Training checkpoint at step 9000
Browse files- trainer_state.json +366 -6
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
-
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2888,6 +2888,366 @@
|
|
| 2888 |
"eval_samples_per_second": 3.199,
|
| 2889 |
"eval_steps_per_second": 1.6,
|
| 2890 |
"step": 8000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2891 |
}
|
| 2892 |
],
|
| 2893 |
"logging_steps": 25,
|
|
@@ -2907,7 +3267,7 @@
|
|
| 2907 |
"attributes": {}
|
| 2908 |
}
|
| 2909 |
},
|
| 2910 |
-
"total_flos": 2.
|
| 2911 |
"train_batch_size": 1,
|
| 2912 |
"trial_name": null,
|
| 2913 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 9000,
|
| 3 |
+
"best_metric": 2.410008430480957,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-9000",
|
| 5 |
+
"epoch": 0.18,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 9000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2888 |
"eval_samples_per_second": 3.199,
|
| 2889 |
"eval_steps_per_second": 1.6,
|
| 2890 |
"step": 8000
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 0.1605,
|
| 2894 |
+
"grad_norm": 0.5861319558312379,
|
| 2895 |
+
"learning_rate": 9.328000000000001e-06,
|
| 2896 |
+
"loss": 2.3883,
|
| 2897 |
+
"step": 8025
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 0.161,
|
| 2901 |
+
"grad_norm": 0.5836976562991806,
|
| 2902 |
+
"learning_rate": 9.322444444444445e-06,
|
| 2903 |
+
"loss": 2.3858,
|
| 2904 |
+
"step": 8050
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 0.1615,
|
| 2908 |
+
"grad_norm": 0.5844356099514875,
|
| 2909 |
+
"learning_rate": 9.31688888888889e-06,
|
| 2910 |
+
"loss": 2.408,
|
| 2911 |
+
"step": 8075
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 0.162,
|
| 2915 |
+
"grad_norm": 0.5898038882596441,
|
| 2916 |
+
"learning_rate": 9.311333333333335e-06,
|
| 2917 |
+
"loss": 2.3979,
|
| 2918 |
+
"step": 8100
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 0.162,
|
| 2922 |
+
"eval_loss": 2.4123263359069824,
|
| 2923 |
+
"eval_runtime": 31.7798,
|
| 2924 |
+
"eval_samples_per_second": 3.21,
|
| 2925 |
+
"eval_steps_per_second": 1.605,
|
| 2926 |
+
"step": 8100
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 0.1625,
|
| 2930 |
+
"grad_norm": 0.6072648398087778,
|
| 2931 |
+
"learning_rate": 9.305777777777779e-06,
|
| 2932 |
+
"loss": 2.3904,
|
| 2933 |
+
"step": 8125
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 0.163,
|
| 2937 |
+
"grad_norm": 0.5947190221089934,
|
| 2938 |
+
"learning_rate": 9.300222222222222e-06,
|
| 2939 |
+
"loss": 2.3908,
|
| 2940 |
+
"step": 8150
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 0.1635,
|
| 2944 |
+
"grad_norm": 0.5923294532719955,
|
| 2945 |
+
"learning_rate": 9.294666666666668e-06,
|
| 2946 |
+
"loss": 2.3994,
|
| 2947 |
+
"step": 8175
|
| 2948 |
+
},
|
| 2949 |
+
{
|
| 2950 |
+
"epoch": 0.164,
|
| 2951 |
+
"grad_norm": 0.6238957997579533,
|
| 2952 |
+
"learning_rate": 9.289111111111113e-06,
|
| 2953 |
+
"loss": 2.3935,
|
| 2954 |
+
"step": 8200
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 0.164,
|
| 2958 |
+
"eval_loss": 2.4118340015411377,
|
| 2959 |
+
"eval_runtime": 31.8145,
|
| 2960 |
+
"eval_samples_per_second": 3.206,
|
| 2961 |
+
"eval_steps_per_second": 1.603,
|
| 2962 |
+
"step": 8200
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 0.1645,
|
| 2966 |
+
"grad_norm": 0.576622489198895,
|
| 2967 |
+
"learning_rate": 9.283555555555556e-06,
|
| 2968 |
+
"loss": 2.396,
|
| 2969 |
+
"step": 8225
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 0.165,
|
| 2973 |
+
"grad_norm": 0.6185118704471244,
|
| 2974 |
+
"learning_rate": 9.278e-06,
|
| 2975 |
+
"loss": 2.4035,
|
| 2976 |
+
"step": 8250
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 0.1655,
|
| 2980 |
+
"grad_norm": 0.5796535805449304,
|
| 2981 |
+
"learning_rate": 9.272444444444445e-06,
|
| 2982 |
+
"loss": 2.3943,
|
| 2983 |
+
"step": 8275
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 0.166,
|
| 2987 |
+
"grad_norm": 0.6173375014397958,
|
| 2988 |
+
"learning_rate": 9.26688888888889e-06,
|
| 2989 |
+
"loss": 2.3935,
|
| 2990 |
+
"step": 8300
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 0.166,
|
| 2994 |
+
"eval_loss": 2.4114973545074463,
|
| 2995 |
+
"eval_runtime": 31.7754,
|
| 2996 |
+
"eval_samples_per_second": 3.21,
|
| 2997 |
+
"eval_steps_per_second": 1.605,
|
| 2998 |
+
"step": 8300
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.1665,
|
| 3002 |
+
"grad_norm": 0.5618534321843206,
|
| 3003 |
+
"learning_rate": 9.261333333333334e-06,
|
| 3004 |
+
"loss": 2.3974,
|
| 3005 |
+
"step": 8325
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 0.167,
|
| 3009 |
+
"grad_norm": 0.6009214777241336,
|
| 3010 |
+
"learning_rate": 9.25577777777778e-06,
|
| 3011 |
+
"loss": 2.4,
|
| 3012 |
+
"step": 8350
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.1675,
|
| 3016 |
+
"grad_norm": 0.5772198441104387,
|
| 3017 |
+
"learning_rate": 9.250222222222223e-06,
|
| 3018 |
+
"loss": 2.3991,
|
| 3019 |
+
"step": 8375
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 0.168,
|
| 3023 |
+
"grad_norm": 0.5740163940994337,
|
| 3024 |
+
"learning_rate": 9.244666666666668e-06,
|
| 3025 |
+
"loss": 2.3947,
|
| 3026 |
+
"step": 8400
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.168,
|
| 3030 |
+
"eval_loss": 2.411425828933716,
|
| 3031 |
+
"eval_runtime": 31.5099,
|
| 3032 |
+
"eval_samples_per_second": 3.237,
|
| 3033 |
+
"eval_steps_per_second": 1.619,
|
| 3034 |
+
"step": 8400
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.1685,
|
| 3038 |
+
"grad_norm": 0.5687873679002051,
|
| 3039 |
+
"learning_rate": 9.239111111111112e-06,
|
| 3040 |
+
"loss": 2.3966,
|
| 3041 |
+
"step": 8425
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.169,
|
| 3045 |
+
"grad_norm": 0.5610136891748577,
|
| 3046 |
+
"learning_rate": 9.233555555555557e-06,
|
| 3047 |
+
"loss": 2.3998,
|
| 3048 |
+
"step": 8450
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.1695,
|
| 3052 |
+
"grad_norm": 0.6032713755890403,
|
| 3053 |
+
"learning_rate": 9.228e-06,
|
| 3054 |
+
"loss": 2.3943,
|
| 3055 |
+
"step": 8475
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.17,
|
| 3059 |
+
"grad_norm": 0.5964144518891603,
|
| 3060 |
+
"learning_rate": 9.222444444444446e-06,
|
| 3061 |
+
"loss": 2.3883,
|
| 3062 |
+
"step": 8500
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.17,
|
| 3066 |
+
"eval_loss": 2.411017656326294,
|
| 3067 |
+
"eval_runtime": 31.5307,
|
| 3068 |
+
"eval_samples_per_second": 3.235,
|
| 3069 |
+
"eval_steps_per_second": 1.617,
|
| 3070 |
+
"step": 8500
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 0.1705,
|
| 3074 |
+
"grad_norm": 0.6150332993234658,
|
| 3075 |
+
"learning_rate": 9.21688888888889e-06,
|
| 3076 |
+
"loss": 2.3947,
|
| 3077 |
+
"step": 8525
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 0.171,
|
| 3081 |
+
"grad_norm": 0.5996705331900282,
|
| 3082 |
+
"learning_rate": 9.211333333333334e-06,
|
| 3083 |
+
"loss": 2.3767,
|
| 3084 |
+
"step": 8550
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 0.1715,
|
| 3088 |
+
"grad_norm": 0.5824632831455251,
|
| 3089 |
+
"learning_rate": 9.20577777777778e-06,
|
| 3090 |
+
"loss": 2.3872,
|
| 3091 |
+
"step": 8575
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 0.172,
|
| 3095 |
+
"grad_norm": 0.606207861483595,
|
| 3096 |
+
"learning_rate": 9.200222222222223e-06,
|
| 3097 |
+
"loss": 2.4039,
|
| 3098 |
+
"step": 8600
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.172,
|
| 3102 |
+
"eval_loss": 2.4107751846313477,
|
| 3103 |
+
"eval_runtime": 31.4387,
|
| 3104 |
+
"eval_samples_per_second": 3.244,
|
| 3105 |
+
"eval_steps_per_second": 1.622,
|
| 3106 |
+
"step": 8600
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 0.1725,
|
| 3110 |
+
"grad_norm": 0.576823131255562,
|
| 3111 |
+
"learning_rate": 9.194666666666667e-06,
|
| 3112 |
+
"loss": 2.3954,
|
| 3113 |
+
"step": 8625
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 0.173,
|
| 3117 |
+
"grad_norm": 0.56597712239854,
|
| 3118 |
+
"learning_rate": 9.189111111111112e-06,
|
| 3119 |
+
"loss": 2.4072,
|
| 3120 |
+
"step": 8650
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 0.1735,
|
| 3124 |
+
"grad_norm": 0.5825959007699376,
|
| 3125 |
+
"learning_rate": 9.183555555555557e-06,
|
| 3126 |
+
"loss": 2.4081,
|
| 3127 |
+
"step": 8675
|
| 3128 |
+
},
|
| 3129 |
+
{
|
| 3130 |
+
"epoch": 0.174,
|
| 3131 |
+
"grad_norm": 0.5776918671405765,
|
| 3132 |
+
"learning_rate": 9.178000000000001e-06,
|
| 3133 |
+
"loss": 2.4091,
|
| 3134 |
+
"step": 8700
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 0.174,
|
| 3138 |
+
"eval_loss": 2.410761594772339,
|
| 3139 |
+
"eval_runtime": 31.7246,
|
| 3140 |
+
"eval_samples_per_second": 3.215,
|
| 3141 |
+
"eval_steps_per_second": 1.608,
|
| 3142 |
+
"step": 8700
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 0.1745,
|
| 3146 |
+
"grad_norm": 0.6256369047041809,
|
| 3147 |
+
"learning_rate": 9.172444444444444e-06,
|
| 3148 |
+
"loss": 2.3953,
|
| 3149 |
+
"step": 8725
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 0.175,
|
| 3153 |
+
"grad_norm": 0.5964709475887552,
|
| 3154 |
+
"learning_rate": 9.16688888888889e-06,
|
| 3155 |
+
"loss": 2.39,
|
| 3156 |
+
"step": 8750
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 0.1755,
|
| 3160 |
+
"grad_norm": 0.5775755843795828,
|
| 3161 |
+
"learning_rate": 9.161333333333335e-06,
|
| 3162 |
+
"loss": 2.391,
|
| 3163 |
+
"step": 8775
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 0.176,
|
| 3167 |
+
"grad_norm": 0.6655706627980364,
|
| 3168 |
+
"learning_rate": 9.155777777777779e-06,
|
| 3169 |
+
"loss": 2.4048,
|
| 3170 |
+
"step": 8800
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 0.176,
|
| 3174 |
+
"eval_loss": 2.4105958938598633,
|
| 3175 |
+
"eval_runtime": 31.4248,
|
| 3176 |
+
"eval_samples_per_second": 3.246,
|
| 3177 |
+
"eval_steps_per_second": 1.623,
|
| 3178 |
+
"step": 8800
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 0.1765,
|
| 3182 |
+
"grad_norm": 0.5865172878151053,
|
| 3183 |
+
"learning_rate": 9.150222222222222e-06,
|
| 3184 |
+
"loss": 2.3878,
|
| 3185 |
+
"step": 8825
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 0.177,
|
| 3189 |
+
"grad_norm": 0.584391124965856,
|
| 3190 |
+
"learning_rate": 9.144666666666667e-06,
|
| 3191 |
+
"loss": 2.401,
|
| 3192 |
+
"step": 8850
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 0.1775,
|
| 3196 |
+
"grad_norm": 0.5726598382185046,
|
| 3197 |
+
"learning_rate": 9.139111111111113e-06,
|
| 3198 |
+
"loss": 2.4018,
|
| 3199 |
+
"step": 8875
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 0.178,
|
| 3203 |
+
"grad_norm": 0.5690725395770588,
|
| 3204 |
+
"learning_rate": 9.133555555555556e-06,
|
| 3205 |
+
"loss": 2.4034,
|
| 3206 |
+
"step": 8900
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 0.178,
|
| 3210 |
+
"eval_loss": 2.4101033210754395,
|
| 3211 |
+
"eval_runtime": 31.4686,
|
| 3212 |
+
"eval_samples_per_second": 3.241,
|
| 3213 |
+
"eval_steps_per_second": 1.621,
|
| 3214 |
+
"step": 8900
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 0.1785,
|
| 3218 |
+
"grad_norm": 0.5978143013011991,
|
| 3219 |
+
"learning_rate": 9.128e-06,
|
| 3220 |
+
"loss": 2.4014,
|
| 3221 |
+
"step": 8925
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 0.179,
|
| 3225 |
+
"grad_norm": 0.6085180927490662,
|
| 3226 |
+
"learning_rate": 9.122444444444445e-06,
|
| 3227 |
+
"loss": 2.3924,
|
| 3228 |
+
"step": 8950
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 0.1795,
|
| 3232 |
+
"grad_norm": 0.5720265034599029,
|
| 3233 |
+
"learning_rate": 9.11688888888889e-06,
|
| 3234 |
+
"loss": 2.3977,
|
| 3235 |
+
"step": 8975
|
| 3236 |
+
},
|
| 3237 |
+
{
|
| 3238 |
+
"epoch": 0.18,
|
| 3239 |
+
"grad_norm": 0.5739306861609581,
|
| 3240 |
+
"learning_rate": 9.111333333333334e-06,
|
| 3241 |
+
"loss": 2.3992,
|
| 3242 |
+
"step": 9000
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 0.18,
|
| 3246 |
+
"eval_loss": 2.410008430480957,
|
| 3247 |
+
"eval_runtime": 32.192,
|
| 3248 |
+
"eval_samples_per_second": 3.168,
|
| 3249 |
+
"eval_steps_per_second": 1.584,
|
| 3250 |
+
"step": 9000
|
| 3251 |
}
|
| 3252 |
],
|
| 3253 |
"logging_steps": 25,
|
|
|
|
| 3267 |
"attributes": {}
|
| 3268 |
}
|
| 3269 |
},
|
| 3270 |
+
"total_flos": 2.8648820684944835e+19,
|
| 3271 |
"train_batch_size": 1,
|
| 3272 |
"trial_name": null,
|
| 3273 |
"trial_params": null
|