diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
-  "best_metric": 6.580160140991211,
-  "best_model_checkpoint": "./output_v2/34bCodellama_CodeLlama-34b-Python-hf_unnatural-instructions_standardized/checkpoint-2200",
-  "epoch": 0.02138874035596975,
-  "global_step": 2800,
+  "best_metric": 6.423073768615723,
+  "best_model_checkpoint": "./output_v2/34bCodellama_CodeLlama-34b-Python-hf_unnatural-instructions_standardized/checkpoint-3200",
+  "epoch": 0.02902757619738752,
+  "global_step": 3800,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -17067,11 +17067,6110 @@
       "train_runtime": 7855.0688,
       "train_samples_per_second": 3.819,
       "train_steps_per_second": 3.819
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0757,
+      "step": 2801
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.8944,
+      "step": 2802
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8365,
+      "step": 2803
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.873,
+      "step": 2804
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3793,
+      "step": 2805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1923,
+      "step": 2806
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.2306,
+      "step": 2807
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.4439,
+      "step": 2808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3377,
+      "step": 2809
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8737,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4191,
+      "step": 2811
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.892,
+      "step": 2812
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4568,
+      "step": 2813
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0652,
+      "step": 2814
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.6912,
+      "step": 2815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.9033,
+      "step": 2816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4433,
+      "step": 2817
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7591,
+      "step": 2818
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4458,
+      "step": 2819
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3721,
+      "step": 2820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4111,
+      "step": 2821
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0999,
+      "step": 2822
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.5154,
+      "step": 2823
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1967,
+      "step": 2824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8437,
+      "step": 2825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.759,
+      "step": 2826
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6223,
+      "step": 2827
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3442,
+      "step": 2828
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1991,
+      "step": 2829
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5302,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1725,
+      "step": 2831
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8107,
+      "step": 2832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.7891,
+      "step": 2833
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5592,
+      "step": 2834
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8792,
+      "step": 2835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2387,
+      "step": 2836
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9369,
+      "step": 2837
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2098,
+      "step": 2838
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6645,
+      "step": 2839
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2349,
+      "step": 2840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8613,
+      "step": 2841
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5482,
+      "step": 2842
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5438,
+      "step": 2843
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6133,
+      "step": 2844
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.9983,
+      "step": 2845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8706,
+      "step": 2846
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9685,
+      "step": 2847
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.04,
+      "step": 2848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6498,
+      "step": 2849
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6786,
+      "step": 2850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.789,
+      "step": 2851
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 10.1116,
+      "step": 2852
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.7085,
+      "step": 2853
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1083,
+      "step": 2854
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0795,
+      "step": 2855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8677,
+      "step": 2856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1663,
+      "step": 2857
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5379,
+      "step": 2858
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4923,
+      "step": 2859
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1456,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1686,
+      "step": 2861
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4153,
+      "step": 2862
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.17,
+      "step": 2863
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3301,
+      "step": 2864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7106,
+      "step": 2865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.463,
+      "step": 2866
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.614,
+      "step": 2867
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1451,
+      "step": 2868
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6817,
+      "step": 2869
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9577,
+      "step": 2870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.6171,
+      "step": 2871
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5129,
+      "step": 2872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3386,
+      "step": 2873
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1176,
+      "step": 2874
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.9075,
+      "step": 2875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.667,
+      "step": 2876
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.8097,
+      "step": 2877
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7707,
+      "step": 2878
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7571,
+      "step": 2879
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0732,
+      "step": 2880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5586,
+      "step": 2881
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8045,
+      "step": 2882
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.4442,
+      "step": 2883
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.418,
+      "step": 2884
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7253,
+      "step": 2885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4227,
+      "step": 2886
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9878,
+      "step": 2887
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.8351,
+      "step": 2888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1715,
+      "step": 2889
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1207,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0396,
+      "step": 2891
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7162,
+      "step": 2892
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2566,
+      "step": 2893
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4087,
+      "step": 2894
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.4509,
+      "step": 2895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8165,
+      "step": 2896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9842,
+      "step": 2897
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.574,
+      "step": 2898
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4512,
+      "step": 2899
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9904,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.6141,
+      "step": 2901
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9528,
+      "step": 2902
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.9028,
+      "step": 2903
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3868,
+      "step": 2904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0461,
+      "step": 2905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.5161,
+      "step": 2906
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.107,
+      "step": 2907
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7174,
+      "step": 2908
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7671,
+      "step": 2909
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6803,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5357,
+      "step": 2911
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6898,
+      "step": 2912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8564,
+      "step": 2913
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1457,
+      "step": 2914
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3408,
+      "step": 2915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.6095,
+      "step": 2916
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.866,
+      "step": 2917
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7324,
+      "step": 2918
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4026,
+      "step": 2919
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1467,
+      "step": 2920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2496,
+      "step": 2921
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.5265,
+      "step": 2922
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8246,
+      "step": 2923
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.5775,
+      "step": 2924
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.2886,
+      "step": 2925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3076,
+      "step": 2926
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7975,
+      "step": 2927
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.9898,
+      "step": 2928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7256,
+      "step": 2929
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7971,
+      "step": 2930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5884,
+      "step": 2931
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.0749,
+      "step": 2932
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6183,
+      "step": 2933
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.0019,
+      "step": 2934
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1706,
+      "step": 2935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4075,
+      "step": 2936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.4092,
+      "step": 2937
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9424,
+      "step": 2938
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9025,
+      "step": 2939
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7751,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.988,
+      "step": 2941
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1208,
+      "step": 2942
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1999,
+      "step": 2943
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2129,
+      "step": 2944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4431,
+      "step": 2945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1979,
+      "step": 2946
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8246,
+      "step": 2947
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4876,
+      "step": 2948
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.7158,
+      "step": 2949
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.3611,
+      "step": 2950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.9919,
+      "step": 2951
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4136,
+      "step": 2952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.801,
+      "step": 2953
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.6503,
+      "step": 2954
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.553,
+      "step": 2955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.3536,
+      "step": 2956
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8516,
+      "step": 2957
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 9.9344,
+      "step": 2958
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8727,
+      "step": 2959
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9155,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9684,
+      "step": 2961
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0399,
+      "step": 2962
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.4298,
+      "step": 2963
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4559,
+      "step": 2964
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0361,
+      "step": 2965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0081,
+      "step": 2966
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6698,
+      "step": 2967
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.3355,
+      "step": 2968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.7555,
+      "step": 2969
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.29,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4079,
+      "step": 2971
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0588,
+      "step": 2972
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2956,
+      "step": 2973
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7318,
+      "step": 2974
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8846,
+      "step": 2975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5159,
+      "step": 2976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7629,
+      "step": 2977
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2039,
+      "step": 2978
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.311,
+      "step": 2979
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9786,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7138,
+      "step": 2981
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4827,
+      "step": 2982
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5264,
+      "step": 2983
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8153,
+      "step": 2984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3512,
+      "step": 2985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1515,
+      "step": 2986
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1689,
+      "step": 2987
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8027,
+      "step": 2988
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7472,
+      "step": 2989
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0236,
+      "step": 2990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1741,
+      "step": 2991
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8884,
+      "step": 2992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3912,
+      "step": 2993
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2109,
+      "step": 2994
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1693,
+      "step": 2995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8166,
+      "step": 2996
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4902,
+      "step": 2997
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3921,
+      "step": 2998
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.8787,
+      "step": 2999
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1304,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 6.659167289733887,
+      "eval_runtime": 22.4512,
+      "eval_samples_per_second": 2.227,
+      "eval_steps_per_second": 1.114,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "mmlu_eval_accuracy": 0.2525477994227994,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.07142857142857142,
+      "mmlu_eval_accuracy_astronomy": 0.3125,
+      "mmlu_eval_accuracy_business_ethics": 0.4444444444444444,
+      "mmlu_loss": 3.973116703033447,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.4366,
+      "step": 3001
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1239,
+      "step": 3002
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.419,
+      "step": 3003
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7662,
+      "step": 3004
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1246,
+      "step": 3005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3498,
+      "step": 3006
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1245,
+      "step": 3007
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6488,
+      "step": 3008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3641,
+      "step": 3009
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7842,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.802,
+      "step": 3011
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1015,
+      "step": 3012
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9015,
+      "step": 3013
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8768,
+      "step": 3014
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7296,
+      "step": 3015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4175,
+      "step": 3016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3568,
+      "step": 3017
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5869,
+      "step": 3018
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5304,
+      "step": 3019
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1129,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8128,
+      "step": 3021
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1541,
+      "step": 3022
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3002,
+      "step": 3023
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.0616,
+      "step": 3024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3152,
+      "step": 3025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4433,
+      "step": 3026
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.8406,
+      "step": 3027
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.2653,
+      "step": 3028
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7041,
+      "step": 3029
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3463,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7164,
+      "step": 3031
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9969,
+      "step": 3032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1306,
+      "step": 3033
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0204,
+      "step": 3034
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6743,
+      "step": 3035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3186,
+      "step": 3036
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.5673,
+      "step": 3037
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1416,
+      "step": 3038
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1956,
+      "step": 3039
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6836,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0172,
+      "step": 3041
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.796,
+      "step": 3042
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6945,
+      "step": 3043
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.5079,
+      "step": 3044
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.399,
+      "step": 3045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0071,
+      "step": 3046
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.4171,
+      "step": 3047
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0925,
+      "step": 3048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.6842,
+      "step": 3049
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.2442,
+      "step": 3050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8023,
+      "step": 3051
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7147,
+      "step": 3052
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9681,
+      "step": 3053
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 9.1882,
+      "step": 3054
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.9869,
+      "step": 3055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.0705,
+      "step": 3056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8802,
+      "step": 3057
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.8399,
+      "step": 3058
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.6438,
+      "step": 3059
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0608,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.763,
+      "step": 3061
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.355,
+      "step": 3062
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5504,
+      "step": 3063
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1582,
+      "step": 3064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1922,
+      "step": 3065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.0706,
+      "step": 3066
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.807,
+      "step": 3067
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0305,
+      "step": 3068
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0774,
+      "step": 3069
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4469,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1158,
+      "step": 3071
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8087,
+      "step": 3072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.5672,
+      "step": 3073
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.5864,
+      "step": 3074
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7665,
+      "step": 3075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2434,
+      "step": 3076
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3988,
+      "step": 3077
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0307,
+      "step": 3078
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.6561,
+      "step": 3079
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8158,
+      "step": 3080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8384,
+      "step": 3081
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5515,
+      "step": 3082
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8108,
+      "step": 3083
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.2536,
+      "step": 3084
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.2506,
+      "step": 3085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1605,
+      "step": 3086
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4572,
+      "step": 3087
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.3312,
+      "step": 3088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1481,
+      "step": 3089
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.3304,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.2617,
+      "step": 3091
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3148,
+      "step": 3092
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4781,
+      "step": 3093
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.327,
+      "step": 3094
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3288,
+      "step": 3095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2354,
+      "step": 3096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4802,
+      "step": 3097
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 9.1659,
+      "step": 3098
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 1.9887,
+      "step": 3099
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 1.9497,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2668,
+      "step": 3101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.458,
+      "step": 3102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.9919,
+      "step": 3103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0408,
+      "step": 3104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.9966,
+      "step": 3105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 9.1371,
+      "step": 3106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0939,
+      "step": 3107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2532,
+      "step": 3108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7343,
+      "step": 3109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.156,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2223,
+      "step": 3111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.6182,
+      "step": 3112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4423,
+      "step": 3113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3855,
+      "step": 3114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.2115,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6719,
+      "step": 3116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5285,
+      "step": 3117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.0171,
+      "step": 3118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2127,
+      "step": 3119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8153,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1838,
+      "step": 3121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.711,
+      "step": 3122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1568,
+      "step": 3123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3741,
+      "step": 3124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2741,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.1653,
+      "step": 3126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.9722,
+      "step": 3127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.9181,
+      "step": 3128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.384,
+      "step": 3129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.1491,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.8641,
+      "step": 3131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.6125,
+      "step": 3132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1702,
+      "step": 3133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.4853,
+      "step": 3134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7929,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8747,
+      "step": 3136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.2659,
+      "step": 3137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.0685,
+      "step": 3138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2297,
+      "step": 3139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0715,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2201,
+      "step": 3141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2134,
+      "step": 3142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6098,
+      "step": 3143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.2036,
+      "step": 3144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2481,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4508,
+      "step": 3146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.1454,
+      "step": 3147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7573,
+      "step": 3148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2946,
+      "step": 3149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0121,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.498,
+      "step": 3151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4016,
+      "step": 3152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5934,
+      "step": 3153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.737,
+      "step": 3154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9874,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7714,
+      "step": 3156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3529,
+      "step": 3157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7673,
+      "step": 3158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3835,
+      "step": 3159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.0336,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2998,
+      "step": 3161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0446,
+      "step": 3162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5225,
+      "step": 3163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1391,
+      "step": 3164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7669,
+      "step": 3165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.118,
+      "step": 3166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7045,
+      "step": 3167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.077,
+      "step": 3168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.0697,
+      "step": 3169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8445,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4044,
+      "step": 3171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9817,
+      "step": 3172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.2508,
+      "step": 3173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.6302,
+      "step": 3174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.46,
+      "step": 3175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.8094,
+      "step": 3176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1174,
+      "step": 3177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.5756,
+      "step": 3178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4215,
+      "step": 3179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.5957,
+      "step": 3180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3534,
+      "step": 3181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9342,
+      "step": 3182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.8227,
+      "step": 3183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.1404,
+      "step": 3184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.9897,
+      "step": 3185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7286,
+      "step": 3186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.5414,
+      "step": 3187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7523,
+      "step": 3188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4033,
+      "step": 3189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8593,
+      "step": 3190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6095,
+      "step": 3191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.504,
+      "step": 3192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.6776,
+      "step": 3193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0174,
+      "step": 3194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7759,
+      "step": 3195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5685,
+      "step": 3196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2786,
+      "step": 3197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.7794,
+      "step": 3198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.5827,
+      "step": 3199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3282,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 6.423073768615723,
+      "eval_runtime": 22.4644,
+      "eval_samples_per_second": 2.226,
+      "eval_steps_per_second": 1.113,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "mmlu_eval_accuracy": 0.2525477994227994,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.07142857142857142,
+      "mmlu_eval_accuracy_astronomy": 0.3125,
+      "mmlu_eval_accuracy_business_ethics": 0.4444444444444444,
+      "mmlu_loss": 3.9956862831115725,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3005,
+      "step": 3201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0131,
+      "step": 3202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.9222,
+      "step": 3203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6755,
+      "step": 3204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.8386,
+      "step": 3205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7854,
+      "step": 3206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4073,
+      "step": 3207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.4278,
+      "step": 3208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5322,
+      "step": 3209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.1508,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4289,
+      "step": 3211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 9.0088,
+      "step": 3212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6227,
+      "step": 3213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.232,
+      "step": 3214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4584,
+      "step": 3215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8538,
+      "step": 3216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.1044,
+      "step": 3217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3851,
+      "step": 3218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7661,
+      "step": 3219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.4478,
+      "step": 3220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0658,
+      "step": 3221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.0961,
+      "step": 3222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.5636,
+      "step": 3223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6945,
+      "step": 3224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.5381,
+      "step": 3225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2431,
+      "step": 3226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3755,
+      "step": 3227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.4123,
+      "step": 3228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.0414,
+      "step": 3229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.6732,
+      "step": 3230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8904,
+      "step": 3231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7829,
+      "step": 3232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8394,
+      "step": 3233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.2565,
+      "step": 3234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.7535,
+      "step": 3235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.8878,
+      "step": 3236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9086,
+      "step": 3237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7491,
+      "step": 3238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.0775,
+      "step": 3239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.3575,
+      "step": 3240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.7583,
+      "step": 3241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.7107,
+      "step": 3242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.3212,
+      "step": 3243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.5662,
+      "step": 3244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.9553,
+      "step": 3245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.3713,
+      "step": 3246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4519,
+      "step": 3247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.0003,
+      "step": 3248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.2796,
+      "step": 3249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.3262,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.0106,
+      "step": 3251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.8502,
+      "step": 3252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.3491,
+      "step": 3253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 2.2728,
+      "step": 3254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.822,
+      "step": 3255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.6077,
+      "step": 3256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9328,
+      "step": 3257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.8518,
+      "step": 3258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.6541,
+      "step": 3259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.6976,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 8.9285,
+      "step": 3261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.9624,
+      "step": 3262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.6754,
+      "step": 3263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.4891,
+      "step": 3264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.7557,
+      "step": 3265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9059,
+      "step": 3266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 6.2432,
+      "step": 3267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 5.9467,
+      "step": 3268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 7.6248,
+      "step": 3269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 3.6632,
+      "step": 3270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.5671,
+      "step": 3271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 4.9754,
+      "step": 3272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.2562,
+      "step": 3273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6304,
+      "step": 3274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2176,
+      "step": 3275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0867,
+      "step": 3276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6769,
+      "step": 3277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.1474,
+      "step": 3278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5615,
+      "step": 3279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3285,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2536,
+      "step": 3281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.1851,
+      "step": 3282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1866,
+      "step": 3283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.26,
+      "step": 3284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1826,
+      "step": 3285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3643,
+      "step": 3286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5651,
+      "step": 3287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7203,
+      "step": 3288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7083,
+      "step": 3289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1364,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1367,
+      "step": 3291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.5968,
+      "step": 3292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.287,
+      "step": 3293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.5756,
+      "step": 3294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3841,
+      "step": 3295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3323,
+      "step": 3296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0486,
+      "step": 3297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6838,
+      "step": 3298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.5777,
+      "step": 3299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.4593,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9308,
+      "step": 3301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.6334,
+      "step": 3302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6559,
+      "step": 3303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7031,
+      "step": 3304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.2146,
+      "step": 3305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.6907,
+      "step": 3306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.8058,
+      "step": 3307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7805,
+      "step": 3308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.557,
+      "step": 3309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.2331,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1702,
+      "step": 3311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2958,
+      "step": 3312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6121,
+      "step": 3313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.9686,
+      "step": 3314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0866,
+      "step": 3315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2124,
+      "step": 3316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9969,
+      "step": 3317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.3417,
+      "step": 3318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7311,
+      "step": 3319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.8008,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.2469,
+      "step": 3321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.4888,
+      "step": 3322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2527,
+      "step": 3323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7326,
+      "step": 3324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4268,
+      "step": 3325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3635,
+      "step": 3326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8848,
+      "step": 3327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.2306,
+      "step": 3328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4185,
+      "step": 3329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0911,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0277,
+      "step": 3331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4066,
+      "step": 3332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.9085,
+      "step": 3333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2858,
+      "step": 3334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.4789,
+      "step": 3335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.8016,
+      "step": 3336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9443,
+      "step": 3337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.8456,
+      "step": 3338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0709,
+      "step": 3339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7879,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.6286,
+      "step": 3341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.7528,
+      "step": 3342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.4875,
+      "step": 3343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4383,
+      "step": 3344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7478,
+      "step": 3345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.8581,
+      "step": 3346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.4681,
+      "step": 3347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.179,
+      "step": 3348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.772,
+      "step": 3349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5266,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.962,
+      "step": 3351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1986,
+      "step": 3352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8946,
+      "step": 3353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8563,
+      "step": 3354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0887,
+      "step": 3355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.9487,
+      "step": 3356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9164,
+      "step": 3357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6243,
+      "step": 3358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1646,
+      "step": 3359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3753,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6102,
+      "step": 3361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8946,
+      "step": 3362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1349,
+      "step": 3363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7417,
+      "step": 3364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4195,
+      "step": 3365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0547,
+      "step": 3366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3341,
+      "step": 3367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4345,
+      "step": 3368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.3088,
+      "step": 3369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7109,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.6865,
+      "step": 3371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5466,
+      "step": 3372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.4801,
+      "step": 3373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7442,
+      "step": 3374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0129,
+      "step": 3375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2859,
+      "step": 3376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0064,
+      "step": 3377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1277,
+      "step": 3378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.9071,
+      "step": 3379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3575,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7016,
+      "step": 3381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.151,
+      "step": 3382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.4008,
+      "step": 3383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3565,
+      "step": 3384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3863,
+      "step": 3385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7621,
+      "step": 3386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3799,
+      "step": 3387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8051,
+      "step": 3388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1864,
+      "step": 3389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4464,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.649,
+      "step": 3391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7819,
+      "step": 3392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4867,
+      "step": 3393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0732,
+      "step": 3394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2601,
+      "step": 3395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.8808,
+      "step": 3396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9649,
+      "step": 3397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3081,
+      "step": 3398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7762,
+      "step": 3399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0978,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 6.543734073638916,
+      "eval_runtime": 22.8771,
+      "eval_samples_per_second": 2.186,
+      "eval_steps_per_second": 1.093,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "mmlu_eval_accuracy": 0.2525477994227994,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.07142857142857142,
+      "mmlu_eval_accuracy_astronomy": 0.3125,
+      "mmlu_eval_accuracy_business_ethics": 0.4444444444444444,
+      "mmlu_loss": 3.872052774429321,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7529,
+      "step": 3401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2717,
+      "step": 3402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.9914,
+      "step": 3403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1987,
+      "step": 3404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.68,
+      "step": 3405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9154,
+      "step": 3406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5814,
+      "step": 3407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3381,
+      "step": 3408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1775,
+      "step": 3409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9658,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.6455,
+      "step": 3411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3931,
+      "step": 3412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.3601,
+      "step": 3413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.4806,
+      "step": 3414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2846,
+      "step": 3415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3628,
+      "step": 3416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7838,
+      "step": 3417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0117,
+      "step": 3418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1784,
+      "step": 3419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.925,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.6457,
+      "step": 3421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.5869,
+      "step": 3422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8333,
+      "step": 3423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.1381,
+      "step": 3424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.8135,
+      "step": 3425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5245,
+      "step": 3426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0547,
+      "step": 3427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7663,
+      "step": 3428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1778,
+      "step": 3429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.4051,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1333,
+      "step": 3431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3353,
+      "step": 3432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.3009,
+      "step": 3433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.7293,
+      "step": 3434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5752,
+      "step": 3435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3992,
+      "step": 3436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.8915,
+      "step": 3437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4496,
+      "step": 3438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8425,
+      "step": 3439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.7227,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4708,
+      "step": 3441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.923,
+      "step": 3442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.5959,
+      "step": 3443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9186,
+      "step": 3444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.2626,
+      "step": 3445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2608,
+      "step": 3446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.9251,
+      "step": 3447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0449,
+      "step": 3448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0141,
+      "step": 3449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.3608,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9488,
+      "step": 3451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7488,
+      "step": 3452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6528,
+      "step": 3453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9995,
+      "step": 3454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.7377,
+      "step": 3455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.6786,
+      "step": 3456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2383,
+      "step": 3457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0354,
+      "step": 3458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9902,
+      "step": 3459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.7258,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5172,
+      "step": 3461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.988,
+      "step": 3462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4772,
+      "step": 3463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5068,
+      "step": 3464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8593,
+      "step": 3465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9398,
+      "step": 3466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0332,
+      "step": 3467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7405,
+      "step": 3468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.6347,
+      "step": 3469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0263,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.6087,
+      "step": 3471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7333,
+      "step": 3472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1251,
+      "step": 3473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5823,
+      "step": 3474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1494,
+      "step": 3475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3429,
+      "step": 3476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4662,
+      "step": 3477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.5451,
+      "step": 3478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.972,
+      "step": 3479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.6314,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8037,
+      "step": 3481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3445,
+      "step": 3482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4575,
+      "step": 3483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0294,
+      "step": 3484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.3406,
+      "step": 3485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3541,
+      "step": 3486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2848,
+      "step": 3487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.2269,
+      "step": 3488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2968,
+      "step": 3489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1864,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9224,
+      "step": 3491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6688,
+      "step": 3492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.19,
+      "step": 3493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4404,
+      "step": 3494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.901,
+      "step": 3495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.1823,
+      "step": 3496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7348,
+      "step": 3497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7604,
+      "step": 3498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2457,
+      "step": 3499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.1082,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.6028,
+      "step": 3501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8972,
+      "step": 3502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0901,
+      "step": 3503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1925,
+      "step": 3504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.2407,
+      "step": 3505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2848,
+      "step": 3506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1169,
+      "step": 3507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2106,
+      "step": 3508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7564,
+      "step": 3509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6047,
+      "step": 3510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3407,
+      "step": 3511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9848,
+      "step": 3512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7671,
+      "step": 3513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6396,
+      "step": 3514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.699,
+      "step": 3515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.2749,
+      "step": 3516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5383,
+      "step": 3517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.4338,
+      "step": 3518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3005,
+      "step": 3519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1728,
+      "step": 3520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0289,
+      "step": 3521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.5751,
+      "step": 3522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0577,
+      "step": 3523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6819,
+      "step": 3524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5446,
+      "step": 3525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0514,
+      "step": 3526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6815,
+      "step": 3527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6218,
+      "step": 3528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4958,
+      "step": 3529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.9706,
+      "step": 3530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9316,
+      "step": 3531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.479,
+      "step": 3532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2855,
+      "step": 3533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2544,
+      "step": 3534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9837,
+      "step": 3535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.978,
+      "step": 3536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7618,
+      "step": 3537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.5827,
+      "step": 3538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8986,
+      "step": 3539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.9653,
+      "step": 3540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.8958,
+      "step": 3541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4007,
+      "step": 3542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0079,
+      "step": 3543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.1045,
+      "step": 3544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0049,
+      "step": 3545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7175,
+      "step": 3546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.4449,
+      "step": 3547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.8293,
+      "step": 3548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9874,
+      "step": 3549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6799,
+      "step": 3550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4073,
+      "step": 3551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.2746,
+      "step": 3552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4299,
+      "step": 3553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7892,
+      "step": 3554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.123,
+      "step": 3555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.6298,
+      "step": 3556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3416,
+      "step": 3557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.746,
+      "step": 3558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8938,
+      "step": 3559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6562,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7783,
+      "step": 3561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.291,
+      "step": 3562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1611,
+      "step": 3563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4299,
+      "step": 3564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.341,
+      "step": 3565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0554,
+      "step": 3566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3736,
+      "step": 3567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4401,
+      "step": 3568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0594,
+      "step": 3569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1593,
+      "step": 3570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3663,
+      "step": 3571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6513,
+      "step": 3572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0796,
+      "step": 3573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.593,
+      "step": 3574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9869,
+      "step": 3575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4861,
+      "step": 3576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.4145,
+      "step": 3577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.9537,
+      "step": 3578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3876,
+      "step": 3579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0458,
+      "step": 3580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1648,
+      "step": 3581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0825,
+      "step": 3582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1962,
+      "step": 3583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9999,
+      "step": 3584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.8877,
+      "step": 3585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.387,
+      "step": 3586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.3146,
+      "step": 3587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2283,
+      "step": 3588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1697,
+      "step": 3589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1183,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7817,
+      "step": 3591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.103,
+      "step": 3592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2412,
+      "step": 3593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9878,
+      "step": 3594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3351,
+      "step": 3595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7125,
+      "step": 3596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0872,
+      "step": 3597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.6162,
+      "step": 3598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.699,
+      "step": 3599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7063,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 6.505133152008057,
+      "eval_runtime": 22.6179,
+      "eval_samples_per_second": 2.211,
+      "eval_steps_per_second": 1.105,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "mmlu_eval_accuracy": 0.2525477994227994,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.07142857142857142,
+      "mmlu_eval_accuracy_astronomy": 0.3125,
+      "mmlu_eval_accuracy_business_ethics": 0.4444444444444444,
+      "mmlu_loss": 4.059268207550049,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.79,
+      "step": 3601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9946,
+      "step": 3602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5461,
+      "step": 3603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5258,
+      "step": 3604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7346,
+      "step": 3605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3883,
+      "step": 3606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5788,
+      "step": 3607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.2204,
+      "step": 3608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1632,
+      "step": 3609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7381,
+      "step": 3610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1473,
+      "step": 3611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.9152,
+      "step": 3612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4622,
+      "step": 3613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.6216,
+      "step": 3614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6317,
+      "step": 3615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4409,
+      "step": 3616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0236,
+      "step": 3617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8894,
+      "step": 3618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0822,
+      "step": 3619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.953,
+      "step": 3620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1938,
+      "step": 3621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3474,
+      "step": 3622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4535,
+      "step": 3623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3665,
+      "step": 3624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.4139,
+      "step": 3625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.6508,
+      "step": 3626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.5838,
+      "step": 3627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9303,
+      "step": 3628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3722,
+      "step": 3629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.4449,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.3978,
+      "step": 3631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5792,
+      "step": 3632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.4508,
+      "step": 3633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.5883,
+      "step": 3634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1239,
+      "step": 3635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.2981,
+      "step": 3636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7042,
+      "step": 3637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1361,
+      "step": 3638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.9068,
+      "step": 3639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.3976,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0685,
+      "step": 3641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5028,
+      "step": 3642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0945,
+      "step": 3643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0476,
+      "step": 3644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7249,
+      "step": 3645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.8524,
+      "step": 3646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1725,
+      "step": 3647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.1051,
+      "step": 3648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7044,
+      "step": 3649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0646,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.5899,
+      "step": 3651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2098,
+      "step": 3652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2737,
+      "step": 3653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7012,
+      "step": 3654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.392,
+      "step": 3655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1024,
+      "step": 3656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8084,
+      "step": 3657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0021,
+      "step": 3658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.0005,
+      "step": 3659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0075,
+      "step": 3660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0208,
+      "step": 3661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1425,
+      "step": 3662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2731,
+      "step": 3663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0704,
+      "step": 3664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.7608,
+      "step": 3665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.126,
+      "step": 3666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1159,
+      "step": 3667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.301,
+      "step": 3668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7234,
+      "step": 3669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.9938,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1246,
+      "step": 3671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2179,
+      "step": 3672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4044,
+      "step": 3673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0196,
+      "step": 3674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.5111,
+      "step": 3675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6913,
+      "step": 3676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.3374,
+      "step": 3677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7363,
+      "step": 3678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.0937,
+      "step": 3679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.691,
+      "step": 3680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.1534,
+      "step": 3681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2765,
+      "step": 3682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.635,
+      "step": 3683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4933,
+      "step": 3684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3922,
+      "step": 3685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2925,
+      "step": 3686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.646,
+      "step": 3687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.2651,
+      "step": 3688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7927,
+      "step": 3689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.7202,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5883,
+      "step": 3691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1781,
+      "step": 3692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5795,
+      "step": 3693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.1326,
+      "step": 3694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.2378,
+      "step": 3695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.462,
+      "step": 3696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.7268,
+      "step": 3697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.9041,
+      "step": 3698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.8232,
+      "step": 3699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 1.7176,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.854,
+      "step": 3701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 1.6464,
+      "step": 3702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.264,
+      "step": 3703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.8032,
+      "step": 3704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.7097,
+      "step": 3705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.068,
+      "step": 3706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.3674,
+      "step": 3707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5067,
+      "step": 3708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.4841,
+      "step": 3709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.2839,
+      "step": 3710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0894,
+      "step": 3711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.5614,
+      "step": 3712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.7785,
+      "step": 3713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8865,
+      "step": 3714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0697,
+      "step": 3715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0635,
+      "step": 3716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0788,
+      "step": 3717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8854,
+      "step": 3718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.3054,
+      "step": 3719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7088,
+      "step": 3720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8188,
+      "step": 3721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4379,
+      "step": 3722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8049,
+      "step": 3723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4344,
+      "step": 3724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.647,
+      "step": 3725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5758,
+      "step": 3726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.9208,
+      "step": 3727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.4727,
+      "step": 3728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2176,
+      "step": 3729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0123,
+      "step": 3730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 9.404,
+      "step": 3731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.0537,
+      "step": 3732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.8559,
+      "step": 3733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2705,
+      "step": 3734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4536,
+      "step": 3735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.5407,
+      "step": 3736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7118,
+      "step": 3737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.4877,
+      "step": 3738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1927,
+      "step": 3739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.3252,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.9464,
+      "step": 3741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.3133,
+      "step": 3742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.427,
+      "step": 3743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.8029,
+      "step": 3744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8428,
+      "step": 3745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3284,
+      "step": 3746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.2628,
+      "step": 3747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.466,
+      "step": 3748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.3269,
+      "step": 3749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.1805,
+      "step": 3750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.5579,
+      "step": 3751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.581,
+      "step": 3752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.0164,
+      "step": 3753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.9736,
+      "step": 3754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.8441,
+      "step": 3755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3202,
+      "step": 3756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.9395,
+      "step": 3757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.2659,
+      "step": 3758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5779,
+      "step": 3759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.1052,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.0075,
+      "step": 3761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.5628,
+      "step": 3762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7005,
+      "step": 3763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.979,
+      "step": 3764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.9874,
+      "step": 3765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.8137,
+      "step": 3766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1964,
+      "step": 3767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.9047,
+      "step": 3768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.2724,
+      "step": 3769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.8014,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.7738,
+      "step": 3771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.926,
+      "step": 3772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.94,
+      "step": 3773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.1657,
+      "step": 3774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9987,
+      "step": 3775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5263,
+      "step": 3776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.0671,
+      "step": 3777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.1628,
+      "step": 3778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4194,
+      "step": 3779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5225,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.3863,
+      "step": 3781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.7951,
+      "step": 3782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5864,
+      "step": 3783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.177,
+      "step": 3784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.2334,
+      "step": 3785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.8444,
+      "step": 3786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.7108,
+      "step": 3787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6521,
+      "step": 3788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.4604,
+      "step": 3789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.12,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 6.3275,
+      "step": 3791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.9743,
+      "step": 3792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 5.6064,
+      "step": 3793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.6108,
+      "step": 3794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 3.0104,
+      "step": 3795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.2739,
+      "step": 3796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 2.8316,
+      "step": 3797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 7.5044,
+      "step": 3798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 8.838,
+      "step": 3799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 4.5967,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 6.464095592498779,
+      "eval_runtime": 22.5612,
+      "eval_samples_per_second": 2.216,
+      "eval_steps_per_second": 1.108,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "mmlu_eval_accuracy": 0.2525477994227994,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.07142857142857142,
+      "mmlu_eval_accuracy_astronomy": 0.3125,
+      "mmlu_eval_accuracy_business_ethics": 0.4444444444444444,
+      "mmlu_loss": 3.589159059524536,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "step": 3800,
+      "total_flos": 6.198225528943411e+16,
+      "train_loss": 1.5118552861401908,
+      "train_runtime": 2842.6735,
+      "train_samples_per_second": 10.553,
+      "train_steps_per_second": 10.553
     }
   ],
   "max_steps": 30000,
   "num_train_epochs": 1,
-  "total_flos": 4.660001608148582e+16,
+  "total_flos": 6.198225528943411e+16,
   "trial_name": null,
   "trial_params": null
 }