93500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3572e-05,
+      "loss": 1.0661,
+      "step": 93600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3564307692307692e-05,
+      "loss": 1.0276,
+      "step": 93700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3556615384615386e-05,
+      "loss": 1.0925,
+      "step": 93800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3548923076923077e-05,
+      "loss": 1.0776,
+      "step": 93900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3541230769230772e-05,
+      "loss": 1.0673,
+      "step": 94000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.353353846153846e-05,
+      "loss": 1.0458,
+      "step": 94100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3525846153846154e-05,
+      "loss": 1.0402,
+      "step": 94200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.351815384615385e-05,
+      "loss": 1.0908,
+      "step": 94300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.351046153846154e-05,
+      "loss": 1.0471,
+      "step": 94400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.350276923076923e-05,
+      "loss": 1.1027,
+      "step": 94500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.349507692307692e-05,
+      "loss": 1.0453,
+      "step": 94600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3487461538461536e-05,
+      "loss": 1.0569,
+      "step": 94700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.347976923076923e-05,
+      "loss": 1.0914,
+      "step": 94800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3472076923076925e-05,
+      "loss": 1.0569,
+      "step": 94900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3464384615384616e-05,
+      "loss": 1.0841,
+      "step": 95000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.5133154392242432,
+      "eval_runtime": 130.7321,
+      "eval_samples_per_second": 15.298,
+      "eval_steps_per_second": 0.956,
+      "step": 95000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3456692307692308e-05,
+      "loss": 1.0889,
+      "step": 95100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3449e-05,
+      "loss": 1.0605,
+      "step": 95200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3441307692307693e-05,
+      "loss": 1.081,
+      "step": 95300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3433615384615388e-05,
+      "loss": 1.0624,
+      "step": 95400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.342592307692308e-05,
+      "loss": 1.0428,
+      "step": 95500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.341823076923077e-05,
+      "loss": 1.0496,
+      "step": 95600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.341053846153846e-05,
+      "loss": 0.9938,
+      "step": 95700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3402846153846155e-05,
+      "loss": 1.0766,
+      "step": 95800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.339515384615385e-05,
+      "loss": 1.0204,
+      "step": 95900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3387461538461537e-05,
+      "loss": 1.1,
+      "step": 96000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3379769230769232e-05,
+      "loss": 1.0333,
+      "step": 96100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3372076923076923e-05,
+      "loss": 1.0397,
+      "step": 96200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3364384615384617e-05,
+      "loss": 1.0832,
+      "step": 96300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3356692307692305e-05,
+      "loss": 1.0113,
+      "step": 96400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3349e-05,
+      "loss": 1.0573,
+      "step": 96500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3341307692307694e-05,
+      "loss": 1.1054,
+      "step": 96600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3333615384615385e-05,
+      "loss": 1.0595,
+      "step": 96700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.332592307692308e-05,
+      "loss": 1.0476,
+      "step": 96800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3318307692307694e-05,
+      "loss": 1.049,
+      "step": 96900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3310615384615382e-05,
+      "loss": 1.001,
+      "step": 97000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3302923076923077e-05,
+      "loss": 1.1202,
+      "step": 97100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.329523076923077e-05,
+      "loss": 1.0364,
+      "step": 97200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3287538461538462e-05,
+      "loss": 1.0396,
+      "step": 97300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3279846153846157e-05,
+      "loss": 1.0358,
+      "step": 97400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3272153846153844e-05,
+      "loss": 1.0217,
+      "step": 97500
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.4849553108215332,
+      "eval_runtime": 130.7595,
+      "eval_samples_per_second": 15.295,
+      "eval_steps_per_second": 0.956,
+      "step": 97500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.326446153846154e-05,
+      "loss": 1.0203,
+      "step": 97600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3256769230769233e-05,
+      "loss": 1.0379,
+      "step": 97700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3249076923076924e-05,
+      "loss": 1.0661,
+      "step": 97800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3241384615384615e-05,
+      "loss": 1.0563,
+      "step": 97900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3233692307692306e-05,
+      "loss": 1.0128,
+      "step": 98000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3226e-05,
+      "loss": 1.0703,
+      "step": 98100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3218307692307695e-05,
+      "loss": 1.0794,
+      "step": 98200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3210615384615383e-05,
+      "loss": 0.9987,
+      "step": 98300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3202923076923078e-05,
+      "loss": 1.0746,
+      "step": 98400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.319523076923077e-05,
+      "loss": 1.0439,
+      "step": 98500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3187538461538463e-05,
+      "loss": 1.1001,
+      "step": 98600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3179846153846158e-05,
+      "loss": 1.0955,
+      "step": 98700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3172153846153845e-05,
+      "loss": 0.9996,
+      "step": 98800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.316446153846154e-05,
+      "loss": 1.0914,
+      "step": 98900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.315676923076923e-05,
+      "loss": 1.0,
+      "step": 99000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3149076923076925e-05,
+      "loss": 1.0505,
+      "step": 99100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3141384615384616e-05,
+      "loss": 1.0647,
+      "step": 99200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.313376923076923e-05,
+      "loss": 0.9942,
+      "step": 99300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3126076923076922e-05,
+      "loss": 1.0691,
+      "step": 99400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3118384615384617e-05,
+      "loss": 1.0445,
+      "step": 99500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3110692307692308e-05,
+      "loss": 1.0339,
+      "step": 99600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3103000000000002e-05,
+      "loss": 1.0494,
+      "step": 99700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3095307692307693e-05,
+      "loss": 1.0224,
+      "step": 99800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3087615384615384e-05,
+      "loss": 1.0361,
+      "step": 99900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.307992307692308e-05,
+      "loss": 1.0622,
+      "step": 100000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.4965757131576538,
+      "eval_runtime": 131.5445,
+      "eval_samples_per_second": 15.204,
+      "eval_steps_per_second": 0.95,
+      "step": 100000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.307223076923077e-05,
+      "loss": 1.0562,
+      "step": 100100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.306453846153846e-05,
+      "loss": 1.0154,
+      "step": 100200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3056846153846152e-05,
+      "loss": 1.0653,
+      "step": 100300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3049153846153847e-05,
+      "loss": 1.0374,
+      "step": 100400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.304146153846154e-05,
+      "loss": 1.037,
+      "step": 100500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3033769230769232e-05,
+      "loss": 1.0262,
+      "step": 100600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3026076923076923e-05,
+      "loss": 1.0138,
+      "step": 100700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3018384615384614e-05,
+      "loss": 1.0329,
+      "step": 100800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.301069230769231e-05,
+      "loss": 1.0512,
+      "step": 100900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3003000000000003e-05,
+      "loss": 1.0413,
+      "step": 101000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.299530769230769e-05,
+      "loss": 1.071,
+      "step": 101100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2987615384615385e-05,
+      "loss": 1.0552,
+      "step": 101200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2979923076923076e-05,
+      "loss": 1.0374,
+      "step": 101300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.297223076923077e-05,
+      "loss": 1.0478,
+      "step": 101400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2964538461538462e-05,
+      "loss": 1.0089,
+      "step": 101500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2956846153846153e-05,
+      "loss": 1.052,
+      "step": 101600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2949153846153848e-05,
+      "loss": 1.0577,
+      "step": 101700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.294146153846154e-05,
+      "loss": 0.9886,
+      "step": 101800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2933769230769233e-05,
+      "loss": 1.0488,
+      "step": 101900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2926076923076924e-05,
+      "loss": 1.0836,
+      "step": 102000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.291846153846154e-05,
+      "loss": 0.9899,
+      "step": 102100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2910846153846157e-05,
+      "loss": 1.0581,
+      "step": 102200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2903153846153845e-05,
+      "loss": 1.052,
+      "step": 102300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.289546153846154e-05,
+      "loss": 1.0239,
+      "step": 102400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.288776923076923e-05,
+      "loss": 1.0289,
+      "step": 102500
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.5171343088150024,
+      "eval_runtime": 130.3864,
+      "eval_samples_per_second": 15.339,
+      "eval_steps_per_second": 0.959,
+      "step": 102500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2880076923076925e-05,
+      "loss": 1.0614,
+      "step": 102600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2872384615384616e-05,
+      "loss": 1.0706,
+      "step": 102700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2864692307692307e-05,
+      "loss": 1.0212,
+      "step": 102800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2857e-05,
+      "loss": 1.0903,
+      "step": 102900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2849307692307693e-05,
+      "loss": 1.0526,
+      "step": 103000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2841615384615384e-05,
+      "loss": 1.0334,
+      "step": 103100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2833923076923078e-05,
+      "loss": 1.0641,
+      "step": 103200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.282623076923077e-05,
+      "loss": 1.074,
+      "step": 103300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2818538461538464e-05,
+      "loss": 1.0184,
+      "step": 103400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2810846153846155e-05,
+      "loss": 1.002,
+      "step": 103500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.2803153846153846e-05,
+      "loss": 1.07,
+      "step": 103600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.279546153846154e-05,
+      "loss": 1.0578,
+      "step": 103700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.278776923076923e-05,
+      "loss": 1.024,
+      "step": 103800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2780076923076926e-05,
+      "loss": 1.0531,
+      "step": 103900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2772384615384614e-05,
+      "loss": 1.0332,
+      "step": 104000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2764692307692308e-05,
+      "loss": 1.0458,
+      "step": 104100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2757e-05,
+      "loss": 1.0424,
+      "step": 104200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2749307692307694e-05,
+      "loss": 1.0296,
+      "step": 104300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2741615384615385e-05,
+      "loss": 1.0204,
+      "step": 104400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2733923076923076e-05,
+      "loss": 1.0803,
+      "step": 104500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.272623076923077e-05,
+      "loss": 1.0599,
+      "step": 104600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.271853846153846e-05,
+      "loss": 1.0386,
+      "step": 104700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2710846153846156e-05,
+      "loss": 1.0471,
+      "step": 104800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2703153846153847e-05,
+      "loss": 1.0195,
+      "step": 104900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2695461538461538e-05,
+      "loss": 1.0214,
+      "step": 105000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 1.4789345264434814,
+      "eval_runtime": 131.8697,
+      "eval_samples_per_second": 15.166,
+      "eval_steps_per_second": 0.948,
+      "step": 105000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2687769230769232e-05,
+      "loss": 1.026,
+      "step": 105100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2680076923076923e-05,
+      "loss": 1.0055,
+      "step": 105200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2672384615384615e-05,
+      "loss": 1.0401,
+      "step": 105300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266469230769231e-05,
+      "loss": 1.032,
+      "step": 105400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2657e-05,
+      "loss": 1.0534,
+      "step": 105500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2649307692307695e-05,
+      "loss": 1.0616,
+      "step": 105600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2641615384615386e-05,
+      "loss": 1.0306,
+      "step": 105700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2633923076923077e-05,
+      "loss": 1.0417,
+      "step": 105800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.262623076923077e-05,
+      "loss": 1.0986,
+      "step": 105900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2618538461538462e-05,
+      "loss": 1.0547,
+      "step": 106000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2610846153846157e-05,
+      "loss": 0.9948,
+      "step": 106100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2603153846153844e-05,
+      "loss": 1.0511,
+      "step": 106200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.259546153846154e-05,
+      "loss": 1.0581,
+      "step": 106300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2587769230769233e-05,
+      "loss": 1.0873,
+      "step": 106400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2580153846153848e-05,
+      "loss": 1.0129,
+      "step": 106500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.257246153846154e-05,
+      "loss": 1.0894,
+      "step": 106600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2564769230769234e-05,
+      "loss": 1.0456,
+      "step": 106700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.255707692307692e-05,
+      "loss": 1.0401,
+      "step": 106800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2549384615384616e-05,
+      "loss": 1.0046,
+      "step": 106900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.254169230769231e-05,
+      "loss": 1.0458,
+      "step": 107000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2534e-05,
+      "loss": 1.1005,
+      "step": 107100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2526307692307693e-05,
+      "loss": 1.0811,
+      "step": 107200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2518615384615384e-05,
+      "loss": 1.0432,
+      "step": 107300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2510923076923078e-05,
+      "loss": 1.0284,
+      "step": 107400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.250323076923077e-05,
+      "loss": 1.063,
+      "step": 107500
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.453118920326233,
+      "eval_runtime": 130.5926,
+      "eval_samples_per_second": 15.315,
+      "eval_steps_per_second": 0.957,
+      "step": 107500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.249553846153846e-05,
+      "loss": 1.0785,
+      "step": 107600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2487846153846155e-05,
+      "loss": 1.0598,
+      "step": 107700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2480153846153846e-05,
+      "loss": 1.0308,
+      "step": 107800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.247246153846154e-05,
+      "loss": 1.0533,
+      "step": 107900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.246476923076923e-05,
+      "loss": 0.9922,
+      "step": 108000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2457076923076922e-05,
+      "loss": 1.0052,
+      "step": 108100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2449384615384617e-05,
+      "loss": 1.0277,
+      "step": 108200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2441692307692308e-05,
+      "loss": 1.0651,
+      "step": 108300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2434000000000002e-05,
+      "loss": 1.0417,
+      "step": 108400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.242630769230769e-05,
+      "loss": 1.0646,
+      "step": 108500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2418615384615385e-05,
+      "loss": 1.0578,
+      "step": 108600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.241092307692308e-05,
+      "loss": 0.9878,
+      "step": 108700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.240323076923077e-05,
+      "loss": 1.0452,
+      "step": 108800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.239553846153846e-05,
+      "loss": 1.0509,
+      "step": 108900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2387846153846152e-05,
+      "loss": 1.0109,
+      "step": 109000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2380153846153847e-05,
+      "loss": 1.0986,
+      "step": 109100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.237246153846154e-05,
+      "loss": 1.0414,
+      "step": 109200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2364769230769232e-05,
+      "loss": 1.0552,
+      "step": 109300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2357076923076923e-05,
+      "loss": 1.0379,
+      "step": 109400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2349384615384614e-05,
+      "loss": 1.0582,
+      "step": 109500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.234169230769231e-05,
+      "loss": 1.0766,
+      "step": 109600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2334000000000003e-05,
+      "loss": 1.0861,
+      "step": 109700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.232630769230769e-05,
+      "loss": 1.0278,
+      "step": 109800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2318615384615386e-05,
+      "loss": 1.0299,
+      "step": 109900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2310923076923077e-05,
+      "loss": 1.0349,
+      "step": 110000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 1.5061930418014526,
+      "eval_runtime": 131.3793,
+      "eval_samples_per_second": 15.223,
+      "eval_steps_per_second": 0.951,
+      "step": 110000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.230323076923077e-05,
+      "loss": 1.0435,
+      "step": 110100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2295538461538462e-05,
+      "loss": 1.0475,
+      "step": 110200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2287846153846153e-05,
+      "loss": 1.0105,
+      "step": 110300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2280153846153848e-05,
+      "loss": 1.0701,
+      "step": 110400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.227246153846154e-05,
+      "loss": 1.0523,
+      "step": 110500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2264846153846154e-05,
+      "loss": 1.0491,
+      "step": 110600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2257153846153848e-05,
+      "loss": 1.0137,
+      "step": 110700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2249538461538463e-05,
+      "loss": 1.0346,
+      "step": 110800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2241846153846157e-05,
+      "loss": 1.0541,
+      "step": 110900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2234153846153845e-05,
+      "loss": 1.0454,
+      "step": 111000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.222646153846154e-05,
+      "loss": 1.0393,
+      "step": 111100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.221876923076923e-05,
+      "loss": 1.0686,
+      "step": 111200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2211076923076925e-05,
+      "loss": 1.0225,
+      "step": 111300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2203384615384613e-05,
+      "loss": 1.0231,
+      "step": 111400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2195692307692307e-05,
+      "loss": 0.9744,
+      "step": 111500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2188e-05,
+      "loss": 1.0253,
+      "step": 111600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2180307692307693e-05,
+      "loss": 1.0015,
+      "step": 111700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2172615384615387e-05,
+      "loss": 1.0802,
+      "step": 111800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2164923076923075e-05,
+      "loss": 1.0627,
+      "step": 111900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.215723076923077e-05,
+      "loss": 1.057,
+      "step": 112000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2149538461538464e-05,
+      "loss": 1.0569,
+      "step": 112100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2141846153846155e-05,
+      "loss": 1.0432,
+      "step": 112200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2134153846153846e-05,
+      "loss": 1.0215,
+      "step": 112300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2126461538461537e-05,
+      "loss": 1.0231,
+      "step": 112400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.211876923076923e-05,
+      "loss": 1.0696,
+      "step": 112500
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.5029948949813843,
+      "eval_runtime": 131.6828,
+      "eval_samples_per_second": 15.188,
+      "eval_steps_per_second": 0.949,
+      "step": 112500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2111076923076926e-05,
+      "loss": 1.0459,
+      "step": 112600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2103384615384614e-05,
+      "loss": 1.0528,
+      "step": 112700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2095692307692308e-05,
+      "loss": 1.0254,
+      "step": 112800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2088076923076923e-05,
+      "loss": 0.9991,
+      "step": 112900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2080384615384614e-05,
+      "loss": 1.0054,
+      "step": 113000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.207269230769231e-05,
+      "loss": 1.0626,
+      "step": 113100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2065000000000003e-05,
+      "loss": 1.0288,
+      "step": 113200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.205730769230769e-05,
+      "loss": 1.0465,
+      "step": 113300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2049615384615385e-05,
+      "loss": 1.0748,
+      "step": 113400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2041923076923076e-05,
+      "loss": 1.0285,
+      "step": 113500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.203423076923077e-05,
+      "loss": 1.0319,
+      "step": 113600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2026538461538462e-05,
+      "loss": 1.0477,
+      "step": 113700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2018846153846153e-05,
+      "loss": 1.0583,
+      "step": 113800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2011153846153847e-05,
+      "loss": 1.0649,
+      "step": 113900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.200346153846154e-05,
+      "loss": 0.9828,
+      "step": 114000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1995769230769233e-05,
+      "loss": 1.07,
+      "step": 114100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1988076923076924e-05,
+      "loss": 1.0722,
+      "step": 114200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1980384615384615e-05,
+      "loss": 1.0299,
+      "step": 114300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.197269230769231e-05,
+      "loss": 1.0127,
+      "step": 114400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1965e-05,
+      "loss": 0.999,
+      "step": 114500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.195730769230769e-05,
+      "loss": 1.0291,
+      "step": 114600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1949615384615383e-05,
+      "loss": 1.0417,
+      "step": 114700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1941923076923077e-05,
+      "loss": 1.0009,
+      "step": 114800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1934230769230772e-05,
+      "loss": 1.0252,
+      "step": 114900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1926538461538463e-05,
+      "loss": 1.0883,
+      "step": 115000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 1.4825422763824463,
+      "eval_runtime": 131.0903,
+      "eval_samples_per_second": 15.257,
+      "eval_steps_per_second": 0.954,
+      "step": 115000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1918923076923078e-05,
+      "loss": 1.0086,
+      "step": 115100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.191123076923077e-05,
+      "loss": 1.027,
+      "step": 115200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1903538461538463e-05,
+      "loss": 1.032,
+      "step": 115300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1895846153846154e-05,
+      "loss": 1.0432,
+      "step": 115400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.188815384615385e-05,
+      "loss": 1.0631,
+      "step": 115500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1880461538461536e-05,
+      "loss": 1.0215,
+      "step": 115600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.187276923076923e-05,
+      "loss": 1.0088,
+      "step": 115700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1865076923076922e-05,
+      "loss": 1.0515,
+      "step": 115800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1857384615384616e-05,
+      "loss": 1.0058,
+      "step": 115900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.184969230769231e-05,
+      "loss": 1.0086,
+      "step": 116000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1842e-05,
+      "loss": 1.0397,
+      "step": 116100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1834307692307693e-05,
+      "loss": 1.016,
+      "step": 116200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1826615384615384e-05,
+      "loss": 1.0141,
+      "step": 116300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.181892307692308e-05,
+      "loss": 1.0329,
+      "step": 116400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.181123076923077e-05,
+      "loss": 1.0405,
+      "step": 116500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.180353846153846e-05,
+      "loss": 1.0623,
+      "step": 116600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1795846153846155e-05,
+      "loss": 1.0103,
+      "step": 116700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1788153846153846e-05,
+      "loss": 1.0166,
+      "step": 116800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.178046153846154e-05,
+      "loss": 1.0348,
+      "step": 116900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1772769230769232e-05,
+      "loss": 1.0755,
+      "step": 117000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1765076923076923e-05,
+      "loss": 1.045,
+      "step": 117100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1757384615384617e-05,
+      "loss": 1.0848,
+      "step": 117200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.174969230769231e-05,
+      "loss": 0.9988,
+      "step": 117300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1742e-05,
+      "loss": 1.0156,
+      "step": 117400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1734307692307694e-05,
+      "loss": 1.0223,
+      "step": 117500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 1.4751300811767578,
+      "eval_runtime": 131.1386,
+      "eval_samples_per_second": 15.251,
+      "eval_steps_per_second": 0.953,
+      "step": 117500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.172669230769231e-05,
+      "loss": 0.9964,
+      "step": 117600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1719e-05,
+      "loss": 1.0262,
+      "step": 117700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1711307692307694e-05,
+      "loss": 1.0444,
+      "step": 117800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1703615384615385e-05,
+      "loss": 1.0184,
+      "step": 117900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1695923076923077e-05,
+      "loss": 1.0043,
+      "step": 118000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.168823076923077e-05,
+      "loss": 1.0518,
+      "step": 118100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1680538461538462e-05,
+      "loss": 1.0169,
+      "step": 118200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1672846153846157e-05,
+      "loss": 1.0528,
+      "step": 118300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1665153846153844e-05,
+      "loss": 1.0483,
+      "step": 118400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.165746153846154e-05,
+      "loss": 0.9975,
+      "step": 118500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1649769230769233e-05,
+      "loss": 1.0597,
+      "step": 118600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1642076923076924e-05,
+      "loss": 1.0243,
+      "step": 118700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1634384615384615e-05,
+      "loss": 1.0682,
+      "step": 118800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1626692307692306e-05,
+      "loss": 1.0388,
+      "step": 118900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1619e-05,
+      "loss": 1.0389,
+      "step": 119000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1611307692307692e-05,
+      "loss": 1.0553,
+      "step": 119100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1603615384615386e-05,
+      "loss": 1.0605,
+      "step": 119200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1595923076923077e-05,
+      "loss": 1.0535,
+      "step": 119300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.158823076923077e-05,
+      "loss": 1.0755,
+      "step": 119400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1580538461538463e-05,
+      "loss": 1.0496,
+      "step": 119500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1572846153846154e-05,
+      "loss": 1.0765,
+      "step": 119600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1565153846153845e-05,
+      "loss": 1.0178,
+      "step": 119700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.155746153846154e-05,
+      "loss": 1.0908,
+      "step": 119800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.154976923076923e-05,
+      "loss": 1.0777,
+      "step": 119900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1542076923076925e-05,
+      "loss": 1.037,
+      "step": 120000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.4711098670959473,
+      "eval_runtime": 131.0971,
+      "eval_samples_per_second": 15.256,
+      "eval_steps_per_second": 0.953,
+      "step": 120000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.153446153846154e-05,
+      "loss": 1.0124,
+      "step": 120100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.152676923076923e-05,
+      "loss": 1.0612,
+      "step": 120200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1519076923076922e-05,
+      "loss": 1.0137,
+      "step": 120300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1511384615384617e-05,
+      "loss": 1.0632,
+      "step": 120400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1503692307692308e-05,
+      "loss": 1.0147,
+      "step": 120500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1496000000000002e-05,
+      "loss": 0.9996,
+      "step": 120600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.148830769230769e-05,
+      "loss": 1.0112,
+      "step": 120700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1480615384615384e-05,
+      "loss": 1.0685,
+      "step": 120800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.147292307692308e-05,
+      "loss": 1.0242,
+      "step": 120900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.146523076923077e-05,
+      "loss": 1.0138,
+      "step": 121000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1457538461538464e-05,
+      "loss": 1.0233,
+      "step": 121100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1449846153846152e-05,
+      "loss": 1.0172,
+      "step": 121200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1442153846153847e-05,
+      "loss": 1.0176,
+      "step": 121300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.143446153846154e-05,
+      "loss": 1.0573,
+      "step": 121400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1426769230769232e-05,
+      "loss": 1.0253,
+      "step": 121500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1419076923076923e-05,
+      "loss": 1.0018,
+      "step": 121600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1411384615384614e-05,
+      "loss": 1.005,
+      "step": 121700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.140369230769231e-05,
+      "loss": 1.0124,
+      "step": 121800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1396e-05,
+      "loss": 1.032,
+      "step": 121900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1388307692307694e-05,
+      "loss": 1.0673,
+      "step": 122000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1380615384615385e-05,
+      "loss": 0.9895,
+      "step": 122100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1372923076923076e-05,
+      "loss": 1.0222,
+      "step": 122200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.136523076923077e-05,
+      "loss": 1.036,
+      "step": 122300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1357538461538462e-05,
+      "loss": 1.0493,
+      "step": 122400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1349846153846153e-05,
+      "loss": 1.0439,
+      "step": 122500
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 1.4662411212921143,
+      "eval_runtime": 130.5315,
+      "eval_samples_per_second": 15.322,
+      "eval_steps_per_second": 0.958,
+      "step": 122500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1342153846153848e-05,
+      "loss": 1.0191,
+      "step": 122600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.133446153846154e-05,
+      "loss": 0.9951,
+      "step": 122700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1326769230769233e-05,
+      "loss": 1.0653,
+      "step": 122800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1319153846153848e-05,
+      "loss": 1.01,
+      "step": 122900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.131146153846154e-05,
+      "loss": 1.0066,
+      "step": 123000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.130376923076923e-05,
+      "loss": 0.9778,
+      "step": 123100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1296076923076925e-05,
+      "loss": 1.0007,
+      "step": 123200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1288384615384616e-05,
+      "loss": 1.0282,
+      "step": 123300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.128069230769231e-05,
+      "loss": 1.0017,
+      "step": 123400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1272999999999998e-05,
+      "loss": 1.0045,
+      "step": 123500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1265307692307692e-05,
+      "loss": 1.0039,
+      "step": 123600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1257692307692307e-05,
+      "loss": 1.0088,
+      "step": 123700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.125e-05,
+      "loss": 1.0472,
+      "step": 123800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1242307692307693e-05,
+      "loss": 0.9917,
+      "step": 123900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1234615384615387e-05,
+      "loss": 1.0214,
+      "step": 124000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1226923076923075e-05,
+      "loss": 0.985,
+      "step": 124100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.121923076923077e-05,
+      "loss": 0.9957,
+      "step": 124200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1211538461538464e-05,
+      "loss": 1.0001,
+      "step": 124300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1203846153846155e-05,
+      "loss": 1.0304,
+      "step": 124400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1196153846153846e-05,
+      "loss": 1.0028,
+      "step": 124500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1188461538461537e-05,
+      "loss": 1.0464,
+      "step": 124600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.118076923076923e-05,
+      "loss": 1.0036,
+      "step": 124700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1173076923076926e-05,
+      "loss": 1.0259,
+      "step": 124800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1165384615384617e-05,
+      "loss": 1.0343,
+      "step": 124900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1157692307692308e-05,
+      "loss": 1.0326,
+      "step": 125000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 1.4721828699111938,
+      "eval_runtime": 130.4999,
+      "eval_samples_per_second": 15.326,
+      "eval_steps_per_second": 0.958,
+      "step": 125000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.115e-05,
+      "loss": 0.998,
+      "step": 125100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1142307692307694e-05,
+      "loss": 1.0264,
+      "step": 125200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1134615384615388e-05,
+      "loss": 1.023,
+      "step": 125300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1126923076923076e-05,
+      "loss": 1.0262,
+      "step": 125400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.111923076923077e-05,
+      "loss": 0.9739,
+      "step": 125500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.111153846153846e-05,
+      "loss": 1.0129,
+      "step": 125600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1103846153846156e-05,
+      "loss": 1.0116,
+      "step": 125700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1096153846153847e-05,
+      "loss": 1.0289,
+      "step": 125800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1088461538461538e-05,
+      "loss": 1.0326,
+      "step": 125900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1080769230769232e-05,
+      "loss": 1.0667,
+      "step": 126000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1073076923076923e-05,
+      "loss": 0.9951,
+      "step": 126100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1065384615384618e-05,
+      "loss": 1.0183,
+      "step": 126200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1057692307692306e-05,
+      "loss": 1.0292,
+      "step": 126300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.105e-05,
+      "loss": 1.0403,
+      "step": 126400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1042307692307695e-05,
+      "loss": 0.9678,
+      "step": 126500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1034615384615386e-05,
+      "loss": 1.0269,
+      "step": 126600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1026923076923077e-05,
+      "loss": 1.0218,
+      "step": 126700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1019230769230768e-05,
+      "loss": 1.024,
+      "step": 126800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1011538461538462e-05,
+      "loss": 1.043,
+      "step": 126900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1003846153846157e-05,
+      "loss": 1.0356,
+      "step": 127000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0996153846153844e-05,
+      "loss": 1.0121,
+      "step": 127100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.098846153846154e-05,
+      "loss": 1.0147,
+      "step": 127200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.098076923076923e-05,
+      "loss": 1.0235,
+      "step": 127300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0973076923076924e-05,
+      "loss": 1.0079,
+      "step": 127400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.096538461538462e-05,
+      "loss": 0.9801,
+      "step": 127500
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 1.4877568483352661,
+      "eval_runtime": 131.1668,
+      "eval_samples_per_second": 15.248,
+      "eval_steps_per_second": 0.953,
+      "step": 127500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0957692307692307e-05,
+      "loss": 0.9847,
+      "step": 127600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.095e-05,
+      "loss": 0.9857,
+      "step": 127700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0942307692307692e-05,
+      "loss": 1.0369,
+      "step": 127800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0934615384615387e-05,
+      "loss": 0.9894,
+      "step": 127900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0926923076923078e-05,
+      "loss": 1.0431,
+      "step": 128000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.091923076923077e-05,
+      "loss": 0.9832,
+      "step": 128100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0911538461538463e-05,
+      "loss": 1.0054,
+      "step": 128200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0903846153846154e-05,
+      "loss": 1.0595,
+      "step": 128300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.089615384615385e-05,
+      "loss": 1.0279,
+      "step": 128400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0888538461538464e-05,
+      "loss": 1.0642,
+      "step": 128500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0880846153846155e-05,
+      "loss": 1.0543,
+      "step": 128600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0873153846153846e-05,
+      "loss": 1.023,
+      "step": 128700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.086546153846154e-05,
+      "loss": 1.0353,
+      "step": 128800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.085776923076923e-05,
+      "loss": 1.0085,
+      "step": 128900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0850076923076922e-05,
+      "loss": 1.0047,
+      "step": 129000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0842384615384613e-05,
+      "loss": 0.9984,
+      "step": 129100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0834692307692308e-05,
+      "loss": 1.0108,
+      "step": 129200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0827000000000002e-05,
+      "loss": 1.0039,
+      "step": 129300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0819307692307693e-05,
+      "loss": 1.0144,
+      "step": 129400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0811615384615385e-05,
+      "loss": 0.998,
+      "step": 129500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0803923076923076e-05,
+      "loss": 1.0098,
+      "step": 129600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.079623076923077e-05,
+      "loss": 0.9844,
+      "step": 129700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0788538461538465e-05,
+      "loss": 1.0114,
+      "step": 129800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0780846153846152e-05,
+      "loss": 1.0157,
+      "step": 129900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0773153846153847e-05,
+      "loss": 1.0749,
+      "step": 130000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 1.4509004354476929,
+      "eval_runtime": 130.8161,
+      "eval_samples_per_second": 15.289,
+      "eval_steps_per_second": 0.956,
+      "step": 130000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0765461538461538e-05,
+      "loss": 1.0207,
+      "step": 130100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0757769230769232e-05,
+      "loss": 1.0396,
+      "step": 130200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0750076923076923e-05,
+      "loss": 1.0425,
+      "step": 130300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0742384615384614e-05,
+      "loss": 1.0311,
+      "step": 130400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.073469230769231e-05,
+      "loss": 0.9855,
+      "step": 130500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0727e-05,
+      "loss": 1.0301,
+      "step": 130600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0719307692307694e-05,
+      "loss": 1.0148,
+      "step": 130700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0711615384615385e-05,
+      "loss": 1.0143,
+      "step": 130800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0703923076923077e-05,
+      "loss": 1.0186,
+      "step": 130900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069630769230769e-05,
+      "loss": 1.0274,
+      "step": 131000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0688615384615386e-05,
+      "loss": 1.0439,
+      "step": 131100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0681e-05,
+      "loss": 1.0244,
+      "step": 131200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0673307692307692e-05,
+      "loss": 1.0216,
+      "step": 131300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0665615384615386e-05,
+      "loss": 1.0411,
+      "step": 131400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0657923076923077e-05,
+      "loss": 0.9774,
+      "step": 131500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.065023076923077e-05,
+      "loss": 1.0346,
+      "step": 131600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0642538461538463e-05,
+      "loss": 1.0283,
+      "step": 131700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0634846153846154e-05,
+      "loss": 1.0329,
+      "step": 131800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0627153846153845e-05,
+      "loss": 1.0054,
+      "step": 131900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.061946153846154e-05,
+      "loss": 1.0139,
+      "step": 132000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.061176923076923e-05,
+      "loss": 1.0035,
+      "step": 132100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0604076923076925e-05,
+      "loss": 1.0331,
+      "step": 132200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0596384615384616e-05,
+      "loss": 0.9911,
+      "step": 132300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0588692307692307e-05,
+      "loss": 1.0,
+      "step": 132400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0581e-05,
+      "loss": 1.0365,
+      "step": 132500
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 1.4605334997177124,
+      "eval_runtime": 131.3026,
+      "eval_samples_per_second": 15.232,
+      "eval_steps_per_second": 0.952,
+      "step": 132500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0573307692307693e-05,
+      "loss": 1.0007,
+      "step": 132600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0565615384615387e-05,
+      "loss": 1.0113,
+      "step": 132700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0557923076923075e-05,
+      "loss": 0.9897,
+      "step": 132800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.055023076923077e-05,
+      "loss": 0.9773,
+      "step": 132900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0542538461538464e-05,
+      "loss": 0.9882,
+      "step": 133000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0534846153846155e-05,
+      "loss": 0.9862,
+      "step": 133100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.052715384615385e-05,
+      "loss": 1.0293,
+      "step": 133200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0519461538461537e-05,
+      "loss": 1.0282,
+      "step": 133300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.051176923076923e-05,
+      "loss": 1.0075,
+      "step": 133400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0504076923076923e-05,
+      "loss": 1.0775,
+      "step": 133500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0496384615384617e-05,
+      "loss": 0.9816,
+      "step": 133600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0488692307692308e-05,
+      "loss": 1.0413,
+      "step": 133700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0481e-05,
+      "loss": 1.0046,
+      "step": 133800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0473307692307694e-05,
+      "loss": 1.0362,
+      "step": 133900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0465615384615385e-05,
+      "loss": 1.055,
+      "step": 134000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0457923076923076e-05,
+      "loss": 1.0554,
+      "step": 134100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.045023076923077e-05,
+      "loss": 1.0076,
+      "step": 134200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.044253846153846e-05,
+      "loss": 0.9869,
+      "step": 134300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0434846153846156e-05,
+      "loss": 1.0075,
+      "step": 134400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0427153846153847e-05,
+      "loss": 1.0065,
+      "step": 134500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0419461538461538e-05,
+      "loss": 1.0058,
+      "step": 134600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0411769230769232e-05,
+      "loss": 0.9671,
+      "step": 134700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0404076923076924e-05,
+      "loss": 1.0053,
+      "step": 134800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0396384615384618e-05,
+      "loss": 1.0237,
+      "step": 134900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0388769230769233e-05,
+      "loss": 0.9749,
+      "step": 135000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 1.4492900371551514,
+      "eval_runtime": 133.2679,
+      "eval_samples_per_second": 15.007,
+      "eval_steps_per_second": 0.938,
+      "step": 135000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0381076923076924e-05,
+      "loss": 0.9948,
+      "step": 135100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0373384615384615e-05,
+      "loss": 1.0331,
+      "step": 135200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.036569230769231e-05,
+      "loss": 1.0025,
+      "step": 135300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0358e-05,
+      "loss": 0.9685,
+      "step": 135400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0350307692307695e-05,
+      "loss": 1.0279,
+      "step": 135500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0342615384615383e-05,
+      "loss": 0.9705,
+      "step": 135600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0334923076923077e-05,
+      "loss": 1.0227,
+      "step": 135700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.032723076923077e-05,
+      "loss": 1.003,
+      "step": 135800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0319538461538463e-05,
+      "loss": 1.0223,
+      "step": 135900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0311846153846154e-05,
+      "loss": 1.0172,
+      "step": 136000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0304153846153845e-05,
+      "loss": 0.9724,
+      "step": 136100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.029646153846154e-05,
+      "loss": 1.0163,
+      "step": 136200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.028876923076923e-05,
+      "loss": 0.9664,
+      "step": 136300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0281076923076925e-05,
+      "loss": 1.0317,
+      "step": 136400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0273384615384616e-05,
+      "loss": 0.9646,
+      "step": 136500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0265692307692307e-05,
+      "loss": 0.992,
+      "step": 136600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0258e-05,
+      "loss": 0.9725,
+      "step": 136700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0250307692307693e-05,
+      "loss": 1.0418,
+      "step": 136800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0242615384615384e-05,
+      "loss": 1.0399,
+      "step": 136900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0234923076923078e-05,
+      "loss": 0.9571,
+      "step": 137000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.022723076923077e-05,
+      "loss": 1.0018,
+      "step": 137100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0219538461538464e-05,
+      "loss": 1.0318,
+      "step": 137200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.021184615384615e-05,
+      "loss": 0.9832,
+      "step": 137300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0204153846153846e-05,
+      "loss": 1.0088,
+      "step": 137400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.019646153846154e-05,
+      "loss": 0.9713,
+      "step": 137500
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.4767963886260986,
+      "eval_runtime": 131.6777,
+      "eval_samples_per_second": 15.189,
+      "eval_steps_per_second": 0.949,
+      "step": 137500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.018876923076923e-05,
+      "loss": 1.0253,
+      "step": 137600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0181076923076926e-05,
+      "loss": 0.9831,
+      "step": 137700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0173384615384614e-05,
+      "loss": 1.0635,
+      "step": 137800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0165692307692308e-05,
+      "loss": 0.9882,
+      "step": 137900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0158000000000002e-05,
+      "loss": 1.0407,
+      "step": 138000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0150307692307694e-05,
+      "loss": 1.0194,
+      "step": 138100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0142615384615385e-05,
+      "loss": 0.9932,
+      "step": 138200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0134923076923076e-05,
+      "loss": 1.0449,
+      "step": 138300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.012723076923077e-05,
+      "loss": 1.0167,
+      "step": 138400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0119538461538465e-05,
+      "loss": 1.0063,
+      "step": 138500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0111846153846152e-05,
+      "loss": 1.0573,
+      "step": 138600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0104153846153847e-05,
+      "loss": 1.0055,
+      "step": 138700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0096461538461538e-05,
+      "loss": 1.0177,
+      "step": 138800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0088769230769232e-05,
+      "loss": 0.9851,
+      "step": 138900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0081076923076923e-05,
+      "loss": 1.0725,
+      "step": 139000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0073384615384615e-05,
+      "loss": 0.9983,
+      "step": 139100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.006569230769231e-05,
+      "loss": 1.0367,
+      "step": 139200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0058076923076924e-05,
+      "loss": 0.9765,
+      "step": 139300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0050384615384615e-05,
+      "loss": 0.9908,
+      "step": 139400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.004269230769231e-05,
+      "loss": 1.0029,
+      "step": 139500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0035e-05,
+      "loss": 0.9956,
+      "step": 139600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.002730769230769e-05,
+      "loss": 0.9935,
+      "step": 139700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0019615384615386e-05,
+      "loss": 0.9898,
+      "step": 139800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0011923076923077e-05,
+      "loss": 0.9609,
+      "step": 139900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.000423076923077e-05,
+      "loss": 1.0125,
+      "step": 140000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.4210669994354248,
+      "eval_runtime": 134.4062,
+      "eval_samples_per_second": 14.88,
+      "eval_steps_per_second": 0.93,
+      "step": 140000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999653846153846e-05,
+      "loss": 0.9824,
+      "step": 140100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9988846153846154e-05,
+      "loss": 1.0082,
+      "step": 140200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9981153846153848e-05,
+      "loss": 0.9837,
+      "step": 140300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.997346153846154e-05,
+      "loss": 0.9824,
+      "step": 140400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996576923076923e-05,
+      "loss": 1.0082,
+      "step": 140500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995807692307692e-05,
+      "loss": 0.9946,
+      "step": 140600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9950384615384616e-05,
+      "loss": 1.0133,
+      "step": 140700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.994269230769231e-05,
+      "loss": 0.9901,
+      "step": 140800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9935e-05,
+      "loss": 0.9914,
+      "step": 140900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9927307692307692e-05,
+      "loss": 0.975,
+      "step": 141000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9919615384615384e-05,
+      "loss": 0.9974,
+      "step": 141100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9911923076923078e-05,
+      "loss": 0.9901,
+      "step": 141200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9904230769230773e-05,
+      "loss": 1.0295,
+      "step": 141300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.989653846153846e-05,
+      "loss": 0.9812,
+      "step": 141400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9888846153846155e-05,
+      "loss": 1.0311,
+      "step": 141500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9881153846153846e-05,
+      "loss": 1.0137,
+      "step": 141600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.987346153846154e-05,
+      "loss": 0.9755,
+      "step": 141700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9865846153846155e-05,
+      "loss": 0.9764,
+      "step": 141800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.985815384615385e-05,
+      "loss": 0.9911,
+      "step": 141900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9850461538461537e-05,
+      "loss": 0.9971,
+      "step": 142000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.984276923076923e-05,
+      "loss": 0.9727,
+      "step": 142100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9835076923076923e-05,
+      "loss": 0.9903,
+      "step": 142200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9827384615384617e-05,
+      "loss": 1.0374,
+      "step": 142300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9819692307692308e-05,
+      "loss": 1.0081,
+      "step": 142400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9812e-05,
+      "loss": 0.9891,
+      "step": 142500
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 1.4216349124908447,
+      "eval_runtime": 131.0588,
+      "eval_samples_per_second": 15.26,
+      "eval_steps_per_second": 0.954,
+      "step": 142500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9804307692307694e-05,
+      "loss": 0.9951,
+      "step": 142600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9796615384615385e-05,
+      "loss": 0.934,
+      "step": 142700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978892307692308e-05,
+      "loss": 1.0546,
+      "step": 142800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9781230769230767e-05,
+      "loss": 0.9473,
+      "step": 142900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.977353846153846e-05,
+      "loss": 1.0,
+      "step": 143000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9765846153846156e-05,
+      "loss": 1.0006,
+      "step": 143100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9758153846153847e-05,
+      "loss": 0.9861,
+      "step": 143200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9750461538461538e-05,
+      "loss": 0.9727,
+      "step": 143300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.974276923076923e-05,
+      "loss": 0.9964,
+      "step": 143400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9735076923076924e-05,
+      "loss": 1.0153,
+      "step": 143500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9727384615384618e-05,
+      "loss": 1.049,
+      "step": 143600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9719692307692306e-05,
+      "loss": 1.0257,
+      "step": 143700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9712e-05,
+      "loss": 1.0012,
+      "step": 143800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.970430769230769e-05,
+      "loss": 0.988,
+      "step": 143900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9696615384615386e-05,
+      "loss": 0.9966,
+      "step": 144000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.968892307692308e-05,
+      "loss": 1.026,
+      "step": 144100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9681230769230768e-05,
+      "loss": 0.9818,
+      "step": 144200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9673538461538462e-05,
+      "loss": 0.9691,
+      "step": 144300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9665846153846154e-05,
+      "loss": 1.0213,
+      "step": 144400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9658153846153848e-05,
+      "loss": 1.0186,
+      "step": 144500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.965046153846154e-05,
+      "loss": 1.0222,
+      "step": 144600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.964276923076923e-05,
+      "loss": 0.9992,
+      "step": 144700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9635076923076925e-05,
+      "loss": 1.0155,
+      "step": 144800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9627384615384616e-05,
+      "loss": 1.0228,
+      "step": 144900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9619692307692307e-05,
+      "loss": 0.989,
+      "step": 145000
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.4293444156646729,
+      "eval_runtime": 130.8969,
+      "eval_samples_per_second": 15.279,
+      "eval_steps_per_second": 0.955,
+      "step": 145000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9612e-05,
+      "loss": 1.0208,
+      "step": 145100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9604307692307692e-05,
+      "loss": 0.9946,
+      "step": 145200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9596615384615387e-05,
+      "loss": 1.0486,
+      "step": 145300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9588923076923078e-05,
+      "loss": 1.0064,
+      "step": 145400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.958123076923077e-05,
+      "loss": 1.0004,
+      "step": 145500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.957353846153846e-05,
+      "loss": 0.9822,
+      "step": 145600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9565846153846155e-05,
+      "loss": 0.9481,
+      "step": 145700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.955815384615385e-05,
+      "loss": 1.0259,
+      "step": 145800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9550538461538464e-05,
+      "loss": 0.9862,
+      "step": 145900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9542846153846155e-05,
+      "loss": 1.0273,
+      "step": 146000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9535153846153846e-05,
+      "loss": 0.9816,
+      "step": 146100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9527461538461537e-05,
+      "loss": 0.9852,
+      "step": 146200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9519846153846155e-05,
+      "loss": 1.0016,
+      "step": 146300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9512153846153846e-05,
+      "loss": 0.9901,
+      "step": 146400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.950446153846154e-05,
+      "loss": 1.0171,
+      "step": 146500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.949676923076923e-05,
+      "loss": 1.0302,
+      "step": 146600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9489076923076923e-05,
+      "loss": 1.0019,
+      "step": 146700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9481384615384614e-05,
+      "loss": 0.9764,
+      "step": 146800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.947369230769231e-05,
+      "loss": 1.0151,
+      "step": 146900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466000000000003e-05,
+      "loss": 1.0138,
+      "step": 147000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945830769230769e-05,
+      "loss": 1.0381,
+      "step": 147100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450615384615385e-05,
+      "loss": 1.0505,
+      "step": 147200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9442923076923076e-05,
+      "loss": 0.972,
+      "step": 147300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.943523076923077e-05,
+      "loss": 1.0091,
+      "step": 147400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9427538461538462e-05,
+      "loss": 0.9877,
+      "step": 147500
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 1.4279431104660034,
+      "eval_runtime": 131.0153,
+      "eval_samples_per_second": 15.265,
+      "eval_steps_per_second": 0.954,
+      "step": 147500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9419846153846153e-05,
+      "loss": 0.9672,
+      "step": 147600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9412153846153847e-05,
+      "loss": 0.9884,
+      "step": 147700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.940446153846154e-05,
+      "loss": 0.9805,
+      "step": 147800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9396769230769233e-05,
+      "loss": 1.0176,
+      "step": 147900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9389076923076924e-05,
+      "loss": 0.9794,
+      "step": 148000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9381384615384615e-05,
+      "loss": 0.9927,
+      "step": 148100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.937369230769231e-05,
+      "loss": 1.0027,
+      "step": 148200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9366e-05,
+      "loss": 0.9799,
+      "step": 148300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.935830769230769e-05,
+      "loss": 1.0118,
+      "step": 148400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9350615384615386e-05,
+      "loss": 0.9646,
+      "step": 148500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9342923076923077e-05,
+      "loss": 0.9482,
+      "step": 148600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.933523076923077e-05,
+      "loss": 1.0143,
+      "step": 148700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.932753846153846e-05,
+      "loss": 0.9995,
+      "step": 148800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9319846153846154e-05,
+      "loss": 0.9745,
+      "step": 148900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9312153846153848e-05,
+      "loss": 0.9696,
+      "step": 149000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.930446153846154e-05,
+      "loss": 1.0219,
+      "step": 149100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9296769230769234e-05,
+      "loss": 0.9842,
+      "step": 149200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.928907692307692e-05,
+      "loss": 0.9924,
+      "step": 149300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9281384615384616e-05,
+      "loss": 0.989,
+      "step": 149400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9273692307692307e-05,
+      "loss": 1.001,
+      "step": 149500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9266e-05,
+      "loss": 0.9924,
+      "step": 149600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9258307692307693e-05,
+      "loss": 0.9899,
+      "step": 149700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9250615384615384e-05,
+      "loss": 0.9713,
+      "step": 149800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9242923076923078e-05,
+      "loss": 0.9859,
+      "step": 149900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.923523076923077e-05,
+      "loss": 0.9931,
+      "step": 150000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.4378827810287476,
+      "eval_runtime": 133.3671,
+      "eval_samples_per_second": 14.996,
+      "eval_steps_per_second": 0.937,
+      "step": 150000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.922753846153846e-05,
+      "loss": 0.9621,
+      "step": 150100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9219846153846155e-05,
+      "loss": 0.9647,
+      "step": 150200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9212153846153846e-05,
+      "loss": 0.9931,
+      "step": 150300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.920446153846154e-05,
+      "loss": 0.9697,
+      "step": 150400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9196846153846155e-05,
+      "loss": 1.0164,
+      "step": 150500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9189153846153846e-05,
+      "loss": 1.0045,
+      "step": 150600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9181461538461537e-05,
+      "loss": 0.9456,
+      "step": 150700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9173769230769232e-05,
+      "loss": 0.9536,
+      "step": 150800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9166076923076923e-05,
+      "loss": 0.9984,
+      "step": 150900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9158384615384617e-05,
+      "loss": 1.0069,
+      "step": 151000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.915069230769231e-05,
+      "loss": 1.0199,
+      "step": 151100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9143e-05,
+      "loss": 1.0159,
+      "step": 151200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9135307692307694e-05,
+      "loss": 0.9888,
+      "step": 151300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9127615384615385e-05,
+      "loss": 0.9908,
+      "step": 151400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.911992307692308e-05,
+      "loss": 0.955,
+      "step": 151500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9112230769230767e-05,
+      "loss": 0.9853,
+      "step": 151600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.910453846153846e-05,
+      "loss": 1.0065,
+      "step": 151700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9096846153846156e-05,
+      "loss": 1.0325,
+      "step": 151800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9089153846153847e-05,
+      "loss": 1.0066,
+      "step": 151900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9081461538461538e-05,
+      "loss": 0.9908,
+      "step": 152000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.907376923076923e-05,
+      "loss": 0.965,
+      "step": 152100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9066076923076924e-05,
+      "loss": 0.9591,
+      "step": 152200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.905838461538462e-05,
+      "loss": 0.9885,
+      "step": 152300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.905069230769231e-05,
+      "loss": 1.0307,
+      "step": 152400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9043076923076924e-05,
+      "loss": 0.9696,
+      "step": 152500
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 1.4165639877319336,
+      "eval_runtime": 134.2423,
+      "eval_samples_per_second": 14.898,
+      "eval_steps_per_second": 0.931,
+      "step": 152500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9035384615384615e-05,
+      "loss": 1.0054,
+      "step": 152600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9027692307692306e-05,
+      "loss": 1.0055,
+      "step": 152700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.902e-05,
+      "loss": 0.9974,
+      "step": 152800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9012307692307695e-05,
+      "loss": 0.9786,
+      "step": 152900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9004615384615386e-05,
+      "loss": 1.0015,
+      "step": 153000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8996923076923077e-05,
+      "loss": 1.0,
+      "step": 153100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.898923076923077e-05,
+      "loss": 1.0208,
+      "step": 153200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8981538461538463e-05,
+      "loss": 1.0258,
+      "step": 153300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8973846153846154e-05,
+      "loss": 1.0148,
+      "step": 153400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8966153846153845e-05,
+      "loss": 0.9728,
+      "step": 153500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.895846153846154e-05,
+      "loss": 0.9891,
+      "step": 153600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.895076923076923e-05,
+      "loss": 1.0143,
+      "step": 153700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8943076923076925e-05,
+      "loss": 1.0121,
+      "step": 153800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8935384615384613e-05,
+      "loss": 1.0004,
+      "step": 153900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8927692307692307e-05,
+      "loss": 0.9878,
+      "step": 154000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8920000000000002e-05,
+      "loss": 0.9582,
+      "step": 154100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8912384615384617e-05,
+      "loss": 0.9905,
+      "step": 154200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8904692307692308e-05,
+      "loss": 0.9727,
+      "step": 154300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8897000000000002e-05,
+      "loss": 0.9799,
+      "step": 154400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.888930769230769e-05,
+      "loss": 1.0302,
+      "step": 154500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8881615384615384e-05,
+      "loss": 1.0026,
+      "step": 154600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.887392307692308e-05,
+      "loss": 0.9551,
+      "step": 154700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.886623076923077e-05,
+      "loss": 0.9783,
+      "step": 154800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.885853846153846e-05,
+      "loss": 1.0122,
+      "step": 154900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8850846153846152e-05,
+      "loss": 0.9756,
+      "step": 155000
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.4003106355667114,
+      "eval_runtime": 133.1059,
+      "eval_samples_per_second": 15.026,
+      "eval_steps_per_second": 0.939,
+      "step": 155000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8843230769230767e-05,
+      "loss": 0.9939,
+      "step": 155100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.883553846153846e-05,
+      "loss": 1.0398,
+      "step": 155200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8827846153846156e-05,
+      "loss": 0.9771,
+      "step": 155300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8820153846153847e-05,
+      "loss": 1.0001,
+      "step": 155400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8812538461538462e-05,
+      "loss": 1.0112,
+      "step": 155500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8804846153846156e-05,
+      "loss": 1.0212,
+      "step": 155600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8797153846153844e-05,
+      "loss": 0.9463,
+      "step": 155700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.878946153846154e-05,
+      "loss": 0.9984,
+      "step": 155800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8781769230769233e-05,
+      "loss": 1.0031,
+      "step": 155900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8774076923076924e-05,
+      "loss": 0.9997,
+      "step": 156000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8766384615384615e-05,
+      "loss": 0.9731,
+      "step": 156100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8758692307692306e-05,
+      "loss": 1.0562,
+      "step": 156200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8751e-05,
+      "loss": 1.0001,
+      "step": 156300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8743307692307695e-05,
+      "loss": 0.9474,
+      "step": 156400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8735615384615383e-05,
+      "loss": 0.9897,
+      "step": 156500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8727923076923077e-05,
+      "loss": 0.9608,
+      "step": 156600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8720230769230768e-05,
+      "loss": 0.9569,
+      "step": 156700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8712538461538463e-05,
+      "loss": 0.9768,
+      "step": 156800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8704846153846157e-05,
+      "loss": 0.9703,
+      "step": 156900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8697153846153845e-05,
+      "loss": 1.0176,
+      "step": 157000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868946153846154e-05,
+      "loss": 1.0043,
+      "step": 157100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868176923076923e-05,
+      "loss": 1.0112,
+      "step": 157200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8674076923076925e-05,
+      "loss": 0.9674,
+      "step": 157300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8666384615384616e-05,
+      "loss": 0.9741,
+      "step": 157400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8658692307692307e-05,
+      "loss": 0.9697,
+      "step": 157500
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.4179853200912476,
+      "eval_runtime": 135.6443,
+      "eval_samples_per_second": 14.744,
+      "eval_steps_per_second": 0.922,
+      "step": 157500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8651e-05,
+      "loss": 0.9744,
+      "step": 157600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8643307692307693e-05,
+      "loss": 0.9926,
+      "step": 157700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8635615384615387e-05,
+      "loss": 0.9807,
+      "step": 157800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8627923076923078e-05,
+      "loss": 0.9859,
+      "step": 157900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.862023076923077e-05,
+      "loss": 0.9707,
+      "step": 158000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8612538461538464e-05,
+      "loss": 0.9887,
+      "step": 158100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8604846153846155e-05,
+      "loss": 1.0477,
+      "step": 158200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8597153846153846e-05,
+      "loss": 0.974,
+      "step": 158300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8589461538461537e-05,
+      "loss": 0.9955,
+      "step": 158400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8581846153846155e-05,
+      "loss": 0.971,
+      "step": 158500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8574153846153846e-05,
+      "loss": 1.0255,
+      "step": 158600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.856646153846154e-05,
+      "loss": 0.9527,
+      "step": 158700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8558769230769232e-05,
+      "loss": 0.999,
+      "step": 158800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8551076923076923e-05,
+      "loss": 0.9982,
+      "step": 158900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8543384615384614e-05,
+      "loss": 0.9915,
+      "step": 159000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.853569230769231e-05,
+      "loss": 0.9559,
+      "step": 159100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8528000000000003e-05,
+      "loss": 1.0121,
+      "step": 159200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.852030769230769e-05,
+      "loss": 0.9682,
+      "step": 159300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8512615384615385e-05,
+      "loss": 0.9854,
+      "step": 159400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8504923076923076e-05,
+      "loss": 0.9941,
+      "step": 159500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.849723076923077e-05,
+      "loss": 0.9676,
+      "step": 159600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.848953846153846e-05,
+      "loss": 0.9719,
+      "step": 159700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8481846153846153e-05,
+      "loss": 1.0092,
+      "step": 159800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8474153846153847e-05,
+      "loss": 1.0141,
+      "step": 159900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8466461538461538e-05,
+      "loss": 0.9869,
+      "step": 160000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.408742070198059,
+      "eval_runtime": 132.0269,
+      "eval_samples_per_second": 15.148,
+      "eval_steps_per_second": 0.947,
+      "step": 160000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8458769230769233e-05,
+      "loss": 0.9774,
+      "step": 160100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8451076923076924e-05,
+      "loss": 0.9549,
+      "step": 160200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8443384615384615e-05,
+      "loss": 0.9527,
+      "step": 160300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.843569230769231e-05,
+      "loss": 0.9913,
+      "step": 160400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8428e-05,
+      "loss": 1.0313,
+      "step": 160500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.842030769230769e-05,
+      "loss": 0.9904,
+      "step": 160600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8412615384615386e-05,
+      "loss": 0.9734,
+      "step": 160700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8404923076923077e-05,
+      "loss": 1.0205,
+      "step": 160800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.839723076923077e-05,
+      "loss": 0.9937,
+      "step": 160900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8389538461538463e-05,
+      "loss": 0.9561,
+      "step": 161000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8381846153846154e-05,
+      "loss": 0.9968,
+      "step": 161100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8374153846153845e-05,
+      "loss": 0.9797,
+      "step": 161200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.836646153846154e-05,
+      "loss": 0.9887,
+      "step": 161300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8358769230769234e-05,
+      "loss": 1.0019,
+      "step": 161400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.835107692307692e-05,
+      "loss": 0.9812,
+      "step": 161500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8343384615384616e-05,
+      "loss": 0.9695,
+      "step": 161600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8335692307692307e-05,
+      "loss": 0.9817,
+      "step": 161700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8328e-05,
+      "loss": 0.9729,
+      "step": 161800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8320307692307692e-05,
+      "loss": 0.9775,
+      "step": 161900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8312615384615384e-05,
+      "loss": 1.0108,
+      "step": 162000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8304923076923078e-05,
+      "loss": 0.9748,
+      "step": 162100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829723076923077e-05,
+      "loss": 0.9944,
+      "step": 162200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8289538461538464e-05,
+      "loss": 1.0178,
+      "step": 162300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8281846153846155e-05,
+      "loss": 1.0137,
+      "step": 162400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8274153846153846e-05,
+      "loss": 0.9665,
+      "step": 162500
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 1.41000235080719,
+      "eval_runtime": 131.9671,
+      "eval_samples_per_second": 15.155,
+      "eval_steps_per_second": 0.947,
+      "step": 162500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.826646153846154e-05,
+      "loss": 1.0077,
+      "step": 162600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.825876923076923e-05,
+      "loss": 0.9965,
+      "step": 162700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8251076923076922e-05,
+      "loss": 0.9813,
+      "step": 162800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8243384615384617e-05,
+      "loss": 0.9987,
+      "step": 162900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8235692307692308e-05,
+      "loss": 0.9704,
+      "step": 163000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8228000000000002e-05,
+      "loss": 0.9858,
+      "step": 163100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.822030769230769e-05,
+      "loss": 0.9481,
+      "step": 163200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8212615384615384e-05,
+      "loss": 0.9844,
+      "step": 163300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820492307692308e-05,
+      "loss": 0.9675,
+      "step": 163400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.819723076923077e-05,
+      "loss": 1.0017,
+      "step": 163500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8189538461538465e-05,
+      "loss": 0.975,
+      "step": 163600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8181846153846152e-05,
+      "loss": 0.991,
+      "step": 163700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8174153846153847e-05,
+      "loss": 0.9776,
+      "step": 163800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8166461538461538e-05,
+      "loss": 0.9855,
+      "step": 163900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8158769230769232e-05,
+      "loss": 0.9699,
+      "step": 164000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8151076923076923e-05,
+      "loss": 1.0117,
+      "step": 164100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8143384615384614e-05,
+      "loss": 0.9284,
+      "step": 164200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.813576923076923e-05,
+      "loss": 0.9329,
+      "step": 164300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8128076923076924e-05,
+      "loss": 0.9726,
+      "step": 164400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8120384615384615e-05,
+      "loss": 0.9773,
+      "step": 164500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.811269230769231e-05,
+      "loss": 0.9853,
+      "step": 164600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8105e-05,
+      "loss": 1.0239,
+      "step": 164700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.809730769230769e-05,
+      "loss": 0.9703,
+      "step": 164800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8089615384615386e-05,
+      "loss": 0.9713,
+      "step": 164900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8081923076923077e-05,
+      "loss": 0.9922,
+      "step": 165000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 1.412420392036438,
+      "eval_runtime": 131.103,
+      "eval_samples_per_second": 15.255,
+      "eval_steps_per_second": 0.953,
+      "step": 165000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8074230769230768e-05,
+      "loss": 0.9724,
+      "step": 165100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8066538461538462e-05,
+      "loss": 0.9886,
+      "step": 165200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8058846153846154e-05,
+      "loss": 0.9581,
+      "step": 165300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8051153846153848e-05,
+      "loss": 0.9659,
+      "step": 165400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.804346153846154e-05,
+      "loss": 0.9828,
+      "step": 165500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.803576923076923e-05,
+      "loss": 1.0208,
+      "step": 165600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8028076923076925e-05,
+      "loss": 0.9672,
+      "step": 165700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8020384615384616e-05,
+      "loss": 1.0415,
+      "step": 165800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.801269230769231e-05,
+      "loss": 0.9638,
+      "step": 165900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8004999999999998e-05,
+      "loss": 0.9718,
+      "step": 166000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7997307692307692e-05,
+      "loss": 0.9438,
+      "step": 166100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7989615384615387e-05,
+      "loss": 0.9384,
+      "step": 166200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7981923076923078e-05,
+      "loss": 0.9839,
+      "step": 166300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.797423076923077e-05,
+      "loss": 0.9957,
+      "step": 166400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.796653846153846e-05,
+      "loss": 1.0052,
+      "step": 166500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7958846153846155e-05,
+      "loss": 0.9588,
+      "step": 166600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.795115384615385e-05,
+      "loss": 1.0039,
+      "step": 166700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.794346153846154e-05,
+      "loss": 0.9388,
+      "step": 166800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.793576923076923e-05,
+      "loss": 0.9337,
+      "step": 166900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7928076923076922e-05,
+      "loss": 1.0016,
+      "step": 167000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7920384615384617e-05,
+      "loss": 0.9276,
+      "step": 167100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7912692307692308e-05,
+      "loss": 1.0339,
+      "step": 167200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7905e-05,
+      "loss": 0.9877,
+      "step": 167300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7897307692307693e-05,
+      "loss": 0.9601,
+      "step": 167400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7889615384615384e-05,
+      "loss": 0.9449,
+      "step": 167500
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.391361117362976,
+      "eval_runtime": 131.4095,
+      "eval_samples_per_second": 15.22,
+      "eval_steps_per_second": 0.951,
+      "step": 167500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.788192307692308e-05,
+      "loss": 1.003,
+      "step": 167600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.787423076923077e-05,
+      "loss": 0.9832,
+      "step": 167700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.786653846153846e-05,
+      "loss": 0.9668,
+      "step": 167800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7858846153846155e-05,
+      "loss": 0.9504,
+      "step": 167900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7851153846153847e-05,
+      "loss": 0.9868,
+      "step": 168000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.784346153846154e-05,
+      "loss": 0.9915,
+      "step": 168100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.783576923076923e-05,
+      "loss": 0.9757,
+      "step": 168200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7828076923076923e-05,
+      "loss": 0.9891,
+      "step": 168300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7820384615384618e-05,
+      "loss": 1.0094,
+      "step": 168400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7812769230769232e-05,
+      "loss": 0.9738,
+      "step": 168500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7805076923076924e-05,
+      "loss": 0.9726,
+      "step": 168600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7797384615384618e-05,
+      "loss": 1.0166,
+      "step": 168700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7789692307692306e-05,
+      "loss": 0.9606,
+      "step": 168800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7782e-05,
+      "loss": 0.9522,
+      "step": 168900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7774307692307695e-05,
+      "loss": 0.9933,
+      "step": 169000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7766615384615386e-05,
+      "loss": 0.9875,
+      "step": 169100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7758923076923077e-05,
+      "loss": 0.9537,
+      "step": 169200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7751230769230768e-05,
+      "loss": 0.9658,
+      "step": 169300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7743538461538462e-05,
+      "loss": 0.9767,
+      "step": 169400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7735846153846157e-05,
+      "loss": 0.987,
+      "step": 169500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7728153846153844e-05,
+      "loss": 0.9656,
+      "step": 169600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.772046153846154e-05,
+      "loss": 0.996,
+      "step": 169700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.771276923076923e-05,
+      "loss": 0.9771,
+      "step": 169800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7705076923076925e-05,
+      "loss": 0.9635,
+      "step": 169900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.769738461538462e-05,
+      "loss": 0.944,
+      "step": 170000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 1.3790719509124756,
+      "eval_runtime": 130.9357,
+      "eval_samples_per_second": 15.275,
+      "eval_steps_per_second": 0.955,
+      "step": 170000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7689692307692307e-05,
+      "loss": 0.9885,
+      "step": 170100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7682e-05,
+      "loss": 0.9832,
+      "step": 170200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7674307692307692e-05,
+      "loss": 0.9918,
+      "step": 170300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7666615384615387e-05,
+      "loss": 0.9674,
+      "step": 170400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7658923076923074e-05,
+      "loss": 0.9649,
+      "step": 170500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.765123076923077e-05,
+      "loss": 0.9999,
+      "step": 170600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.764369230769231e-05,
+      "loss": 0.9779,
+      "step": 170700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7636e-05,
+      "loss": 0.967,
+      "step": 170800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7628307692307693e-05,
+      "loss": 1.0084,
+      "step": 170900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7620615384615384e-05,
+      "loss": 0.9926,
+      "step": 171000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.761292307692308e-05,
+      "loss": 0.9871,
+      "step": 171100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.760523076923077e-05,
+      "loss": 1.0015,
+      "step": 171200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.759753846153846e-05,
+      "loss": 0.9334,
+      "step": 171300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7589846153846155e-05,
+      "loss": 1.0183,
+      "step": 171400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7582153846153846e-05,
+      "loss": 0.9815,
+      "step": 171500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.757446153846154e-05,
+      "loss": 0.9603,
+      "step": 171600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.756676923076923e-05,
+      "loss": 0.9867,
+      "step": 171700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7559076923076923e-05,
+      "loss": 0.9657,
+      "step": 171800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7551384615384617e-05,
+      "loss": 0.9347,
+      "step": 171900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.754369230769231e-05,
+      "loss": 0.9888,
+      "step": 172000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7536e-05,
+      "loss": 0.9489,
+      "step": 172100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.752830769230769e-05,
+      "loss": 0.9625,
+      "step": 172200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7520615384615385e-05,
+      "loss": 0.9573,
+      "step": 172300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.751292307692308e-05,
+      "loss": 0.9624,
+      "step": 172400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.750523076923077e-05,
+      "loss": 0.9443,
+      "step": 172500
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 1.4071744680404663,
+      "eval_runtime": 133.195,
+      "eval_samples_per_second": 15.016,
+      "eval_steps_per_second": 0.938,
+      "step": 172500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.749753846153846e-05,
+      "loss": 0.989,
+      "step": 172600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7489846153846153e-05,
+      "loss": 0.9775,
+      "step": 172700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7482153846153847e-05,
+      "loss": 1.0061,
+      "step": 172800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.747446153846154e-05,
+      "loss": 0.9709,
+      "step": 172900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.746676923076923e-05,
+      "loss": 0.9512,
+      "step": 173000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7459076923076924e-05,
+      "loss": 0.9243,
+      "step": 173100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7451384615384615e-05,
+      "loss": 0.9923,
+      "step": 173200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.744369230769231e-05,
+      "loss": 0.9402,
+      "step": 173300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7436e-05,
+      "loss": 0.9499,
+      "step": 173400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.742830769230769e-05,
+      "loss": 1.0136,
+      "step": 173500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7420615384615386e-05,
+      "loss": 0.9722,
+      "step": 173600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7412923076923077e-05,
+      "loss": 0.9667,
+      "step": 173700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.740523076923077e-05,
+      "loss": 0.9718,
+      "step": 173800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7397538461538463e-05,
+      "loss": 0.9728,
+      "step": 173900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7389846153846154e-05,
+      "loss": 1.0082,
+      "step": 174000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382153846153848e-05,
+      "loss": 0.9944,
+      "step": 174100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737446153846154e-05,
+      "loss": 0.9627,
+      "step": 174200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736676923076923e-05,
+      "loss": 0.9718,
+      "step": 174300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735907692307692e-05,
+      "loss": 0.9638,
+      "step": 174400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351384615384616e-05,
+      "loss": 0.9309,
+      "step": 174500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734369230769231e-05,
+      "loss": 0.9436,
+      "step": 174600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336076923076925e-05,
+      "loss": 1.0075,
+      "step": 174700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328384615384616e-05,
+      "loss": 0.9734,
+      "step": 174800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320692307692307e-05,
+      "loss": 0.9683,
+      "step": 174900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7313e-05,
+      "loss": 0.9669,
+      "step": 175000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.3755971193313599,
+      "eval_runtime": 135.8413,
+      "eval_samples_per_second": 14.723,
+      "eval_steps_per_second": 0.92,
+      "step": 175000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305307692307693e-05,
+      "loss": 0.9708,
+      "step": 175100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297615384615387e-05,
+      "loss": 0.9911,
+      "step": 175200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289923076923075e-05,
+      "loss": 0.9741,
+      "step": 175300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728223076923077e-05,
+      "loss": 0.9461,
+      "step": 175400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727453846153846e-05,
+      "loss": 0.9782,
+      "step": 175500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266846153846155e-05,
+      "loss": 0.982,
+      "step": 175600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725923076923077e-05,
+      "loss": 0.9369,
+      "step": 175700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251538461538464e-05,
+      "loss": 0.98,
+      "step": 175800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243846153846152e-05,
+      "loss": 0.9957,
+      "step": 175900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7236153846153846e-05,
+      "loss": 0.9959,
+      "step": 176000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7228461538461538e-05,
+      "loss": 0.9887,
+      "step": 176100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7220769230769232e-05,
+      "loss": 0.9849,
+      "step": 176200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7213076923076923e-05,
+      "loss": 0.9466,
+      "step": 176300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7205384615384614e-05,
+      "loss": 0.9692,
+      "step": 176400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.719769230769231e-05,
+      "loss": 0.9683,
+      "step": 176500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.719e-05,
+      "loss": 0.9611,
+      "step": 176600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7182307692307694e-05,
+      "loss": 0.9791,
+      "step": 176700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7174615384615385e-05,
+      "loss": 0.9647,
+      "step": 176800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7166923076923076e-05,
+      "loss": 0.9834,
+      "step": 176900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.715923076923077e-05,
+      "loss": 0.9199,
+      "step": 177000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7151538461538462e-05,
+      "loss": 1.0197,
+      "step": 177100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7143846153846153e-05,
+      "loss": 0.9566,
+      "step": 177200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7136153846153847e-05,
+      "loss": 0.9433,
+      "step": 177300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.712846153846154e-05,
+      "loss": 0.9865,
+      "step": 177400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7120769230769233e-05,
+      "loss": 0.9363,
+      "step": 177500
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.3896074295043945,
+      "eval_runtime": 132.1921,
+      "eval_samples_per_second": 15.129,
+      "eval_steps_per_second": 0.946,
+      "step": 177500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7113076923076924e-05,
+      "loss": 0.9905,
+      "step": 177600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7105384615384615e-05,
+      "loss": 1.0024,
+      "step": 177700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.709769230769231e-05,
+      "loss": 0.9964,
+      "step": 177800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.709e-05,
+      "loss": 0.9867,
+      "step": 177900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7082307692307695e-05,
+      "loss": 0.9373,
+      "step": 178000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7074615384615383e-05,
+      "loss": 0.9647,
+      "step": 178100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7066923076923077e-05,
+      "loss": 0.9845,
+      "step": 178200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.705923076923077e-05,
+      "loss": 0.968,
+      "step": 178300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7051538461538463e-05,
+      "loss": 1.0642,
+      "step": 178400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7043846153846154e-05,
+      "loss": 0.9015,
+      "step": 178500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7036153846153845e-05,
+      "loss": 0.9717,
+      "step": 178600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.702846153846154e-05,
+      "loss": 0.9658,
+      "step": 178700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.702076923076923e-05,
+      "loss": 0.9705,
+      "step": 178800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7013076923076925e-05,
+      "loss": 1.0008,
+      "step": 178900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7005384615384616e-05,
+      "loss": 0.9806,
+      "step": 179000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6997692307692307e-05,
+      "loss": 0.9539,
+      "step": 179100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.699e-05,
+      "loss": 0.921,
+      "step": 179200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6982307692307693e-05,
+      "loss": 0.9557,
+      "step": 179300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6974615384615384e-05,
+      "loss": 0.9668,
+      "step": 179400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6966923076923078e-05,
+      "loss": 0.9333,
+      "step": 179500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6959307692307693e-05,
+      "loss": 0.9883,
+      "step": 179600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6951615384615384e-05,
+      "loss": 0.9703,
+      "step": 179700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.694392307692308e-05,
+      "loss": 0.9284,
+      "step": 179800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.693623076923077e-05,
+      "loss": 0.9983,
+      "step": 179900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.692853846153846e-05,
+      "loss": 0.9284,
+      "step": 180000
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 1.3655064105987549,
+      "eval_runtime": 131.4163,
+      "eval_samples_per_second": 15.219,
+      "eval_steps_per_second": 0.951,
+      "step": 180000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6920846153846155e-05,
+      "loss": 0.9442,
+      "step": 180100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6913153846153846e-05,
+      "loss": 0.9865,
+      "step": 180200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.690546153846154e-05,
+      "loss": 0.9983,
+      "step": 180300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.689776923076923e-05,
+      "loss": 0.9643,
+      "step": 180400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6890076923076923e-05,
+      "loss": 0.9664,
+      "step": 180500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6882384615384617e-05,
+      "loss": 0.9423,
+      "step": 180600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.687469230769231e-05,
+      "loss": 0.9722,
+      "step": 180700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6867e-05,
+      "loss": 0.9748,
+      "step": 180800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.685930769230769e-05,
+      "loss": 0.9516,
+      "step": 180900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6851615384615385e-05,
+      "loss": 0.9431,
+      "step": 181000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.684392307692308e-05,
+      "loss": 1.0381,
+      "step": 181100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.683623076923077e-05,
+      "loss": 0.9987,
+      "step": 181200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6828538461538462e-05,
+      "loss": 0.9364,
+      "step": 181300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6820846153846153e-05,
+      "loss": 0.9649,
+      "step": 181400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6813153846153847e-05,
+      "loss": 0.9634,
+      "step": 181500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.680546153846154e-05,
+      "loss": 0.979,
+      "step": 181600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.679776923076923e-05,
+      "loss": 0.9553,
+      "step": 181700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6790076923076924e-05,
+      "loss": 0.9662,
+      "step": 181800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6782384615384615e-05,
+      "loss": 0.9804,
+      "step": 181900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677469230769231e-05,
+      "loss": 0.9598,
+      "step": 182000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6767e-05,
+      "loss": 0.9793,
+      "step": 182100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675930769230769e-05,
+      "loss": 0.9902,
+      "step": 182200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6751615384615386e-05,
+      "loss": 0.9878,
+      "step": 182300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6743923076923077e-05,
+      "loss": 0.9606,
+      "step": 182400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673623076923077e-05,
+      "loss": 0.9432,
+      "step": 182500
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.3623239994049072,
+      "eval_runtime": 133.1746,
+      "eval_samples_per_second": 15.018,
+      "eval_steps_per_second": 0.939,
+      "step": 182500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672853846153846e-05,
+      "loss": 0.9636,
+      "step": 182600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6720846153846154e-05,
+      "loss": 0.9773,
+      "step": 182700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6713153846153848e-05,
+      "loss": 0.9479,
+      "step": 182800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670546153846154e-05,
+      "loss": 0.9771,
+      "step": 182900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669776923076923e-05,
+      "loss": 0.9487,
+      "step": 183000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669007692307692e-05,
+      "loss": 0.9433,
+      "step": 183100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6682384615384616e-05,
+      "loss": 1.0182,
+      "step": 183200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.667469230769231e-05,
+      "loss": 0.9602,
+      "step": 183300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6667e-05,
+      "loss": 0.9695,
+      "step": 183400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6659307692307693e-05,
+      "loss": 0.954,
+      "step": 183500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6651615384615384e-05,
+      "loss": 1.0222,
+      "step": 183600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6644e-05,
+      "loss": 0.9658,
+      "step": 183700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6636307692307693e-05,
+      "loss": 0.9547,
+      "step": 183800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6628615384615387e-05,
+      "loss": 0.9519,
+      "step": 183900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.662092307692308e-05,
+      "loss": 0.9458,
+      "step": 184000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.661323076923077e-05,
+      "loss": 0.9238,
+      "step": 184100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.660553846153846e-05,
+      "loss": 0.9996,
+      "step": 184200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6597846153846155e-05,
+      "loss": 0.9508,
+      "step": 184300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.659015384615385e-05,
+      "loss": 0.918,
+      "step": 184400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6582461538461537e-05,
+      "loss": 0.9612,
+      "step": 184500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6574769230769232e-05,
+      "loss": 0.9675,
+      "step": 184600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6567076923076923e-05,
+      "loss": 1.0102,
+      "step": 184700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6559384615384617e-05,
+      "loss": 0.9606,
+      "step": 184800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6551692307692305e-05,
+      "loss": 0.9755,
+      "step": 184900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6544e-05,
+      "loss": 0.9893,
+      "step": 185000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 1.3565884828567505,
+      "eval_runtime": 139.9559,
+      "eval_samples_per_second": 14.29,
+      "eval_steps_per_second": 0.893,
+      "step": 185000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6536307692307694e-05,
+      "loss": 0.9898,
+      "step": 185100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6528615384615385e-05,
+      "loss": 0.957,
+      "step": 185200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.652092307692308e-05,
+      "loss": 0.9273,
+      "step": 185300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6513230769230767e-05,
+      "loss": 0.978,
+      "step": 185400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.650553846153846e-05,
+      "loss": 0.9636,
+      "step": 185500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6497846153846156e-05,
+      "loss": 0.958,
+      "step": 185600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6490153846153847e-05,
+      "loss": 0.9808,
+      "step": 185700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6482461538461538e-05,
+      "loss": 0.9225,
+      "step": 185800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.647476923076923e-05,
+      "loss": 0.9736,
+      "step": 185900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6467076923076924e-05,
+      "loss": 0.9718,
+      "step": 186000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6459384615384618e-05,
+      "loss": 0.9537,
+      "step": 186100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6451769230769233e-05,
+      "loss": 0.9387,
+      "step": 186200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6444076923076924e-05,
+      "loss": 0.9921,
+      "step": 186300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6436384615384615e-05,
+      "loss": 0.9505,
+      "step": 186400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6428692307692306e-05,
+      "loss": 0.9912,
+      "step": 186500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6421e-05,
+      "loss": 0.9637,
+      "step": 186600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6413307692307695e-05,
+      "loss": 0.9648,
+      "step": 186700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6405615384615383e-05,
+      "loss": 0.968,
+      "step": 186800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6397923076923077e-05,
+      "loss": 0.9671,
+      "step": 186900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.639023076923077e-05,
+      "loss": 0.9385,
+      "step": 187000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6382538461538463e-05,
+      "loss": 0.9691,
+      "step": 187100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6374846153846157e-05,
+      "loss": 0.9524,
+      "step": 187200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6367153846153845e-05,
+      "loss": 0.953,
+      "step": 187300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.635946153846154e-05,
+      "loss": 0.9853,
+      "step": 187400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.635176923076923e-05,
+      "loss": 0.953,
+      "step": 187500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.371340036392212,
+      "eval_runtime": 140.9087,
+      "eval_samples_per_second": 14.194,
+      "eval_steps_per_second": 0.887,
+      "step": 187500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6344076923076925e-05,
+      "loss": 0.9591,
+      "step": 187600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6336384615384613e-05,
+      "loss": 0.97,
+      "step": 187700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6328692307692307e-05,
+      "loss": 0.9672,
+      "step": 187800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6321000000000002e-05,
+      "loss": 0.9874,
+      "step": 187900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6313307692307693e-05,
+      "loss": 0.9598,
+      "step": 188000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6305615384615384e-05,
+      "loss": 0.9763,
+      "step": 188100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6297923076923075e-05,
+      "loss": 0.944,
+      "step": 188200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.629023076923077e-05,
+      "loss": 0.9571,
+      "step": 188300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6282538461538464e-05,
+      "loss": 0.9588,
+      "step": 188400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6274846153846155e-05,
+      "loss": 0.9607,
+      "step": 188500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6267153846153846e-05,
+      "loss": 0.9579,
+      "step": 188600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6259461538461537e-05,
+      "loss": 0.9721,
+      "step": 188700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6251846153846152e-05,
+      "loss": 0.9816,
+      "step": 188800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6244153846153847e-05,
+      "loss": 0.9506,
+      "step": 188900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.623646153846154e-05,
+      "loss": 0.9825,
+      "step": 189000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6228769230769232e-05,
+      "loss": 0.9721,
+      "step": 189100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6221076923076923e-05,
+      "loss": 0.9711,
+      "step": 189200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6213384615384614e-05,
+      "loss": 0.9696,
+      "step": 189300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.620569230769231e-05,
+      "loss": 0.9563,
+      "step": 189400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6198000000000003e-05,
+      "loss": 0.9352,
+      "step": 189500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.619030769230769e-05,
+      "loss": 0.9673,
+      "step": 189600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6182615384615385e-05,
+      "loss": 0.9652,
+      "step": 189700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6174923076923076e-05,
+      "loss": 0.9467,
+      "step": 189800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.616723076923077e-05,
+      "loss": 0.9391,
+      "step": 189900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6159615384615386e-05,
+      "loss": 0.9467,
+      "step": 190000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.3749462366104126,
+      "eval_runtime": 131.7303,
+      "eval_samples_per_second": 15.183,
+      "eval_steps_per_second": 0.949,
+      "step": 190000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.615192307692308e-05,
+      "loss": 0.923,
+      "step": 190100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6144230769230768e-05,
+      "loss": 0.9901,
+      "step": 190200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6136538461538462e-05,
+      "loss": 0.9372,
+      "step": 190300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6128846153846153e-05,
+      "loss": 0.9687,
+      "step": 190400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6121153846153848e-05,
+      "loss": 0.9398,
+      "step": 190500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.611346153846154e-05,
+      "loss": 0.9872,
+      "step": 190600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.610576923076923e-05,
+      "loss": 0.967,
+      "step": 190700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6098076923076924e-05,
+      "loss": 0.9923,
+      "step": 190800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6090384615384616e-05,
+      "loss": 0.9723,
+      "step": 190900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6082692307692307e-05,
+      "loss": 0.9512,
+      "step": 191000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6075e-05,
+      "loss": 0.9297,
+      "step": 191100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6067307692307692e-05,
+      "loss": 0.9593,
+      "step": 191200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6059615384615387e-05,
+      "loss": 0.9539,
+      "step": 191300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6051923076923078e-05,
+      "loss": 0.9674,
+      "step": 191400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.604423076923077e-05,
+      "loss": 0.9473,
+      "step": 191500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6036538461538463e-05,
+      "loss": 0.9954,
+      "step": 191600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6028846153846154e-05,
+      "loss": 0.9827,
+      "step": 191700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.602115384615385e-05,
+      "loss": 0.987,
+      "step": 191800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6013461538461537e-05,
+      "loss": 0.9517,
+      "step": 191900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.600576923076923e-05,
+      "loss": 0.9772,
+      "step": 192000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5998076923076922e-05,
+      "loss": 0.9919,
+      "step": 192100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5990384615384617e-05,
+      "loss": 0.9501,
+      "step": 192200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5982692307692308e-05,
+      "loss": 0.9456,
+      "step": 192300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5975e-05,
+      "loss": 0.9733,
+      "step": 192400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5967307692307693e-05,
+      "loss": 0.959,
+      "step": 192500
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.361405611038208,
+      "eval_runtime": 132.9526,
+      "eval_samples_per_second": 15.043,
+      "eval_steps_per_second": 0.94,
+      "step": 192500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5959615384615384e-05,
+      "loss": 0.9728,
+      "step": 192600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.595192307692308e-05,
+      "loss": 0.9455,
+      "step": 192700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.594423076923077e-05,
+      "loss": 1.0118,
+      "step": 192800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.593653846153846e-05,
+      "loss": 0.9246,
+      "step": 192900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5928846153846155e-05,
+      "loss": 0.9864,
+      "step": 193000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5921153846153846e-05,
+      "loss": 0.9297,
+      "step": 193100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5913461538461537e-05,
+      "loss": 0.9416,
+      "step": 193200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5905769230769232e-05,
+      "loss": 0.9363,
+      "step": 193300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5898076923076923e-05,
+      "loss": 0.9736,
+      "step": 193400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5890384615384617e-05,
+      "loss": 0.9861,
+      "step": 193500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.588269230769231e-05,
+      "loss": 0.9604,
+      "step": 193600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5875e-05,
+      "loss": 0.9373,
+      "step": 193700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5867307692307694e-05,
+      "loss": 0.9488,
+      "step": 193800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5859615384615385e-05,
+      "loss": 0.9454,
+      "step": 193900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.585192307692308e-05,
+      "loss": 0.9423,
+      "step": 194000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5844230769230767e-05,
+      "loss": 0.9533,
+      "step": 194100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5836615384615382e-05,
+      "loss": 0.9598,
+      "step": 194200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5828923076923077e-05,
+      "loss": 0.9681,
+      "step": 194300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.582123076923077e-05,
+      "loss": 0.9367,
+      "step": 194400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5813538461538462e-05,
+      "loss": 0.9673,
+      "step": 194500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5805846153846157e-05,
+      "loss": 0.9738,
+      "step": 194600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5798153846153844e-05,
+      "loss": 0.9586,
+      "step": 194700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.579046153846154e-05,
+      "loss": 0.9515,
+      "step": 194800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.578276923076923e-05,
+      "loss": 0.959,
+      "step": 194900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5775076923076924e-05,
+      "loss": 0.9592,
+      "step": 195000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.3450126647949219,
+      "eval_runtime": 137.8736,
+      "eval_samples_per_second": 14.506,
+      "eval_steps_per_second": 0.907,
+      "step": 195000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5767384615384615e-05,
+      "loss": 0.9442,
+      "step": 195100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5759692307692307e-05,
+      "loss": 0.9347,
+      "step": 195200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5752e-05,
+      "loss": 0.9847,
+      "step": 195300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5744307692307692e-05,
+      "loss": 0.9273,
+      "step": 195400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5736615384615387e-05,
+      "loss": 0.9211,
+      "step": 195500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5728923076923078e-05,
+      "loss": 0.9566,
+      "step": 195600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.572123076923077e-05,
+      "loss": 0.9406,
+      "step": 195700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5713538461538463e-05,
+      "loss": 0.9384,
+      "step": 195800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5705846153846154e-05,
+      "loss": 0.9457,
+      "step": 195900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5698153846153845e-05,
+      "loss": 0.9526,
+      "step": 196000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.569046153846154e-05,
+      "loss": 0.948,
+      "step": 196100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.568276923076923e-05,
+      "loss": 0.9415,
+      "step": 196200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5675076923076925e-05,
+      "loss": 0.9493,
+      "step": 196300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5667384615384613e-05,
+      "loss": 0.9162,
+      "step": 196400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5659692307692307e-05,
+      "loss": 0.9423,
+      "step": 196500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5652000000000002e-05,
+      "loss": 0.9142,
+      "step": 196600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5644307692307693e-05,
+      "loss": 0.9326,
+      "step": 196700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5636615384615387e-05,
+      "loss": 0.9446,
+      "step": 196800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5628923076923075e-05,
+      "loss": 0.8877,
+      "step": 196900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.562123076923077e-05,
+      "loss": 0.9581,
+      "step": 197000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5613538461538464e-05,
+      "loss": 1.0145,
+      "step": 197100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5605846153846155e-05,
+      "loss": 0.9916,
+      "step": 197200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5598153846153846e-05,
+      "loss": 0.9574,
+      "step": 197300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5590461538461537e-05,
+      "loss": 0.9637,
+      "step": 197400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5582769230769232e-05,
+      "loss": 0.9632,
+      "step": 197500
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 1.3249619007110596,
+      "eval_runtime": 141.3503,
+      "eval_samples_per_second": 14.149,
+      "eval_steps_per_second": 0.884,
+      "step": 197500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5575076923076923e-05,
+      "loss": 0.9912,
+      "step": 197600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5567384615384614e-05,
+      "loss": 0.9349,
+      "step": 197700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5559769230769232e-05,
+      "loss": 0.9486,
+      "step": 197800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5552076923076923e-05,
+      "loss": 0.9649,
+      "step": 197900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5544384615384614e-05,
+      "loss": 0.9199,
+      "step": 198000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.553669230769231e-05,
+      "loss": 0.9687,
+      "step": 198100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5529e-05,
+      "loss": 0.9732,
+      "step": 198200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.552130769230769e-05,
+      "loss": 0.9582,
+      "step": 198300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5513615384615385e-05,
+      "loss": 0.9144,
+      "step": 198400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5505923076923077e-05,
+      "loss": 0.9924,
+      "step": 198500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.549823076923077e-05,
+      "loss": 0.9957,
+      "step": 198600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5490538461538462e-05,
+      "loss": 0.9647,
+      "step": 198700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5482846153846153e-05,
+      "loss": 0.9104,
+      "step": 198800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5475153846153848e-05,
+      "loss": 0.8968,
+      "step": 198900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.546746153846154e-05,
+      "loss": 0.9262,
+      "step": 199000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5459769230769233e-05,
+      "loss": 0.9638,
+      "step": 199100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.545207692307692e-05,
+      "loss": 0.9507,
+      "step": 199200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5444384615384615e-05,
+      "loss": 0.9571,
+      "step": 199300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.543669230769231e-05,
+      "loss": 0.9727,
+      "step": 199400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5429e-05,
+      "loss": 0.9588,
+      "step": 199500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5421307692307692e-05,
+      "loss": 0.932,
+      "step": 199600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5413615384615383e-05,
+      "loss": 0.966,
+      "step": 199700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5405923076923077e-05,
+      "loss": 0.9479,
+      "step": 199800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5398230769230772e-05,
+      "loss": 0.9638,
+      "step": 199900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5390538461538463e-05,
+      "loss": 0.9442,
+      "step": 200000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.3430618047714233,
+      "eval_runtime": 132.3695,
+      "eval_samples_per_second": 15.109,
+      "eval_steps_per_second": 0.944,
+      "step": 200000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5382846153846154e-05,
+      "loss": 0.9234,
+      "step": 200100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5375153846153845e-05,
+      "loss": 0.99,
+      "step": 200200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.536746153846154e-05,
+      "loss": 0.967,
+      "step": 200300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5359846153846154e-05,
+      "loss": 0.9394,
+      "step": 200400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.535215384615385e-05,
+      "loss": 0.9542,
+      "step": 200500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5344461538461537e-05,
+      "loss": 0.9219,
+      "step": 200600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.533676923076923e-05,
+      "loss": 0.9259,
+      "step": 200700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5329076923076922e-05,
+      "loss": 0.9319,
+      "step": 200800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5321384615384617e-05,
+      "loss": 0.9458,
+      "step": 200900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.531369230769231e-05,
+      "loss": 0.9168,
+      "step": 201000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5306e-05,
+      "loss": 0.9654,
+      "step": 201100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5298307692307693e-05,
+      "loss": 0.9474,
+      "step": 201200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5290615384615384e-05,
+      "loss": 0.94,
+      "step": 201300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.528292307692308e-05,
+      "loss": 0.9773,
+      "step": 201400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5275230769230767e-05,
+      "loss": 0.9838,
+      "step": 201500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.526753846153846e-05,
+      "loss": 0.9702,
+      "step": 201600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5259846153846155e-05,
+      "loss": 0.8796,
+      "step": 201700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5252153846153847e-05,
+      "loss": 0.9391,
+      "step": 201800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.524446153846154e-05,
+      "loss": 0.9725,
+      "step": 201900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.523676923076923e-05,
+      "loss": 0.9575,
+      "step": 202000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5229076923076923e-05,
+      "loss": 0.9789,
+      "step": 202100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5221384615384618e-05,
+      "loss": 0.9425,
+      "step": 202200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5213692307692307e-05,
+      "loss": 0.9923,
+      "step": 202300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5206000000000001e-05,
+      "loss": 0.9589,
+      "step": 202400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5198384615384616e-05,
+      "loss": 0.9715,
+      "step": 202500
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 1.344556450843811,
+      "eval_runtime": 132.2707,
+      "eval_samples_per_second": 15.121,
+      "eval_steps_per_second": 0.945,
+      "step": 202500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5190692307692307e-05,
+      "loss": 0.9635,
+      "step": 202600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5183e-05,
+      "loss": 0.9402,
+      "step": 202700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5175307692307695e-05,
+      "loss": 0.9543,
+      "step": 202800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5167615384615384e-05,
+      "loss": 0.9247,
+      "step": 202900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5159923076923078e-05,
+      "loss": 0.9212,
+      "step": 203000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5152230769230768e-05,
+      "loss": 0.9715,
+      "step": 203100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5144538461538462e-05,
+      "loss": 0.92,
+      "step": 203200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5136846153846155e-05,
+      "loss": 0.9685,
+      "step": 203300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5129153846153846e-05,
+      "loss": 0.9293,
+      "step": 203400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5121461538461539e-05,
+      "loss": 0.9299,
+      "step": 203500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.511376923076923e-05,
+      "loss": 0.9328,
+      "step": 203600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5106153846153845e-05,
+      "loss": 0.9564,
+      "step": 203700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.509846153846154e-05,
+      "loss": 0.9779,
+      "step": 203800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5090769230769232e-05,
+      "loss": 0.9795,
+      "step": 203900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5083076923076923e-05,
+      "loss": 0.9706,
+      "step": 204000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5075384615384616e-05,
+      "loss": 0.9322,
+      "step": 204100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5067692307692307e-05,
+      "loss": 0.9486,
+      "step": 204200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.506e-05,
+      "loss": 0.947,
+      "step": 204300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5052307692307694e-05,
+      "loss": 0.941,
+      "step": 204400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5044615384615384e-05,
+      "loss": 0.9407,
+      "step": 204500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5036923076923078e-05,
+      "loss": 0.9725,
+      "step": 204600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.502923076923077e-05,
+      "loss": 0.9348,
+      "step": 204700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5021538461538462e-05,
+      "loss": 0.9493,
+      "step": 204800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5013846153846156e-05,
+      "loss": 0.9685,
+      "step": 204900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5006153846153846e-05,
+      "loss": 0.9406,
+      "step": 205000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.3243491649627686,
+      "eval_runtime": 132.3857,
+      "eval_samples_per_second": 15.107,
+      "eval_steps_per_second": 0.944,
+      "step": 205000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4998461538461539e-05,
+      "loss": 0.9397,
+      "step": 205100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4990769230769231e-05,
+      "loss": 0.9279,
+      "step": 205200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4983076923076924e-05,
+      "loss": 0.9859,
+      "step": 205300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4975384615384615e-05,
+      "loss": 0.9369,
+      "step": 205400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4967692307692308e-05,
+      "loss": 0.9374,
+      "step": 205500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4959999999999999e-05,
+      "loss": 0.9088,
+      "step": 205600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4952307692307694e-05,
+      "loss": 0.9119,
+      "step": 205700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4944615384615385e-05,
+      "loss": 0.9531,
+      "step": 205800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4936923076923077e-05,
+      "loss": 0.9567,
+      "step": 205900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.492923076923077e-05,
+      "loss": 0.9319,
+      "step": 206000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4921538461538461e-05,
+      "loss": 0.9294,
+      "step": 206100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4913846153846154e-05,
+      "loss": 0.9343,
+      "step": 206200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4906153846153847e-05,
+      "loss": 0.9376,
+      "step": 206300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.489846153846154e-05,
+      "loss": 0.8987,
+      "step": 206400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.489076923076923e-05,
+      "loss": 0.9609,
+      "step": 206500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4883076923076923e-05,
+      "loss": 0.9309,
+      "step": 206600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4875384615384614e-05,
+      "loss": 0.9459,
+      "step": 206700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4867692307692309e-05,
+      "loss": 0.9164,
+      "step": 206800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.486e-05,
+      "loss": 0.9069,
+      "step": 206900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4852307692307693e-05,
+      "loss": 0.9725,
+      "step": 207000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4844615384615386e-05,
+      "loss": 0.9293,
+      "step": 207100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4836923076923077e-05,
+      "loss": 0.9347,
+      "step": 207200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.482923076923077e-05,
+      "loss": 0.9394,
+      "step": 207300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4821538461538462e-05,
+      "loss": 0.9506,
+      "step": 207400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4813846153846155e-05,
+      "loss": 0.9277,
+      "step": 207500
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 1.3164209127426147,
+      "eval_runtime": 131.5056,
+      "eval_samples_per_second": 15.208,
+      "eval_steps_per_second": 0.951,
+      "step": 207500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4806153846153846e-05,
+      "loss": 0.9352,
+      "step": 207600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4798461538461539e-05,
+      "loss": 0.9524,
+      "step": 207700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.479076923076923e-05,
+      "loss": 0.9562,
+      "step": 207800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4783076923076924e-05,
+      "loss": 0.9474,
+      "step": 207900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4775384615384615e-05,
+      "loss": 0.9495,
+      "step": 208000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4767692307692308e-05,
+      "loss": 0.9593,
+      "step": 208100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 0.9721,
+      "step": 208200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4752307692307692e-05,
+      "loss": 0.9732,
+      "step": 208300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4744615384615385e-05,
+      "loss": 0.983,
+      "step": 208400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4736923076923078e-05,
+      "loss": 0.915,
+      "step": 208500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.472923076923077e-05,
+      "loss": 0.9684,
+      "step": 208600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4721538461538461e-05,
+      "loss": 0.9619,
+      "step": 208700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4713846153846154e-05,
+      "loss": 0.9628,
+      "step": 208800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4706153846153845e-05,
+      "loss": 0.9574,
+      "step": 208900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.469846153846154e-05,
+      "loss": 0.9406,
+      "step": 209000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4690769230769231e-05,
+      "loss": 0.9784,
+      "step": 209100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4683076923076924e-05,
+      "loss": 0.9737,
+      "step": 209200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4675384615384615e-05,
+      "loss": 0.9306,
+      "step": 209300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4667692307692307e-05,
+      "loss": 0.9245,
+      "step": 209400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4660000000000002e-05,
+      "loss": 0.9869,
+      "step": 209500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4652307692307693e-05,
+      "loss": 0.9628,
+      "step": 209600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4644615384615386e-05,
+      "loss": 0.9235,
+      "step": 209700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4637e-05,
+      "loss": 0.923,
+      "step": 209800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4629307692307692e-05,
+      "loss": 0.9764,
+      "step": 209900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4621615384615384e-05,
+      "loss": 0.9151,
+      "step": 210000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 1.3465312719345093,
+      "eval_runtime": 131.0431,
+      "eval_samples_per_second": 15.262,
+      "eval_steps_per_second": 0.954,
+      "step": 210000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4613923076923077e-05,
+      "loss": 0.9143,
+      "step": 210100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.460623076923077e-05,
+      "loss": 0.9586,
+      "step": 210200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4598538461538463e-05,
+      "loss": 0.8972,
+      "step": 210300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4590846153846154e-05,
+      "loss": 0.9608,
+      "step": 210400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4583153846153847e-05,
+      "loss": 0.9486,
+      "step": 210500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4575461538461538e-05,
+      "loss": 0.9229,
+      "step": 210600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4567769230769232e-05,
+      "loss": 0.963,
+      "step": 210700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4560076923076923e-05,
+      "loss": 0.9718,
+      "step": 210800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4552384615384616e-05,
+      "loss": 0.9201,
+      "step": 210900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4544692307692307e-05,
+      "loss": 0.9327,
+      "step": 211000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4537e-05,
+      "loss": 0.9285,
+      "step": 211100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4529307692307693e-05,
+      "loss": 0.959,
+      "step": 211200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4521615384615385e-05,
+      "loss": 0.9084,
+      "step": 211300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4513923076923078e-05,
+      "loss": 0.9299,
+      "step": 211400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.450623076923077e-05,
+      "loss": 0.9783,
+      "step": 211500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4498538461538462e-05,
+      "loss": 0.9433,
+      "step": 211600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4490846153846153e-05,
+      "loss": 0.909,
+      "step": 211700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4483153846153848e-05,
+      "loss": 0.9339,
+      "step": 211800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4475538461538462e-05,
+      "loss": 0.9643,
+      "step": 211900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4467846153846154e-05,
+      "loss": 0.95,
+      "step": 212000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4460153846153846e-05,
+      "loss": 0.9535,
+      "step": 212100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4452461538461539e-05,
+      "loss": 0.961,
+      "step": 212200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4444769230769232e-05,
+      "loss": 0.9249,
+      "step": 212300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4437076923076925e-05,
+      "loss": 0.9144,
+      "step": 212400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4429384615384616e-05,
+      "loss": 0.9572,
+      "step": 212500
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.284892201423645,
+      "eval_runtime": 132.2339,
+      "eval_samples_per_second": 15.125,
+      "eval_steps_per_second": 0.945,
+      "step": 212500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4421692307692308e-05,
+      "loss": 0.921,
+      "step": 212600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4414e-05,
+      "loss": 0.9245,
+      "step": 212700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4406307692307692e-05,
+      "loss": 0.9346,
+      "step": 212800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4398615384615385e-05,
+      "loss": 0.9134,
+      "step": 212900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4390923076923078e-05,
+      "loss": 0.9699,
+      "step": 213000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4383230769230769e-05,
+      "loss": 0.9391,
+      "step": 213100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4375538461538462e-05,
+      "loss": 0.889,
+      "step": 213200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4367923076923077e-05,
+      "loss": 0.9044,
+      "step": 213300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.436023076923077e-05,
+      "loss": 0.9357,
+      "step": 213400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4352538461538462e-05,
+      "loss": 0.9213,
+      "step": 213500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4344846153846155e-05,
+      "loss": 0.8873,
+      "step": 213600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4337153846153846e-05,
+      "loss": 0.9226,
+      "step": 213700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4329461538461539e-05,
+      "loss": 0.9333,
+      "step": 213800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.432176923076923e-05,
+      "loss": 0.9499,
+      "step": 213900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4314076923076924e-05,
+      "loss": 0.9058,
+      "step": 214000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4306384615384615e-05,
+      "loss": 0.9065,
+      "step": 214100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4298692307692308e-05,
+      "loss": 0.9669,
+      "step": 214200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4291000000000001e-05,
+      "loss": 0.9501,
+      "step": 214300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4283307692307692e-05,
+      "loss": 0.9314,
+      "step": 214400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4275615384615385e-05,
+      "loss": 0.9806,
+      "step": 214500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4267923076923078e-05,
+      "loss": 0.9218,
+      "step": 214600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.426023076923077e-05,
+      "loss": 0.9527,
+      "step": 214700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4252538461538461e-05,
+      "loss": 0.9532,
+      "step": 214800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4244846153846154e-05,
+      "loss": 0.9297,
+      "step": 214900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4237153846153845e-05,
+      "loss": 0.9124,
+      "step": 215000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.3008171319961548,
+      "eval_runtime": 132.8543,
+      "eval_samples_per_second": 15.054,
+      "eval_steps_per_second": 0.941,
+      "step": 215000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.422946153846154e-05,
+      "loss": 0.9525,
+      "step": 215100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.422176923076923e-05,
+      "loss": 0.9309,
+      "step": 215200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4214076923076924e-05,
+      "loss": 0.9221,
+      "step": 215300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4206384615384616e-05,
+      "loss": 0.8993,
+      "step": 215400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4198692307692307e-05,
+      "loss": 0.926,
+      "step": 215500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4191e-05,
+      "loss": 0.987,
+      "step": 215600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4183307692307693e-05,
+      "loss": 0.9387,
+      "step": 215700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4175615384615386e-05,
+      "loss": 0.9592,
+      "step": 215800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4167923076923077e-05,
+      "loss": 0.9053,
+      "step": 215900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.416023076923077e-05,
+      "loss": 0.949,
+      "step": 216000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.415253846153846e-05,
+      "loss": 0.9734,
+      "step": 216100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4144846153846155e-05,
+      "loss": 0.9241,
+      "step": 216200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4137153846153846e-05,
+      "loss": 0.9074,
+      "step": 216300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4129461538461539e-05,
+      "loss": 0.9399,
+      "step": 216400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.412176923076923e-05,
+      "loss": 0.9142,
+      "step": 216500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4114076923076923e-05,
+      "loss": 0.9372,
+      "step": 216600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4106384615384617e-05,
+      "loss": 0.9612,
+      "step": 216700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4098692307692308e-05,
+      "loss": 0.9469,
+      "step": 216800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4091000000000001e-05,
+      "loss": 0.9253,
+      "step": 216900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4083384615384616e-05,
+      "loss": 0.9451,
+      "step": 217000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4075692307692307e-05,
+      "loss": 0.8985,
+      "step": 217100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4068e-05,
+      "loss": 0.967,
+      "step": 217200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4060307692307693e-05,
+      "loss": 0.9302,
+      "step": 217300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4052615384615385e-05,
+      "loss": 0.9502,
+      "step": 217400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4044923076923078e-05,
+      "loss": 0.9485,
+      "step": 217500
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 1.326869010925293,
+      "eval_runtime": 132.1409,
+      "eval_samples_per_second": 15.135,
+      "eval_steps_per_second": 0.946,
+      "step": 217500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.403723076923077e-05,
+      "loss": 0.9198,
+      "step": 217600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4029538461538462e-05,
+      "loss": 0.9632,
+      "step": 217700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4021846153846153e-05,
+      "loss": 0.9808,
+      "step": 217800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4014153846153848e-05,
+      "loss": 0.9639,
+      "step": 217900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4006461538461539e-05,
+      "loss": 0.9127,
+      "step": 218000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3998769230769231e-05,
+      "loss": 0.9569,
+      "step": 218100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3991076923076922e-05,
+      "loss": 0.9631,
+      "step": 218200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3983384615384615e-05,
+      "loss": 0.957,
+      "step": 218300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3975692307692308e-05,
+      "loss": 0.9416,
+      "step": 218400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3968e-05,
+      "loss": 0.9098,
+      "step": 218500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3960307692307694e-05,
+      "loss": 0.966,
+      "step": 218600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3952615384615385e-05,
+      "loss": 0.9617,
+      "step": 218700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3944923076923077e-05,
+      "loss": 0.8882,
+      "step": 218800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3937230769230768e-05,
+      "loss": 0.9637,
+      "step": 218900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3929538461538463e-05,
+      "loss": 0.9702,
+      "step": 219000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3921846153846154e-05,
+      "loss": 0.9515,
+      "step": 219100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3914153846153847e-05,
+      "loss": 0.9784,
+      "step": 219200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3906461538461538e-05,
+      "loss": 0.9037,
+      "step": 219300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389876923076923e-05,
+      "loss": 0.9472,
+      "step": 219400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3891076923076923e-05,
+      "loss": 0.9371,
+      "step": 219500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3883384615384616e-05,
+      "loss": 0.9285,
+      "step": 219600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3875692307692309e-05,
+      "loss": 0.917,
+      "step": 219700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3868e-05,
+      "loss": 0.9543,
+      "step": 219800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3860307692307693e-05,
+      "loss": 0.936,
+      "step": 219900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3852615384615386e-05,
+      "loss": 0.9039,
+      "step": 220000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 1.2885546684265137,
+      "eval_runtime": 132.6753,
+      "eval_samples_per_second": 15.074,
+      "eval_steps_per_second": 0.942,
+      "step": 220000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3844923076923078e-05,
+      "loss": 0.9238,
+      "step": 220100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.383723076923077e-05,
+      "loss": 0.9759,
+      "step": 220200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3829538461538462e-05,
+      "loss": 0.9312,
+      "step": 220300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3821846153846153e-05,
+      "loss": 0.9054,
+      "step": 220400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3814153846153846e-05,
+      "loss": 0.9308,
+      "step": 220500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3806461538461539e-05,
+      "loss": 0.941,
+      "step": 220600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3798769230769232e-05,
+      "loss": 0.9246,
+      "step": 220700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3791153846153846e-05,
+      "loss": 0.9101,
+      "step": 220800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.378346153846154e-05,
+      "loss": 0.9316,
+      "step": 220900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.377576923076923e-05,
+      "loss": 0.9118,
+      "step": 221000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3768076923076923e-05,
+      "loss": 0.9408,
+      "step": 221100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3760384615384616e-05,
+      "loss": 0.9209,
+      "step": 221200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3752692307692309e-05,
+      "loss": 0.9285,
+      "step": 221300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3745e-05,
+      "loss": 0.9235,
+      "step": 221400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3737307692307692e-05,
+      "loss": 0.9352,
+      "step": 221500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3729615384615384e-05,
+      "loss": 0.908,
+      "step": 221600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3721923076923078e-05,
+      "loss": 0.9434,
+      "step": 221700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.371423076923077e-05,
+      "loss": 0.8978,
+      "step": 221800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3706538461538462e-05,
+      "loss": 0.936,
+      "step": 221900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3698846153846155e-05,
+      "loss": 0.928,
+      "step": 222000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3691153846153846e-05,
+      "loss": 0.9238,
+      "step": 222100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3683461538461538e-05,
+      "loss": 0.9426,
+      "step": 222200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3675769230769231e-05,
+      "loss": 0.885,
+      "step": 222300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3668076923076924e-05,
+      "loss": 0.9343,
+      "step": 222400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3660384615384615e-05,
+      "loss": 0.949,
+      "step": 222500
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.3161851167678833,
+      "eval_runtime": 132.9957,
+      "eval_samples_per_second": 15.038,
+      "eval_steps_per_second": 0.94,
+      "step": 222500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3652692307692308e-05,
+      "loss": 0.9593,
+      "step": 222600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3644999999999999e-05,
+      "loss": 0.9495,
+      "step": 222700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3637307692307693e-05,
+      "loss": 0.9686,
+      "step": 222800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3629615384615386e-05,
+      "loss": 0.9271,
+      "step": 222900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3621923076923077e-05,
+      "loss": 0.9266,
+      "step": 223000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.361423076923077e-05,
+      "loss": 0.8693,
+      "step": 223100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3606538461538461e-05,
+      "loss": 0.8614,
+      "step": 223200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3598846153846154e-05,
+      "loss": 0.8325,
+      "step": 223300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.359123076923077e-05,
+      "loss": 0.8687,
+      "step": 223400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3583538461538462e-05,
+      "loss": 0.8601,
+      "step": 223500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3575846153846154e-05,
+      "loss": 0.8395,
+      "step": 223600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3568153846153847e-05,
+      "loss": 0.8539,
+      "step": 223700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3560461538461538e-05,
+      "loss": 0.8643,
+      "step": 223800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3552769230769233e-05,
+      "loss": 0.8342,
+      "step": 223900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3545076923076924e-05,
+      "loss": 0.8713,
+      "step": 224000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3537384615384616e-05,
+      "loss": 0.8722,
+      "step": 224100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3529692307692308e-05,
+      "loss": 0.8124,
+      "step": 224200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3522e-05,
+      "loss": 0.8268,
+      "step": 224300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3514307692307691e-05,
+      "loss": 0.8459,
+      "step": 224400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3506615384615386e-05,
+      "loss": 0.8638,
+      "step": 224500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3498923076923077e-05,
+      "loss": 0.8581,
+      "step": 224600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3491307692307693e-05,
+      "loss": 0.844,
+      "step": 224700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3483615384615385e-05,
+      "loss": 0.8655,
+      "step": 224800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3475923076923077e-05,
+      "loss": 0.8905,
+      "step": 224900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3468230769230768e-05,
+      "loss": 0.848,
+      "step": 225000
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 1.2882050275802612,
+      "eval_runtime": 132.1203,
+      "eval_samples_per_second": 15.138,
+      "eval_steps_per_second": 0.946,
+      "step": 225000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3460538461538463e-05,
+      "loss": 0.8415,
+      "step": 225100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3452846153846154e-05,
+      "loss": 0.841,
+      "step": 225200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3445153846153847e-05,
+      "loss": 0.8712,
+      "step": 225300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3437461538461538e-05,
+      "loss": 0.8462,
+      "step": 225400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.342976923076923e-05,
+      "loss": 0.8502,
+      "step": 225500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3422076923076923e-05,
+      "loss": 0.8723,
+      "step": 225600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3414384615384616e-05,
+      "loss": 0.8614,
+      "step": 225700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3406692307692309e-05,
+      "loss": 0.8481,
+      "step": 225800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3399e-05,
+      "loss": 0.8541,
+      "step": 225900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3391307692307693e-05,
+      "loss": 0.8245,
+      "step": 226000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3383615384615384e-05,
+      "loss": 0.8706,
+      "step": 226100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3375923076923078e-05,
+      "loss": 0.8832,
+      "step": 226200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.336823076923077e-05,
+      "loss": 0.8743,
+      "step": 226300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3360538461538462e-05,
+      "loss": 0.847,
+      "step": 226400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3352846153846153e-05,
+      "loss": 0.8402,
+      "step": 226500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3345153846153846e-05,
+      "loss": 0.8193,
+      "step": 226600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3337461538461539e-05,
+      "loss": 0.8722,
+      "step": 226700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3329769230769232e-05,
+      "loss": 0.8682,
+      "step": 226800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3322076923076924e-05,
+      "loss": 0.8725,
+      "step": 226900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3314384615384615e-05,
+      "loss": 0.8551,
+      "step": 227000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3306692307692308e-05,
+      "loss": 0.8475,
+      "step": 227100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3299000000000001e-05,
+      "loss": 0.8731,
+      "step": 227200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3291307692307694e-05,
+      "loss": 0.8985,
+      "step": 227300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3283615384615385e-05,
+      "loss": 0.8445,
+      "step": 227400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3275923076923078e-05,
+      "loss": 0.8176,
+      "step": 227500
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 1.280893087387085,
+      "eval_runtime": 132.5442,
+      "eval_samples_per_second": 15.089,
+      "eval_steps_per_second": 0.943,
+      "step": 227500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3268230769230769e-05,
+      "loss": 0.8688,
+      "step": 227600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3260538461538461e-05,
+      "loss": 0.8787,
+      "step": 227700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3252846153846154e-05,
+      "loss": 0.8287,
+      "step": 227800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3245153846153847e-05,
+      "loss": 0.8257,
+      "step": 227900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3237538461538462e-05,
+      "loss": 0.8201,
+      "step": 228000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3229846153846155e-05,
+      "loss": 0.8624,
+      "step": 228100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3222153846153846e-05,
+      "loss": 0.8401,
+      "step": 228200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3214461538461538e-05,
+      "loss": 0.8358,
+      "step": 228300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3206769230769231e-05,
+      "loss": 0.8698,
+      "step": 228400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3199076923076924e-05,
+      "loss": 0.8296,
+      "step": 228500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3191384615384615e-05,
+      "loss": 0.8769,
+      "step": 228600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3183692307692308e-05,
+      "loss": 0.8759,
+      "step": 228700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3175999999999999e-05,
+      "loss": 0.8398,
+      "step": 228800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3168307692307693e-05,
+      "loss": 0.8636,
+      "step": 228900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3160615384615386e-05,
+      "loss": 0.8673,
+      "step": 229000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3152923076923077e-05,
+      "loss": 0.833,
+      "step": 229100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.314523076923077e-05,
+      "loss": 0.872,
+      "step": 229200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3137538461538461e-05,
+      "loss": 0.8442,
+      "step": 229300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3129846153846154e-05,
+      "loss": 0.8612,
+      "step": 229400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3122153846153847e-05,
+      "loss": 0.8531,
+      "step": 229500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.311446153846154e-05,
+      "loss": 0.8662,
+      "step": 229600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.310676923076923e-05,
+      "loss": 0.8534,
+      "step": 229700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3099076923076923e-05,
+      "loss": 0.8496,
+      "step": 229800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3091384615384614e-05,
+      "loss": 0.8509,
+      "step": 229900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3083692307692309e-05,
+      "loss": 0.8591,
+      "step": 230000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 1.2731475830078125,
+      "eval_runtime": 131.7106,
+      "eval_samples_per_second": 15.185,
+      "eval_steps_per_second": 0.949,
+      "step": 230000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3076000000000002e-05,
+      "loss": 0.8854,
+      "step": 230100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3068307692307693e-05,
+      "loss": 0.8756,
+      "step": 230200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3060615384615385e-05,
+      "loss": 0.8841,
+      "step": 230300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3052923076923076e-05,
+      "loss": 0.8573,
+      "step": 230400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.304523076923077e-05,
+      "loss": 0.8901,
+      "step": 230500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3037538461538462e-05,
+      "loss": 0.8664,
+      "step": 230600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3029846153846155e-05,
+      "loss": 0.854,
+      "step": 230700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3022153846153846e-05,
+      "loss": 0.853,
+      "step": 230800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3014461538461539e-05,
+      "loss": 0.8484,
+      "step": 230900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.300676923076923e-05,
+      "loss": 0.8437,
+      "step": 231000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2999076923076924e-05,
+      "loss": 0.8485,
+      "step": 231100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2991384615384615e-05,
+      "loss": 0.8372,
+      "step": 231200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2983692307692308e-05,
+      "loss": 0.8442,
+      "step": 231300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2976e-05,
+      "loss": 0.8734,
+      "step": 231400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2968307692307692e-05,
+      "loss": 0.8632,
+      "step": 231500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2960615384615386e-05,
+      "loss": 0.8138,
+      "step": 231600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2952923076923077e-05,
+      "loss": 0.847,
+      "step": 231700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.294523076923077e-05,
+      "loss": 0.8221,
+      "step": 231800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2937538461538461e-05,
+      "loss": 0.8941,
+      "step": 231900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2929846153846154e-05,
+      "loss": 0.8367,
+      "step": 232000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2922153846153845e-05,
+      "loss": 0.8771,
+      "step": 232100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.291446153846154e-05,
+      "loss": 0.858,
+      "step": 232200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.290676923076923e-05,
+      "loss": 0.8255,
+      "step": 232300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2899076923076923e-05,
+      "loss": 0.8391,
+      "step": 232400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2891384615384616e-05,
+      "loss": 0.872,
+      "step": 232500
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 1.2969560623168945,
+      "eval_runtime": 131.1496,
+      "eval_samples_per_second": 15.25,
+      "eval_steps_per_second": 0.953,
+      "step": 232500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2883692307692307e-05,
+      "loss": 0.8546,
+      "step": 232600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2876000000000002e-05,
+      "loss": 0.8934,
+      "step": 232700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2868307692307693e-05,
+      "loss": 0.8594,
+      "step": 232800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2860615384615386e-05,
+      "loss": 0.8798,
+      "step": 232900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2852923076923077e-05,
+      "loss": 0.8501,
+      "step": 233000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.284523076923077e-05,
+      "loss": 0.8206,
+      "step": 233100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.283753846153846e-05,
+      "loss": 0.8462,
+      "step": 233200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2829846153846155e-05,
+      "loss": 0.853,
+      "step": 233300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2822153846153846e-05,
+      "loss": 0.816,
+      "step": 233400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2814461538461539e-05,
+      "loss": 0.8433,
+      "step": 233500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2806769230769232e-05,
+      "loss": 0.861,
+      "step": 233600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2799076923076923e-05,
+      "loss": 0.8256,
+      "step": 233700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2791461538461538e-05,
+      "loss": 0.8198,
+      "step": 233800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2783769230769232e-05,
+      "loss": 0.866,
+      "step": 233900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2776076923076923e-05,
+      "loss": 0.8516,
+      "step": 234000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2768384615384616e-05,
+      "loss": 0.8842,
+      "step": 234100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2760692307692307e-05,
+      "loss": 0.8628,
+      "step": 234200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2753e-05,
+      "loss": 0.8317,
+      "step": 234300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2745307692307694e-05,
+      "loss": 0.8435,
+      "step": 234400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2737615384615385e-05,
+      "loss": 0.8449,
+      "step": 234500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.273e-05,
+      "loss": 0.8405,
+      "step": 234600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2722307692307693e-05,
+      "loss": 0.8561,
+      "step": 234700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2714615384615384e-05,
+      "loss": 0.8672,
+      "step": 234800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2706923076923077e-05,
+      "loss": 0.8857,
+      "step": 234900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.269923076923077e-05,
+      "loss": 0.873,
+      "step": 235000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 1.2799692153930664,
+      "eval_runtime": 133.3921,
+      "eval_samples_per_second": 14.993,
+      "eval_steps_per_second": 0.937,
+      "step": 235000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2691538461538462e-05,
+      "loss": 0.8443,
+      "step": 235100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2683846153846155e-05,
+      "loss": 0.8289,
+      "step": 235200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2676153846153846e-05,
+      "loss": 0.8678,
+      "step": 235300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2668461538461539e-05,
+      "loss": 0.8386,
+      "step": 235400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2660769230769232e-05,
+      "loss": 0.8757,
+      "step": 235500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2653076923076924e-05,
+      "loss": 0.9056,
+      "step": 235600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2645384615384616e-05,
+      "loss": 0.8528,
+      "step": 235700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2637692307692308e-05,
+      "loss": 0.8788,
+      "step": 235800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.263e-05,
+      "loss": 0.8169,
+      "step": 235900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2622307692307692e-05,
+      "loss": 0.8123,
+      "step": 236000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2614615384615385e-05,
+      "loss": 0.8263,
+      "step": 236100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2606923076923078e-05,
+      "loss": 0.8449,
+      "step": 236200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2599307692307693e-05,
+      "loss": 0.8339,
+      "step": 236300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2591615384615385e-05,
+      "loss": 0.8555,
+      "step": 236400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2583923076923076e-05,
+      "loss": 0.7763,
+      "step": 236500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2576230769230769e-05,
+      "loss": 0.8891,
+      "step": 236600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2568538461538462e-05,
+      "loss": 0.8539,
+      "step": 236700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2560846153846155e-05,
+      "loss": 0.8594,
+      "step": 236800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2553153846153846e-05,
+      "loss": 0.8545,
+      "step": 236900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2545461538461539e-05,
+      "loss": 0.8253,
+      "step": 237000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.253776923076923e-05,
+      "loss": 0.8311,
+      "step": 237100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2530076923076924e-05,
+      "loss": 0.8729,
+      "step": 237200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2522384615384617e-05,
+      "loss": 0.8455,
+      "step": 237300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2514692307692308e-05,
+      "loss": 0.8461,
+      "step": 237400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2507e-05,
+      "loss": 0.832,
+      "step": 237500
+    },
+    {
+      "epoch": 1.06,
+      "eval_loss": 1.2913880348205566,
+      "eval_runtime": 132.2703,
+      "eval_samples_per_second": 15.121,
+      "eval_steps_per_second": 0.945,
+      "step": 237500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2499307692307692e-05,
+      "loss": 0.8687,
+      "step": 237600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2491615384615385e-05,
+      "loss": 0.8842,
+      "step": 237700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2483923076923077e-05,
+      "loss": 0.8592,
+      "step": 237800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.247623076923077e-05,
+      "loss": 0.8761,
+      "step": 237900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2468538461538461e-05,
+      "loss": 0.8917,
+      "step": 238000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2460846153846154e-05,
+      "loss": 0.8573,
+      "step": 238100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2453153846153845e-05,
+      "loss": 0.868,
+      "step": 238200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.244546153846154e-05,
+      "loss": 0.8433,
+      "step": 238300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2437769230769232e-05,
+      "loss": 0.855,
+      "step": 238400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2430076923076923e-05,
+      "loss": 0.8429,
+      "step": 238500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2422384615384616e-05,
+      "loss": 0.8353,
+      "step": 238600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2414692307692307e-05,
+      "loss": 0.879,
+      "step": 238700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2407000000000002e-05,
+      "loss": 0.8264,
+      "step": 238800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2399307692307693e-05,
+      "loss": 0.882,
+      "step": 238900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2391615384615386e-05,
+      "loss": 0.8371,
+      "step": 239000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2383923076923077e-05,
+      "loss": 0.8453,
+      "step": 239100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.237623076923077e-05,
+      "loss": 0.8562,
+      "step": 239200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.236853846153846e-05,
+      "loss": 0.8546,
+      "step": 239300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2360846153846155e-05,
+      "loss": 0.8521,
+      "step": 239400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2353153846153846e-05,
+      "loss": 0.8449,
+      "step": 239500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2345461538461539e-05,
+      "loss": 0.85,
+      "step": 239600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2337769230769232e-05,
+      "loss": 0.8369,
+      "step": 239700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2330076923076923e-05,
+      "loss": 0.8359,
+      "step": 239800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2322384615384617e-05,
+      "loss": 0.8465,
+      "step": 239900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2314692307692308e-05,
+      "loss": 0.8583,
+      "step": 240000
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 1.2713655233383179,
+      "eval_runtime": 132.2608,
+      "eval_samples_per_second": 15.122,
+      "eval_steps_per_second": 0.945,
+      "step": 240000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2307000000000001e-05,
+      "loss": 0.8482,
+      "step": 240100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2299307692307692e-05,
+      "loss": 0.8519,
+      "step": 240200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2291615384615385e-05,
+      "loss": 0.8489,
+      "step": 240300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2283923076923076e-05,
+      "loss": 0.8438,
+      "step": 240400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.227623076923077e-05,
+      "loss": 0.8173,
+      "step": 240500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2268538461538461e-05,
+      "loss": 0.8539,
+      "step": 240600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2260846153846154e-05,
+      "loss": 0.8533,
+      "step": 240700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2253153846153847e-05,
+      "loss": 0.8629,
+      "step": 240800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2245461538461538e-05,
+      "loss": 0.8811,
+      "step": 240900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2237769230769232e-05,
+      "loss": 0.8391,
+      "step": 241000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2230076923076924e-05,
+      "loss": 0.8545,
+      "step": 241100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2222384615384616e-05,
+      "loss": 0.8708,
+      "step": 241200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2214692307692307e-05,
+      "loss": 0.8195,
+      "step": 241300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2207e-05,
+      "loss": 0.8647,
+      "step": 241400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2199307692307691e-05,
+      "loss": 0.8507,
+      "step": 241500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2191615384615386e-05,
+      "loss": 0.8608,
+      "step": 241600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2183923076923077e-05,
+      "loss": 0.8592,
+      "step": 241700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.217623076923077e-05,
+      "loss": 0.8589,
+      "step": 241800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2168615384615384e-05,
+      "loss": 0.8492,
+      "step": 241900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2160923076923077e-05,
+      "loss": 0.8463,
+      "step": 242000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.215323076923077e-05,
+      "loss": 0.8861,
+      "step": 242100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2145538461538463e-05,
+      "loss": 0.8547,
+      "step": 242200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2137846153846154e-05,
+      "loss": 0.8401,
+      "step": 242300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2130153846153847e-05,
+      "loss": 0.8561,
+      "step": 242400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2122461538461538e-05,
+      "loss": 0.8744,
+      "step": 242500
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 1.261897087097168,
+      "eval_runtime": 134.7027,
+      "eval_samples_per_second": 14.848,
+      "eval_steps_per_second": 0.928,
+      "step": 242500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.211476923076923e-05,
+      "loss": 0.8373,
+      "step": 242600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2107076923076923e-05,
+      "loss": 0.8551,
+      "step": 242700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2099384615384616e-05,
+      "loss": 0.8903,
+      "step": 242800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2091692307692309e-05,
+      "loss": 0.8774,
+      "step": 242900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2084e-05,
+      "loss": 0.8465,
+      "step": 243000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2076307692307693e-05,
+      "loss": 0.8196,
+      "step": 243100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2068615384615385e-05,
+      "loss": 0.8382,
+      "step": 243200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2060923076923078e-05,
+      "loss": 0.8343,
+      "step": 243300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.205323076923077e-05,
+      "loss": 0.8473,
+      "step": 243400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2045538461538462e-05,
+      "loss": 0.8571,
+      "step": 243500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2037846153846153e-05,
+      "loss": 0.814,
+      "step": 243600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2030153846153846e-05,
+      "loss": 0.8468,
+      "step": 243700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2022461538461539e-05,
+      "loss": 0.8544,
+      "step": 243800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2014769230769231e-05,
+      "loss": 0.8662,
+      "step": 243900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2007076923076924e-05,
+      "loss": 0.8564,
+      "step": 244000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1999384615384615e-05,
+      "loss": 0.8697,
+      "step": 244100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1991692307692308e-05,
+      "loss": 0.842,
+      "step": 244200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1984e-05,
+      "loss": 0.859,
+      "step": 244300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1976307692307694e-05,
+      "loss": 0.8539,
+      "step": 244400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1968615384615385e-05,
+      "loss": 0.8128,
+      "step": 244500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1960923076923077e-05,
+      "loss": 0.8525,
+      "step": 244600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1953230769230768e-05,
+      "loss": 0.8313,
+      "step": 244700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1945538461538461e-05,
+      "loss": 0.8385,
+      "step": 244800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1937846153846154e-05,
+      "loss": 0.8356,
+      "step": 244900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1930153846153847e-05,
+      "loss": 0.8334,
+      "step": 245000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 1.2770377397537231,
+      "eval_runtime": 138.4711,
+      "eval_samples_per_second": 14.443,
+      "eval_steps_per_second": 0.903,
+      "step": 245000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.192246153846154e-05,
+      "loss": 0.8148,
+      "step": 245100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.191476923076923e-05,
+      "loss": 0.8854,
+      "step": 245200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1907076923076923e-05,
+      "loss": 0.8824,
+      "step": 245300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1899384615384616e-05,
+      "loss": 0.8002,
+      "step": 245400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1891692307692309e-05,
+      "loss": 0.8619,
+      "step": 245500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1884e-05,
+      "loss": 0.8653,
+      "step": 245600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1876307692307693e-05,
+      "loss": 0.8154,
+      "step": 245700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1868615384615384e-05,
+      "loss": 0.8276,
+      "step": 245800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1861e-05,
+      "loss": 0.8467,
+      "step": 245900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1853307692307693e-05,
+      "loss": 0.8226,
+      "step": 246000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1845615384615386e-05,
+      "loss": 0.8823,
+      "step": 246100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1837923076923077e-05,
+      "loss": 0.8298,
+      "step": 246200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.183023076923077e-05,
+      "loss": 0.8182,
+      "step": 246300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1822538461538461e-05,
+      "loss": 0.8211,
+      "step": 246400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1814846153846155e-05,
+      "loss": 0.8352,
+      "step": 246500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1807153846153846e-05,
+      "loss": 0.8488,
+      "step": 246600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.179946153846154e-05,
+      "loss": 0.8263,
+      "step": 246700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.179176923076923e-05,
+      "loss": 0.815,
+      "step": 246800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1784076923076923e-05,
+      "loss": 0.862,
+      "step": 246900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1776384615384614e-05,
+      "loss": 0.8494,
+      "step": 247000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1768692307692309e-05,
+      "loss": 0.8554,
+      "step": 247100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1761000000000001e-05,
+      "loss": 0.8595,
+      "step": 247200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1753307692307692e-05,
+      "loss": 0.8391,
+      "step": 247300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1745615384615385e-05,
+      "loss": 0.8272,
+      "step": 247400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1737923076923076e-05,
+      "loss": 0.8383,
+      "step": 247500
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 1.2777844667434692,
+      "eval_runtime": 132.1617,
+      "eval_samples_per_second": 15.133,
+      "eval_steps_per_second": 0.946,
+      "step": 247500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.173023076923077e-05,
+      "loss": 0.8375,
+      "step": 247600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1722538461538462e-05,
+      "loss": 0.8327,
+      "step": 247700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1714846153846155e-05,
+      "loss": 0.8172,
+      "step": 247800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1707153846153846e-05,
+      "loss": 0.8451,
+      "step": 247900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1699461538461539e-05,
+      "loss": 0.8145,
+      "step": 248000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.169176923076923e-05,
+      "loss": 0.8517,
+      "step": 248100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1684076923076924e-05,
+      "loss": 0.8758,
+      "step": 248200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1676384615384617e-05,
+      "loss": 0.8384,
+      "step": 248300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1668692307692308e-05,
+      "loss": 0.8415,
+      "step": 248400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1661076923076923e-05,
+      "loss": 0.8438,
+      "step": 248500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1653384615384616e-05,
+      "loss": 0.8542,
+      "step": 248600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1645692307692307e-05,
+      "loss": 0.8203,
+      "step": 248700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1638000000000001e-05,
+      "loss": 0.8453,
+      "step": 248800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1630384615384616e-05,
+      "loss": 0.8661,
+      "step": 248900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1622692307692309e-05,
+      "loss": 0.8928,
+      "step": 249000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1615e-05,
+      "loss": 0.8898,
+      "step": 249100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1607307692307693e-05,
+      "loss": 0.8499,
+      "step": 249200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1599615384615385e-05,
+      "loss": 0.8409,
+      "step": 249300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1591923076923078e-05,
+      "loss": 0.8634,
+      "step": 249400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1584230769230769e-05,
+      "loss": 0.8703,
+      "step": 249500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1576538461538462e-05,
+      "loss": 0.8255,
+      "step": 249600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1568846153846153e-05,
+      "loss": 0.849,
+      "step": 249700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1561153846153846e-05,
+      "loss": 0.8535,
+      "step": 249800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.155346153846154e-05,
+      "loss": 0.8543,
+      "step": 249900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1545769230769231e-05,
+      "loss": 0.8404,
+      "step": 250000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 1.2656991481781006,
+      "eval_runtime": 131.9624,
+      "eval_samples_per_second": 15.156,
+      "eval_steps_per_second": 0.947,
+      "step": 250000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1538076923076924e-05,
+      "loss": 0.8763,
+      "step": 250100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1530384615384615e-05,
+      "loss": 0.9062,
+      "step": 250200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1522692307692308e-05,
+      "loss": 0.8425,
+      "step": 250300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1515e-05,
+      "loss": 0.8324,
+      "step": 250400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1507307692307693e-05,
+      "loss": 0.829,
+      "step": 250500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1499615384615385e-05,
+      "loss": 0.8327,
+      "step": 250600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1491923076923077e-05,
+      "loss": 0.8323,
+      "step": 250700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1484230769230768e-05,
+      "loss": 0.8173,
+      "step": 250800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1476538461538461e-05,
+      "loss": 0.8419,
+      "step": 250900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1468846153846154e-05,
+      "loss": 0.8588,
+      "step": 251000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1461153846153847e-05,
+      "loss": 0.846,
+      "step": 251100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.145346153846154e-05,
+      "loss": 0.8607,
+      "step": 251200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.144576923076923e-05,
+      "loss": 0.8687,
+      "step": 251300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1438076923076923e-05,
+      "loss": 0.8407,
+      "step": 251400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1430384615384616e-05,
+      "loss": 0.8292,
+      "step": 251500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1422692307692309e-05,
+      "loss": 0.8432,
+      "step": 251600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1415e-05,
+      "loss": 0.8531,
+      "step": 251700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1407307692307693e-05,
+      "loss": 0.8585,
+      "step": 251800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1399615384615384e-05,
+      "loss": 0.8212,
+      "step": 251900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1391923076923077e-05,
+      "loss": 0.8381,
+      "step": 252000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.138423076923077e-05,
+      "loss": 0.8199,
+      "step": 252100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1376538461538462e-05,
+      "loss": 0.8571,
+      "step": 252200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1368846153846155e-05,
+      "loss": 0.8325,
+      "step": 252300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1361153846153846e-05,
+      "loss": 0.8592,
+      "step": 252400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1353461538461539e-05,
+      "loss": 0.8811,
+      "step": 252500
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 1.2745730876922607,
+      "eval_runtime": 131.6225,
+      "eval_samples_per_second": 15.195,
+      "eval_steps_per_second": 0.95,
+      "step": 252500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1345769230769232e-05,
+      "loss": 0.8395,
+      "step": 252600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1338076923076924e-05,
+      "loss": 0.8128,
+      "step": 252700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1330384615384615e-05,
+      "loss": 0.8275,
+      "step": 252800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1322692307692308e-05,
+      "loss": 0.8384,
+      "step": 252900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1315e-05,
+      "loss": 0.8237,
+      "step": 253000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1307307692307692e-05,
+      "loss": 0.8419,
+      "step": 253100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1299692307692309e-05,
+      "loss": 0.8687,
+      "step": 253200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1292000000000001e-05,
+      "loss": 0.8368,
+      "step": 253300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1284307692307692e-05,
+      "loss": 0.8444,
+      "step": 253400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1276615384615385e-05,
+      "loss": 0.8677,
+      "step": 253500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1268923076923076e-05,
+      "loss": 0.8563,
+      "step": 253600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.126123076923077e-05,
+      "loss": 0.8036,
+      "step": 253700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1253538461538462e-05,
+      "loss": 0.8466,
+      "step": 253800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1245846153846155e-05,
+      "loss": 0.8283,
+      "step": 253900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1238153846153846e-05,
+      "loss": 0.8691,
+      "step": 254000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1230461538461538e-05,
+      "loss": 0.8842,
+      "step": 254100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.122276923076923e-05,
+      "loss": 0.8451,
+      "step": 254200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1215076923076924e-05,
+      "loss": 0.8366,
+      "step": 254300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1207384615384617e-05,
+      "loss": 0.8346,
+      "step": 254400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1199692307692308e-05,
+      "loss": 0.8423,
+      "step": 254500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1192076923076923e-05,
+      "loss": 0.842,
+      "step": 254600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1184384615384615e-05,
+      "loss": 0.8268,
+      "step": 254700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1176692307692307e-05,
+      "loss": 0.8343,
+      "step": 254800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1169000000000001e-05,
+      "loss": 0.8594,
+      "step": 254900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1161307692307692e-05,
+      "loss": 0.8648,
+      "step": 255000
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 1.2777912616729736,
+      "eval_runtime": 131.9497,
+      "eval_samples_per_second": 15.157,
+      "eval_steps_per_second": 0.947,
+      "step": 255000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1153692307692309e-05,
+      "loss": 0.8455,
+      "step": 255100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1146e-05,
+      "loss": 0.8699,
+      "step": 255200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1138307692307692e-05,
+      "loss": 0.8006,
+      "step": 255300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1130615384615384e-05,
+      "loss": 0.8438,
+      "step": 255400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1122923076923078e-05,
+      "loss": 0.8497,
+      "step": 255500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1115230769230769e-05,
+      "loss": 0.8605,
+      "step": 255600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1107538461538462e-05,
+      "loss": 0.8175,
+      "step": 255700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1099846153846153e-05,
+      "loss": 0.858,
+      "step": 255800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1092153846153846e-05,
+      "loss": 0.8286,
+      "step": 255900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.108446153846154e-05,
+      "loss": 0.8447,
+      "step": 256000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1076769230769231e-05,
+      "loss": 0.8715,
+      "step": 256100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1069076923076924e-05,
+      "loss": 0.8517,
+      "step": 256200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1061384615384615e-05,
+      "loss": 0.8606,
+      "step": 256300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1053692307692308e-05,
+      "loss": 0.8084,
+      "step": 256400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1046e-05,
+      "loss": 0.821,
+      "step": 256500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1038307692307693e-05,
+      "loss": 0.8012,
+      "step": 256600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1030615384615384e-05,
+      "loss": 0.8444,
+      "step": 256700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1022923076923077e-05,
+      "loss": 0.8436,
+      "step": 256800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1015230769230768e-05,
+      "loss": 0.8445,
+      "step": 256900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1007538461538461e-05,
+      "loss": 0.8257,
+      "step": 257000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.0999846153846156e-05,
+      "loss": 0.828,
+      "step": 257100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.0992153846153847e-05,
+      "loss": 0.873,
+      "step": 257200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.098446153846154e-05,
+      "loss": 0.8437,
+      "step": 257300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.097676923076923e-05,
+      "loss": 0.8464,
+      "step": 257400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.0969076923076923e-05,
+      "loss": 0.8509,
+      "step": 257500
+    },
+    {
+      "epoch": 1.15,
+      "eval_loss": 1.2446132898330688,
+      "eval_runtime": 135.4639,
+      "eval_samples_per_second": 14.764,
+      "eval_steps_per_second": 0.923,
+      "step": 257500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0961384615384616e-05,
+      "loss": 0.8181,
+      "step": 257600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0953692307692309e-05,
+      "loss": 0.8769,
+      "step": 257700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0946e-05,
+      "loss": 0.8317,
+      "step": 257800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0938384615384616e-05,
+      "loss": 0.843,
+      "step": 257900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0930692307692308e-05,
+      "loss": 0.8486,
+      "step": 258000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0923e-05,
+      "loss": 0.8549,
+      "step": 258100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0915307692307693e-05,
+      "loss": 0.8452,
+      "step": 258200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0907615384615386e-05,
+      "loss": 0.8635,
+      "step": 258300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0899923076923077e-05,
+      "loss": 0.814,
+      "step": 258400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.089223076923077e-05,
+      "loss": 0.8066,
+      "step": 258500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.088453846153846e-05,
+      "loss": 0.8564,
+      "step": 258600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0876846153846154e-05,
+      "loss": 0.8629,
+      "step": 258700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0869153846153846e-05,
+      "loss": 0.8144,
+      "step": 258800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0861461538461539e-05,
+      "loss": 0.8562,
+      "step": 258900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0853769230769232e-05,
+      "loss": 0.8667,
+      "step": 259000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0846076923076923e-05,
+      "loss": 0.8473,
+      "step": 259100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0838384615384616e-05,
+      "loss": 0.8393,
+      "step": 259200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0830692307692308e-05,
+      "loss": 0.8128,
+      "step": 259300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0823000000000001e-05,
+      "loss": 0.8175,
+      "step": 259400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0815307692307692e-05,
+      "loss": 0.8268,
+      "step": 259500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0807615384615385e-05,
+      "loss": 0.7883,
+      "step": 259600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0799923076923076e-05,
+      "loss": 0.8665,
+      "step": 259700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.0792230769230769e-05,
+      "loss": 0.8019,
+      "step": 259800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0784538461538462e-05,
+      "loss": 0.854,
+      "step": 259900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0776846153846154e-05,
+      "loss": 0.8435,
+      "step": 260000
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 1.264352560043335,
+      "eval_runtime": 131.5627,
+      "eval_samples_per_second": 15.202,
+      "eval_steps_per_second": 0.95,
+      "step": 260000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0769153846153846e-05,
+      "loss": 0.8404,
+      "step": 260100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0761461538461538e-05,
+      "loss": 0.8431,
+      "step": 260200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0753769230769231e-05,
+      "loss": 0.8584,
+      "step": 260300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0746076923076924e-05,
+      "loss": 0.8642,
+      "step": 260400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0738384615384617e-05,
+      "loss": 0.8304,
+      "step": 260500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0730692307692308e-05,
+      "loss": 0.8504,
+      "step": 260600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0723e-05,
+      "loss": 0.8287,
+      "step": 260700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0715307692307692e-05,
+      "loss": 0.8598,
+      "step": 260800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0707615384615386e-05,
+      "loss": 0.8349,
+      "step": 260900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0699923076923077e-05,
+      "loss": 0.832,
+      "step": 261000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.069223076923077e-05,
+      "loss": 0.8454,
+      "step": 261100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0684538461538461e-05,
+      "loss": 0.8624,
+      "step": 261200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0676846153846154e-05,
+      "loss": 0.8329,
+      "step": 261300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0669153846153847e-05,
+      "loss": 0.8546,
+      "step": 261400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.066146153846154e-05,
+      "loss": 0.8463,
+      "step": 261500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0653769230769232e-05,
+      "loss": 0.8309,
+      "step": 261600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0646076923076923e-05,
+      "loss": 0.811,
+      "step": 261700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0638384615384616e-05,
+      "loss": 0.8493,
+      "step": 261800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0630692307692307e-05,
+      "loss": 0.8479,
+      "step": 261900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0623000000000001e-05,
+      "loss": 0.8364,
+      "step": 262000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0615307692307693e-05,
+      "loss": 0.8186,
+      "step": 262100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0607615384615385e-05,
+      "loss": 0.8364,
+      "step": 262200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0599923076923076e-05,
+      "loss": 0.8405,
+      "step": 262300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.059223076923077e-05,
+      "loss": 0.8524,
+      "step": 262400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.058453846153846e-05,
+      "loss": 0.8056,
+      "step": 262500
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 1.2480024099349976,
+      "eval_runtime": 133.4968,
+      "eval_samples_per_second": 14.982,
+      "eval_steps_per_second": 0.936,
+      "step": 262500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0576846153846155e-05,
+      "loss": 0.8418,
+      "step": 262600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0569153846153847e-05,
+      "loss": 0.8285,
+      "step": 262700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0561461538461539e-05,
+      "loss": 0.8065,
+      "step": 262800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0553769230769231e-05,
+      "loss": 0.8304,
+      "step": 262900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0546076923076922e-05,
+      "loss": 0.87,
+      "step": 263000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0538384615384617e-05,
+      "loss": 0.842,
+      "step": 263100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0530692307692308e-05,
+      "loss": 0.8223,
+      "step": 263200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0523e-05,
+      "loss": 0.8562,
+      "step": 263300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0515307692307692e-05,
+      "loss": 0.8629,
+      "step": 263400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0507615384615385e-05,
+      "loss": 0.8626,
+      "step": 263500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0499923076923076e-05,
+      "loss": 0.8593,
+      "step": 263600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.049223076923077e-05,
+      "loss": 0.823,
+      "step": 263700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0484538461538463e-05,
+      "loss": 0.8176,
+      "step": 263800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0476846153846154e-05,
+      "loss": 0.8179,
+      "step": 263900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0469230769230769e-05,
+      "loss": 0.8311,
+      "step": 264000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0461538461538462e-05,
+      "loss": 0.8176,
+      "step": 264100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0453846153846154e-05,
+      "loss": 0.8343,
+      "step": 264200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0446153846153847e-05,
+      "loss": 0.8463,
+      "step": 264300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0438461538461538e-05,
+      "loss": 0.8558,
+      "step": 264400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0430769230769231e-05,
+      "loss": 0.8043,
+      "step": 264500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0423076923076924e-05,
+      "loss": 0.8312,
+      "step": 264600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0415384615384615e-05,
+      "loss": 0.8352,
+      "step": 264700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.040769230769231e-05,
+      "loss": 0.845,
+      "step": 264800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.04e-05,
+      "loss": 0.8377,
+      "step": 264900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0392307692307693e-05,
+      "loss": 0.8346,
+      "step": 265000
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 1.2594469785690308,
+      "eval_runtime": 132.3194,
+      "eval_samples_per_second": 15.115,
+      "eval_steps_per_second": 0.945,
+      "step": 265000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0384692307692308e-05,
+      "loss": 0.8371,
+      "step": 265100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0376999999999999e-05,
+      "loss": 0.8441,
+      "step": 265200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0369307692307692e-05,
+      "loss": 0.8486,
+      "step": 265300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0361615384615386e-05,
+      "loss": 0.8528,
+      "step": 265400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0353923076923077e-05,
+      "loss": 0.8234,
+      "step": 265500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.034623076923077e-05,
+      "loss": 0.8463,
+      "step": 265600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0338538461538461e-05,
+      "loss": 0.8108,
+      "step": 265700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0330846153846154e-05,
+      "loss": 0.8216,
+      "step": 265800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0323153846153847e-05,
+      "loss": 0.8409,
+      "step": 265900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.031546153846154e-05,
+      "loss": 0.8039,
+      "step": 266000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.030776923076923e-05,
+      "loss": 0.8413,
+      "step": 266100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0300076923076923e-05,
+      "loss": 0.833,
+      "step": 266200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0292384615384614e-05,
+      "loss": 0.8211,
+      "step": 266300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0284692307692307e-05,
+      "loss": 0.8131,
+      "step": 266400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0277e-05,
+      "loss": 0.885,
+      "step": 266500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0269307692307693e-05,
+      "loss": 0.8657,
+      "step": 266600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0261615384615386e-05,
+      "loss": 0.8684,
+      "step": 266700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0253923076923077e-05,
+      "loss": 0.8108,
+      "step": 266800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.024623076923077e-05,
+      "loss": 0.8542,
+      "step": 266900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0238538461538462e-05,
+      "loss": 0.8512,
+      "step": 267000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0230846153846155e-05,
+      "loss": 0.8843,
+      "step": 267100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0223153846153846e-05,
+      "loss": 0.7962,
+      "step": 267200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0215461538461539e-05,
+      "loss": 0.8364,
+      "step": 267300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.020776923076923e-05,
+      "loss": 0.807,
+      "step": 267400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0200076923076923e-05,
+      "loss": 0.8415,
+      "step": 267500
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.247600793838501,
+      "eval_runtime": 132.3149,
+      "eval_samples_per_second": 15.115,
+      "eval_steps_per_second": 0.945,
+      "step": 267500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0192384615384615e-05,
+      "loss": 0.8134,
+      "step": 267600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0184692307692308e-05,
+      "loss": 0.818,
+      "step": 267700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0177000000000001e-05,
+      "loss": 0.8388,
+      "step": 267800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0169307692307692e-05,
+      "loss": 0.834,
+      "step": 267900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0161615384615385e-05,
+      "loss": 0.8437,
+      "step": 268000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0153923076923078e-05,
+      "loss": 0.8144,
+      "step": 268100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.014623076923077e-05,
+      "loss": 0.8656,
+      "step": 268200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0138538461538461e-05,
+      "loss": 0.8494,
+      "step": 268300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0130846153846154e-05,
+      "loss": 0.8117,
+      "step": 268400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0123153846153845e-05,
+      "loss": 0.8337,
+      "step": 268500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.011546153846154e-05,
+      "loss": 0.8401,
+      "step": 268600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0107769230769231e-05,
+      "loss": 0.8419,
+      "step": 268700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0100076923076924e-05,
+      "loss": 0.8246,
+      "step": 268800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0092384615384615e-05,
+      "loss": 0.8571,
+      "step": 268900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0084692307692307e-05,
+      "loss": 0.8436,
+      "step": 269000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0077e-05,
+      "loss": 0.8099,
+      "step": 269100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0069307692307693e-05,
+      "loss": 0.8256,
+      "step": 269200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0061692307692308e-05,
+      "loss": 0.8189,
+      "step": 269300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0054e-05,
+      "loss": 0.8501,
+      "step": 269400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0046307692307692e-05,
+      "loss": 0.852,
+      "step": 269500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0038615384615384e-05,
+      "loss": 0.8077,
+      "step": 269600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0030923076923077e-05,
+      "loss": 0.8478,
+      "step": 269700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.002323076923077e-05,
+      "loss": 0.8442,
+      "step": 269800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0015538461538463e-05,
+      "loss": 0.8051,
+      "step": 269900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0007846153846154e-05,
+      "loss": 0.8378,
+      "step": 270000
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 1.2303105592727661,
+      "eval_runtime": 132.2431,
+      "eval_samples_per_second": 15.124,
+      "eval_steps_per_second": 0.945,
+      "step": 270000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0000153846153847e-05,
+      "loss": 0.8169,
+      "step": 270100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.992461538461538e-06,
+      "loss": 0.836,
+      "step": 270200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.984769230769232e-06,
+      "loss": 0.8578,
+      "step": 270300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.977076923076923e-06,
+      "loss": 0.8534,
+      "step": 270400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.969384615384616e-06,
+      "loss": 0.8648,
+      "step": 270500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.961692307692307e-06,
+      "loss": 0.8372,
+      "step": 270600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.954e-06,
+      "loss": 0.8123,
+      "step": 270700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.946307692307691e-06,
+      "loss": 0.8504,
+      "step": 270800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.938615384615385e-06,
+      "loss": 0.8391,
+      "step": 270900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.930923076923078e-06,
+      "loss": 0.8425,
+      "step": 271000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.92323076923077e-06,
+      "loss": 0.8434,
+      "step": 271100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.915538461538462e-06,
+      "loss": 0.8297,
+      "step": 271200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.907846153846153e-06,
+      "loss": 0.8623,
+      "step": 271300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.900153846153848e-06,
+      "loss": 0.8318,
+      "step": 271400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.892461538461539e-06,
+      "loss": 0.8403,
+      "step": 271500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.884769230769231e-06,
+      "loss": 0.8157,
+      "step": 271600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.877076923076923e-06,
+      "loss": 0.8414,
+      "step": 271700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.869384615384615e-06,
+      "loss": 0.8016,
+      "step": 271800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.861692307692306e-06,
+      "loss": 0.8406,
+      "step": 271900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.854000000000001e-06,
+      "loss": 0.8774,
+      "step": 272000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.846307692307694e-06,
+      "loss": 0.8421,
+      "step": 272100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.838615384615385e-06,
+      "loss": 0.8481,
+      "step": 272200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.830923076923077e-06,
+      "loss": 0.8365,
+      "step": 272300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.823230769230769e-06,
+      "loss": 0.8358,
+      "step": 272400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.815538461538463e-06,
+      "loss": 0.8511,
+      "step": 272500
+    },
+    {
+      "epoch": 1.22,
+      "eval_loss": 1.2428909540176392,
+      "eval_runtime": 131.8398,
+      "eval_samples_per_second": 15.17,
+      "eval_steps_per_second": 0.948,
+      "step": 272500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.807846153846154e-06,
+      "loss": 0.8327,
+      "step": 272600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.800153846153847e-06,
+      "loss": 0.8535,
+      "step": 272700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.792461538461538e-06,
+      "loss": 0.837,
+      "step": 272800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.78476923076923e-06,
+      "loss": 0.8512,
+      "step": 272900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.777076923076924e-06,
+      "loss": 0.8463,
+      "step": 273000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.769384615384616e-06,
+      "loss": 0.8311,
+      "step": 273100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.761692307692307e-06,
+      "loss": 0.833,
+      "step": 273200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754e-06,
+      "loss": 0.8039,
+      "step": 273300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.746384615384615e-06,
+      "loss": 0.8388,
+      "step": 273400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.738692307692308e-06,
+      "loss": 0.8554,
+      "step": 273500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.731e-06,
+      "loss": 0.7997,
+      "step": 273600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.723307692307693e-06,
+      "loss": 0.8099,
+      "step": 273700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.715615384615384e-06,
+      "loss": 0.808,
+      "step": 273800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.707923076923077e-06,
+      "loss": 0.8608,
+      "step": 273900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.700230769230768e-06,
+      "loss": 0.8589,
+      "step": 274000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.692538461538461e-06,
+      "loss": 0.8446,
+      "step": 274100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.684846153846155e-06,
+      "loss": 0.8374,
+      "step": 274200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.677153846153847e-06,
+      "loss": 0.8538,
+      "step": 274300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.669538461538461e-06,
+      "loss": 0.8084,
+      "step": 274400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.661846153846154e-06,
+      "loss": 0.8323,
+      "step": 274500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.654153846153845e-06,
+      "loss": 0.8617,
+      "step": 274600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.646461538461538e-06,
+      "loss": 0.8182,
+      "step": 274700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.63876923076923e-06,
+      "loss": 0.8756,
+      "step": 274800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.631076923076924e-06,
+      "loss": 0.8647,
+      "step": 274900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.623384615384616e-06,
+      "loss": 0.8477,
+      "step": 275000
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 1.2365238666534424,
+      "eval_runtime": 132.0791,
+      "eval_samples_per_second": 15.142,
+      "eval_steps_per_second": 0.946,
+      "step": 275000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.615692307692307e-06,
+      "loss": 0.8501,
+      "step": 275100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.608e-06,
+      "loss": 0.8341,
+      "step": 275200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.600307692307693e-06,
+      "loss": 0.865,
+      "step": 275300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.592615384615386e-06,
+      "loss": 0.8129,
+      "step": 275400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.584923076923077e-06,
+      "loss": 0.818,
+      "step": 275500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.57723076923077e-06,
+      "loss": 0.8075,
+      "step": 275600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.56953846153846e-06,
+      "loss": 0.8413,
+      "step": 275700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.561846153846155e-06,
+      "loss": 0.7984,
+      "step": 275800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.554153846153846e-06,
+      "loss": 0.7835,
+      "step": 275900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.546461538461539e-06,
+      "loss": 0.808,
+      "step": 276000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.538769230769232e-06,
+      "loss": 0.8283,
+      "step": 276100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.531076923076923e-06,
+      "loss": 0.8863,
+      "step": 276200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.523384615384616e-06,
+      "loss": 0.8156,
+      "step": 276300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.515692307692308e-06,
+      "loss": 0.8332,
+      "step": 276400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.508076923076923e-06,
+      "loss": 0.821,
+      "step": 276500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.500384615384616e-06,
+      "loss": 0.8313,
+      "step": 276600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.492692307692307e-06,
+      "loss": 0.8366,
+      "step": 276700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.485e-06,
+      "loss": 0.8322,
+      "step": 276800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.477307692307693e-06,
+      "loss": 0.8795,
+      "step": 276900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.469615384615385e-06,
+      "loss": 0.8227,
+      "step": 277000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.461923076923078e-06,
+      "loss": 0.8388,
+      "step": 277100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.45423076923077e-06,
+      "loss": 0.8575,
+      "step": 277200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.446538461538462e-06,
+      "loss": 0.8394,
+      "step": 277300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.438846153846153e-06,
+      "loss": 0.7968,
+      "step": 277400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.431153846153848e-06,
+      "loss": 0.8303,
+      "step": 277500
+    },
+    {
+      "epoch": 1.24,
+      "eval_loss": 1.2384347915649414,
+      "eval_runtime": 135.3439,
+      "eval_samples_per_second": 14.777,
+      "eval_steps_per_second": 0.924,
+      "step": 277500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.423461538461539e-06,
+      "loss": 0.8643,
+      "step": 277600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.415769230769231e-06,
+      "loss": 0.822,
+      "step": 277700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.408076923076922e-06,
+      "loss": 0.861,
+      "step": 277800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.400384615384615e-06,
+      "loss": 0.8249,
+      "step": 277900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.392692307692306e-06,
+      "loss": 0.8441,
+      "step": 278000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.385e-06,
+      "loss": 0.8705,
+      "step": 278100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.377307692307694e-06,
+      "loss": 0.8769,
+      "step": 278200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.369615384615385e-06,
+      "loss": 0.7936,
+      "step": 278300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.361923076923077e-06,
+      "loss": 0.8166,
+      "step": 278400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.354230769230768e-06,
+      "loss": 0.8471,
+      "step": 278500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.346538461538463e-06,
+      "loss": 0.8313,
+      "step": 278600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.338846153846154e-06,
+      "loss": 0.8408,
+      "step": 278700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.331153846153847e-06,
+      "loss": 0.8333,
+      "step": 278800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.323461538461538e-06,
+      "loss": 0.8407,
+      "step": 278900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.31576923076923e-06,
+      "loss": 0.8305,
+      "step": 279000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.308076923076922e-06,
+      "loss": 0.8096,
+      "step": 279100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.300384615384616e-06,
+      "loss": 0.8711,
+      "step": 279200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.292692307692309e-06,
+      "loss": 0.8008,
+      "step": 279300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.285e-06,
+      "loss": 0.8109,
+      "step": 279400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.277307692307693e-06,
+      "loss": 0.8444,
+      "step": 279500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.269615384615384e-06,
+      "loss": 0.8363,
+      "step": 279600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.261923076923078e-06,
+      "loss": 0.8553,
+      "step": 279700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.25423076923077e-06,
+      "loss": 0.8379,
+      "step": 279800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.246538461538462e-06,
+      "loss": 0.8295,
+      "step": 279900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.238846153846153e-06,
+      "loss": 0.809,
+      "step": 280000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 1.240295648574829,
+      "eval_runtime": 132.8395,
+      "eval_samples_per_second": 15.056,
+      "eval_steps_per_second": 0.941,
+      "step": 280000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.231153846153846e-06,
+      "loss": 0.8441,
+      "step": 280100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.223461538461539e-06,
+      "loss": 0.844,
+      "step": 280200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.215769230769232e-06,
+      "loss": 0.8067,
+      "step": 280300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.208076923076923e-06,
+      "loss": 0.8038,
+      "step": 280400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.20046153846154e-06,
+      "loss": 0.824,
+      "step": 280500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.19276923076923e-06,
+      "loss": 0.8303,
+      "step": 280600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.185076923076923e-06,
+      "loss": 0.8351,
+      "step": 280700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.177384615384616e-06,
+      "loss": 0.8534,
+      "step": 280800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.169692307692309e-06,
+      "loss": 0.8226,
+      "step": 280900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.162e-06,
+      "loss": 0.8334,
+      "step": 281000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.154307692307692e-06,
+      "loss": 0.7293,
+      "step": 281100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.146615384615384e-06,
+      "loss": 0.8203,
+      "step": 281200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.139e-06,
+      "loss": 0.8609,
+      "step": 281300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.131307692307693e-06,
+      "loss": 0.8492,
+      "step": 281400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.123615384615386e-06,
+      "loss": 0.888,
+      "step": 281500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.115923076923077e-06,
+      "loss": 0.8829,
+      "step": 281600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.10823076923077e-06,
+      "loss": 0.832,
+      "step": 281700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.10053846153846e-06,
+      "loss": 0.8616,
+      "step": 281800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.092846153846153e-06,
+      "loss": 0.7989,
+      "step": 281900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.085153846153846e-06,
+      "loss": 0.8387,
+      "step": 282000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.077461538461539e-06,
+      "loss": 0.7876,
+      "step": 282100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.069769230769232e-06,
+      "loss": 0.8333,
+      "step": 282200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.062076923076923e-06,
+      "loss": 0.8173,
+      "step": 282300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.054384615384615e-06,
+      "loss": 0.8484,
+      "step": 282400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.046692307692308e-06,
+      "loss": 0.838,
+      "step": 282500
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 1.2172142267227173,
+      "eval_runtime": 131.9496,
+      "eval_samples_per_second": 15.157,
+      "eval_steps_per_second": 0.947,
+      "step": 282500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.039000000000001e-06,
+      "loss": 0.8249,
+      "step": 282600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.031307692307692e-06,
+      "loss": 0.8048,
+      "step": 282700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.023615384615385e-06,
+      "loss": 0.8655,
+      "step": 282800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.015923076923076e-06,
+      "loss": 0.84,
+      "step": 282900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.00823076923077e-06,
+      "loss": 0.8156,
+      "step": 283000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.000538461538462e-06,
+      "loss": 0.8358,
+      "step": 283100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.992846153846154e-06,
+      "loss": 0.823,
+      "step": 283200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.985153846153847e-06,
+      "loss": 0.7899,
+      "step": 283300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.977461538461538e-06,
+      "loss": 0.8758,
+      "step": 283400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.969769230769231e-06,
+      "loss": 0.8139,
+      "step": 283500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.962076923076924e-06,
+      "loss": 0.8401,
+      "step": 283600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.954384615384616e-06,
+      "loss": 0.802,
+      "step": 283700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.946692307692308e-06,
+      "loss": 0.8225,
+      "step": 283800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.939e-06,
+      "loss": 0.8581,
+      "step": 283900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.931307692307691e-06,
+      "loss": 0.8339,
+      "step": 284000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.923615384615386e-06,
+      "loss": 0.8255,
+      "step": 284100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.915923076923077e-06,
+      "loss": 0.8026,
+      "step": 284200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.90823076923077e-06,
+      "loss": 0.8084,
+      "step": 284300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.900538461538462e-06,
+      "loss": 0.8541,
+      "step": 284400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.892846153846154e-06,
+      "loss": 0.8272,
+      "step": 284500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.885153846153846e-06,
+      "loss": 0.8428,
+      "step": 284600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.877461538461539e-06,
+      "loss": 0.8365,
+      "step": 284700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.869769230769232e-06,
+      "loss": 0.8622,
+      "step": 284800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.862153846153847e-06,
+      "loss": 0.8484,
+      "step": 284900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.854461538461538e-06,
+      "loss": 0.8136,
+      "step": 285000
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 1.2236456871032715,
+      "eval_runtime": 131.8674,
+      "eval_samples_per_second": 15.167,
+      "eval_steps_per_second": 0.948,
+      "step": 285000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.84676923076923e-06,
+      "loss": 0.8562,
+      "step": 285100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.839076923076923e-06,
+      "loss": 0.7899,
+      "step": 285200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.831384615384616e-06,
+      "loss": 0.8076,
+      "step": 285300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.823692307692309e-06,
+      "loss": 0.8109,
+      "step": 285400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.816076923076924e-06,
+      "loss": 0.7985,
+      "step": 285500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.808384615384615e-06,
+      "loss": 0.8183,
+      "step": 285600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.800692307692308e-06,
+      "loss": 0.8726,
+      "step": 285700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.793e-06,
+      "loss": 0.8252,
+      "step": 285800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.785307692307693e-06,
+      "loss": 0.8465,
+      "step": 285900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.777615384615384e-06,
+      "loss": 0.847,
+      "step": 286000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.769923076923077e-06,
+      "loss": 0.8319,
+      "step": 286100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.76223076923077e-06,
+      "loss": 0.8185,
+      "step": 286200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.75453846153846e-06,
+      "loss": 0.7928,
+      "step": 286300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.746846153846155e-06,
+      "loss": 0.8266,
+      "step": 286400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.739153846153846e-06,
+      "loss": 0.8326,
+      "step": 286500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.731461538461539e-06,
+      "loss": 0.8282,
+      "step": 286600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.72376923076923e-06,
+      "loss": 0.8282,
+      "step": 286700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.716076923076923e-06,
+      "loss": 0.8377,
+      "step": 286800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.708384615384616e-06,
+      "loss": 0.8031,
+      "step": 286900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.700692307692309e-06,
+      "loss": 0.8183,
+      "step": 287000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.693e-06,
+      "loss": 0.8163,
+      "step": 287100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.685307692307692e-06,
+      "loss": 0.8643,
+      "step": 287200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.677615384615385e-06,
+      "loss": 0.8284,
+      "step": 287300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.669923076923076e-06,
+      "loss": 0.8347,
+      "step": 287400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.66223076923077e-06,
+      "loss": 0.8114,
+      "step": 287500
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 1.234848141670227,
+      "eval_runtime": 131.7418,
+      "eval_samples_per_second": 15.181,
+      "eval_steps_per_second": 0.949,
+      "step": 287500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.654538461538462e-06,
+      "loss": 0.8213,
+      "step": 287600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.646846153846155e-06,
+      "loss": 0.8449,
+      "step": 287700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.639153846153846e-06,
+      "loss": 0.8052,
+      "step": 287800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.631461538461538e-06,
+      "loss": 0.8075,
+      "step": 287900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.623769230769231e-06,
+      "loss": 0.8404,
+      "step": 288000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.616076923076924e-06,
+      "loss": 0.8674,
+      "step": 288100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.608384615384615e-06,
+      "loss": 0.8285,
+      "step": 288200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.600692307692308e-06,
+      "loss": 0.7849,
+      "step": 288300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.593e-06,
+      "loss": 0.8722,
+      "step": 288400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.585307692307692e-06,
+      "loss": 0.7875,
+      "step": 288500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.577615384615386e-06,
+      "loss": 0.8296,
+      "step": 288600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.569923076923077e-06,
+      "loss": 0.8552,
+      "step": 288700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.56223076923077e-06,
+      "loss": 0.8704,
+      "step": 288800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.554538461538461e-06,
+      "loss": 0.8443,
+      "step": 288900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.546846153846154e-06,
+      "loss": 0.8046,
+      "step": 289000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.539153846153847e-06,
+      "loss": 0.7961,
+      "step": 289100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.53146153846154e-06,
+      "loss": 0.8262,
+      "step": 289200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.52376923076923e-06,
+      "loss": 0.8524,
+      "step": 289300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.516076923076923e-06,
+      "loss": 0.8373,
+      "step": 289400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.508384615384614e-06,
+      "loss": 0.8763,
+      "step": 289500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.500692307692307e-06,
+      "loss": 0.8117,
+      "step": 289600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.493000000000002e-06,
+      "loss": 0.825,
+      "step": 289700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.485307692307693e-06,
+      "loss": 0.8042,
+      "step": 289800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.477615384615385e-06,
+      "loss": 0.8371,
+      "step": 289900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.469923076923076e-06,
+      "loss": 0.8035,
+      "step": 290000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 1.2349365949630737,
+      "eval_runtime": 132.0671,
+      "eval_samples_per_second": 15.144,
+      "eval_steps_per_second": 0.946,
+      "step": 290000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.462307692307691e-06,
+      "loss": 0.8331,
+      "step": 290100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.454615384615386e-06,
+      "loss": 0.8533,
+      "step": 290200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.446923076923077e-06,
+      "loss": 0.8315,
+      "step": 290300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.43923076923077e-06,
+      "loss": 0.8192,
+      "step": 290400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.431538461538462e-06,
+      "loss": 0.8056,
+      "step": 290500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.423923076923077e-06,
+      "loss": 0.818,
+      "step": 290600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.416230769230768e-06,
+      "loss": 0.8435,
+      "step": 290700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.408538461538463e-06,
+      "loss": 0.7893,
+      "step": 290800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.400846153846154e-06,
+      "loss": 0.829,
+      "step": 290900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.393153846153847e-06,
+      "loss": 0.7861,
+      "step": 291000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.385461538461538e-06,
+      "loss": 0.8119,
+      "step": 291100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.37776923076923e-06,
+      "loss": 0.7994,
+      "step": 291200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.370076923076923e-06,
+      "loss": 0.8287,
+      "step": 291300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.362384615384616e-06,
+      "loss": 0.844,
+      "step": 291400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.354692307692309e-06,
+      "loss": 0.8067,
+      "step": 291500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.347e-06,
+      "loss": 0.7906,
+      "step": 291600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.339307692307693e-06,
+      "loss": 0.8305,
+      "step": 291700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.331615384615384e-06,
+      "loss": 0.8132,
+      "step": 291800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.323923076923078e-06,
+      "loss": 0.8139,
+      "step": 291900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.31623076923077e-06,
+      "loss": 0.8258,
+      "step": 292000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.308538461538462e-06,
+      "loss": 0.8306,
+      "step": 292100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.300846153846153e-06,
+      "loss": 0.8444,
+      "step": 292200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.293153846153846e-06,
+      "loss": 0.8257,
+      "step": 292300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.285461538461539e-06,
+      "loss": 0.8771,
+      "step": 292400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.277769230769231e-06,
+      "loss": 0.8474,
+      "step": 292500
+    },
+    {
+      "epoch": 1.31,
+      "eval_loss": 1.2180542945861816,
+      "eval_runtime": 131.5314,
+      "eval_samples_per_second": 15.205,
+      "eval_steps_per_second": 0.95,
+      "step": 292500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.270076923076924e-06,
+      "loss": 0.7949,
+      "step": 292600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.262384615384615e-06,
+      "loss": 0.8166,
+      "step": 292700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.254692307692308e-06,
+      "loss": 0.7694,
+      "step": 292800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.246999999999999e-06,
+      "loss": 0.8173,
+      "step": 292900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.239307692307694e-06,
+      "loss": 0.8346,
+      "step": 293000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.231615384615385e-06,
+      "loss": 0.8258,
+      "step": 293100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.223923076923077e-06,
+      "loss": 0.8358,
+      "step": 293200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.216230769230769e-06,
+      "loss": 0.8057,
+      "step": 293300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.208538461538461e-06,
+      "loss": 0.8307,
+      "step": 293400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.200846153846154e-06,
+      "loss": 0.81,
+      "step": 293500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.193153846153847e-06,
+      "loss": 0.7997,
+      "step": 293600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.18546153846154e-06,
+      "loss": 0.8358,
+      "step": 293700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.17776923076923e-06,
+      "loss": 0.8113,
+      "step": 293800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.170076923076923e-06,
+      "loss": 0.8182,
+      "step": 293900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.162384615384615e-06,
+      "loss": 0.8563,
+      "step": 294000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.154692307692309e-06,
+      "loss": 0.8024,
+      "step": 294100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.147e-06,
+      "loss": 0.814,
+      "step": 294200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.139307692307693e-06,
+      "loss": 0.8462,
+      "step": 294300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.131615384615384e-06,
+      "loss": 0.8162,
+      "step": 294400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.123923076923077e-06,
+      "loss": 0.8075,
+      "step": 294500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.11623076923077e-06,
+      "loss": 0.8394,
+      "step": 294600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.108538461538462e-06,
+      "loss": 0.8189,
+      "step": 294700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.100846153846155e-06,
+      "loss": 0.7869,
+      "step": 294800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.093153846153846e-06,
+      "loss": 0.8603,
+      "step": 294900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.085461538461539e-06,
+      "loss": 0.8382,
+      "step": 295000
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 1.2197514772415161,
+      "eval_runtime": 131.6668,
+      "eval_samples_per_second": 15.19,
+      "eval_steps_per_second": 0.949,
+      "step": 295000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.07776923076923e-06,
+      "loss": 0.8162,
+      "step": 295100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.070076923076924e-06,
+      "loss": 0.8301,
+      "step": 295200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.062384615384616e-06,
+      "loss": 0.8308,
+      "step": 295300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.054692307692308e-06,
+      "loss": 0.8088,
+      "step": 295400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.047e-06,
+      "loss": 0.8523,
+      "step": 295500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.039307692307692e-06,
+      "loss": 0.8348,
+      "step": 295600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.031615384615385e-06,
+      "loss": 0.8338,
+      "step": 295700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.023923076923078e-06,
+      "loss": 0.8428,
+      "step": 295800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.016230769230769e-06,
+      "loss": 0.8308,
+      "step": 295900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.008538461538462e-06,
+      "loss": 0.8215,
+      "step": 296000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.000846153846154e-06,
+      "loss": 0.834,
+      "step": 296100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.993153846153845e-06,
+      "loss": 0.8268,
+      "step": 296200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.985538461538462e-06,
+      "loss": 0.8375,
+      "step": 296300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.977846153846155e-06,
+      "loss": 0.847,
+      "step": 296400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.970153846153846e-06,
+      "loss": 0.8349,
+      "step": 296500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.962461538461539e-06,
+      "loss": 0.8599,
+      "step": 296600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.95476923076923e-06,
+      "loss": 0.8632,
+      "step": 296700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.947076923076922e-06,
+      "loss": 0.8191,
+      "step": 296800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.939384615384617e-06,
+      "loss": 0.8395,
+      "step": 296900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.931692307692308e-06,
+      "loss": 0.8435,
+      "step": 297000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.924e-06,
+      "loss": 0.8071,
+      "step": 297100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.916307692307692e-06,
+      "loss": 0.7951,
+      "step": 297200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.908615384615385e-06,
+      "loss": 0.8132,
+      "step": 297300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.900923076923077e-06,
+      "loss": 0.8423,
+      "step": 297400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.89323076923077e-06,
+      "loss": 0.8229,
+      "step": 297500
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 1.2216039896011353,
+      "eval_runtime": 132.2441,
+      "eval_samples_per_second": 15.124,
+      "eval_steps_per_second": 0.945,
+      "step": 297500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.885538461538461e-06,
+      "loss": 0.8117,
+      "step": 297600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.877846153846154e-06,
+      "loss": 0.8595,
+      "step": 297700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.870153846153845e-06,
+      "loss": 0.8167,
+      "step": 297800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.86246153846154e-06,
+      "loss": 0.833,
+      "step": 297900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.854769230769232e-06,
+      "loss": 0.8292,
+      "step": 298000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.847153846153847e-06,
+      "loss": 0.8178,
+      "step": 298100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.839461538461538e-06,
+      "loss": 0.8452,
+      "step": 298200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.831769230769231e-06,
+      "loss": 0.7997,
+      "step": 298300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.824076923076922e-06,
+      "loss": 0.7769,
+      "step": 298400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.816384615384617e-06,
+      "loss": 0.8452,
+      "step": 298500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.808692307692308e-06,
+      "loss": 0.8176,
+      "step": 298600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.801e-06,
+      "loss": 0.8232,
+      "step": 298700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.793307692307693e-06,
+      "loss": 0.8013,
+      "step": 298800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.785615384615384e-06,
+      "loss": 0.7823,
+      "step": 298900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.777923076923077e-06,
+      "loss": 0.82,
+      "step": 299000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.77023076923077e-06,
+      "loss": 0.8044,
+      "step": 299100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.762538461538463e-06,
+      "loss": 0.8419,
+      "step": 299200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.754846153846154e-06,
+      "loss": 0.7954,
+      "step": 299300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.747153846153846e-06,
+      "loss": 0.8211,
+      "step": 299400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.739461538461537e-06,
+      "loss": 0.8726,
+      "step": 299500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.731769230769232e-06,
+      "loss": 0.8393,
+      "step": 299600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.724076923076923e-06,
+      "loss": 0.8365,
+      "step": 299700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.716384615384616e-06,
+      "loss": 0.816,
+      "step": 299800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.708692307692309e-06,
+      "loss": 0.8449,
+      "step": 299900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.701e-06,
+      "loss": 0.8227,
+      "step": 300000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 1.222668170928955,
+      "eval_runtime": 135.4139,
+      "eval_samples_per_second": 14.77,
+      "eval_steps_per_second": 0.923,
+      "step": 300000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.693307692307692e-06,
+      "loss": 0.8181,
+      "step": 300100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.685615384615385e-06,
+      "loss": 0.8225,
+      "step": 300200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.677923076923078e-06,
+      "loss": 0.8031,
+      "step": 300300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.670230769230769e-06,
+      "loss": 0.8507,
+      "step": 300400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.662538461538462e-06,
+      "loss": 0.831,
+      "step": 300500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.654846153846153e-06,
+      "loss": 0.7881,
+      "step": 300600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.647153846153847e-06,
+      "loss": 0.848,
+      "step": 300700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.639461538461538e-06,
+      "loss": 0.8258,
+      "step": 300800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.631769230769231e-06,
+      "loss": 0.7829,
+      "step": 300900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.624076923076923e-06,
+      "loss": 0.8514,
+      "step": 301000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.616384615384615e-06,
+      "loss": 0.8145,
+      "step": 301100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.608692307692307e-06,
+      "loss": 0.7871,
+      "step": 301200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.601000000000001e-06,
+      "loss": 0.8357,
+      "step": 301300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5933076923076925e-06,
+      "loss": 0.804,
+      "step": 301400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5856153846153844e-06,
+      "loss": 0.8161,
+      "step": 301500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.577923076923076e-06,
+      "loss": 0.8227,
+      "step": 301600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.570230769230769e-06,
+      "loss": 0.7941,
+      "step": 301700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.562538461538463e-06,
+      "loss": 0.8171,
+      "step": 301800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.554846153846155e-06,
+      "loss": 0.8255,
+      "step": 301900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.547153846153847e-06,
+      "loss": 0.7844,
+      "step": 302000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5395384615384614e-06,
+      "loss": 0.84,
+      "step": 302100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.531846153846153e-06,
+      "loss": 0.8406,
+      "step": 302200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.524153846153845e-06,
+      "loss": 0.8144,
+      "step": 302300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.516461538461539e-06,
+      "loss": 0.8066,
+      "step": 302400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.508769230769232e-06,
+      "loss": 0.8239,
+      "step": 302500
+    },
+    {
+      "epoch": 1.36,
+      "eval_loss": 1.2094714641571045,
+      "eval_runtime": 131.7264,
+      "eval_samples_per_second": 15.183,
+      "eval_steps_per_second": 0.949,
+      "step": 302500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.501076923076924e-06,
+      "loss": 0.8081,
+      "step": 302600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4933846153846155e-06,
+      "loss": 0.8131,
+      "step": 302700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.485692307692308e-06,
+      "loss": 0.7914,
+      "step": 302800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.478e-06,
+      "loss": 0.8708,
+      "step": 302900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.470384615384615e-06,
+      "loss": 0.8146,
+      "step": 303000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.462692307692308e-06,
+      "loss": 0.8356,
+      "step": 303100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.455e-06,
+      "loss": 0.8154,
+      "step": 303200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4473076923076925e-06,
+      "loss": 0.7952,
+      "step": 303300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.439615384615385e-06,
+      "loss": 0.8127,
+      "step": 303400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.431923076923077e-06,
+      "loss": 0.82,
+      "step": 303500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.42423076923077e-06,
+      "loss": 0.8218,
+      "step": 303600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.416538461538462e-06,
+      "loss": 0.8144,
+      "step": 303700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.408846153846154e-06,
+      "loss": 0.8038,
+      "step": 303800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.401153846153847e-06,
+      "loss": 0.8266,
+      "step": 303900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3934615384615386e-06,
+      "loss": 0.8173,
+      "step": 304000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3857692307692305e-06,
+      "loss": 0.8139,
+      "step": 304100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.378076923076923e-06,
+      "loss": 0.8606,
+      "step": 304200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.370384615384615e-06,
+      "loss": 0.8594,
+      "step": 304300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.362692307692307e-06,
+      "loss": 0.8139,
+      "step": 304400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.355000000000001e-06,
+      "loss": 0.7841,
+      "step": 304500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.347307692307693e-06,
+      "loss": 0.8213,
+      "step": 304600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.3396153846153854e-06,
+      "loss": 0.8373,
+      "step": 304700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.331923076923077e-06,
+      "loss": 0.8428,
+      "step": 304800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.324230769230769e-06,
+      "loss": 0.8144,
+      "step": 304900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.316538461538462e-06,
+      "loss": 0.8247,
+      "step": 305000
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 1.2055754661560059,
+      "eval_runtime": 132.297,
+      "eval_samples_per_second": 15.118,
+      "eval_steps_per_second": 0.945,
+      "step": 305000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.308846153846154e-06,
+      "loss": 0.8651,
+      "step": 305100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.301153846153846e-06,
+      "loss": 0.8507,
+      "step": 305200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.293461538461539e-06,
+      "loss": 0.8517,
+      "step": 305300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.285769230769231e-06,
+      "loss": 0.8279,
+      "step": 305400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.278153846153846e-06,
+      "loss": 0.8398,
+      "step": 305500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.270461538461539e-06,
+      "loss": 0.8008,
+      "step": 305600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.262769230769231e-06,
+      "loss": 0.8279,
+      "step": 305700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.255076923076923e-06,
+      "loss": 0.8199,
+      "step": 305800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.247384615384616e-06,
+      "loss": 0.8192,
+      "step": 305900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.239692307692308e-06,
+      "loss": 0.829,
+      "step": 306000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.232076923076923e-06,
+      "loss": 0.8303,
+      "step": 306100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.224384615384616e-06,
+      "loss": 0.7759,
+      "step": 306200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.216692307692308e-06,
+      "loss": 0.8482,
+      "step": 306300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.209000000000001e-06,
+      "loss": 0.8269,
+      "step": 306400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.201307692307693e-06,
+      "loss": 0.8401,
+      "step": 306500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.193615384615385e-06,
+      "loss": 0.8355,
+      "step": 306600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.186076923076923e-06,
+      "loss": 0.853,
+      "step": 306700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.178384615384616e-06,
+      "loss": 0.8039,
+      "step": 306800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.170692307692308e-06,
+      "loss": 0.8468,
+      "step": 306900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.163e-06,
+      "loss": 0.8146,
+      "step": 307000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.155307692307693e-06,
+      "loss": 0.8371,
+      "step": 307100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.1476153846153846e-06,
+      "loss": 0.8299,
+      "step": 307200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.14e-06,
+      "loss": 0.8436,
+      "step": 307300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.132307692307693e-06,
+      "loss": 0.8287,
+      "step": 307400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.124615384615385e-06,
+      "loss": 0.817,
+      "step": 307500
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 1.194072961807251,
+      "eval_runtime": 131.6666,
+      "eval_samples_per_second": 15.19,
+      "eval_steps_per_second": 0.949,
+      "step": 307500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.116923076923077e-06,
+      "loss": 0.8225,
+      "step": 307600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.10923076923077e-06,
+      "loss": 0.8144,
+      "step": 307700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.1015384615384616e-06,
+      "loss": 0.8126,
+      "step": 307800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.093846153846154e-06,
+      "loss": 0.8262,
+      "step": 307900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.086153846153846e-06,
+      "loss": 0.8235,
+      "step": 308000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.078461538461538e-06,
+      "loss": 0.8518,
+      "step": 308100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.070769230769231e-06,
+      "loss": 0.8463,
+      "step": 308200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.063076923076923e-06,
+      "loss": 0.8021,
+      "step": 308300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.055384615384615e-06,
+      "loss": 0.8228,
+      "step": 308400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0476923076923084e-06,
+      "loss": 0.8066,
+      "step": 308500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.04e-06,
+      "loss": 0.816,
+      "step": 308600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.032307692307693e-06,
+      "loss": 0.8135,
+      "step": 308700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.024615384615385e-06,
+      "loss": 0.8167,
+      "step": 308800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.016923076923077e-06,
+      "loss": 0.8163,
+      "step": 308900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.00923076923077e-06,
+      "loss": 0.8048,
+      "step": 309000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.001538461538462e-06,
+      "loss": 0.8134,
+      "step": 309100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.993846153846154e-06,
+      "loss": 0.8183,
+      "step": 309200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.986153846153846e-06,
+      "loss": 0.8088,
+      "step": 309300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.978461538461538e-06,
+      "loss": 0.8126,
+      "step": 309400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.97076923076923e-06,
+      "loss": 0.7844,
+      "step": 309500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.963076923076923e-06,
+      "loss": 0.7913,
+      "step": 309600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.955384615384616e-06,
+      "loss": 0.819,
+      "step": 309700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.9476923076923085e-06,
+      "loss": 0.8278,
+      "step": 309800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.8231,
+      "step": 309900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.932307692307692e-06,
+      "loss": 0.8668,
+      "step": 310000
+    },
+    {
+      "epoch": 1.39,
+      "eval_loss": 1.1910014152526855,
+      "eval_runtime": 132.9377,
+      "eval_samples_per_second": 15.045,
+      "eval_steps_per_second": 0.94,
+      "step": 310000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.924615384615385e-06,
+      "loss": 0.7895,
+      "step": 310100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.916923076923077e-06,
+      "loss": 0.811,
+      "step": 310200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.909230769230769e-06,
+      "loss": 0.8103,
+      "step": 310300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.901538461538462e-06,
+      "loss": 0.7814,
+      "step": 310400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.893846153846154e-06,
+      "loss": 0.8365,
+      "step": 310500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.8861538461538465e-06,
+      "loss": 0.8063,
+      "step": 310600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.878461538461538e-06,
+      "loss": 0.8152,
+      "step": 310700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.87076923076923e-06,
+      "loss": 0.819,
+      "step": 310800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.863076923076924e-06,
+      "loss": 0.8247,
+      "step": 310900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.855384615384616e-06,
+      "loss": 0.8148,
+      "step": 311000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.847692307692308e-06,
+      "loss": 0.8643,
+      "step": 311100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 0.8017,
+      "step": 311200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.8323076923076925e-06,
+      "loss": 0.7952,
+      "step": 311300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.8246153846153844e-06,
+      "loss": 0.8286,
+      "step": 311400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.816923076923077e-06,
+      "loss": 0.8162,
+      "step": 311500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.809230769230769e-06,
+      "loss": 0.8347,
+      "step": 311600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.801538461538462e-06,
+      "loss": 0.7963,
+      "step": 311700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.793846153846154e-06,
+      "loss": 0.8546,
+      "step": 311800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.786153846153846e-06,
+      "loss": 0.7862,
+      "step": 311900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.7784615384615385e-06,
+      "loss": 0.8043,
+      "step": 312000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.770769230769231e-06,
+      "loss": 0.8111,
+      "step": 312100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.763076923076923e-06,
+      "loss": 0.8204,
+      "step": 312200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.755384615384616e-06,
+      "loss": 0.8281,
+      "step": 312300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.747692307692308e-06,
+      "loss": 0.817,
+      "step": 312400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.74e-06,
+      "loss": 0.8297,
+      "step": 312500
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 1.1995940208435059,
+      "eval_runtime": 132.3156,
+      "eval_samples_per_second": 15.115,
+      "eval_steps_per_second": 0.945,
+      "step": 312500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.732307692307693e-06,
+      "loss": 0.8178,
+      "step": 312600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.7246153846153845e-06,
+      "loss": 0.8297,
+      "step": 312700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.716923076923077e-06,
+      "loss": 0.8359,
+      "step": 312800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.709230769230769e-06,
+      "loss": 0.819,
+      "step": 312900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.701538461538461e-06,
+      "loss": 0.8298,
+      "step": 313000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.693846153846154e-06,
+      "loss": 0.8169,
+      "step": 313100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.686153846153846e-06,
+      "loss": 0.8423,
+      "step": 313200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6785384615384616e-06,
+      "loss": 0.8319,
+      "step": 313300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.670846153846154e-06,
+      "loss": 0.8384,
+      "step": 313400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.663153846153846e-06,
+      "loss": 0.8062,
+      "step": 313500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.655461538461538e-06,
+      "loss": 0.817,
+      "step": 313600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.647769230769231e-06,
+      "loss": 0.8029,
+      "step": 313700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.640076923076923e-06,
+      "loss": 0.8244,
+      "step": 313800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.632384615384615e-06,
+      "loss": 0.7968,
+      "step": 313900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.6246923076923076e-06,
+      "loss": 0.7851,
+      "step": 314000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.617e-06,
+      "loss": 0.806,
+      "step": 314100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.609307692307693e-06,
+      "loss": 0.7901,
+      "step": 314200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.601615384615385e-06,
+      "loss": 0.7778,
+      "step": 314300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.593923076923077e-06,
+      "loss": 0.8252,
+      "step": 314400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.58623076923077e-06,
+      "loss": 0.8057,
+      "step": 314500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.578538461538462e-06,
+      "loss": 0.7818,
+      "step": 314600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.570846153846154e-06,
+      "loss": 0.8187,
+      "step": 314700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.563153846153846e-06,
+      "loss": 0.8173,
+      "step": 314800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.555461538461538e-06,
+      "loss": 0.7959,
+      "step": 314900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.547769230769231e-06,
+      "loss": 0.8346,
+      "step": 315000
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 1.2003310918807983,
+      "eval_runtime": 136.8224,
+      "eval_samples_per_second": 14.617,
+      "eval_steps_per_second": 0.914,
+      "step": 315000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.540076923076923e-06,
+      "loss": 0.7953,
+      "step": 315100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.532384615384616e-06,
+      "loss": 0.7976,
+      "step": 315200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.5246923076923085e-06,
+      "loss": 0.7968,
+      "step": 315300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.5170000000000005e-06,
+      "loss": 0.8069,
+      "step": 315400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.509307692307692e-06,
+      "loss": 0.8135,
+      "step": 315500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.501615384615385e-06,
+      "loss": 0.82,
+      "step": 315600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.493923076923077e-06,
+      "loss": 0.8609,
+      "step": 315700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.486230769230769e-06,
+      "loss": 0.7993,
+      "step": 315800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.478538461538462e-06,
+      "loss": 0.7947,
+      "step": 315900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.470846153846154e-06,
+      "loss": 0.7962,
+      "step": 316000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.4631538461538465e-06,
+      "loss": 0.7945,
+      "step": 316100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.455461538461538e-06,
+      "loss": 0.7864,
+      "step": 316200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.44776923076923e-06,
+      "loss": 0.8187,
+      "step": 316300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.440076923076924e-06,
+      "loss": 0.8219,
+      "step": 316400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.432384615384616e-06,
+      "loss": 0.8228,
+      "step": 316500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.424692307692308e-06,
+      "loss": 0.8242,
+      "step": 316600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.4170000000000006e-06,
+      "loss": 0.8288,
+      "step": 316700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.4093076923076925e-06,
+      "loss": 0.8167,
+      "step": 316800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.401615384615384e-06,
+      "loss": 0.8132,
+      "step": 316900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.393923076923077e-06,
+      "loss": 0.8292,
+      "step": 317000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.386230769230769e-06,
+      "loss": 0.8292,
+      "step": 317100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.378538461538462e-06,
+      "loss": 0.846,
+      "step": 317200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.370846153846154e-06,
+      "loss": 0.8006,
+      "step": 317300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.363153846153846e-06,
+      "loss": 0.801,
+      "step": 317400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.3554615384615385e-06,
+      "loss": 0.8161,
+      "step": 317500
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 1.1925783157348633,
+      "eval_runtime": 131.8354,
+      "eval_samples_per_second": 15.17,
+      "eval_steps_per_second": 0.948,
+      "step": 317500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.347769230769231e-06,
+      "loss": 0.8192,
+      "step": 317600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.340076923076923e-06,
+      "loss": 0.8127,
+      "step": 317700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.332384615384616e-06,
+      "loss": 0.8167,
+      "step": 317800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.324692307692308e-06,
+      "loss": 0.8056,
+      "step": 317900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.317000000000001e-06,
+      "loss": 0.8174,
+      "step": 318000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.309307692307693e-06,
+      "loss": 0.7855,
+      "step": 318100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.3016153846153845e-06,
+      "loss": 0.8136,
+      "step": 318200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.293923076923077e-06,
+      "loss": 0.8313,
+      "step": 318300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.286230769230769e-06,
+      "loss": 0.7554,
+      "step": 318400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.278538461538461e-06,
+      "loss": 0.8055,
+      "step": 318500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.270846153846154e-06,
+      "loss": 0.7956,
+      "step": 318600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.263153846153846e-06,
+      "loss": 0.8347,
+      "step": 318700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.255461538461539e-06,
+      "loss": 0.8281,
+      "step": 318800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.247769230769231e-06,
+      "loss": 0.7821,
+      "step": 318900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.240076923076923e-06,
+      "loss": 0.8026,
+      "step": 319000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.232384615384616e-06,
+      "loss": 0.8158,
+      "step": 319100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.224692307692308e-06,
+      "loss": 0.8443,
+      "step": 319200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.217e-06,
+      "loss": 0.8411,
+      "step": 319300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.209307692307693e-06,
+      "loss": 0.8165,
+      "step": 319400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.201615384615385e-06,
+      "loss": 0.8154,
+      "step": 319500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.1939230769230766e-06,
+      "loss": 0.7964,
+      "step": 319600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.186230769230769e-06,
+      "loss": 0.8163,
+      "step": 319700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.178538461538461e-06,
+      "loss": 0.779,
+      "step": 319800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.170923076923077e-06,
+      "loss": 0.8385,
+      "step": 319900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.16323076923077e-06,
+      "loss": 0.8222,
+      "step": 320000
+    },
+    {
+      "epoch": 1.43,
+      "eval_loss": 1.2052990198135376,
+      "eval_runtime": 132.3385,
+      "eval_samples_per_second": 15.113,
+      "eval_steps_per_second": 0.945,
+      "step": 320000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.155538461538462e-06,
+      "loss": 0.82,
+      "step": 320100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.1478461538461536e-06,
+      "loss": 0.8119,
+      "step": 320200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.140153846153846e-06,
+      "loss": 0.8105,
+      "step": 320300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.132461538461538e-06,
+      "loss": 0.8089,
+      "step": 320400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.124769230769231e-06,
+      "loss": 0.8244,
+      "step": 320500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.117076923076923e-06,
+      "loss": 0.7852,
+      "step": 320600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.109384615384615e-06,
+      "loss": 0.815,
+      "step": 320700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.1016923076923085e-06,
+      "loss": 0.8097,
+      "step": 320800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.0940000000000004e-06,
+      "loss": 0.8298,
+      "step": 320900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.086307692307692e-06,
+      "loss": 0.7696,
+      "step": 321000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.078615384615385e-06,
+      "loss": 0.8141,
+      "step": 321100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.070923076923077e-06,
+      "loss": 0.8071,
+      "step": 321200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.063307692307692e-06,
+      "loss": 0.7894,
+      "step": 321300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.055615384615385e-06,
+      "loss": 0.8342,
+      "step": 321400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.047923076923077e-06,
+      "loss": 0.8497,
+      "step": 321500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.040230769230769e-06,
+      "loss": 0.7959,
+      "step": 321600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.032538461538462e-06,
+      "loss": 0.8402,
+      "step": 321700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.024846153846154e-06,
+      "loss": 0.815,
+      "step": 321800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.017153846153847e-06,
+      "loss": 0.8097,
+      "step": 321900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.009461538461539e-06,
+      "loss": 0.8318,
+      "step": 322000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.001769230769231e-06,
+      "loss": 0.8284,
+      "step": 322100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9940769230769235e-06,
+      "loss": 0.8246,
+      "step": 322200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.986384615384615e-06,
+      "loss": 0.8236,
+      "step": 322300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.978692307692307e-06,
+      "loss": 0.8298,
+      "step": 322400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.971e-06,
+      "loss": 0.8109,
+      "step": 322500
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 1.1790719032287598,
+      "eval_runtime": 131.4868,
+      "eval_samples_per_second": 15.211,
+      "eval_steps_per_second": 0.951,
+      "step": 322500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.963307692307692e-06,
+      "loss": 0.7944,
+      "step": 322600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.955615384615384e-06,
+      "loss": 0.8302,
+      "step": 322700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9479230769230776e-06,
+      "loss": 0.7988,
+      "step": 322800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9402307692307695e-06,
+      "loss": 0.8292,
+      "step": 322900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.932538461538462e-06,
+      "loss": 0.7974,
+      "step": 323000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.924846153846154e-06,
+      "loss": 0.8232,
+      "step": 323100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.917153846153846e-06,
+      "loss": 0.8159,
+      "step": 323200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.909461538461539e-06,
+      "loss": 0.7865,
+      "step": 323300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.901769230769231e-06,
+      "loss": 0.7934,
+      "step": 323400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.894076923076923e-06,
+      "loss": 0.7916,
+      "step": 323500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8863846153846155e-06,
+      "loss": 0.7922,
+      "step": 323600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8786923076923074e-06,
+      "loss": 0.8021,
+      "step": 323700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.871e-06,
+      "loss": 0.807,
+      "step": 323800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.863307692307692e-06,
+      "loss": 0.8025,
+      "step": 323900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.855615384615385e-06,
+      "loss": 0.8075,
+      "step": 324000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.847923076923078e-06,
+      "loss": 0.7703,
+      "step": 324100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.84023076923077e-06,
+      "loss": 0.7923,
+      "step": 324200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8325384615384615e-06,
+      "loss": 0.7957,
+      "step": 324300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.824846153846154e-06,
+      "loss": 0.7996,
+      "step": 324400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.817153846153846e-06,
+      "loss": 0.814,
+      "step": 324500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.809461538461538e-06,
+      "loss": 0.8212,
+      "step": 324600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.801769230769231e-06,
+      "loss": 0.8196,
+      "step": 324700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.794076923076923e-06,
+      "loss": 0.8004,
+      "step": 324800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.786384615384616e-06,
+      "loss": 0.7948,
+      "step": 324900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.778769230769231e-06,
+      "loss": 0.8398,
+      "step": 325000
+    },
+    {
+      "epoch": 1.46,
+      "eval_loss": 1.191308856010437,
+      "eval_runtime": 133.0455,
+      "eval_samples_per_second": 15.032,
+      "eval_steps_per_second": 0.94,
+      "step": 325000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.771076923076923e-06,
+      "loss": 0.7591,
+      "step": 325100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.763384615384616e-06,
+      "loss": 0.8131,
+      "step": 325200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.755692307692308e-06,
+      "loss": 0.8029,
+      "step": 325300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.748e-06,
+      "loss": 0.8295,
+      "step": 325400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.740307692307693e-06,
+      "loss": 0.8321,
+      "step": 325500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.7326153846153846e-06,
+      "loss": 0.7849,
+      "step": 325600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.7249230769230765e-06,
+      "loss": 0.8264,
+      "step": 325700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.717230769230769e-06,
+      "loss": 0.7985,
+      "step": 325800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.709538461538461e-06,
+      "loss": 0.8254,
+      "step": 325900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.701846153846154e-06,
+      "loss": 0.7844,
+      "step": 326000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.694153846153847e-06,
+      "loss": 0.8149,
+      "step": 326100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.686461538461539e-06,
+      "loss": 0.8346,
+      "step": 326200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.678769230769231e-06,
+      "loss": 0.7824,
+      "step": 326300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.671076923076923e-06,
+      "loss": 0.8118,
+      "step": 326400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.663384615384615e-06,
+      "loss": 0.8367,
+      "step": 326500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.655769230769231e-06,
+      "loss": 0.7974,
+      "step": 326600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.648076923076923e-06,
+      "loss": 0.8409,
+      "step": 326700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.640384615384616e-06,
+      "loss": 0.8204,
+      "step": 326800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6326923076923084e-06,
+      "loss": 0.8254,
+      "step": 326900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.625e-06,
+      "loss": 0.7967,
+      "step": 327000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.617307692307692e-06,
+      "loss": 0.7778,
+      "step": 327100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.609615384615385e-06,
+      "loss": 0.7961,
+      "step": 327200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.601923076923077e-06,
+      "loss": 0.8304,
+      "step": 327300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.594230769230769e-06,
+      "loss": 0.8376,
+      "step": 327400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.586538461538462e-06,
+      "loss": 0.8245,
+      "step": 327500
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 1.1887290477752686,
+      "eval_runtime": 132.4601,
+      "eval_samples_per_second": 15.099,
+      "eval_steps_per_second": 0.944,
+      "step": 327500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.578846153846154e-06,
+      "loss": 0.8113,
+      "step": 327600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.571153846153846e-06,
+      "loss": 0.8029,
+      "step": 327700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.563461538461538e-06,
+      "loss": 0.811,
+      "step": 327800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.555769230769231e-06,
+      "loss": 0.7939,
+      "step": 327900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.548076923076924e-06,
+      "loss": 0.8169,
+      "step": 328000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.540384615384616e-06,
+      "loss": 0.7843,
+      "step": 328100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.532692307692308e-06,
+      "loss": 0.7948,
+      "step": 328200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5250000000000005e-06,
+      "loss": 0.8158,
+      "step": 328300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.517307692307692e-06,
+      "loss": 0.81,
+      "step": 328400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.509615384615385e-06,
+      "loss": 0.8093,
+      "step": 328500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.501923076923077e-06,
+      "loss": 0.8471,
+      "step": 328600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.494230769230769e-06,
+      "loss": 0.8256,
+      "step": 328700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.486538461538462e-06,
+      "loss": 0.8001,
+      "step": 328800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.478846153846154e-06,
+      "loss": 0.8049,
+      "step": 328900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.471153846153846e-06,
+      "loss": 0.8355,
+      "step": 329000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.463461538461539e-06,
+      "loss": 0.7961,
+      "step": 329100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.455769230769231e-06,
+      "loss": 0.8122,
+      "step": 329200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.448076923076923e-06,
+      "loss": 0.8182,
+      "step": 329300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.440384615384616e-06,
+      "loss": 0.8159,
+      "step": 329400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.432692307692308e-06,
+      "loss": 0.7853,
+      "step": 329500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4250000000000006e-06,
+      "loss": 0.8245,
+      "step": 329600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4173076923076925e-06,
+      "loss": 0.8022,
+      "step": 329700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4096153846153844e-06,
+      "loss": 0.828,
+      "step": 329800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.401923076923077e-06,
+      "loss": 0.7769,
+      "step": 329900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.394230769230769e-06,
+      "loss": 0.8062,
+      "step": 330000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.1920084953308105,
+      "eval_runtime": 133.0555,
+      "eval_samples_per_second": 15.031,
+      "eval_steps_per_second": 0.939,
+      "step": 330000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.386538461538461e-06,
+      "loss": 0.7865,
+      "step": 330100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.378846153846154e-06,
+      "loss": 0.7571,
+      "step": 330200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.371153846153847e-06,
+      "loss": 0.7531,
+      "step": 330300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.3634615384615385e-06,
+      "loss": 0.8076,
+      "step": 330400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.355769230769231e-06,
+      "loss": 0.8131,
+      "step": 330500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.348076923076923e-06,
+      "loss": 0.8576,
+      "step": 330600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.340384615384616e-06,
+      "loss": 0.8288,
+      "step": 330700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.332692307692308e-06,
+      "loss": 0.7675,
+      "step": 330800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.325e-06,
+      "loss": 0.7982,
+      "step": 330900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.317307692307693e-06,
+      "loss": 0.8088,
+      "step": 331000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.3096153846153845e-06,
+      "loss": 0.7876,
+      "step": 331100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3019230769230765e-06,
+      "loss": 0.8215,
+      "step": 331200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.294230769230769e-06,
+      "loss": 0.8048,
+      "step": 331300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.286538461538461e-06,
+      "loss": 0.7656,
+      "step": 331400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.278846153846155e-06,
+      "loss": 0.8381,
+      "step": 331500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.271153846153847e-06,
+      "loss": 0.7926,
+      "step": 331600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.263461538461539e-06,
+      "loss": 0.7969,
+      "step": 331700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.255769230769231e-06,
+      "loss": 0.7788,
+      "step": 331800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.248076923076923e-06,
+      "loss": 0.7681,
+      "step": 331900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.240384615384615e-06,
+      "loss": 0.7972,
+      "step": 332000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.232692307692308e-06,
+      "loss": 0.7804,
+      "step": 332100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.225e-06,
+      "loss": 0.8242,
+      "step": 332200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.217307692307692e-06,
+      "loss": 0.8025,
+      "step": 332300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.209615384615385e-06,
+      "loss": 0.8058,
+      "step": 332400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.2019230769230766e-06,
+      "loss": 0.8212,
+      "step": 332500
+    },
+    {
+      "epoch": 1.49,
+      "eval_loss": 1.1807278394699097,
+      "eval_runtime": 132.4863,
+      "eval_samples_per_second": 15.096,
+      "eval_steps_per_second": 0.943,
+      "step": 332500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.194230769230769e-06,
+      "loss": 0.7731,
+      "step": 332600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.186538461538462e-06,
+      "loss": 0.7809,
+      "step": 332700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.178846153846154e-06,
+      "loss": 0.8024,
+      "step": 332800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.171153846153847e-06,
+      "loss": 0.8119,
+      "step": 332900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.163461538461539e-06,
+      "loss": 0.8079,
+      "step": 333000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.155769230769231e-06,
+      "loss": 0.8255,
+      "step": 333100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.1480769230769234e-06,
+      "loss": 0.8487,
+      "step": 333200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.140384615384615e-06,
+      "loss": 0.8129,
+      "step": 333300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.132692307692307e-06,
+      "loss": 0.818,
+      "step": 333400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.125e-06,
+      "loss": 0.7926,
+      "step": 333500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.117307692307692e-06,
+      "loss": 0.7992,
+      "step": 333600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.109615384615385e-06,
+      "loss": 0.8096,
+      "step": 333700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.101923076923077e-06,
+      "loss": 0.8097,
+      "step": 333800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0942307692307695e-06,
+      "loss": 0.8043,
+      "step": 333900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.086538461538462e-06,
+      "loss": 0.7693,
+      "step": 334000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.078846153846154e-06,
+      "loss": 0.81,
+      "step": 334100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.071153846153846e-06,
+      "loss": 0.8302,
+      "step": 334200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.063461538461539e-06,
+      "loss": 0.8368,
+      "step": 334300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.055769230769231e-06,
+      "loss": 0.8229,
+      "step": 334400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0480769230769235e-06,
+      "loss": 0.8264,
+      "step": 334500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0403846153846155e-06,
+      "loss": 0.8041,
+      "step": 334600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.032692307692307e-06,
+      "loss": 0.8376,
+      "step": 334700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.025e-06,
+      "loss": 0.7926,
+      "step": 334800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.017307692307692e-06,
+      "loss": 0.8175,
+      "step": 334900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.009615384615384e-06,
+      "loss": 0.8018,
+      "step": 335000
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 1.1677196025848389,
+      "eval_runtime": 131.8347,
+      "eval_samples_per_second": 15.171,
+      "eval_steps_per_second": 0.948,
+      "step": 335000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.001923076923078e-06,
+      "loss": 0.8164,
+      "step": 335100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9942307692307696e-06,
+      "loss": 0.8067,
+      "step": 335200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9865384615384615e-06,
+      "loss": 0.8061,
+      "step": 335300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.978846153846154e-06,
+      "loss": 0.7983,
+      "step": 335400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.971153846153846e-06,
+      "loss": 0.7957,
+      "step": 335500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.963461538461539e-06,
+      "loss": 0.8241,
+      "step": 335600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.955769230769231e-06,
+      "loss": 0.8074,
+      "step": 335700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.948076923076923e-06,
+      "loss": 0.808,
+      "step": 335800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.940384615384616e-06,
+      "loss": 0.8251,
+      "step": 335900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9326923076923075e-06,
+      "loss": 0.8041,
+      "step": 336000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9249999999999994e-06,
+      "loss": 0.812,
+      "step": 336100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.917307692307692e-06,
+      "loss": 0.7872,
+      "step": 336200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.909615384615385e-06,
+      "loss": 0.8177,
+      "step": 336300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.901923076923078e-06,
+      "loss": 0.818,
+      "step": 336400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.89423076923077e-06,
+      "loss": 0.7959,
+      "step": 336500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.886538461538462e-06,
+      "loss": 0.8085,
+      "step": 336600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.878846153846154e-06,
+      "loss": 0.8104,
+      "step": 336700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.871153846153846e-06,
+      "loss": 0.8177,
+      "step": 336800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.863461538461538e-06,
+      "loss": 0.7842,
+      "step": 336900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.855769230769231e-06,
+      "loss": 0.8348,
+      "step": 337000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.848076923076923e-06,
+      "loss": 0.7822,
+      "step": 337100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.840384615384615e-06,
+      "loss": 0.8137,
+      "step": 337200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.832692307692308e-06,
+      "loss": 0.8062,
+      "step": 337300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.8249999999999995e-06,
+      "loss": 0.8122,
+      "step": 337400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.817307692307693e-06,
+      "loss": 0.7755,
+      "step": 337500
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 1.1613469123840332,
+      "eval_runtime": 132.3471,
+      "eval_samples_per_second": 15.112,
+      "eval_steps_per_second": 0.944,
+      "step": 337500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.809615384615385e-06,
+      "loss": 0.8479,
+      "step": 337600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.801923076923077e-06,
+      "loss": 0.8076,
+      "step": 337700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.79423076923077e-06,
+      "loss": 0.806,
+      "step": 337800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.786615384615385e-06,
+      "loss": 0.7965,
+      "step": 337900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7789230769230766e-06,
+      "loss": 0.8154,
+      "step": 338000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.771230769230769e-06,
+      "loss": 0.7988,
+      "step": 338100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.763538461538461e-06,
+      "loss": 0.7784,
+      "step": 338200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.755846153846154e-06,
+      "loss": 0.8067,
+      "step": 338300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.748153846153847e-06,
+      "loss": 0.8016,
+      "step": 338400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.740461538461539e-06,
+      "loss": 0.8182,
+      "step": 338500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.732769230769231e-06,
+      "loss": 0.8084,
+      "step": 338600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.725076923076923e-06,
+      "loss": 0.8219,
+      "step": 338700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.717384615384615e-06,
+      "loss": 0.7887,
+      "step": 338800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.709692307692308e-06,
+      "loss": 0.8035,
+      "step": 338900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.702e-06,
+      "loss": 0.7787,
+      "step": 339000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.694307692307692e-06,
+      "loss": 0.7785,
+      "step": 339100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.686615384615385e-06,
+      "loss": 0.7902,
+      "step": 339200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.678923076923077e-06,
+      "loss": 0.7969,
+      "step": 339300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.6712307692307694e-06,
+      "loss": 0.7851,
+      "step": 339400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.663538461538462e-06,
+      "loss": 0.7952,
+      "step": 339500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.655846153846154e-06,
+      "loss": 0.7922,
+      "step": 339600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.648153846153846e-06,
+      "loss": 0.8395,
+      "step": 339700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.640461538461539e-06,
+      "loss": 0.7873,
+      "step": 339800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.632769230769231e-06,
+      "loss": 0.8373,
+      "step": 339900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.6250769230769235e-06,
+      "loss": 0.8158,
+      "step": 340000
+    },
+    {
+      "epoch": 1.52,
+      "eval_loss": 1.1748687028884888,
+      "eval_runtime": 131.5299,
+      "eval_samples_per_second": 15.206,
+      "eval_steps_per_second": 0.95,
+      "step": 340000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.6173846153846155e-06,
+      "loss": 0.8118,
+      "step": 340100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.609692307692307e-06,
+      "loss": 0.823,
+      "step": 340200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.602e-06,
+      "loss": 0.8113,
+      "step": 340300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.594307692307692e-06,
+      "loss": 0.8074,
+      "step": 340400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.586615384615384e-06,
+      "loss": 0.7793,
+      "step": 340500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.578923076923078e-06,
+      "loss": 0.7742,
+      "step": 340600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5712307692307695e-06,
+      "loss": 0.8072,
+      "step": 340700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.563538461538462e-06,
+      "loss": 0.7709,
+      "step": 340800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.555923076923077e-06,
+      "loss": 0.7905,
+      "step": 340900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.548230769230769e-06,
+      "loss": 0.7997,
+      "step": 341000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.540538461538461e-06,
+      "loss": 0.784,
+      "step": 341100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.532846153846154e-06,
+      "loss": 0.8535,
+      "step": 341200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.525153846153846e-06,
+      "loss": 0.8355,
+      "step": 341300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.517461538461539e-06,
+      "loss": 0.83,
+      "step": 341400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.509769230769231e-06,
+      "loss": 0.7774,
+      "step": 341500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.502076923076923e-06,
+      "loss": 0.8052,
+      "step": 341600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.494461538461539e-06,
+      "loss": 0.8296,
+      "step": 341700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.486769230769231e-06,
+      "loss": 0.8149,
+      "step": 341800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.479076923076923e-06,
+      "loss": 0.7916,
+      "step": 341900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4713846153846155e-06,
+      "loss": 0.8454,
+      "step": 342000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4636923076923074e-06,
+      "loss": 0.8084,
+      "step": 342100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.456e-06,
+      "loss": 0.8069,
+      "step": 342200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.448307692307693e-06,
+      "loss": 0.8006,
+      "step": 342300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.440615384615385e-06,
+      "loss": 0.8221,
+      "step": 342400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.432923076923077e-06,
+      "loss": 0.7884,
+      "step": 342500
+    },
+    {
+      "epoch": 1.54,
+      "eval_loss": 1.179386019706726,
+      "eval_runtime": 132.4973,
+      "eval_samples_per_second": 15.095,
+      "eval_steps_per_second": 0.943,
+      "step": 342500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.42523076923077e-06,
+      "loss": 0.8414,
+      "step": 342600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4175384615384615e-06,
+      "loss": 0.8344,
+      "step": 342700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.409846153846154e-06,
+      "loss": 0.8364,
+      "step": 342800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.402153846153846e-06,
+      "loss": 0.7911,
+      "step": 342900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.394461538461538e-06,
+      "loss": 0.801,
+      "step": 343000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.386769230769231e-06,
+      "loss": 0.7901,
+      "step": 343100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.379076923076923e-06,
+      "loss": 0.8126,
+      "step": 343200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.371384615384615e-06,
+      "loss": 0.798,
+      "step": 343300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.363692307692308e-06,
+      "loss": 0.8,
+      "step": 343400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.356e-06,
+      "loss": 0.7811,
+      "step": 343500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.348307692307693e-06,
+      "loss": 0.8086,
+      "step": 343600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.340615384615385e-06,
+      "loss": 0.7778,
+      "step": 343700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.332923076923077e-06,
+      "loss": 0.7513,
+      "step": 343800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.32523076923077e-06,
+      "loss": 0.8279,
+      "step": 343900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.3176153846153845e-06,
+      "loss": 0.8045,
+      "step": 344000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.3099230769230765e-06,
+      "loss": 0.7834,
+      "step": 344100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.30223076923077e-06,
+      "loss": 0.7802,
+      "step": 344200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.294538461538462e-06,
+      "loss": 0.8061,
+      "step": 344300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.286846153846154e-06,
+      "loss": 0.8364,
+      "step": 344400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.279153846153847e-06,
+      "loss": 0.7944,
+      "step": 344500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.271461538461539e-06,
+      "loss": 0.8031,
+      "step": 344600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2637692307692306e-06,
+      "loss": 0.8076,
+      "step": 344700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.256076923076923e-06,
+      "loss": 0.7824,
+      "step": 344800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.248384615384615e-06,
+      "loss": 0.7921,
+      "step": 344900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.240692307692308e-06,
+      "loss": 0.7872,
+      "step": 345000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 1.1694403886795044,
+      "eval_runtime": 131.7885,
+      "eval_samples_per_second": 15.176,
+      "eval_steps_per_second": 0.948,
+      "step": 345000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233076923076924e-06,
+      "loss": 0.7839,
+      "step": 345100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225384615384616e-06,
+      "loss": 0.7979,
+      "step": 345200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.217692307692308e-06,
+      "loss": 0.8071,
+      "step": 345300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.21e-06,
+      "loss": 0.8099,
+      "step": 345400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.202307692307692e-06,
+      "loss": 0.8053,
+      "step": 345500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.194615384615385e-06,
+      "loss": 0.7731,
+      "step": 345600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.186923076923077e-06,
+      "loss": 0.8033,
+      "step": 345700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.179230769230769e-06,
+      "loss": 0.8154,
+      "step": 345800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.171538461538462e-06,
+      "loss": 0.8251,
+      "step": 345900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.163846153846154e-06,
+      "loss": 0.7755,
+      "step": 346000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.1561538461538455e-06,
+      "loss": 0.7724,
+      "step": 346100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.148461538461539e-06,
+      "loss": 0.78,
+      "step": 346200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.140769230769231e-06,
+      "loss": 0.7701,
+      "step": 346300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.133076923076924e-06,
+      "loss": 0.7737,
+      "step": 346400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.125384615384616e-06,
+      "loss": 0.8519,
+      "step": 346500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.117692307692308e-06,
+      "loss": 0.7835,
+      "step": 346600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.1100000000000005e-06,
+      "loss": 0.8116,
+      "step": 346700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.102307692307692e-06,
+      "loss": 0.8487,
+      "step": 346800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.094615384615384e-06,
+      "loss": 0.8283,
+      "step": 346900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.086923076923077e-06,
+      "loss": 0.841,
+      "step": 347000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.079230769230769e-06,
+      "loss": 0.7785,
+      "step": 347100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.071538461538461e-06,
+      "loss": 0.7646,
+      "step": 347200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.063846153846154e-06,
+      "loss": 0.7968,
+      "step": 347300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0561538461538465e-06,
+      "loss": 0.7707,
+      "step": 347400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.048461538461539e-06,
+      "loss": 0.8158,
+      "step": 347500
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 1.1668165922164917,
+      "eval_runtime": 132.6943,
+      "eval_samples_per_second": 15.072,
+      "eval_steps_per_second": 0.942,
+      "step": 347500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.040769230769231e-06,
+      "loss": 0.814,
+      "step": 347600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033076923076923e-06,
+      "loss": 0.7941,
+      "step": 347700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025384615384616e-06,
+      "loss": 0.797,
+      "step": 347800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.017692307692308e-06,
+      "loss": 0.765,
+      "step": 347900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.01e-06,
+      "loss": 0.8025,
+      "step": 348000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0023076923076925e-06,
+      "loss": 0.8135,
+      "step": 348100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.994615384615384e-06,
+      "loss": 0.8065,
+      "step": 348200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.986923076923076e-06,
+      "loss": 0.7959,
+      "step": 348300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.979230769230769e-06,
+      "loss": 0.8197,
+      "step": 348400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.971538461538461e-06,
+      "loss": 0.7863,
+      "step": 348500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.963846153846155e-06,
+      "loss": 0.7794,
+      "step": 348600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.956153846153847e-06,
+      "loss": 0.7961,
+      "step": 348700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.9484615384615385e-06,
+      "loss": 0.7973,
+      "step": 348800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.940769230769231e-06,
+      "loss": 0.8172,
+      "step": 348900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.933076923076923e-06,
+      "loss": 0.7904,
+      "step": 349000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.925384615384615e-06,
+      "loss": 0.8118,
+      "step": 349100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.917692307692308e-06,
+      "loss": 0.8039,
+      "step": 349200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.91e-06,
+      "loss": 0.7761,
+      "step": 349300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.902307692307693e-06,
+      "loss": 0.8032,
+      "step": 349400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8946153846153845e-06,
+      "loss": 0.7888,
+      "step": 349500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8869230769230765e-06,
+      "loss": 0.7927,
+      "step": 349600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.879230769230769e-06,
+      "loss": 0.8429,
+      "step": 349700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.871538461538462e-06,
+      "loss": 0.7773,
+      "step": 349800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.863846153846154e-06,
+      "loss": 0.7485,
+      "step": 349900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.856153846153847e-06,
+      "loss": 0.7947,
+      "step": 350000
+    },
+    {
+      "epoch": 1.57,
+      "eval_loss": 1.1623356342315674,
+      "eval_runtime": 134.5991,
+      "eval_samples_per_second": 14.859,
+      "eval_steps_per_second": 0.929,
+      "step": 350000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.848461538461539e-06,
+      "loss": 0.8028,
+      "step": 350100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8407692307692305e-06,
+      "loss": 0.8049,
+      "step": 350200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833076923076923e-06,
+      "loss": 0.7857,
+      "step": 350300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.825384615384615e-06,
+      "loss": 0.7788,
+      "step": 350400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.817692307692308e-06,
+      "loss": 0.8147,
+      "step": 350500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.81e-06,
+      "loss": 0.7769,
+      "step": 350600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8023076923076923e-06,
+      "loss": 0.8173,
+      "step": 350700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.794615384615385e-06,
+      "loss": 0.8107,
+      "step": 350800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.786923076923077e-06,
+      "loss": 0.7968,
+      "step": 350900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.779230769230769e-06,
+      "loss": 0.821,
+      "step": 351000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.7715384615384617e-06,
+      "loss": 0.8165,
+      "step": 351100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.763846153846154e-06,
+      "loss": 0.7936,
+      "step": 351200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.756153846153847e-06,
+      "loss": 0.7863,
+      "step": 351300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7484615384615387e-06,
+      "loss": 0.7781,
+      "step": 351400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7407692307692307e-06,
+      "loss": 0.7847,
+      "step": 351500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.733076923076923e-06,
+      "loss": 0.8239,
+      "step": 351600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7253846153846154e-06,
+      "loss": 0.7924,
+      "step": 351700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.717692307692308e-06,
+      "loss": 0.7859,
+      "step": 351800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.71e-06,
+      "loss": 0.8278,
+      "step": 351900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7023846153846153e-06,
+      "loss": 0.7637,
+      "step": 352000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.694692307692308e-06,
+      "loss": 0.7822,
+      "step": 352100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.687e-06,
+      "loss": 0.7964,
+      "step": 352200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6793076923076924e-06,
+      "loss": 0.8048,
+      "step": 352300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6716153846153847e-06,
+      "loss": 0.7806,
+      "step": 352400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.663923076923077e-06,
+      "loss": 0.8003,
+      "step": 352500
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 1.1615023612976074,
+      "eval_runtime": 133.389,
+      "eval_samples_per_second": 14.994,
+      "eval_steps_per_second": 0.937,
+      "step": 352500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6562307692307694e-06,
+      "loss": 0.8179,
+      "step": 352600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6485384615384618e-06,
+      "loss": 0.8278,
+      "step": 352700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.640846153846154e-06,
+      "loss": 0.7906,
+      "step": 352800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.633153846153846e-06,
+      "loss": 0.8127,
+      "step": 352900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6254615384615384e-06,
+      "loss": 0.7738,
+      "step": 353000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6177692307692307e-06,
+      "loss": 0.8023,
+      "step": 353100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6100769230769235e-06,
+      "loss": 0.7901,
+      "step": 353200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6023846153846154e-06,
+      "loss": 0.7893,
+      "step": 353300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.5946923076923078e-06,
+      "loss": 0.8406,
+      "step": 353400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.587e-06,
+      "loss": 0.8053,
+      "step": 353500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5793076923076925e-06,
+      "loss": 0.7748,
+      "step": 353600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5716153846153844e-06,
+      "loss": 0.7931,
+      "step": 353700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.563923076923077e-06,
+      "loss": 0.7899,
+      "step": 353800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5562307692307695e-06,
+      "loss": 0.795,
+      "step": 353900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.548538461538462e-06,
+      "loss": 0.7946,
+      "step": 354000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.540846153846154e-06,
+      "loss": 0.7889,
+      "step": 354100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.533153846153846e-06,
+      "loss": 0.8048,
+      "step": 354200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5254615384615385e-06,
+      "loss": 0.7979,
+      "step": 354300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.517769230769231e-06,
+      "loss": 0.7963,
+      "step": 354400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.510076923076923e-06,
+      "loss": 0.8032,
+      "step": 354500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5023846153846155e-06,
+      "loss": 0.8165,
+      "step": 354600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.494769230769231e-06,
+      "loss": 0.7665,
+      "step": 354700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.487076923076923e-06,
+      "loss": 0.7761,
+      "step": 354800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4793846153846155e-06,
+      "loss": 0.8144,
+      "step": 354900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.471692307692308e-06,
+      "loss": 0.805,
+      "step": 355000
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 1.1481444835662842,
+      "eval_runtime": 134.5313,
+      "eval_samples_per_second": 14.866,
+      "eval_steps_per_second": 0.929,
+      "step": 355000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4639999999999998e-06,
+      "loss": 0.7784,
+      "step": 355100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4563076923076926e-06,
+      "loss": 0.8096,
+      "step": 355200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.448615384615385e-06,
+      "loss": 0.7699,
+      "step": 355300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4409230769230773e-06,
+      "loss": 0.7903,
+      "step": 355400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.433230769230769e-06,
+      "loss": 0.8206,
+      "step": 355500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4255384615384615e-06,
+      "loss": 0.8328,
+      "step": 355600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417846153846154e-06,
+      "loss": 0.7966,
+      "step": 355700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4101538461538462e-06,
+      "loss": 0.7958,
+      "step": 355800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4024615384615386e-06,
+      "loss": 0.8068,
+      "step": 355900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.394769230769231e-06,
+      "loss": 0.8188,
+      "step": 356000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3870769230769233e-06,
+      "loss": 0.8154,
+      "step": 356100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.379384615384615e-06,
+      "loss": 0.7639,
+      "step": 356200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3716923076923075e-06,
+      "loss": 0.8064,
+      "step": 356300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3640000000000003e-06,
+      "loss": 0.8079,
+      "step": 356400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3563076923076927e-06,
+      "loss": 0.7942,
+      "step": 356500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3486153846153846e-06,
+      "loss": 0.8116,
+      "step": 356600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.340923076923077e-06,
+      "loss": 0.8187,
+      "step": 356700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3332307692307693e-06,
+      "loss": 0.824,
+      "step": 356800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3255384615384616e-06,
+      "loss": 0.8216,
+      "step": 356900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.317846153846154e-06,
+      "loss": 0.812,
+      "step": 357000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3101538461538463e-06,
+      "loss": 0.7907,
+      "step": 357100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3024615384615387e-06,
+      "loss": 0.7735,
+      "step": 357200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2947692307692306e-06,
+      "loss": 0.7784,
+      "step": 357300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.287076923076923e-06,
+      "loss": 0.8191,
+      "step": 357400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2793846153846153e-06,
+      "loss": 0.7674,
+      "step": 357500
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 1.155331015586853,
+      "eval_runtime": 132.3805,
+      "eval_samples_per_second": 15.108,
+      "eval_steps_per_second": 0.944,
+      "step": 357500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.271692307692308e-06,
+      "loss": 0.7876,
+      "step": 357600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.264e-06,
+      "loss": 0.8071,
+      "step": 357700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2563076923076924e-06,
+      "loss": 0.7967,
+      "step": 357800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2486153846153847e-06,
+      "loss": 0.8078,
+      "step": 357900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.240923076923077e-06,
+      "loss": 0.7973,
+      "step": 358000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.233230769230769e-06,
+      "loss": 0.7956,
+      "step": 358100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.2255384615384617e-06,
+      "loss": 0.774,
+      "step": 358200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.217846153846154e-06,
+      "loss": 0.7932,
+      "step": 358300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.2101538461538464e-06,
+      "loss": 0.8143,
+      "step": 358400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.2024615384615384e-06,
+      "loss": 0.7907,
+      "step": 358500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1947692307692307e-06,
+      "loss": 0.7765,
+      "step": 358600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.187076923076923e-06,
+      "loss": 0.8048,
+      "step": 358700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1794615384615383e-06,
+      "loss": 0.799,
+      "step": 358800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1717692307692307e-06,
+      "loss": 0.8062,
+      "step": 358900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1640769230769235e-06,
+      "loss": 0.7943,
+      "step": 359000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1563846153846154e-06,
+      "loss": 0.7861,
+      "step": 359100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1486923076923077e-06,
+      "loss": 0.7596,
+      "step": 359200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.141e-06,
+      "loss": 0.8152,
+      "step": 359300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1333076923076924e-06,
+      "loss": 0.7768,
+      "step": 359400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1256153846153844e-06,
+      "loss": 0.8021,
+      "step": 359500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.117923076923077e-06,
+      "loss": 0.825,
+      "step": 359600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1102307692307695e-06,
+      "loss": 0.7908,
+      "step": 359700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.102538461538462e-06,
+      "loss": 0.7834,
+      "step": 359800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0948461538461538e-06,
+      "loss": 0.8126,
+      "step": 359900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.087153846153846e-06,
+      "loss": 0.8287,
+      "step": 360000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.1670494079589844,
+      "eval_runtime": 133.0581,
+      "eval_samples_per_second": 15.031,
+      "eval_steps_per_second": 0.939,
+      "step": 360000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0795384615384614e-06,
+      "loss": 0.7993,
+      "step": 360100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0718461538461537e-06,
+      "loss": 0.7531,
+      "step": 360200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.064153846153846e-06,
+      "loss": 0.8588,
+      "step": 360300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.056461538461539e-06,
+      "loss": 0.7583,
+      "step": 360400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0487692307692308e-06,
+      "loss": 0.7628,
+      "step": 360500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.041076923076923e-06,
+      "loss": 0.784,
+      "step": 360600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0333846153846155e-06,
+      "loss": 0.8043,
+      "step": 360700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.025692307692308e-06,
+      "loss": 0.8197,
+      "step": 360800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0179999999999997e-06,
+      "loss": 0.7617,
+      "step": 360900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0103076923076925e-06,
+      "loss": 0.7772,
+      "step": 361000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.002615384615385e-06,
+      "loss": 0.8274,
+      "step": 361100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.994923076923077e-06,
+      "loss": 0.8333,
+      "step": 361200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.987230769230769e-06,
+      "loss": 0.7862,
+      "step": 361300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9795384615384615e-06,
+      "loss": 0.7994,
+      "step": 361400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.971846153846154e-06,
+      "loss": 0.8154,
+      "step": 361500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9641538461538466e-06,
+      "loss": 0.7954,
+      "step": 361600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9564615384615385e-06,
+      "loss": 0.7946,
+      "step": 361700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.948769230769231e-06,
+      "loss": 0.7803,
+      "step": 361800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9410769230769232e-06,
+      "loss": 0.759,
+      "step": 361900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.933384615384615e-06,
+      "loss": 0.7965,
+      "step": 362000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9256923076923075e-06,
+      "loss": 0.8113,
+      "step": 362100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9180000000000003e-06,
+      "loss": 0.8,
+      "step": 362200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9103076923076926e-06,
+      "loss": 0.7784,
+      "step": 362300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9026153846153845e-06,
+      "loss": 0.7823,
+      "step": 362400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.894923076923077e-06,
+      "loss": 0.8065,
+      "step": 362500
+    },
+    {
+      "epoch": 1.63,
+      "eval_loss": 1.1505937576293945,
+      "eval_runtime": 140.0056,
+      "eval_samples_per_second": 14.285,
+      "eval_steps_per_second": 0.893,
+      "step": 362500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8872307692307692e-06,
+      "loss": 0.77,
+      "step": 362600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8795384615384616e-06,
+      "loss": 0.7727,
+      "step": 362700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.871923076923077e-06,
+      "loss": 0.7872,
+      "step": 362800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.864230769230769e-06,
+      "loss": 0.7612,
+      "step": 362900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8565384615384616e-06,
+      "loss": 0.8135,
+      "step": 363000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.848846153846154e-06,
+      "loss": 0.8211,
+      "step": 363100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8411538461538463e-06,
+      "loss": 0.7879,
+      "step": 363200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8334615384615386e-06,
+      "loss": 0.8213,
+      "step": 363300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8257692307692305e-06,
+      "loss": 0.8152,
+      "step": 363400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.818076923076923e-06,
+      "loss": 0.823,
+      "step": 363500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.8103846153846156e-06,
+      "loss": 0.7969,
+      "step": 363600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.802692307692308e-06,
+      "loss": 0.7961,
+      "step": 363700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.795e-06,
+      "loss": 0.7821,
+      "step": 363800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7873076923076923e-06,
+      "loss": 0.8052,
+      "step": 363900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7796153846153846e-06,
+      "loss": 0.7994,
+      "step": 364000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.771923076923077e-06,
+      "loss": 0.807,
+      "step": 364100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7642307692307693e-06,
+      "loss": 0.7867,
+      "step": 364200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7566153846153846e-06,
+      "loss": 0.7896,
+      "step": 364300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.748923076923077e-06,
+      "loss": 0.7711,
+      "step": 364400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7412307692307693e-06,
+      "loss": 0.7875,
+      "step": 364500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7335384615384616e-06,
+      "loss": 0.8202,
+      "step": 364600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.725846153846154e-06,
+      "loss": 0.8164,
+      "step": 364700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.718153846153846e-06,
+      "loss": 0.7963,
+      "step": 364800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7104615384615383e-06,
+      "loss": 0.7894,
+      "step": 364900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.702769230769231e-06,
+      "loss": 0.7558,
+      "step": 365000
+    },
+    {
+      "epoch": 1.64,
+      "eval_loss": 1.1470996141433716,
+      "eval_runtime": 132.4199,
+      "eval_samples_per_second": 15.103,
+      "eval_steps_per_second": 0.944,
+      "step": 365000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6950769230769234e-06,
+      "loss": 0.7668,
+      "step": 365100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6873846153846153e-06,
+      "loss": 0.7843,
+      "step": 365200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6796923076923076e-06,
+      "loss": 0.801,
+      "step": 365300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.672e-06,
+      "loss": 0.7785,
+      "step": 365400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6643076923076923e-06,
+      "loss": 0.783,
+      "step": 365500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6566153846153847e-06,
+      "loss": 0.807,
+      "step": 365600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.648923076923077e-06,
+      "loss": 0.7822,
+      "step": 365700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6412307692307694e-06,
+      "loss": 0.8248,
+      "step": 365800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6335384615384617e-06,
+      "loss": 0.8026,
+      "step": 365900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6258461538461537e-06,
+      "loss": 0.8037,
+      "step": 366000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6181538461538464e-06,
+      "loss": 0.8116,
+      "step": 366100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6104615384615388e-06,
+      "loss": 0.8012,
+      "step": 366200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6027692307692307e-06,
+      "loss": 0.8025,
+      "step": 366300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.595076923076923e-06,
+      "loss": 0.7849,
+      "step": 366400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5873846153846154e-06,
+      "loss": 0.7881,
+      "step": 366500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5796923076923078e-06,
+      "loss": 0.7869,
+      "step": 366600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.572e-06,
+      "loss": 0.8099,
+      "step": 366700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5643076923076925e-06,
+      "loss": 0.801,
+      "step": 366800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.556615384615385e-06,
+      "loss": 0.8272,
+      "step": 366900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.548923076923077e-06,
+      "loss": 0.7886,
+      "step": 367000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.541230769230769e-06,
+      "loss": 0.7696,
+      "step": 367100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5335384615384614e-06,
+      "loss": 0.8206,
+      "step": 367200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.525846153846154e-06,
+      "loss": 0.8389,
+      "step": 367300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5181538461538465e-06,
+      "loss": 0.796,
+      "step": 367400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5104615384615385e-06,
+      "loss": 0.7927,
+      "step": 367500
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 1.150688648223877,
+      "eval_runtime": 132.6379,
+      "eval_samples_per_second": 15.079,
+      "eval_steps_per_second": 0.942,
+      "step": 367500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.502769230769231e-06,
+      "loss": 0.7984,
+      "step": 367600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.495076923076923e-06,
+      "loss": 0.8053,
+      "step": 367700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.487384615384615e-06,
+      "loss": 0.768,
+      "step": 367800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.479692307692308e-06,
+      "loss": 0.8277,
+      "step": 367900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4720000000000002e-06,
+      "loss": 0.7819,
+      "step": 368000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4643076923076926e-06,
+      "loss": 0.8397,
+      "step": 368100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4566153846153845e-06,
+      "loss": 0.819,
+      "step": 368200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.448923076923077e-06,
+      "loss": 0.7929,
+      "step": 368300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4413076923076925e-06,
+      "loss": 0.7999,
+      "step": 368400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4336153846153845e-06,
+      "loss": 0.7763,
+      "step": 368500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.425923076923077e-06,
+      "loss": 0.7872,
+      "step": 368600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4182307692307696e-06,
+      "loss": 0.7747,
+      "step": 368700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.410538461538462e-06,
+      "loss": 0.7777,
+      "step": 368800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.402846153846154e-06,
+      "loss": 0.8023,
+      "step": 368900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.395153846153846e-06,
+      "loss": 0.7836,
+      "step": 369000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3874615384615386e-06,
+      "loss": 0.8267,
+      "step": 369100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3797692307692305e-06,
+      "loss": 0.7535,
+      "step": 369200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3720769230769232e-06,
+      "loss": 0.8046,
+      "step": 369300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3643846153846156e-06,
+      "loss": 0.8005,
+      "step": 369400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.356692307692308e-06,
+      "loss": 0.7943,
+      "step": 369500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.349e-06,
+      "loss": 0.7664,
+      "step": 369600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3413076923076922e-06,
+      "loss": 0.7715,
+      "step": 369700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3336153846153846e-06,
+      "loss": 0.7609,
+      "step": 369800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3259230769230773e-06,
+      "loss": 0.8006,
+      "step": 369900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3182307692307693e-06,
+      "loss": 0.7946,
+      "step": 370000
+    },
+    {
+      "epoch": 1.66,
+      "eval_loss": 1.1560249328613281,
+      "eval_runtime": 132.5639,
+      "eval_samples_per_second": 15.087,
+      "eval_steps_per_second": 0.943,
+      "step": 370000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3105384615384616e-06,
+      "loss": 0.7796,
+      "step": 370100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.302846153846154e-06,
+      "loss": 0.801,
+      "step": 370200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2951538461538463e-06,
+      "loss": 0.796,
+      "step": 370300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2874615384615382e-06,
+      "loss": 0.7897,
+      "step": 370400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.279769230769231e-06,
+      "loss": 0.7667,
+      "step": 370500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.272153846153846e-06,
+      "loss": 0.7682,
+      "step": 370600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2644615384615386e-06,
+      "loss": 0.7736,
+      "step": 370700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.256769230769231e-06,
+      "loss": 0.7918,
+      "step": 370800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2490769230769233e-06,
+      "loss": 0.7736,
+      "step": 370900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2413846153846153e-06,
+      "loss": 0.778,
+      "step": 371000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2336923076923076e-06,
+      "loss": 0.7967,
+      "step": 371100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.226e-06,
+      "loss": 0.8045,
+      "step": 371200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2183076923076927e-06,
+      "loss": 0.7822,
+      "step": 371300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2106153846153846e-06,
+      "loss": 0.7946,
+      "step": 371400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.202923076923077e-06,
+      "loss": 0.7946,
+      "step": 371500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1952307692307693e-06,
+      "loss": 0.7867,
+      "step": 371600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1875384615384617e-06,
+      "loss": 0.7813,
+      "step": 371700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1798461538461536e-06,
+      "loss": 0.7681,
+      "step": 371800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1721538461538464e-06,
+      "loss": 0.8119,
+      "step": 371900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1644615384615387e-06,
+      "loss": 0.7952,
+      "step": 372000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1567692307692307e-06,
+      "loss": 0.8023,
+      "step": 372100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.149076923076923e-06,
+      "loss": 0.7781,
+      "step": 372200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1413846153846154e-06,
+      "loss": 0.784,
+      "step": 372300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1336923076923077e-06,
+      "loss": 0.7829,
+      "step": 372400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.126e-06,
+      "loss": 0.7979,
+      "step": 372500
+    },
+    {
+      "epoch": 1.67,
+      "eval_loss": 1.147834300994873,
+      "eval_runtime": 131.5858,
+      "eval_samples_per_second": 15.199,
+      "eval_steps_per_second": 0.95,
+      "step": 372500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1183076923076924e-06,
+      "loss": 0.7967,
+      "step": 372600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1106153846153848e-06,
+      "loss": 0.7757,
+      "step": 372700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.102923076923077e-06,
+      "loss": 0.7978,
+      "step": 372800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0953076923076924e-06,
+      "loss": 0.794,
+      "step": 372900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0876153846153847e-06,
+      "loss": 0.7701,
+      "step": 373000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.079923076923077e-06,
+      "loss": 0.7915,
+      "step": 373100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.072230769230769e-06,
+      "loss": 0.7833,
+      "step": 373200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0645384615384618e-06,
+      "loss": 0.8182,
+      "step": 373300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.056846153846154e-06,
+      "loss": 0.7732,
+      "step": 373400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.049153846153846e-06,
+      "loss": 0.7965,
+      "step": 373500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0414615384615384e-06,
+      "loss": 0.826,
+      "step": 373600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0337692307692307e-06,
+      "loss": 0.8053,
+      "step": 373700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.026076923076923e-06,
+      "loss": 0.7808,
+      "step": 373800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0183846153846154e-06,
+      "loss": 0.7494,
+      "step": 373900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0107692307692307e-06,
+      "loss": 0.8035,
+      "step": 374000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0030769230769235e-06,
+      "loss": 0.7595,
+      "step": 374100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9953846153846154e-06,
+      "loss": 0.8059,
+      "step": 374200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9876923076923078e-06,
+      "loss": 0.7949,
+      "step": 374300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.98e-06,
+      "loss": 0.7577,
+      "step": 374400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9723076923076924e-06,
+      "loss": 0.7883,
+      "step": 374500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9646153846153844e-06,
+      "loss": 0.7745,
+      "step": 374600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.956923076923077e-06,
+      "loss": 0.7625,
+      "step": 374700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9492307692307695e-06,
+      "loss": 0.823,
+      "step": 374800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9415384615384614e-06,
+      "loss": 0.7706,
+      "step": 374900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9338461538461538e-06,
+      "loss": 0.7472,
+      "step": 375000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.1514644622802734,
+      "eval_runtime": 132.2515,
+      "eval_samples_per_second": 15.123,
+      "eval_steps_per_second": 0.945,
+      "step": 375000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.926153846153846e-06,
+      "loss": 0.7476,
+      "step": 375100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9184615384615385e-06,
+      "loss": 0.8041,
+      "step": 375200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.910769230769231e-06,
+      "loss": 0.8107,
+      "step": 375300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9030769230769232e-06,
+      "loss": 0.8004,
+      "step": 375400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8953846153846155e-06,
+      "loss": 0.7825,
+      "step": 375500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8876923076923079e-06,
+      "loss": 0.8024,
+      "step": 375600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.88e-06,
+      "loss": 0.8146,
+      "step": 375700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8723076923076923e-06,
+      "loss": 0.7962,
+      "step": 375800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8646153846153847e-06,
+      "loss": 0.8121,
+      "step": 375900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.856923076923077e-06,
+      "loss": 0.7813,
+      "step": 376000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8492307692307694e-06,
+      "loss": 0.7935,
+      "step": 376100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8415384615384615e-06,
+      "loss": 0.7804,
+      "step": 376200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8338461538461539e-06,
+      "loss": 0.7957,
+      "step": 376300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8261538461538462e-06,
+      "loss": 0.8149,
+      "step": 376400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8184615384615384e-06,
+      "loss": 0.7951,
+      "step": 376500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.810769230769231e-06,
+      "loss": 0.7849,
+      "step": 376600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.803076923076923e-06,
+      "loss": 0.7959,
+      "step": 376700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7953846153846154e-06,
+      "loss": 0.8196,
+      "step": 376800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7876923076923078e-06,
+      "loss": 0.7789,
+      "step": 376900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.7889,
+      "step": 377000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7723076923076922e-06,
+      "loss": 0.7854,
+      "step": 377100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7646153846153848e-06,
+      "loss": 0.798,
+      "step": 377200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.756923076923077e-06,
+      "loss": 0.7864,
+      "step": 377300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7492307692307693e-06,
+      "loss": 0.7521,
+      "step": 377400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7415384615384616e-06,
+      "loss": 0.7984,
+      "step": 377500
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 1.148796558380127,
+      "eval_runtime": 135.3287,
+      "eval_samples_per_second": 14.779,
+      "eval_steps_per_second": 0.924,
+      "step": 377500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.733846153846154e-06,
+      "loss": 0.7639,
+      "step": 377600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7261538461538461e-06,
+      "loss": 0.7947,
+      "step": 377700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7184615384615385e-06,
+      "loss": 0.831,
+      "step": 377800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7107692307692308e-06,
+      "loss": 0.804,
+      "step": 377900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.703076923076923e-06,
+      "loss": 0.7622,
+      "step": 378000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6953846153846155e-06,
+      "loss": 0.8009,
+      "step": 378100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6876923076923077e-06,
+      "loss": 0.7849,
+      "step": 378200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6800769230769231e-06,
+      "loss": 0.7825,
+      "step": 378300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6723846153846155e-06,
+      "loss": 0.835,
+      "step": 378400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6646923076923076e-06,
+      "loss": 0.7871,
+      "step": 378500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6570000000000002e-06,
+      "loss": 0.7443,
+      "step": 378600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6493076923076923e-06,
+      "loss": 0.7842,
+      "step": 378700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6416153846153847e-06,
+      "loss": 0.7682,
+      "step": 378800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.633923076923077e-06,
+      "loss": 0.8161,
+      "step": 378900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6262307692307694e-06,
+      "loss": 0.7715,
+      "step": 379000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6185384615384615e-06,
+      "loss": 0.8036,
+      "step": 379100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6108461538461539e-06,
+      "loss": 0.7636,
+      "step": 379200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6031538461538462e-06,
+      "loss": 0.8488,
+      "step": 379300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5954615384615383e-06,
+      "loss": 0.7783,
+      "step": 379400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.587769230769231e-06,
+      "loss": 0.7825,
+      "step": 379500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.580076923076923e-06,
+      "loss": 0.7428,
+      "step": 379600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5723846153846154e-06,
+      "loss": 0.7754,
+      "step": 379700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5646923076923077e-06,
+      "loss": 0.7828,
+      "step": 379800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.557e-06,
+      "loss": 0.7775,
+      "step": 379900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5493076923076922e-06,
+      "loss": 0.8326,
+      "step": 380000
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 1.1469863653182983,
+      "eval_runtime": 132.1004,
+      "eval_samples_per_second": 15.14,
+      "eval_steps_per_second": 0.946,
+      "step": 380000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5416153846153848e-06,
+      "loss": 0.796,
+      "step": 380100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.533923076923077e-06,
+      "loss": 0.7841,
+      "step": 380200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5263076923076924e-06,
+      "loss": 0.8189,
+      "step": 380300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5186153846153847e-06,
+      "loss": 0.7775,
+      "step": 380400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5109230769230769e-06,
+      "loss": 0.784,
+      "step": 380500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5032307692307692e-06,
+      "loss": 0.7684,
+      "step": 380600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4955384615384616e-06,
+      "loss": 0.7834,
+      "step": 380700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4878461538461537e-06,
+      "loss": 0.7983,
+      "step": 380800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4801538461538463e-06,
+      "loss": 0.8022,
+      "step": 380900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4724615384615384e-06,
+      "loss": 0.7869,
+      "step": 381000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4647692307692308e-06,
+      "loss": 0.8154,
+      "step": 381100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4570769230769231e-06,
+      "loss": 0.8036,
+      "step": 381200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4493846153846155e-06,
+      "loss": 0.7981,
+      "step": 381300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4416923076923076e-06,
+      "loss": 0.7791,
+      "step": 381400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4340000000000002e-06,
+      "loss": 0.7858,
+      "step": 381500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4263076923076923e-06,
+      "loss": 0.7597,
+      "step": 381600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4186153846153846e-06,
+      "loss": 0.7744,
+      "step": 381700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.410923076923077e-06,
+      "loss": 0.797,
+      "step": 381800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4032307692307693e-06,
+      "loss": 0.7641,
+      "step": 381900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3955384615384615e-06,
+      "loss": 0.7591,
+      "step": 382000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3878461538461538e-06,
+      "loss": 0.7794,
+      "step": 382100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3801538461538462e-06,
+      "loss": 0.7874,
+      "step": 382200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3724615384615383e-06,
+      "loss": 0.8223,
+      "step": 382300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3647692307692309e-06,
+      "loss": 0.7878,
+      "step": 382400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.357076923076923e-06,
+      "loss": 0.8211,
+      "step": 382500
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 1.1496639251708984,
+      "eval_runtime": 132.1891,
+      "eval_samples_per_second": 15.13,
+      "eval_steps_per_second": 0.946,
+      "step": 382500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3493846153846154e-06,
+      "loss": 0.8205,
+      "step": 382600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3416923076923077e-06,
+      "loss": 0.8047,
+      "step": 382700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.334e-06,
+      "loss": 0.7624,
+      "step": 382800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3263846153846155e-06,
+      "loss": 0.7977,
+      "step": 382900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3186923076923077e-06,
+      "loss": 0.7712,
+      "step": 383000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.311e-06,
+      "loss": 0.7778,
+      "step": 383100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3033076923076924e-06,
+      "loss": 0.7874,
+      "step": 383200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2956153846153847e-06,
+      "loss": 0.7706,
+      "step": 383300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2879230769230769e-06,
+      "loss": 0.7664,
+      "step": 383400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2802307692307692e-06,
+      "loss": 0.8057,
+      "step": 383500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2725384615384616e-06,
+      "loss": 0.792,
+      "step": 383600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2648461538461537e-06,
+      "loss": 0.7891,
+      "step": 383700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2571538461538463e-06,
+      "loss": 0.7899,
+      "step": 383800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2494615384615384e-06,
+      "loss": 0.7938,
+      "step": 383900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2417692307692307e-06,
+      "loss": 0.7925,
+      "step": 384000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.234076923076923e-06,
+      "loss": 0.7961,
+      "step": 384100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2263846153846154e-06,
+      "loss": 0.7884,
+      "step": 384200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2186923076923076e-06,
+      "loss": 0.7513,
+      "step": 384300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2110000000000001e-06,
+      "loss": 0.7836,
+      "step": 384400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2033076923076923e-06,
+      "loss": 0.7991,
+      "step": 384500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1956153846153846e-06,
+      "loss": 0.7558,
+      "step": 384600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.187923076923077e-06,
+      "loss": 0.7792,
+      "step": 384700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1802307692307693e-06,
+      "loss": 0.7988,
+      "step": 384800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1725384615384615e-06,
+      "loss": 0.7718,
+      "step": 384900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.164846153846154e-06,
+      "loss": 0.8031,
+      "step": 385000
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 1.146566390991211,
+      "eval_runtime": 132.6054,
+      "eval_samples_per_second": 15.082,
+      "eval_steps_per_second": 0.943,
+      "step": 385000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1571538461538462e-06,
+      "loss": 0.8213,
+      "step": 385100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1494615384615383e-06,
+      "loss": 0.7975,
+      "step": 385200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1418461538461538e-06,
+      "loss": 0.7783,
+      "step": 385300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1341538461538461e-06,
+      "loss": 0.7922,
+      "step": 385400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1264615384615385e-06,
+      "loss": 0.7798,
+      "step": 385500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1187692307692308e-06,
+      "loss": 0.7836,
+      "step": 385600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.111076923076923e-06,
+      "loss": 0.772,
+      "step": 385700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1033846153846155e-06,
+      "loss": 0.7573,
+      "step": 385800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0956923076923077e-06,
+      "loss": 0.7861,
+      "step": 385900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.088e-06,
+      "loss": 0.7954,
+      "step": 386000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0803076923076924e-06,
+      "loss": 0.795,
+      "step": 386100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0726153846153847e-06,
+      "loss": 0.7894,
+      "step": 386200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.064923076923077e-06,
+      "loss": 0.7777,
+      "step": 386300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0572307692307692e-06,
+      "loss": 0.7869,
+      "step": 386400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0495384615384615e-06,
+      "loss": 0.7962,
+      "step": 386500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0418461538461539e-06,
+      "loss": 0.8036,
+      "step": 386600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0341538461538462e-06,
+      "loss": 0.795,
+      "step": 386700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0264615384615384e-06,
+      "loss": 0.786,
+      "step": 386800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.018769230769231e-06,
+      "loss": 0.7797,
+      "step": 386900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.011076923076923e-06,
+      "loss": 0.8004,
+      "step": 387000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0033846153846154e-06,
+      "loss": 0.7927,
+      "step": 387100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.956923076923078e-07,
+      "loss": 0.8093,
+      "step": 387200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.880000000000001e-07,
+      "loss": 0.7943,
+      "step": 387300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.803076923076922e-07,
+      "loss": 0.8188,
+      "step": 387400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.726153846153848e-07,
+      "loss": 0.7787,
+      "step": 387500
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 1.1474705934524536,
+      "eval_runtime": 133.1402,
+      "eval_samples_per_second": 15.022,
+      "eval_steps_per_second": 0.939,
+      "step": 387500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.64923076923077e-07,
+      "loss": 0.7939,
+      "step": 387600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.572307692307693e-07,
+      "loss": 0.7784,
+      "step": 387700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.496153846153846e-07,
+      "loss": 0.7883,
+      "step": 387800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.419230769230769e-07,
+      "loss": 0.8229,
+      "step": 387900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.342307692307693e-07,
+      "loss": 0.8185,
+      "step": 388000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.265384615384616e-07,
+      "loss": 0.8403,
+      "step": 388100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.188461538461539e-07,
+      "loss": 0.804,
+      "step": 388200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.111538461538462e-07,
+      "loss": 0.7827,
+      "step": 388300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.034615384615386e-07,
+      "loss": 0.7833,
+      "step": 388400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.957692307692308e-07,
+      "loss": 0.7951,
+      "step": 388500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.88076923076923e-07,
+      "loss": 0.788,
+      "step": 388600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.803846153846154e-07,
+      "loss": 0.788,
+      "step": 388700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.726923076923076e-07,
+      "loss": 0.7753,
+      "step": 388800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.65e-07,
+      "loss": 0.7786,
+      "step": 388900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.573076923076923e-07,
+      "loss": 0.7923,
+      "step": 389000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.496153846153846e-07,
+      "loss": 0.7848,
+      "step": 389100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.419230769230769e-07,
+      "loss": 0.7532,
+      "step": 389200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.342307692307693e-07,
+      "loss": 0.7794,
+      "step": 389300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.265384615384616e-07,
+      "loss": 0.8062,
+      "step": 389400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.188461538461539e-07,
+      "loss": 0.8065,
+      "step": 389500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.111538461538462e-07,
+      "loss": 0.8285,
+      "step": 389600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.034615384615386e-07,
+      "loss": 0.783,
+      "step": 389700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.957692307692308e-07,
+      "loss": 0.7781,
+      "step": 389800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.880769230769231e-07,
+      "loss": 0.7854,
+      "step": 389900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803846153846155e-07,
+      "loss": 0.7619,
+      "step": 390000
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 1.1405349969863892,
+      "eval_runtime": 131.7885,
+      "eval_samples_per_second": 15.176,
+      "eval_steps_per_second": 0.948,
+      "step": 390000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.727692307692308e-07,
+      "loss": 0.77,
+      "step": 390100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.650769230769231e-07,
+      "loss": 0.8003,
+      "step": 390200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.573846153846154e-07,
+      "loss": 0.8342,
+      "step": 390300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.496923076923077e-07,
+      "loss": 0.7898,
+      "step": 390400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.420000000000001e-07,
+      "loss": 0.7916,
+      "step": 390500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.343076923076923e-07,
+      "loss": 0.7827,
+      "step": 390600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.266153846153846e-07,
+      "loss": 0.7569,
+      "step": 390700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.18923076923077e-07,
+      "loss": 0.7985,
+      "step": 390800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.112307692307692e-07,
+      "loss": 0.7683,
+      "step": 390900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.035384615384616e-07,
+      "loss": 0.7937,
+      "step": 391000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.958461538461539e-07,
+      "loss": 0.757,
+      "step": 391100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.881538461538462e-07,
+      "loss": 0.7648,
+      "step": 391200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.804615384615385e-07,
+      "loss": 0.7915,
+      "step": 391300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.727692307692309e-07,
+      "loss": 0.8161,
+      "step": 391400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.650769230769231e-07,
+      "loss": 0.8023,
+      "step": 391500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.573846153846154e-07,
+      "loss": 0.8085,
+      "step": 391600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.496923076923077e-07,
+      "loss": 0.7683,
+      "step": 391700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.42e-07,
+      "loss": 0.7803,
+      "step": 391800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.343076923076923e-07,
+      "loss": 0.7876,
+      "step": 391900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.266153846153846e-07,
+      "loss": 0.8039,
+      "step": 392000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.189230769230769e-07,
+      "loss": 0.8001,
+      "step": 392100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.113076923076924e-07,
+      "loss": 0.7965,
+      "step": 392200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.036153846153846e-07,
+      "loss": 0.7638,
+      "step": 392300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.95923076923077e-07,
+      "loss": 0.769,
+      "step": 392400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.882307692307693e-07,
+      "loss": 0.8261,
+      "step": 392500
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 1.142776608467102,
+      "eval_runtime": 132.4957,
+      "eval_samples_per_second": 15.095,
+      "eval_steps_per_second": 0.943,
+      "step": 392500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.805384615384616e-07,
+      "loss": 0.7768,
+      "step": 392600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.728461538461539e-07,
+      "loss": 0.7615,
+      "step": 392700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.651538461538463e-07,
+      "loss": 0.7932,
+      "step": 392800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.574615384615384e-07,
+      "loss": 0.7624,
+      "step": 392900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.497692307692307e-07,
+      "loss": 0.7975,
+      "step": 393000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.420769230769231e-07,
+      "loss": 0.811,
+      "step": 393100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.343846153846153e-07,
+      "loss": 0.8155,
+      "step": 393200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.266923076923077e-07,
+      "loss": 0.8029,
+      "step": 393300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.19e-07,
+      "loss": 0.7897,
+      "step": 393400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.113076923076923e-07,
+      "loss": 0.8171,
+      "step": 393500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.036153846153846e-07,
+      "loss": 0.7708,
+      "step": 393600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.95923076923077e-07,
+      "loss": 0.7763,
+      "step": 393700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.882307692307692e-07,
+      "loss": 0.8022,
+      "step": 393800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.805384615384616e-07,
+      "loss": 0.7896,
+      "step": 393900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7284615384615386e-07,
+      "loss": 0.7555,
+      "step": 394000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.6515384615384615e-07,
+      "loss": 0.7977,
+      "step": 394100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5746153846153845e-07,
+      "loss": 0.7986,
+      "step": 394200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.4976923076923074e-07,
+      "loss": 0.7848,
+      "step": 394300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.420769230769231e-07,
+      "loss": 0.7656,
+      "step": 394400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.343846153846154e-07,
+      "loss": 0.7887,
+      "step": 394500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.2669230769230774e-07,
+      "loss": 0.7754,
+      "step": 394600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1907692307692306e-07,
+      "loss": 0.8247,
+      "step": 394700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1138461538461535e-07,
+      "loss": 0.7725,
+      "step": 394800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.036923076923077e-07,
+      "loss": 0.7607,
+      "step": 394900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.96e-07,
+      "loss": 0.7703,
+      "step": 395000
+    },
+    {
+      "epoch": 1.77,
+      "eval_loss": 1.141352891921997,
+      "eval_runtime": 133.4532,
+      "eval_samples_per_second": 14.987,
+      "eval_steps_per_second": 0.937,
+      "step": 395000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8830769230769235e-07,
+      "loss": 0.7805,
+      "step": 395100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8061538461538464e-07,
+      "loss": 0.7712,
+      "step": 395200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7292307692307694e-07,
+      "loss": 0.7816,
+      "step": 395300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.652307692307693e-07,
+      "loss": 0.7793,
+      "step": 395400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5753846153846153e-07,
+      "loss": 0.7115,
+      "step": 395500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4984615384615383e-07,
+      "loss": 0.7648,
+      "step": 395600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.421538461538462e-07,
+      "loss": 0.7506,
+      "step": 395700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3446153846153847e-07,
+      "loss": 0.7979,
+      "step": 395800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2676923076923077e-07,
+      "loss": 0.7417,
+      "step": 395900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1915384615384614e-07,
+      "loss": 0.7903,
+      "step": 396000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1146153846153844e-07,
+      "loss": 0.7763,
+      "step": 396100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.037692307692308e-07,
+      "loss": 0.7791,
+      "step": 396200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.960769230769231e-07,
+      "loss": 0.7887,
+      "step": 396300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.883846153846154e-07,
+      "loss": 0.7851,
+      "step": 396400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8069230769230773e-07,
+      "loss": 0.7981,
+      "step": 396500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.73e-07,
+      "loss": 0.7995,
+      "step": 396600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.653076923076923e-07,
+      "loss": 0.8106,
+      "step": 396700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.576153846153846e-07,
+      "loss": 0.7634,
+      "step": 396800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.499230769230769e-07,
+      "loss": 0.8249,
+      "step": 396900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.422307692307692e-07,
+      "loss": 0.8166,
+      "step": 397000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3453846153846153e-07,
+      "loss": 0.7968,
+      "step": 397100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2684615384615385e-07,
+      "loss": 0.7673,
+      "step": 397200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1915384615384617e-07,
+      "loss": 0.7869,
+      "step": 397300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1146153846153844e-07,
+      "loss": 0.77,
+      "step": 397400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0376923076923076e-07,
+      "loss": 0.7701,
+      "step": 397500
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 1.1418771743774414,
+      "eval_runtime": 134.5194,
+      "eval_samples_per_second": 14.868,
+      "eval_steps_per_second": 0.929,
+      "step": 397500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9607692307692309e-07,
+      "loss": 0.8229,
+      "step": 397600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.883846153846154e-07,
+      "loss": 0.7781,
+      "step": 397700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8069230769230768e-07,
+      "loss": 0.8093,
+      "step": 397800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.73e-07,
+      "loss": 0.7774,
+      "step": 397900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6530769230769232e-07,
+      "loss": 0.7858,
+      "step": 398000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.5761538461538462e-07,
+      "loss": 0.7885,
+      "step": 398100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4992307692307691e-07,
+      "loss": 0.7738,
+      "step": 398200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4223076923076924e-07,
+      "loss": 0.7755,
+      "step": 398300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3453846153846153e-07,
+      "loss": 0.7877,
+      "step": 398400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2684615384615385e-07,
+      "loss": 0.7785,
+      "step": 398500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1915384615384615e-07,
+      "loss": 0.8049,
+      "step": 398600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1146153846153846e-07,
+      "loss": 0.7722,
+      "step": 398700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.0376923076923077e-07,
+      "loss": 0.7856,
+      "step": 398800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.607692307692308e-08,
+      "loss": 0.7917,
+      "step": 398900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.838461538461539e-08,
+      "loss": 0.782,
+      "step": 399000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.069230769230768e-08,
+      "loss": 0.7903,
+      "step": 399100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3e-08,
+      "loss": 0.7766,
+      "step": 399200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.53076923076923e-08,
+      "loss": 0.7852,
+      "step": 399300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7615384615384617e-08,
+      "loss": 0.7599,
+      "step": 399400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9923076923076926e-08,
+      "loss": 0.7744,
+      "step": 399500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.2230769230769235e-08,
+      "loss": 0.7723,
+      "step": 399600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.453846153846154e-08,
+      "loss": 0.8049,
+      "step": 399700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6846153846153846e-08,
+      "loss": 0.818,
+      "step": 399800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9153846153846155e-08,
+      "loss": 0.7848,
+      "step": 399900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1461538461538463e-08,
+      "loss": 0.7543,
+      "step": 400000
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 1.1417272090911865,
+      "eval_runtime": 132.3963,
+      "eval_samples_per_second": 15.106,
+      "eval_steps_per_second": 0.944,
+      "step": 400000
+    }
+  ],
+  "max_steps": 400000,
+  "num_train_epochs": 2,
+  "total_flos": 3.167022778492335e+19,
+  "trial_name": null,
+  "trial_params": null
+}