Upload fine-tuned checkpoint
Browse files- adapter_model.safetensors +2 -2
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +5 -2813
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bffa9d18e49a345a9b66654a8b379c12896d28ad3d84beb951c5224569da50f1
|
3 |
+
size 13269912
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 7354554
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a94175e53964c6c84a5af9ab8815c4ee3b327e86338e6970e19fe608e3799137
|
3 |
size 7354554
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1351e3cbde5fba28ca4ac8afda43f48563c490779fc18bffc14596d2ecdfa130
|
3 |
size 14244
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e055298148c186988a54be766385f194e9e1ed4ff410a4916a76b1dcb1be936e
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"best_metric": 3.
|
3 |
-
"best_model_checkpoint": "checkpoints/test_1M_1-2025-02-12-12-32/checkpoint-
|
4 |
-
"epoch":
|
5 |
"eval_steps": 10000,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -2815,2814 +2815,6 @@
|
|
2815 |
"eval_samples_per_second": 264.165,
|
2816 |
"eval_steps_per_second": 33.021,
|
2817 |
"step": 10000
|
2818 |
-
},
|
2819 |
-
{
|
2820 |
-
"epoch": 0.7934309457855164,
|
2821 |
-
"grad_norm": 130.07545471191406,
|
2822 |
-
"learning_rate": 2.169432389294147e-07,
|
2823 |
-
"loss": 3.7883,
|
2824 |
-
"step": 10025
|
2825 |
-
},
|
2826 |
-
{
|
2827 |
-
"epoch": 0.7954095765730115,
|
2828 |
-
"grad_norm": 135.5163116455078,
|
2829 |
-
"learning_rate": 2.16860794892427e-07,
|
2830 |
-
"loss": 3.8028,
|
2831 |
-
"step": 10050
|
2832 |
-
},
|
2833 |
-
{
|
2834 |
-
"epoch": 0.7973882073605065,
|
2835 |
-
"grad_norm": 130.33010864257812,
|
2836 |
-
"learning_rate": 2.1677835085543934e-07,
|
2837 |
-
"loss": 3.9204,
|
2838 |
-
"step": 10075
|
2839 |
-
},
|
2840 |
-
{
|
2841 |
-
"epoch": 0.7993668381480016,
|
2842 |
-
"grad_norm": 131.47166442871094,
|
2843 |
-
"learning_rate": 2.1669590681845162e-07,
|
2844 |
-
"loss": 3.7961,
|
2845 |
-
"step": 10100
|
2846 |
-
},
|
2847 |
-
{
|
2848 |
-
"epoch": 0.8013454689354966,
|
2849 |
-
"grad_norm": 126.78470611572266,
|
2850 |
-
"learning_rate": 2.1661346278146396e-07,
|
2851 |
-
"loss": 3.8249,
|
2852 |
-
"step": 10125
|
2853 |
-
},
|
2854 |
-
{
|
2855 |
-
"epoch": 0.8033240997229917,
|
2856 |
-
"grad_norm": 188.20448303222656,
|
2857 |
-
"learning_rate": 2.1653101874447624e-07,
|
2858 |
-
"loss": 3.778,
|
2859 |
-
"step": 10150
|
2860 |
-
},
|
2861 |
-
{
|
2862 |
-
"epoch": 0.8053027305104867,
|
2863 |
-
"grad_norm": 137.7311553955078,
|
2864 |
-
"learning_rate": 2.1644857470748853e-07,
|
2865 |
-
"loss": 3.9346,
|
2866 |
-
"step": 10175
|
2867 |
-
},
|
2868 |
-
{
|
2869 |
-
"epoch": 0.8072813612979818,
|
2870 |
-
"grad_norm": 89.07192993164062,
|
2871 |
-
"learning_rate": 2.1636613067050087e-07,
|
2872 |
-
"loss": 3.7072,
|
2873 |
-
"step": 10200
|
2874 |
-
},
|
2875 |
-
{
|
2876 |
-
"epoch": 0.8092599920854768,
|
2877 |
-
"grad_norm": 96.4891357421875,
|
2878 |
-
"learning_rate": 2.1628368663351315e-07,
|
2879 |
-
"loss": 3.8609,
|
2880 |
-
"step": 10225
|
2881 |
-
},
|
2882 |
-
{
|
2883 |
-
"epoch": 0.8112386228729719,
|
2884 |
-
"grad_norm": 144.67372131347656,
|
2885 |
-
"learning_rate": 2.1620124259652546e-07,
|
2886 |
-
"loss": 3.6157,
|
2887 |
-
"step": 10250
|
2888 |
-
},
|
2889 |
-
{
|
2890 |
-
"epoch": 0.8132172536604669,
|
2891 |
-
"grad_norm": 115.51387023925781,
|
2892 |
-
"learning_rate": 2.1611879855953778e-07,
|
2893 |
-
"loss": 3.7226,
|
2894 |
-
"step": 10275
|
2895 |
-
},
|
2896 |
-
{
|
2897 |
-
"epoch": 0.815195884447962,
|
2898 |
-
"grad_norm": 99.25152587890625,
|
2899 |
-
"learning_rate": 2.160363545225501e-07,
|
2900 |
-
"loss": 3.9583,
|
2901 |
-
"step": 10300
|
2902 |
-
},
|
2903 |
-
{
|
2904 |
-
"epoch": 0.817174515235457,
|
2905 |
-
"grad_norm": 129.25799560546875,
|
2906 |
-
"learning_rate": 2.1595391048556237e-07,
|
2907 |
-
"loss": 3.9272,
|
2908 |
-
"step": 10325
|
2909 |
-
},
|
2910 |
-
{
|
2911 |
-
"epoch": 0.8191531460229521,
|
2912 |
-
"grad_norm": 142.94081115722656,
|
2913 |
-
"learning_rate": 2.158714664485747e-07,
|
2914 |
-
"loss": 4.0403,
|
2915 |
-
"step": 10350
|
2916 |
-
},
|
2917 |
-
{
|
2918 |
-
"epoch": 0.8211317768104471,
|
2919 |
-
"grad_norm": 138.73974609375,
|
2920 |
-
"learning_rate": 2.15789022411587e-07,
|
2921 |
-
"loss": 3.9837,
|
2922 |
-
"step": 10375
|
2923 |
-
},
|
2924 |
-
{
|
2925 |
-
"epoch": 0.8231104075979422,
|
2926 |
-
"grad_norm": 128.65940856933594,
|
2927 |
-
"learning_rate": 2.1570657837459933e-07,
|
2928 |
-
"loss": 3.9414,
|
2929 |
-
"step": 10400
|
2930 |
-
},
|
2931 |
-
{
|
2932 |
-
"epoch": 0.8250890383854372,
|
2933 |
-
"grad_norm": 98.70417022705078,
|
2934 |
-
"learning_rate": 2.1562413433761162e-07,
|
2935 |
-
"loss": 3.9875,
|
2936 |
-
"step": 10425
|
2937 |
-
},
|
2938 |
-
{
|
2939 |
-
"epoch": 0.8270676691729323,
|
2940 |
-
"grad_norm": 95.58749389648438,
|
2941 |
-
"learning_rate": 2.155416903006239e-07,
|
2942 |
-
"loss": 4.0485,
|
2943 |
-
"step": 10450
|
2944 |
-
},
|
2945 |
-
{
|
2946 |
-
"epoch": 0.8290462999604273,
|
2947 |
-
"grad_norm": 115.98033905029297,
|
2948 |
-
"learning_rate": 2.1545924626363624e-07,
|
2949 |
-
"loss": 3.9302,
|
2950 |
-
"step": 10475
|
2951 |
-
},
|
2952 |
-
{
|
2953 |
-
"epoch": 0.8310249307479224,
|
2954 |
-
"grad_norm": 115.73841094970703,
|
2955 |
-
"learning_rate": 2.1537680222664853e-07,
|
2956 |
-
"loss": 3.8931,
|
2957 |
-
"step": 10500
|
2958 |
-
},
|
2959 |
-
{
|
2960 |
-
"epoch": 0.8330035615354174,
|
2961 |
-
"grad_norm": 120.9448013305664,
|
2962 |
-
"learning_rate": 2.1529435818966086e-07,
|
2963 |
-
"loss": 3.9164,
|
2964 |
-
"step": 10525
|
2965 |
-
},
|
2966 |
-
{
|
2967 |
-
"epoch": 0.8349821923229126,
|
2968 |
-
"grad_norm": 108.95042419433594,
|
2969 |
-
"learning_rate": 2.1521191415267315e-07,
|
2970 |
-
"loss": 3.8218,
|
2971 |
-
"step": 10550
|
2972 |
-
},
|
2973 |
-
{
|
2974 |
-
"epoch": 0.8369608231104076,
|
2975 |
-
"grad_norm": 134.7786407470703,
|
2976 |
-
"learning_rate": 2.1512947011568546e-07,
|
2977 |
-
"loss": 3.8017,
|
2978 |
-
"step": 10575
|
2979 |
-
},
|
2980 |
-
{
|
2981 |
-
"epoch": 0.8389394538979027,
|
2982 |
-
"grad_norm": 135.17352294921875,
|
2983 |
-
"learning_rate": 2.1504702607869777e-07,
|
2984 |
-
"loss": 3.7896,
|
2985 |
-
"step": 10600
|
2986 |
-
},
|
2987 |
-
{
|
2988 |
-
"epoch": 0.8409180846853977,
|
2989 |
-
"grad_norm": 107.60875701904297,
|
2990 |
-
"learning_rate": 2.1496458204171008e-07,
|
2991 |
-
"loss": 3.8653,
|
2992 |
-
"step": 10625
|
2993 |
-
},
|
2994 |
-
{
|
2995 |
-
"epoch": 0.8428967154728928,
|
2996 |
-
"grad_norm": 106.85396575927734,
|
2997 |
-
"learning_rate": 2.1488213800472237e-07,
|
2998 |
-
"loss": 3.7688,
|
2999 |
-
"step": 10650
|
3000 |
-
},
|
3001 |
-
{
|
3002 |
-
"epoch": 0.8448753462603878,
|
3003 |
-
"grad_norm": 154.32797241210938,
|
3004 |
-
"learning_rate": 2.147996939677347e-07,
|
3005 |
-
"loss": 3.7928,
|
3006 |
-
"step": 10675
|
3007 |
-
},
|
3008 |
-
{
|
3009 |
-
"epoch": 0.8468539770478829,
|
3010 |
-
"grad_norm": 105.45416259765625,
|
3011 |
-
"learning_rate": 2.14717249930747e-07,
|
3012 |
-
"loss": 3.9773,
|
3013 |
-
"step": 10700
|
3014 |
-
},
|
3015 |
-
{
|
3016 |
-
"epoch": 0.848832607835378,
|
3017 |
-
"grad_norm": 130.95082092285156,
|
3018 |
-
"learning_rate": 2.1463480589375933e-07,
|
3019 |
-
"loss": 3.9007,
|
3020 |
-
"step": 10725
|
3021 |
-
},
|
3022 |
-
{
|
3023 |
-
"epoch": 0.850811238622873,
|
3024 |
-
"grad_norm": 111.79964447021484,
|
3025 |
-
"learning_rate": 2.1455236185677162e-07,
|
3026 |
-
"loss": 3.8503,
|
3027 |
-
"step": 10750
|
3028 |
-
},
|
3029 |
-
{
|
3030 |
-
"epoch": 0.852789869410368,
|
3031 |
-
"grad_norm": 111.66275787353516,
|
3032 |
-
"learning_rate": 2.144699178197839e-07,
|
3033 |
-
"loss": 4.1484,
|
3034 |
-
"step": 10775
|
3035 |
-
},
|
3036 |
-
{
|
3037 |
-
"epoch": 0.8547685001978631,
|
3038 |
-
"grad_norm": 105.59733581542969,
|
3039 |
-
"learning_rate": 2.1438747378279624e-07,
|
3040 |
-
"loss": 3.8871,
|
3041 |
-
"step": 10800
|
3042 |
-
},
|
3043 |
-
{
|
3044 |
-
"epoch": 0.8567471309853582,
|
3045 |
-
"grad_norm": 100.40887451171875,
|
3046 |
-
"learning_rate": 2.1430502974580852e-07,
|
3047 |
-
"loss": 3.8456,
|
3048 |
-
"step": 10825
|
3049 |
-
},
|
3050 |
-
{
|
3051 |
-
"epoch": 0.8587257617728532,
|
3052 |
-
"grad_norm": 129.26177978515625,
|
3053 |
-
"learning_rate": 2.1422258570882084e-07,
|
3054 |
-
"loss": 3.673,
|
3055 |
-
"step": 10850
|
3056 |
-
},
|
3057 |
-
{
|
3058 |
-
"epoch": 0.8607043925603483,
|
3059 |
-
"grad_norm": 144.7709197998047,
|
3060 |
-
"learning_rate": 2.1414014167183315e-07,
|
3061 |
-
"loss": 3.7465,
|
3062 |
-
"step": 10875
|
3063 |
-
},
|
3064 |
-
{
|
3065 |
-
"epoch": 0.8626830233478433,
|
3066 |
-
"grad_norm": 137.7886505126953,
|
3067 |
-
"learning_rate": 2.1405769763484546e-07,
|
3068 |
-
"loss": 3.8944,
|
3069 |
-
"step": 10900
|
3070 |
-
},
|
3071 |
-
{
|
3072 |
-
"epoch": 0.8646616541353384,
|
3073 |
-
"grad_norm": 117.49480438232422,
|
3074 |
-
"learning_rate": 2.1397525359785774e-07,
|
3075 |
-
"loss": 3.7995,
|
3076 |
-
"step": 10925
|
3077 |
-
},
|
3078 |
-
{
|
3079 |
-
"epoch": 0.8666402849228334,
|
3080 |
-
"grad_norm": 168.4739990234375,
|
3081 |
-
"learning_rate": 2.1389280956087008e-07,
|
3082 |
-
"loss": 4.0448,
|
3083 |
-
"step": 10950
|
3084 |
-
},
|
3085 |
-
{
|
3086 |
-
"epoch": 0.8686189157103285,
|
3087 |
-
"grad_norm": 133.91720581054688,
|
3088 |
-
"learning_rate": 2.1381036552388237e-07,
|
3089 |
-
"loss": 3.6778,
|
3090 |
-
"step": 10975
|
3091 |
-
},
|
3092 |
-
{
|
3093 |
-
"epoch": 0.8705975464978235,
|
3094 |
-
"grad_norm": 100.49728393554688,
|
3095 |
-
"learning_rate": 2.137279214868947e-07,
|
3096 |
-
"loss": 3.9251,
|
3097 |
-
"step": 11000
|
3098 |
-
},
|
3099 |
-
{
|
3100 |
-
"epoch": 0.8725761772853186,
|
3101 |
-
"grad_norm": 189.62327575683594,
|
3102 |
-
"learning_rate": 2.13645477449907e-07,
|
3103 |
-
"loss": 3.8032,
|
3104 |
-
"step": 11025
|
3105 |
-
},
|
3106 |
-
{
|
3107 |
-
"epoch": 0.8745548080728136,
|
3108 |
-
"grad_norm": 108.47595977783203,
|
3109 |
-
"learning_rate": 2.135630334129193e-07,
|
3110 |
-
"loss": 3.92,
|
3111 |
-
"step": 11050
|
3112 |
-
},
|
3113 |
-
{
|
3114 |
-
"epoch": 0.8765334388603087,
|
3115 |
-
"grad_norm": 150.95584106445312,
|
3116 |
-
"learning_rate": 2.1348058937593161e-07,
|
3117 |
-
"loss": 3.9559,
|
3118 |
-
"step": 11075
|
3119 |
-
},
|
3120 |
-
{
|
3121 |
-
"epoch": 0.8785120696478037,
|
3122 |
-
"grad_norm": 126.86693572998047,
|
3123 |
-
"learning_rate": 2.133981453389439e-07,
|
3124 |
-
"loss": 3.9322,
|
3125 |
-
"step": 11100
|
3126 |
-
},
|
3127 |
-
{
|
3128 |
-
"epoch": 0.8804907004352988,
|
3129 |
-
"grad_norm": 158.07049560546875,
|
3130 |
-
"learning_rate": 2.1331570130195624e-07,
|
3131 |
-
"loss": 3.7498,
|
3132 |
-
"step": 11125
|
3133 |
-
},
|
3134 |
-
{
|
3135 |
-
"epoch": 0.8824693312227938,
|
3136 |
-
"grad_norm": 156.98699951171875,
|
3137 |
-
"learning_rate": 2.1323325726496852e-07,
|
3138 |
-
"loss": 3.9679,
|
3139 |
-
"step": 11150
|
3140 |
-
},
|
3141 |
-
{
|
3142 |
-
"epoch": 0.8844479620102889,
|
3143 |
-
"grad_norm": 91.1626968383789,
|
3144 |
-
"learning_rate": 2.1315081322798083e-07,
|
3145 |
-
"loss": 3.9669,
|
3146 |
-
"step": 11175
|
3147 |
-
},
|
3148 |
-
{
|
3149 |
-
"epoch": 0.8864265927977839,
|
3150 |
-
"grad_norm": 145.4545135498047,
|
3151 |
-
"learning_rate": 2.1306836919099315e-07,
|
3152 |
-
"loss": 3.9038,
|
3153 |
-
"step": 11200
|
3154 |
-
},
|
3155 |
-
{
|
3156 |
-
"epoch": 0.888405223585279,
|
3157 |
-
"grad_norm": 134.78793334960938,
|
3158 |
-
"learning_rate": 2.1298592515400546e-07,
|
3159 |
-
"loss": 3.9122,
|
3160 |
-
"step": 11225
|
3161 |
-
},
|
3162 |
-
{
|
3163 |
-
"epoch": 0.890383854372774,
|
3164 |
-
"grad_norm": 132.1155242919922,
|
3165 |
-
"learning_rate": 2.1290348111701774e-07,
|
3166 |
-
"loss": 4.0016,
|
3167 |
-
"step": 11250
|
3168 |
-
},
|
3169 |
-
{
|
3170 |
-
"epoch": 0.8923624851602691,
|
3171 |
-
"grad_norm": 116.72430419921875,
|
3172 |
-
"learning_rate": 2.1282103708003008e-07,
|
3173 |
-
"loss": 3.9878,
|
3174 |
-
"step": 11275
|
3175 |
-
},
|
3176 |
-
{
|
3177 |
-
"epoch": 0.8943411159477641,
|
3178 |
-
"grad_norm": 117.89616394042969,
|
3179 |
-
"learning_rate": 2.1273859304304237e-07,
|
3180 |
-
"loss": 3.7834,
|
3181 |
-
"step": 11300
|
3182 |
-
},
|
3183 |
-
{
|
3184 |
-
"epoch": 0.8963197467352592,
|
3185 |
-
"grad_norm": 151.65277099609375,
|
3186 |
-
"learning_rate": 2.126561490060547e-07,
|
3187 |
-
"loss": 3.9696,
|
3188 |
-
"step": 11325
|
3189 |
-
},
|
3190 |
-
{
|
3191 |
-
"epoch": 0.8982983775227542,
|
3192 |
-
"grad_norm": 139.39405822753906,
|
3193 |
-
"learning_rate": 2.12573704969067e-07,
|
3194 |
-
"loss": 3.9585,
|
3195 |
-
"step": 11350
|
3196 |
-
},
|
3197 |
-
{
|
3198 |
-
"epoch": 0.9002770083102493,
|
3199 |
-
"grad_norm": 123.25849914550781,
|
3200 |
-
"learning_rate": 2.124912609320793e-07,
|
3201 |
-
"loss": 3.816,
|
3202 |
-
"step": 11375
|
3203 |
-
},
|
3204 |
-
{
|
3205 |
-
"epoch": 0.9022556390977443,
|
3206 |
-
"grad_norm": 148.54562377929688,
|
3207 |
-
"learning_rate": 2.124088168950916e-07,
|
3208 |
-
"loss": 3.8433,
|
3209 |
-
"step": 11400
|
3210 |
-
},
|
3211 |
-
{
|
3212 |
-
"epoch": 0.9042342698852394,
|
3213 |
-
"grad_norm": 135.60752868652344,
|
3214 |
-
"learning_rate": 2.123263728581039e-07,
|
3215 |
-
"loss": 3.7851,
|
3216 |
-
"step": 11425
|
3217 |
-
},
|
3218 |
-
{
|
3219 |
-
"epoch": 0.9062129006727345,
|
3220 |
-
"grad_norm": 129.5711212158203,
|
3221 |
-
"learning_rate": 2.122439288211162e-07,
|
3222 |
-
"loss": 3.7133,
|
3223 |
-
"step": 11450
|
3224 |
-
},
|
3225 |
-
{
|
3226 |
-
"epoch": 0.9081915314602296,
|
3227 |
-
"grad_norm": 136.88392639160156,
|
3228 |
-
"learning_rate": 2.1216148478412852e-07,
|
3229 |
-
"loss": 3.9864,
|
3230 |
-
"step": 11475
|
3231 |
-
},
|
3232 |
-
{
|
3233 |
-
"epoch": 0.9101701622477246,
|
3234 |
-
"grad_norm": 148.6637725830078,
|
3235 |
-
"learning_rate": 2.1207904074714083e-07,
|
3236 |
-
"loss": 3.6906,
|
3237 |
-
"step": 11500
|
3238 |
-
},
|
3239 |
-
{
|
3240 |
-
"epoch": 0.9121487930352197,
|
3241 |
-
"grad_norm": 164.6747283935547,
|
3242 |
-
"learning_rate": 2.1199659671015314e-07,
|
3243 |
-
"loss": 3.8543,
|
3244 |
-
"step": 11525
|
3245 |
-
},
|
3246 |
-
{
|
3247 |
-
"epoch": 0.9141274238227147,
|
3248 |
-
"grad_norm": 114.64603424072266,
|
3249 |
-
"learning_rate": 2.1191415267316546e-07,
|
3250 |
-
"loss": 3.6973,
|
3251 |
-
"step": 11550
|
3252 |
-
},
|
3253 |
-
{
|
3254 |
-
"epoch": 0.9161060546102098,
|
3255 |
-
"grad_norm": 129.53265380859375,
|
3256 |
-
"learning_rate": 2.1183170863617774e-07,
|
3257 |
-
"loss": 3.5896,
|
3258 |
-
"step": 11575
|
3259 |
-
},
|
3260 |
-
{
|
3261 |
-
"epoch": 0.9180846853977048,
|
3262 |
-
"grad_norm": 145.63973999023438,
|
3263 |
-
"learning_rate": 2.1174926459919008e-07,
|
3264 |
-
"loss": 3.963,
|
3265 |
-
"step": 11600
|
3266 |
-
},
|
3267 |
-
{
|
3268 |
-
"epoch": 0.9200633161851999,
|
3269 |
-
"grad_norm": 132.06729125976562,
|
3270 |
-
"learning_rate": 2.1166682056220236e-07,
|
3271 |
-
"loss": 3.6703,
|
3272 |
-
"step": 11625
|
3273 |
-
},
|
3274 |
-
{
|
3275 |
-
"epoch": 0.9220419469726949,
|
3276 |
-
"grad_norm": 129.76583862304688,
|
3277 |
-
"learning_rate": 2.1158437652521468e-07,
|
3278 |
-
"loss": 3.6965,
|
3279 |
-
"step": 11650
|
3280 |
-
},
|
3281 |
-
{
|
3282 |
-
"epoch": 0.92402057776019,
|
3283 |
-
"grad_norm": 115.21681213378906,
|
3284 |
-
"learning_rate": 2.1150193248822699e-07,
|
3285 |
-
"loss": 3.8711,
|
3286 |
-
"step": 11675
|
3287 |
-
},
|
3288 |
-
{
|
3289 |
-
"epoch": 0.925999208547685,
|
3290 |
-
"grad_norm": 85.92119598388672,
|
3291 |
-
"learning_rate": 2.114194884512393e-07,
|
3292 |
-
"loss": 3.8377,
|
3293 |
-
"step": 11700
|
3294 |
-
},
|
3295 |
-
{
|
3296 |
-
"epoch": 0.9279778393351801,
|
3297 |
-
"grad_norm": 117.11397552490234,
|
3298 |
-
"learning_rate": 2.113370444142516e-07,
|
3299 |
-
"loss": 3.7965,
|
3300 |
-
"step": 11725
|
3301 |
-
},
|
3302 |
-
{
|
3303 |
-
"epoch": 0.9299564701226751,
|
3304 |
-
"grad_norm": 113.50032806396484,
|
3305 |
-
"learning_rate": 2.112546003772639e-07,
|
3306 |
-
"loss": 3.7684,
|
3307 |
-
"step": 11750
|
3308 |
-
},
|
3309 |
-
{
|
3310 |
-
"epoch": 0.9319351009101702,
|
3311 |
-
"grad_norm": 107.85367584228516,
|
3312 |
-
"learning_rate": 2.111721563402762e-07,
|
3313 |
-
"loss": 3.6988,
|
3314 |
-
"step": 11775
|
3315 |
-
},
|
3316 |
-
{
|
3317 |
-
"epoch": 0.9339137316976652,
|
3318 |
-
"grad_norm": 121.42901611328125,
|
3319 |
-
"learning_rate": 2.1108971230328852e-07,
|
3320 |
-
"loss": 3.6745,
|
3321 |
-
"step": 11800
|
3322 |
-
},
|
3323 |
-
{
|
3324 |
-
"epoch": 0.9358923624851603,
|
3325 |
-
"grad_norm": 139.971923828125,
|
3326 |
-
"learning_rate": 2.1100726826630083e-07,
|
3327 |
-
"loss": 3.8403,
|
3328 |
-
"step": 11825
|
3329 |
-
},
|
3330 |
-
{
|
3331 |
-
"epoch": 0.9378709932726553,
|
3332 |
-
"grad_norm": 126.40741729736328,
|
3333 |
-
"learning_rate": 2.1092482422931312e-07,
|
3334 |
-
"loss": 3.7846,
|
3335 |
-
"step": 11850
|
3336 |
-
},
|
3337 |
-
{
|
3338 |
-
"epoch": 0.9398496240601504,
|
3339 |
-
"grad_norm": 110.96858215332031,
|
3340 |
-
"learning_rate": 2.1084238019232545e-07,
|
3341 |
-
"loss": 3.9256,
|
3342 |
-
"step": 11875
|
3343 |
-
},
|
3344 |
-
{
|
3345 |
-
"epoch": 0.9418282548476454,
|
3346 |
-
"grad_norm": 107.72772979736328,
|
3347 |
-
"learning_rate": 2.1075993615533774e-07,
|
3348 |
-
"loss": 3.7888,
|
3349 |
-
"step": 11900
|
3350 |
-
},
|
3351 |
-
{
|
3352 |
-
"epoch": 0.9438068856351405,
|
3353 |
-
"grad_norm": 94.70952606201172,
|
3354 |
-
"learning_rate": 2.1067749211835008e-07,
|
3355 |
-
"loss": 3.8496,
|
3356 |
-
"step": 11925
|
3357 |
-
},
|
3358 |
-
{
|
3359 |
-
"epoch": 0.9457855164226355,
|
3360 |
-
"grad_norm": 178.84500122070312,
|
3361 |
-
"learning_rate": 2.1059504808136236e-07,
|
3362 |
-
"loss": 3.737,
|
3363 |
-
"step": 11950
|
3364 |
-
},
|
3365 |
-
{
|
3366 |
-
"epoch": 0.9477641472101306,
|
3367 |
-
"grad_norm": 179.0364227294922,
|
3368 |
-
"learning_rate": 2.1051260404437467e-07,
|
3369 |
-
"loss": 3.7584,
|
3370 |
-
"step": 11975
|
3371 |
-
},
|
3372 |
-
{
|
3373 |
-
"epoch": 0.9497427779976256,
|
3374 |
-
"grad_norm": 141.7266845703125,
|
3375 |
-
"learning_rate": 2.1043016000738698e-07,
|
3376 |
-
"loss": 3.7948,
|
3377 |
-
"step": 12000
|
3378 |
-
},
|
3379 |
-
{
|
3380 |
-
"epoch": 0.9517214087851207,
|
3381 |
-
"grad_norm": 120.56340026855469,
|
3382 |
-
"learning_rate": 2.1034771597039927e-07,
|
3383 |
-
"loss": 3.8951,
|
3384 |
-
"step": 12025
|
3385 |
-
},
|
3386 |
-
{
|
3387 |
-
"epoch": 0.9537000395726157,
|
3388 |
-
"grad_norm": 113.90511322021484,
|
3389 |
-
"learning_rate": 2.1026527193341158e-07,
|
3390 |
-
"loss": 3.8823,
|
3391 |
-
"step": 12050
|
3392 |
-
},
|
3393 |
-
{
|
3394 |
-
"epoch": 0.9556786703601108,
|
3395 |
-
"grad_norm": 103.57383728027344,
|
3396 |
-
"learning_rate": 2.101828278964239e-07,
|
3397 |
-
"loss": 3.7853,
|
3398 |
-
"step": 12075
|
3399 |
-
},
|
3400 |
-
{
|
3401 |
-
"epoch": 0.9576573011476058,
|
3402 |
-
"grad_norm": 111.0738525390625,
|
3403 |
-
"learning_rate": 2.101003838594362e-07,
|
3404 |
-
"loss": 3.8394,
|
3405 |
-
"step": 12100
|
3406 |
-
},
|
3407 |
-
{
|
3408 |
-
"epoch": 0.9596359319351009,
|
3409 |
-
"grad_norm": 130.9629364013672,
|
3410 |
-
"learning_rate": 2.1001793982244852e-07,
|
3411 |
-
"loss": 3.7752,
|
3412 |
-
"step": 12125
|
3413 |
-
},
|
3414 |
-
{
|
3415 |
-
"epoch": 0.9616145627225959,
|
3416 |
-
"grad_norm": 98.29790496826172,
|
3417 |
-
"learning_rate": 2.0993549578546083e-07,
|
3418 |
-
"loss": 3.8332,
|
3419 |
-
"step": 12150
|
3420 |
-
},
|
3421 |
-
{
|
3422 |
-
"epoch": 0.963593193510091,
|
3423 |
-
"grad_norm": 116.66043853759766,
|
3424 |
-
"learning_rate": 2.098530517484731e-07,
|
3425 |
-
"loss": 3.964,
|
3426 |
-
"step": 12175
|
3427 |
-
},
|
3428 |
-
{
|
3429 |
-
"epoch": 0.965571824297586,
|
3430 |
-
"grad_norm": 179.0226287841797,
|
3431 |
-
"learning_rate": 2.0977060771148545e-07,
|
3432 |
-
"loss": 3.8848,
|
3433 |
-
"step": 12200
|
3434 |
-
},
|
3435 |
-
{
|
3436 |
-
"epoch": 0.9675504550850811,
|
3437 |
-
"grad_norm": 100.40991973876953,
|
3438 |
-
"learning_rate": 2.0968816367449774e-07,
|
3439 |
-
"loss": 3.8461,
|
3440 |
-
"step": 12225
|
3441 |
-
},
|
3442 |
-
{
|
3443 |
-
"epoch": 0.9695290858725761,
|
3444 |
-
"grad_norm": 117.44912719726562,
|
3445 |
-
"learning_rate": 2.0960571963751005e-07,
|
3446 |
-
"loss": 3.723,
|
3447 |
-
"step": 12250
|
3448 |
-
},
|
3449 |
-
{
|
3450 |
-
"epoch": 0.9715077166600712,
|
3451 |
-
"grad_norm": 106.5282974243164,
|
3452 |
-
"learning_rate": 2.0952327560052236e-07,
|
3453 |
-
"loss": 3.8557,
|
3454 |
-
"step": 12275
|
3455 |
-
},
|
3456 |
-
{
|
3457 |
-
"epoch": 0.9734863474475662,
|
3458 |
-
"grad_norm": 111.7376480102539,
|
3459 |
-
"learning_rate": 2.0944083156353467e-07,
|
3460 |
-
"loss": 3.8505,
|
3461 |
-
"step": 12300
|
3462 |
-
},
|
3463 |
-
{
|
3464 |
-
"epoch": 0.9754649782350613,
|
3465 |
-
"grad_norm": 185.3255157470703,
|
3466 |
-
"learning_rate": 2.0935838752654698e-07,
|
3467 |
-
"loss": 3.9339,
|
3468 |
-
"step": 12325
|
3469 |
-
},
|
3470 |
-
{
|
3471 |
-
"epoch": 0.9774436090225563,
|
3472 |
-
"grad_norm": 128.0303955078125,
|
3473 |
-
"learning_rate": 2.0927594348955927e-07,
|
3474 |
-
"loss": 3.9595,
|
3475 |
-
"step": 12350
|
3476 |
-
},
|
3477 |
-
{
|
3478 |
-
"epoch": 0.9794222398100515,
|
3479 |
-
"grad_norm": 117.01809692382812,
|
3480 |
-
"learning_rate": 2.0919349945257158e-07,
|
3481 |
-
"loss": 3.8385,
|
3482 |
-
"step": 12375
|
3483 |
-
},
|
3484 |
-
{
|
3485 |
-
"epoch": 0.9814008705975465,
|
3486 |
-
"grad_norm": 132.46775817871094,
|
3487 |
-
"learning_rate": 2.091110554155839e-07,
|
3488 |
-
"loss": 3.6976,
|
3489 |
-
"step": 12400
|
3490 |
-
},
|
3491 |
-
{
|
3492 |
-
"epoch": 0.9833795013850416,
|
3493 |
-
"grad_norm": 125.96062469482422,
|
3494 |
-
"learning_rate": 2.090286113785962e-07,
|
3495 |
-
"loss": 3.9652,
|
3496 |
-
"step": 12425
|
3497 |
-
},
|
3498 |
-
{
|
3499 |
-
"epoch": 0.9853581321725366,
|
3500 |
-
"grad_norm": 172.13600158691406,
|
3501 |
-
"learning_rate": 2.089461673416085e-07,
|
3502 |
-
"loss": 3.9026,
|
3503 |
-
"step": 12450
|
3504 |
-
},
|
3505 |
-
{
|
3506 |
-
"epoch": 0.9873367629600317,
|
3507 |
-
"grad_norm": 107.95278930664062,
|
3508 |
-
"learning_rate": 2.0886372330462083e-07,
|
3509 |
-
"loss": 3.8457,
|
3510 |
-
"step": 12475
|
3511 |
-
},
|
3512 |
-
{
|
3513 |
-
"epoch": 0.9893153937475268,
|
3514 |
-
"grad_norm": 196.08303833007812,
|
3515 |
-
"learning_rate": 2.087812792676331e-07,
|
3516 |
-
"loss": 3.8319,
|
3517 |
-
"step": 12500
|
3518 |
-
},
|
3519 |
-
{
|
3520 |
-
"epoch": 0.9912940245350218,
|
3521 |
-
"grad_norm": 119.07249450683594,
|
3522 |
-
"learning_rate": 2.0869883523064545e-07,
|
3523 |
-
"loss": 3.7464,
|
3524 |
-
"step": 12525
|
3525 |
-
},
|
3526 |
-
{
|
3527 |
-
"epoch": 0.9932726553225169,
|
3528 |
-
"grad_norm": 118.91383361816406,
|
3529 |
-
"learning_rate": 2.0861639119365773e-07,
|
3530 |
-
"loss": 3.7734,
|
3531 |
-
"step": 12550
|
3532 |
-
},
|
3533 |
-
{
|
3534 |
-
"epoch": 0.9952512861100119,
|
3535 |
-
"grad_norm": 122.37627410888672,
|
3536 |
-
"learning_rate": 2.0853394715667005e-07,
|
3537 |
-
"loss": 3.9263,
|
3538 |
-
"step": 12575
|
3539 |
-
},
|
3540 |
-
{
|
3541 |
-
"epoch": 0.997229916897507,
|
3542 |
-
"grad_norm": 101.24188232421875,
|
3543 |
-
"learning_rate": 2.0845150311968236e-07,
|
3544 |
-
"loss": 3.6865,
|
3545 |
-
"step": 12600
|
3546 |
-
},
|
3547 |
-
{
|
3548 |
-
"epoch": 0.999208547685002,
|
3549 |
-
"grad_norm": 253.74789428710938,
|
3550 |
-
"learning_rate": 2.0836905908269467e-07,
|
3551 |
-
"loss": 3.857,
|
3552 |
-
"step": 12625
|
3553 |
-
},
|
3554 |
-
{
|
3555 |
-
"epoch": 1.001187178472497,
|
3556 |
-
"grad_norm": 121.46297454833984,
|
3557 |
-
"learning_rate": 2.0828661504570695e-07,
|
3558 |
-
"loss": 3.9145,
|
3559 |
-
"step": 12650
|
3560 |
-
},
|
3561 |
-
{
|
3562 |
-
"epoch": 1.003165809259992,
|
3563 |
-
"grad_norm": 115.41104125976562,
|
3564 |
-
"learning_rate": 2.0820417100871927e-07,
|
3565 |
-
"loss": 3.9441,
|
3566 |
-
"step": 12675
|
3567 |
-
},
|
3568 |
-
{
|
3569 |
-
"epoch": 1.0051444400474872,
|
3570 |
-
"grad_norm": 105.03096008300781,
|
3571 |
-
"learning_rate": 2.0812172697173158e-07,
|
3572 |
-
"loss": 3.6926,
|
3573 |
-
"step": 12700
|
3574 |
-
},
|
3575 |
-
{
|
3576 |
-
"epoch": 1.007123070834982,
|
3577 |
-
"grad_norm": 98.86603546142578,
|
3578 |
-
"learning_rate": 2.080392829347439e-07,
|
3579 |
-
"loss": 3.5524,
|
3580 |
-
"step": 12725
|
3581 |
-
},
|
3582 |
-
{
|
3583 |
-
"epoch": 1.0091017016224773,
|
3584 |
-
"grad_norm": 120.75003814697266,
|
3585 |
-
"learning_rate": 2.079568388977562e-07,
|
3586 |
-
"loss": 3.8494,
|
3587 |
-
"step": 12750
|
3588 |
-
},
|
3589 |
-
{
|
3590 |
-
"epoch": 1.0110803324099722,
|
3591 |
-
"grad_norm": 86.03683471679688,
|
3592 |
-
"learning_rate": 2.0787439486076849e-07,
|
3593 |
-
"loss": 3.7635,
|
3594 |
-
"step": 12775
|
3595 |
-
},
|
3596 |
-
{
|
3597 |
-
"epoch": 1.0130589631974674,
|
3598 |
-
"grad_norm": 132.28945922851562,
|
3599 |
-
"learning_rate": 2.0779195082378082e-07,
|
3600 |
-
"loss": 4.007,
|
3601 |
-
"step": 12800
|
3602 |
-
},
|
3603 |
-
{
|
3604 |
-
"epoch": 1.0150375939849625,
|
3605 |
-
"grad_norm": 129.51051330566406,
|
3606 |
-
"learning_rate": 2.077095067867931e-07,
|
3607 |
-
"loss": 3.6065,
|
3608 |
-
"step": 12825
|
3609 |
-
},
|
3610 |
-
{
|
3611 |
-
"epoch": 1.0170162247724575,
|
3612 |
-
"grad_norm": 103.38876342773438,
|
3613 |
-
"learning_rate": 2.0762706274980542e-07,
|
3614 |
-
"loss": 3.7508,
|
3615 |
-
"step": 12850
|
3616 |
-
},
|
3617 |
-
{
|
3618 |
-
"epoch": 1.0189948555599526,
|
3619 |
-
"grad_norm": 163.7755584716797,
|
3620 |
-
"learning_rate": 2.0754461871281773e-07,
|
3621 |
-
"loss": 3.7519,
|
3622 |
-
"step": 12875
|
3623 |
-
},
|
3624 |
-
{
|
3625 |
-
"epoch": 1.0209734863474476,
|
3626 |
-
"grad_norm": 123.19217681884766,
|
3627 |
-
"learning_rate": 2.0746217467583004e-07,
|
3628 |
-
"loss": 3.8441,
|
3629 |
-
"step": 12900
|
3630 |
-
},
|
3631 |
-
{
|
3632 |
-
"epoch": 1.0229521171349427,
|
3633 |
-
"grad_norm": 122.48994445800781,
|
3634 |
-
"learning_rate": 2.0737973063884236e-07,
|
3635 |
-
"loss": 3.8664,
|
3636 |
-
"step": 12925
|
3637 |
-
},
|
3638 |
-
{
|
3639 |
-
"epoch": 1.0249307479224377,
|
3640 |
-
"grad_norm": 97.31283569335938,
|
3641 |
-
"learning_rate": 2.0729728660185467e-07,
|
3642 |
-
"loss": 3.9103,
|
3643 |
-
"step": 12950
|
3644 |
-
},
|
3645 |
-
{
|
3646 |
-
"epoch": 1.0269093787099328,
|
3647 |
-
"grad_norm": 178.35073852539062,
|
3648 |
-
"learning_rate": 2.0721484256486695e-07,
|
3649 |
-
"loss": 3.8718,
|
3650 |
-
"step": 12975
|
3651 |
-
},
|
3652 |
-
{
|
3653 |
-
"epoch": 1.0288880094974278,
|
3654 |
-
"grad_norm": 124.15072631835938,
|
3655 |
-
"learning_rate": 2.0713239852787926e-07,
|
3656 |
-
"loss": 3.7962,
|
3657 |
-
"step": 13000
|
3658 |
-
},
|
3659 |
-
{
|
3660 |
-
"epoch": 1.030866640284923,
|
3661 |
-
"grad_norm": 141.29360961914062,
|
3662 |
-
"learning_rate": 2.0704995449089158e-07,
|
3663 |
-
"loss": 3.8207,
|
3664 |
-
"step": 13025
|
3665 |
-
},
|
3666 |
-
{
|
3667 |
-
"epoch": 1.0328452710724179,
|
3668 |
-
"grad_norm": 117.48384857177734,
|
3669 |
-
"learning_rate": 2.0696751045390386e-07,
|
3670 |
-
"loss": 4.0525,
|
3671 |
-
"step": 13050
|
3672 |
-
},
|
3673 |
-
{
|
3674 |
-
"epoch": 1.034823901859913,
|
3675 |
-
"grad_norm": 124.6396713256836,
|
3676 |
-
"learning_rate": 2.068850664169162e-07,
|
3677 |
-
"loss": 3.8003,
|
3678 |
-
"step": 13075
|
3679 |
-
},
|
3680 |
-
{
|
3681 |
-
"epoch": 1.036802532647408,
|
3682 |
-
"grad_norm": 164.6786346435547,
|
3683 |
-
"learning_rate": 2.0680262237992848e-07,
|
3684 |
-
"loss": 3.6338,
|
3685 |
-
"step": 13100
|
3686 |
-
},
|
3687 |
-
{
|
3688 |
-
"epoch": 1.0387811634349031,
|
3689 |
-
"grad_norm": 144.25732421875,
|
3690 |
-
"learning_rate": 2.0672017834294082e-07,
|
3691 |
-
"loss": 3.9446,
|
3692 |
-
"step": 13125
|
3693 |
-
},
|
3694 |
-
{
|
3695 |
-
"epoch": 1.040759794222398,
|
3696 |
-
"grad_norm": 130.24571228027344,
|
3697 |
-
"learning_rate": 2.066377343059531e-07,
|
3698 |
-
"loss": 3.9012,
|
3699 |
-
"step": 13150
|
3700 |
-
},
|
3701 |
-
{
|
3702 |
-
"epoch": 1.0427384250098932,
|
3703 |
-
"grad_norm": 122.11434173583984,
|
3704 |
-
"learning_rate": 2.0655529026896542e-07,
|
3705 |
-
"loss": 3.7799,
|
3706 |
-
"step": 13175
|
3707 |
-
},
|
3708 |
-
{
|
3709 |
-
"epoch": 1.0447170557973882,
|
3710 |
-
"grad_norm": 161.62745666503906,
|
3711 |
-
"learning_rate": 2.0647284623197773e-07,
|
3712 |
-
"loss": 3.8881,
|
3713 |
-
"step": 13200
|
3714 |
-
},
|
3715 |
-
{
|
3716 |
-
"epoch": 1.0466956865848833,
|
3717 |
-
"grad_norm": 129.13072204589844,
|
3718 |
-
"learning_rate": 2.0639040219499004e-07,
|
3719 |
-
"loss": 3.8295,
|
3720 |
-
"step": 13225
|
3721 |
-
},
|
3722 |
-
{
|
3723 |
-
"epoch": 1.0486743173723783,
|
3724 |
-
"grad_norm": 153.86805725097656,
|
3725 |
-
"learning_rate": 2.0630795815800233e-07,
|
3726 |
-
"loss": 3.7328,
|
3727 |
-
"step": 13250
|
3728 |
-
},
|
3729 |
-
{
|
3730 |
-
"epoch": 1.0506529481598734,
|
3731 |
-
"grad_norm": 173.9022979736328,
|
3732 |
-
"learning_rate": 2.0622551412101464e-07,
|
3733 |
-
"loss": 3.9866,
|
3734 |
-
"step": 13275
|
3735 |
-
},
|
3736 |
-
{
|
3737 |
-
"epoch": 1.0526315789473684,
|
3738 |
-
"grad_norm": 104.5372543334961,
|
3739 |
-
"learning_rate": 2.0614307008402695e-07,
|
3740 |
-
"loss": 3.6121,
|
3741 |
-
"step": 13300
|
3742 |
-
},
|
3743 |
-
{
|
3744 |
-
"epoch": 1.0546102097348635,
|
3745 |
-
"grad_norm": 115.93688201904297,
|
3746 |
-
"learning_rate": 2.0606062604703926e-07,
|
3747 |
-
"loss": 3.8851,
|
3748 |
-
"step": 13325
|
3749 |
-
},
|
3750 |
-
{
|
3751 |
-
"epoch": 1.0565888405223585,
|
3752 |
-
"grad_norm": 140.26748657226562,
|
3753 |
-
"learning_rate": 2.0597818201005157e-07,
|
3754 |
-
"loss": 3.9924,
|
3755 |
-
"step": 13350
|
3756 |
-
},
|
3757 |
-
{
|
3758 |
-
"epoch": 1.0585674713098536,
|
3759 |
-
"grad_norm": 122.87477111816406,
|
3760 |
-
"learning_rate": 2.0589573797306386e-07,
|
3761 |
-
"loss": 3.7768,
|
3762 |
-
"step": 13375
|
3763 |
-
},
|
3764 |
-
{
|
3765 |
-
"epoch": 1.0605461020973486,
|
3766 |
-
"grad_norm": 105.3402099609375,
|
3767 |
-
"learning_rate": 2.058132939360762e-07,
|
3768 |
-
"loss": 3.6371,
|
3769 |
-
"step": 13400
|
3770 |
-
},
|
3771 |
-
{
|
3772 |
-
"epoch": 1.0625247328848437,
|
3773 |
-
"grad_norm": 112.1316146850586,
|
3774 |
-
"learning_rate": 2.0573084989908848e-07,
|
3775 |
-
"loss": 3.8185,
|
3776 |
-
"step": 13425
|
3777 |
-
},
|
3778 |
-
{
|
3779 |
-
"epoch": 1.0645033636723387,
|
3780 |
-
"grad_norm": 105.67694854736328,
|
3781 |
-
"learning_rate": 2.056484058621008e-07,
|
3782 |
-
"loss": 3.7202,
|
3783 |
-
"step": 13450
|
3784 |
-
},
|
3785 |
-
{
|
3786 |
-
"epoch": 1.0664819944598338,
|
3787 |
-
"grad_norm": 160.9032440185547,
|
3788 |
-
"learning_rate": 2.055659618251131e-07,
|
3789 |
-
"loss": 3.6819,
|
3790 |
-
"step": 13475
|
3791 |
-
},
|
3792 |
-
{
|
3793 |
-
"epoch": 1.0684606252473288,
|
3794 |
-
"grad_norm": 197.6964569091797,
|
3795 |
-
"learning_rate": 2.0548351778812542e-07,
|
3796 |
-
"loss": 3.914,
|
3797 |
-
"step": 13500
|
3798 |
-
},
|
3799 |
-
{
|
3800 |
-
"epoch": 1.070439256034824,
|
3801 |
-
"grad_norm": 100.68509674072266,
|
3802 |
-
"learning_rate": 2.0540107375113773e-07,
|
3803 |
-
"loss": 3.7281,
|
3804 |
-
"step": 13525
|
3805 |
-
},
|
3806 |
-
{
|
3807 |
-
"epoch": 1.0724178868223189,
|
3808 |
-
"grad_norm": 121.31140899658203,
|
3809 |
-
"learning_rate": 2.0531862971415004e-07,
|
3810 |
-
"loss": 3.7413,
|
3811 |
-
"step": 13550
|
3812 |
-
},
|
3813 |
-
{
|
3814 |
-
"epoch": 1.074396517609814,
|
3815 |
-
"grad_norm": 176.4757080078125,
|
3816 |
-
"learning_rate": 2.0523618567716232e-07,
|
3817 |
-
"loss": 4.1066,
|
3818 |
-
"step": 13575
|
3819 |
-
},
|
3820 |
-
{
|
3821 |
-
"epoch": 1.076375148397309,
|
3822 |
-
"grad_norm": 98.44918823242188,
|
3823 |
-
"learning_rate": 2.0515374164017464e-07,
|
3824 |
-
"loss": 3.6721,
|
3825 |
-
"step": 13600
|
3826 |
-
},
|
3827 |
-
{
|
3828 |
-
"epoch": 1.0783537791848041,
|
3829 |
-
"grad_norm": 114.9334487915039,
|
3830 |
-
"learning_rate": 2.0507129760318695e-07,
|
3831 |
-
"loss": 3.7888,
|
3832 |
-
"step": 13625
|
3833 |
-
},
|
3834 |
-
{
|
3835 |
-
"epoch": 1.080332409972299,
|
3836 |
-
"grad_norm": 172.9710235595703,
|
3837 |
-
"learning_rate": 2.0498885356619923e-07,
|
3838 |
-
"loss": 3.8811,
|
3839 |
-
"step": 13650
|
3840 |
-
},
|
3841 |
-
{
|
3842 |
-
"epoch": 1.0823110407597942,
|
3843 |
-
"grad_norm": 77.68281555175781,
|
3844 |
-
"learning_rate": 2.0490640952921157e-07,
|
3845 |
-
"loss": 3.7865,
|
3846 |
-
"step": 13675
|
3847 |
-
},
|
3848 |
-
{
|
3849 |
-
"epoch": 1.0842896715472894,
|
3850 |
-
"grad_norm": 117.03499603271484,
|
3851 |
-
"learning_rate": 2.0482396549222386e-07,
|
3852 |
-
"loss": 3.6545,
|
3853 |
-
"step": 13700
|
3854 |
-
},
|
3855 |
-
{
|
3856 |
-
"epoch": 1.0862683023347843,
|
3857 |
-
"grad_norm": 111.88362121582031,
|
3858 |
-
"learning_rate": 2.047415214552362e-07,
|
3859 |
-
"loss": 3.7833,
|
3860 |
-
"step": 13725
|
3861 |
-
},
|
3862 |
-
{
|
3863 |
-
"epoch": 1.0882469331222793,
|
3864 |
-
"grad_norm": 177.05654907226562,
|
3865 |
-
"learning_rate": 2.0465907741824848e-07,
|
3866 |
-
"loss": 3.8182,
|
3867 |
-
"step": 13750
|
3868 |
-
},
|
3869 |
-
{
|
3870 |
-
"epoch": 1.0902255639097744,
|
3871 |
-
"grad_norm": 110.29389953613281,
|
3872 |
-
"learning_rate": 2.045766333812608e-07,
|
3873 |
-
"loss": 3.571,
|
3874 |
-
"step": 13775
|
3875 |
-
},
|
3876 |
-
{
|
3877 |
-
"epoch": 1.0922041946972696,
|
3878 |
-
"grad_norm": 81.29917907714844,
|
3879 |
-
"learning_rate": 2.044941893442731e-07,
|
3880 |
-
"loss": 3.7003,
|
3881 |
-
"step": 13800
|
3882 |
-
},
|
3883 |
-
{
|
3884 |
-
"epoch": 1.0941828254847645,
|
3885 |
-
"grad_norm": 130.32589721679688,
|
3886 |
-
"learning_rate": 2.0441174530728541e-07,
|
3887 |
-
"loss": 3.965,
|
3888 |
-
"step": 13825
|
3889 |
-
},
|
3890 |
-
{
|
3891 |
-
"epoch": 1.0961614562722597,
|
3892 |
-
"grad_norm": 134.07106018066406,
|
3893 |
-
"learning_rate": 2.043293012702977e-07,
|
3894 |
-
"loss": 4.0457,
|
3895 |
-
"step": 13850
|
3896 |
-
},
|
3897 |
-
{
|
3898 |
-
"epoch": 1.0981400870597546,
|
3899 |
-
"grad_norm": 204.80294799804688,
|
3900 |
-
"learning_rate": 2.0424685723331004e-07,
|
3901 |
-
"loss": 3.5721,
|
3902 |
-
"step": 13875
|
3903 |
-
},
|
3904 |
-
{
|
3905 |
-
"epoch": 1.1001187178472498,
|
3906 |
-
"grad_norm": 157.00570678710938,
|
3907 |
-
"learning_rate": 2.0416441319632232e-07,
|
3908 |
-
"loss": 3.8119,
|
3909 |
-
"step": 13900
|
3910 |
-
},
|
3911 |
-
{
|
3912 |
-
"epoch": 1.1020973486347447,
|
3913 |
-
"grad_norm": 118.97384643554688,
|
3914 |
-
"learning_rate": 2.0408196915933463e-07,
|
3915 |
-
"loss": 3.6172,
|
3916 |
-
"step": 13925
|
3917 |
-
},
|
3918 |
-
{
|
3919 |
-
"epoch": 1.10407597942224,
|
3920 |
-
"grad_norm": 147.43740844726562,
|
3921 |
-
"learning_rate": 2.0399952512234695e-07,
|
3922 |
-
"loss": 3.8369,
|
3923 |
-
"step": 13950
|
3924 |
-
},
|
3925 |
-
{
|
3926 |
-
"epoch": 1.1060546102097348,
|
3927 |
-
"grad_norm": 116.36746215820312,
|
3928 |
-
"learning_rate": 2.0391708108535923e-07,
|
3929 |
-
"loss": 3.8669,
|
3930 |
-
"step": 13975
|
3931 |
-
},
|
3932 |
-
{
|
3933 |
-
"epoch": 1.10803324099723,
|
3934 |
-
"grad_norm": 119.54280090332031,
|
3935 |
-
"learning_rate": 2.0383463704837157e-07,
|
3936 |
-
"loss": 3.8281,
|
3937 |
-
"step": 14000
|
3938 |
-
},
|
3939 |
-
{
|
3940 |
-
"epoch": 1.110011871784725,
|
3941 |
-
"grad_norm": 161.6844024658203,
|
3942 |
-
"learning_rate": 2.0375219301138385e-07,
|
3943 |
-
"loss": 3.632,
|
3944 |
-
"step": 14025
|
3945 |
-
},
|
3946 |
-
{
|
3947 |
-
"epoch": 1.11199050257222,
|
3948 |
-
"grad_norm": 115.4676284790039,
|
3949 |
-
"learning_rate": 2.0366974897439617e-07,
|
3950 |
-
"loss": 3.6665,
|
3951 |
-
"step": 14050
|
3952 |
-
},
|
3953 |
-
{
|
3954 |
-
"epoch": 1.113969133359715,
|
3955 |
-
"grad_norm": 149.53504943847656,
|
3956 |
-
"learning_rate": 2.0358730493740848e-07,
|
3957 |
-
"loss": 3.8808,
|
3958 |
-
"step": 14075
|
3959 |
-
},
|
3960 |
-
{
|
3961 |
-
"epoch": 1.1159477641472102,
|
3962 |
-
"grad_norm": 108.68144989013672,
|
3963 |
-
"learning_rate": 2.035048609004208e-07,
|
3964 |
-
"loss": 3.763,
|
3965 |
-
"step": 14100
|
3966 |
-
},
|
3967 |
-
{
|
3968 |
-
"epoch": 1.1179263949347051,
|
3969 |
-
"grad_norm": 140.69000244140625,
|
3970 |
-
"learning_rate": 2.034224168634331e-07,
|
3971 |
-
"loss": 3.8483,
|
3972 |
-
"step": 14125
|
3973 |
-
},
|
3974 |
-
{
|
3975 |
-
"epoch": 1.1199050257222003,
|
3976 |
-
"grad_norm": 116.24478149414062,
|
3977 |
-
"learning_rate": 2.033399728264454e-07,
|
3978 |
-
"loss": 3.6232,
|
3979 |
-
"step": 14150
|
3980 |
-
},
|
3981 |
-
{
|
3982 |
-
"epoch": 1.1218836565096952,
|
3983 |
-
"grad_norm": 151.31192016601562,
|
3984 |
-
"learning_rate": 2.032575287894577e-07,
|
3985 |
-
"loss": 3.9463,
|
3986 |
-
"step": 14175
|
3987 |
-
},
|
3988 |
-
{
|
3989 |
-
"epoch": 1.1238622872971904,
|
3990 |
-
"grad_norm": 104.94828796386719,
|
3991 |
-
"learning_rate": 2.0317508475247004e-07,
|
3992 |
-
"loss": 3.8141,
|
3993 |
-
"step": 14200
|
3994 |
-
},
|
3995 |
-
{
|
3996 |
-
"epoch": 1.1258409180846853,
|
3997 |
-
"grad_norm": 174.0530548095703,
|
3998 |
-
"learning_rate": 2.0309264071548232e-07,
|
3999 |
-
"loss": 3.7278,
|
4000 |
-
"step": 14225
|
4001 |
-
},
|
4002 |
-
{
|
4003 |
-
"epoch": 1.1278195488721805,
|
4004 |
-
"grad_norm": 123.31523895263672,
|
4005 |
-
"learning_rate": 2.030101966784946e-07,
|
4006 |
-
"loss": 3.6594,
|
4007 |
-
"step": 14250
|
4008 |
-
},
|
4009 |
-
{
|
4010 |
-
"epoch": 1.1297981796596754,
|
4011 |
-
"grad_norm": 199.5244140625,
|
4012 |
-
"learning_rate": 2.0292775264150694e-07,
|
4013 |
-
"loss": 3.7397,
|
4014 |
-
"step": 14275
|
4015 |
-
},
|
4016 |
-
{
|
4017 |
-
"epoch": 1.1317768104471706,
|
4018 |
-
"grad_norm": 121.85034942626953,
|
4019 |
-
"learning_rate": 2.0284530860451923e-07,
|
4020 |
-
"loss": 3.7268,
|
4021 |
-
"step": 14300
|
4022 |
-
},
|
4023 |
-
{
|
4024 |
-
"epoch": 1.1337554412346655,
|
4025 |
-
"grad_norm": 100.23025512695312,
|
4026 |
-
"learning_rate": 2.0276286456753157e-07,
|
4027 |
-
"loss": 3.6129,
|
4028 |
-
"step": 14325
|
4029 |
-
},
|
4030 |
-
{
|
4031 |
-
"epoch": 1.1357340720221607,
|
4032 |
-
"grad_norm": 194.6273651123047,
|
4033 |
-
"learning_rate": 2.0268042053054385e-07,
|
4034 |
-
"loss": 3.8529,
|
4035 |
-
"step": 14350
|
4036 |
-
},
|
4037 |
-
{
|
4038 |
-
"epoch": 1.1377127028096556,
|
4039 |
-
"grad_norm": 86.26454162597656,
|
4040 |
-
"learning_rate": 2.0259797649355616e-07,
|
4041 |
-
"loss": 3.7558,
|
4042 |
-
"step": 14375
|
4043 |
-
},
|
4044 |
-
{
|
4045 |
-
"epoch": 1.1396913335971508,
|
4046 |
-
"grad_norm": 94.16887664794922,
|
4047 |
-
"learning_rate": 2.0251553245656847e-07,
|
4048 |
-
"loss": 3.743,
|
4049 |
-
"step": 14400
|
4050 |
-
},
|
4051 |
-
{
|
4052 |
-
"epoch": 1.1416699643846457,
|
4053 |
-
"grad_norm": 112.39076232910156,
|
4054 |
-
"learning_rate": 2.0243308841958079e-07,
|
4055 |
-
"loss": 3.6799,
|
4056 |
-
"step": 14425
|
4057 |
-
},
|
4058 |
-
{
|
4059 |
-
"epoch": 1.143648595172141,
|
4060 |
-
"grad_norm": 146.2572479248047,
|
4061 |
-
"learning_rate": 2.0235064438259307e-07,
|
4062 |
-
"loss": 3.8062,
|
4063 |
-
"step": 14450
|
4064 |
-
},
|
4065 |
-
{
|
4066 |
-
"epoch": 1.1456272259596358,
|
4067 |
-
"grad_norm": 100.57402801513672,
|
4068 |
-
"learning_rate": 2.022682003456054e-07,
|
4069 |
-
"loss": 3.7005,
|
4070 |
-
"step": 14475
|
4071 |
-
},
|
4072 |
-
{
|
4073 |
-
"epoch": 1.147605856747131,
|
4074 |
-
"grad_norm": 110.64191436767578,
|
4075 |
-
"learning_rate": 2.021857563086177e-07,
|
4076 |
-
"loss": 4.007,
|
4077 |
-
"step": 14500
|
4078 |
-
},
|
4079 |
-
{
|
4080 |
-
"epoch": 1.149584487534626,
|
4081 |
-
"grad_norm": 182.17724609375,
|
4082 |
-
"learning_rate": 2.0210331227163003e-07,
|
4083 |
-
"loss": 3.782,
|
4084 |
-
"step": 14525
|
4085 |
-
},
|
4086 |
-
{
|
4087 |
-
"epoch": 1.151563118322121,
|
4088 |
-
"grad_norm": 135.72979736328125,
|
4089 |
-
"learning_rate": 2.0202086823464232e-07,
|
4090 |
-
"loss": 3.6697,
|
4091 |
-
"step": 14550
|
4092 |
-
},
|
4093 |
-
{
|
4094 |
-
"epoch": 1.1535417491096163,
|
4095 |
-
"grad_norm": 101.66889953613281,
|
4096 |
-
"learning_rate": 2.019384241976546e-07,
|
4097 |
-
"loss": 3.7594,
|
4098 |
-
"step": 14575
|
4099 |
-
},
|
4100 |
-
{
|
4101 |
-
"epoch": 1.1555203798971112,
|
4102 |
-
"grad_norm": 160.48109436035156,
|
4103 |
-
"learning_rate": 2.0185598016066694e-07,
|
4104 |
-
"loss": 3.7643,
|
4105 |
-
"step": 14600
|
4106 |
-
},
|
4107 |
-
{
|
4108 |
-
"epoch": 1.1574990106846061,
|
4109 |
-
"grad_norm": 104.34113311767578,
|
4110 |
-
"learning_rate": 2.0177353612367923e-07,
|
4111 |
-
"loss": 3.9202,
|
4112 |
-
"step": 14625
|
4113 |
-
},
|
4114 |
-
{
|
4115 |
-
"epoch": 1.1594776414721013,
|
4116 |
-
"grad_norm": 89.21700286865234,
|
4117 |
-
"learning_rate": 2.0169109208669154e-07,
|
4118 |
-
"loss": 3.5803,
|
4119 |
-
"step": 14650
|
4120 |
-
},
|
4121 |
-
{
|
4122 |
-
"epoch": 1.1614562722595965,
|
4123 |
-
"grad_norm": 111.87966918945312,
|
4124 |
-
"learning_rate": 2.0160864804970385e-07,
|
4125 |
-
"loss": 3.5052,
|
4126 |
-
"step": 14675
|
4127 |
-
},
|
4128 |
-
{
|
4129 |
-
"epoch": 1.1634349030470914,
|
4130 |
-
"grad_norm": 134.26589965820312,
|
4131 |
-
"learning_rate": 2.0152620401271616e-07,
|
4132 |
-
"loss": 3.7542,
|
4133 |
-
"step": 14700
|
4134 |
-
},
|
4135 |
-
{
|
4136 |
-
"epoch": 1.1654135338345863,
|
4137 |
-
"grad_norm": 93.41925811767578,
|
4138 |
-
"learning_rate": 2.0144375997572847e-07,
|
4139 |
-
"loss": 3.8865,
|
4140 |
-
"step": 14725
|
4141 |
-
},
|
4142 |
-
{
|
4143 |
-
"epoch": 1.1673921646220815,
|
4144 |
-
"grad_norm": 121.87728881835938,
|
4145 |
-
"learning_rate": 2.0136131593874078e-07,
|
4146 |
-
"loss": 3.9852,
|
4147 |
-
"step": 14750
|
4148 |
-
},
|
4149 |
-
{
|
4150 |
-
"epoch": 1.1693707954095767,
|
4151 |
-
"grad_norm": 129.56564331054688,
|
4152 |
-
"learning_rate": 2.0127887190175307e-07,
|
4153 |
-
"loss": 3.6025,
|
4154 |
-
"step": 14775
|
4155 |
-
},
|
4156 |
-
{
|
4157 |
-
"epoch": 1.1713494261970716,
|
4158 |
-
"grad_norm": 138.36422729492188,
|
4159 |
-
"learning_rate": 2.011964278647654e-07,
|
4160 |
-
"loss": 3.9298,
|
4161 |
-
"step": 14800
|
4162 |
-
},
|
4163 |
-
{
|
4164 |
-
"epoch": 1.1733280569845668,
|
4165 |
-
"grad_norm": 115.63604736328125,
|
4166 |
-
"learning_rate": 2.011139838277777e-07,
|
4167 |
-
"loss": 3.811,
|
4168 |
-
"step": 14825
|
4169 |
-
},
|
4170 |
-
{
|
4171 |
-
"epoch": 1.1753066877720617,
|
4172 |
-
"grad_norm": 128.9251708984375,
|
4173 |
-
"learning_rate": 2.0103153979078998e-07,
|
4174 |
-
"loss": 4.013,
|
4175 |
-
"step": 14850
|
4176 |
-
},
|
4177 |
-
{
|
4178 |
-
"epoch": 1.1772853185595569,
|
4179 |
-
"grad_norm": 115.66397857666016,
|
4180 |
-
"learning_rate": 2.0094909575380232e-07,
|
4181 |
-
"loss": 3.9727,
|
4182 |
-
"step": 14875
|
4183 |
-
},
|
4184 |
-
{
|
4185 |
-
"epoch": 1.1792639493470518,
|
4186 |
-
"grad_norm": 105.60704040527344,
|
4187 |
-
"learning_rate": 2.008666517168146e-07,
|
4188 |
-
"loss": 4.0072,
|
4189 |
-
"step": 14900
|
4190 |
-
},
|
4191 |
-
{
|
4192 |
-
"epoch": 1.181242580134547,
|
4193 |
-
"grad_norm": 127.380859375,
|
4194 |
-
"learning_rate": 2.0078420767982694e-07,
|
4195 |
-
"loss": 3.763,
|
4196 |
-
"step": 14925
|
4197 |
-
},
|
4198 |
-
{
|
4199 |
-
"epoch": 1.183221210922042,
|
4200 |
-
"grad_norm": 87.45874786376953,
|
4201 |
-
"learning_rate": 2.0070176364283922e-07,
|
4202 |
-
"loss": 3.6529,
|
4203 |
-
"step": 14950
|
4204 |
-
},
|
4205 |
-
{
|
4206 |
-
"epoch": 1.185199841709537,
|
4207 |
-
"grad_norm": 257.6990966796875,
|
4208 |
-
"learning_rate": 2.0061931960585154e-07,
|
4209 |
-
"loss": 3.8049,
|
4210 |
-
"step": 14975
|
4211 |
-
},
|
4212 |
-
{
|
4213 |
-
"epoch": 1.187178472497032,
|
4214 |
-
"grad_norm": 111.80567169189453,
|
4215 |
-
"learning_rate": 2.0053687556886385e-07,
|
4216 |
-
"loss": 3.9666,
|
4217 |
-
"step": 15000
|
4218 |
-
},
|
4219 |
-
{
|
4220 |
-
"epoch": 1.1891571032845272,
|
4221 |
-
"grad_norm": 98.00049591064453,
|
4222 |
-
"learning_rate": 2.0045443153187616e-07,
|
4223 |
-
"loss": 3.773,
|
4224 |
-
"step": 15025
|
4225 |
-
},
|
4226 |
-
{
|
4227 |
-
"epoch": 1.1911357340720221,
|
4228 |
-
"grad_norm": 117.19742584228516,
|
4229 |
-
"learning_rate": 2.0037198749488844e-07,
|
4230 |
-
"loss": 3.8228,
|
4231 |
-
"step": 15050
|
4232 |
-
},
|
4233 |
-
{
|
4234 |
-
"epoch": 1.1931143648595173,
|
4235 |
-
"grad_norm": 99.34817504882812,
|
4236 |
-
"learning_rate": 2.0028954345790078e-07,
|
4237 |
-
"loss": 3.8045,
|
4238 |
-
"step": 15075
|
4239 |
-
},
|
4240 |
-
{
|
4241 |
-
"epoch": 1.1950929956470122,
|
4242 |
-
"grad_norm": 193.61734008789062,
|
4243 |
-
"learning_rate": 2.0020709942091307e-07,
|
4244 |
-
"loss": 3.7231,
|
4245 |
-
"step": 15100
|
4246 |
-
},
|
4247 |
-
{
|
4248 |
-
"epoch": 1.1970716264345074,
|
4249 |
-
"grad_norm": 110.1533203125,
|
4250 |
-
"learning_rate": 2.001246553839254e-07,
|
4251 |
-
"loss": 3.7013,
|
4252 |
-
"step": 15125
|
4253 |
-
},
|
4254 |
-
{
|
4255 |
-
"epoch": 1.1990502572220023,
|
4256 |
-
"grad_norm": 81.41429901123047,
|
4257 |
-
"learning_rate": 2.000422113469377e-07,
|
4258 |
-
"loss": 3.5668,
|
4259 |
-
"step": 15150
|
4260 |
-
},
|
4261 |
-
{
|
4262 |
-
"epoch": 1.2010288880094975,
|
4263 |
-
"grad_norm": 103.47528839111328,
|
4264 |
-
"learning_rate": 1.9995976730994998e-07,
|
4265 |
-
"loss": 3.6627,
|
4266 |
-
"step": 15175
|
4267 |
-
},
|
4268 |
-
{
|
4269 |
-
"epoch": 1.2030075187969924,
|
4270 |
-
"grad_norm": 179.40489196777344,
|
4271 |
-
"learning_rate": 1.9987732327296231e-07,
|
4272 |
-
"loss": 3.8307,
|
4273 |
-
"step": 15200
|
4274 |
-
},
|
4275 |
-
{
|
4276 |
-
"epoch": 1.2049861495844876,
|
4277 |
-
"grad_norm": 108.21305084228516,
|
4278 |
-
"learning_rate": 1.997948792359746e-07,
|
4279 |
-
"loss": 3.7619,
|
4280 |
-
"step": 15225
|
4281 |
-
},
|
4282 |
-
{
|
4283 |
-
"epoch": 1.2069647803719825,
|
4284 |
-
"grad_norm": 206.6326904296875,
|
4285 |
-
"learning_rate": 1.997124351989869e-07,
|
4286 |
-
"loss": 3.7036,
|
4287 |
-
"step": 15250
|
4288 |
-
},
|
4289 |
-
{
|
4290 |
-
"epoch": 1.2089434111594777,
|
4291 |
-
"grad_norm": 115.525390625,
|
4292 |
-
"learning_rate": 1.9962999116199922e-07,
|
4293 |
-
"loss": 3.823,
|
4294 |
-
"step": 15275
|
4295 |
-
},
|
4296 |
-
{
|
4297 |
-
"epoch": 1.2109220419469726,
|
4298 |
-
"grad_norm": 124.59696197509766,
|
4299 |
-
"learning_rate": 1.9954754712501153e-07,
|
4300 |
-
"loss": 3.7983,
|
4301 |
-
"step": 15300
|
4302 |
-
},
|
4303 |
-
{
|
4304 |
-
"epoch": 1.2129006727344678,
|
4305 |
-
"grad_norm": 107.92992401123047,
|
4306 |
-
"learning_rate": 1.9946510308802385e-07,
|
4307 |
-
"loss": 3.7677,
|
4308 |
-
"step": 15325
|
4309 |
-
},
|
4310 |
-
{
|
4311 |
-
"epoch": 1.2148793035219627,
|
4312 |
-
"grad_norm": 164.7980194091797,
|
4313 |
-
"learning_rate": 1.9938265905103616e-07,
|
4314 |
-
"loss": 3.7692,
|
4315 |
-
"step": 15350
|
4316 |
-
},
|
4317 |
-
{
|
4318 |
-
"epoch": 1.2168579343094579,
|
4319 |
-
"grad_norm": 95.48210906982422,
|
4320 |
-
"learning_rate": 1.9930021501404844e-07,
|
4321 |
-
"loss": 3.8568,
|
4322 |
-
"step": 15375
|
4323 |
-
},
|
4324 |
-
{
|
4325 |
-
"epoch": 1.2188365650969528,
|
4326 |
-
"grad_norm": 172.18484497070312,
|
4327 |
-
"learning_rate": 1.9921777097706078e-07,
|
4328 |
-
"loss": 3.596,
|
4329 |
-
"step": 15400
|
4330 |
-
},
|
4331 |
-
{
|
4332 |
-
"epoch": 1.220815195884448,
|
4333 |
-
"grad_norm": 146.89479064941406,
|
4334 |
-
"learning_rate": 1.9913532694007307e-07,
|
4335 |
-
"loss": 3.7992,
|
4336 |
-
"step": 15425
|
4337 |
-
},
|
4338 |
-
{
|
4339 |
-
"epoch": 1.222793826671943,
|
4340 |
-
"grad_norm": 106.06961822509766,
|
4341 |
-
"learning_rate": 1.9905288290308538e-07,
|
4342 |
-
"loss": 3.5689,
|
4343 |
-
"step": 15450
|
4344 |
-
},
|
4345 |
-
{
|
4346 |
-
"epoch": 1.224772457459438,
|
4347 |
-
"grad_norm": 124.81307220458984,
|
4348 |
-
"learning_rate": 1.989704388660977e-07,
|
4349 |
-
"loss": 3.8072,
|
4350 |
-
"step": 15475
|
4351 |
-
},
|
4352 |
-
{
|
4353 |
-
"epoch": 1.226751088246933,
|
4354 |
-
"grad_norm": 144.33468627929688,
|
4355 |
-
"learning_rate": 1.9888799482910997e-07,
|
4356 |
-
"loss": 3.7889,
|
4357 |
-
"step": 15500
|
4358 |
-
},
|
4359 |
-
{
|
4360 |
-
"epoch": 1.2287297190344282,
|
4361 |
-
"grad_norm": 108.416015625,
|
4362 |
-
"learning_rate": 1.988055507921223e-07,
|
4363 |
-
"loss": 3.7345,
|
4364 |
-
"step": 15525
|
4365 |
-
},
|
4366 |
-
{
|
4367 |
-
"epoch": 1.2307083498219233,
|
4368 |
-
"grad_norm": 118.40252685546875,
|
4369 |
-
"learning_rate": 1.987231067551346e-07,
|
4370 |
-
"loss": 3.9306,
|
4371 |
-
"step": 15550
|
4372 |
-
},
|
4373 |
-
{
|
4374 |
-
"epoch": 1.2326869806094183,
|
4375 |
-
"grad_norm": 121.7845230102539,
|
4376 |
-
"learning_rate": 1.986406627181469e-07,
|
4377 |
-
"loss": 3.6828,
|
4378 |
-
"step": 15575
|
4379 |
-
},
|
4380 |
-
{
|
4381 |
-
"epoch": 1.2346656113969132,
|
4382 |
-
"grad_norm": 126.77731323242188,
|
4383 |
-
"learning_rate": 1.9855821868115922e-07,
|
4384 |
-
"loss": 3.8064,
|
4385 |
-
"step": 15600
|
4386 |
-
},
|
4387 |
-
{
|
4388 |
-
"epoch": 1.2366442421844084,
|
4389 |
-
"grad_norm": 128.55198669433594,
|
4390 |
-
"learning_rate": 1.9847577464417153e-07,
|
4391 |
-
"loss": 3.952,
|
4392 |
-
"step": 15625
|
4393 |
-
},
|
4394 |
-
{
|
4395 |
-
"epoch": 1.2386228729719035,
|
4396 |
-
"grad_norm": 99.97991180419922,
|
4397 |
-
"learning_rate": 1.9839333060718382e-07,
|
4398 |
-
"loss": 3.8026,
|
4399 |
-
"step": 15650
|
4400 |
-
},
|
4401 |
-
{
|
4402 |
-
"epoch": 1.2406015037593985,
|
4403 |
-
"grad_norm": 109.74803924560547,
|
4404 |
-
"learning_rate": 1.9831088657019615e-07,
|
4405 |
-
"loss": 3.8499,
|
4406 |
-
"step": 15675
|
4407 |
-
},
|
4408 |
-
{
|
4409 |
-
"epoch": 1.2425801345468936,
|
4410 |
-
"grad_norm": 125.961669921875,
|
4411 |
-
"learning_rate": 1.9822844253320844e-07,
|
4412 |
-
"loss": 3.8659,
|
4413 |
-
"step": 15700
|
4414 |
-
},
|
4415 |
-
{
|
4416 |
-
"epoch": 1.2445587653343886,
|
4417 |
-
"grad_norm": 106.30962371826172,
|
4418 |
-
"learning_rate": 1.9814599849622078e-07,
|
4419 |
-
"loss": 3.7085,
|
4420 |
-
"step": 15725
|
4421 |
-
},
|
4422 |
-
{
|
4423 |
-
"epoch": 1.2465373961218837,
|
4424 |
-
"grad_norm": 101.37333679199219,
|
4425 |
-
"learning_rate": 1.9806355445923306e-07,
|
4426 |
-
"loss": 3.6006,
|
4427 |
-
"step": 15750
|
4428 |
-
},
|
4429 |
-
{
|
4430 |
-
"epoch": 1.2485160269093787,
|
4431 |
-
"grad_norm": 117.87637329101562,
|
4432 |
-
"learning_rate": 1.9798111042224537e-07,
|
4433 |
-
"loss": 3.7463,
|
4434 |
-
"step": 15775
|
4435 |
-
},
|
4436 |
-
{
|
4437 |
-
"epoch": 1.2504946576968738,
|
4438 |
-
"grad_norm": 158.502685546875,
|
4439 |
-
"learning_rate": 1.9789866638525769e-07,
|
4440 |
-
"loss": 3.7699,
|
4441 |
-
"step": 15800
|
4442 |
-
},
|
4443 |
-
{
|
4444 |
-
"epoch": 1.2524732884843688,
|
4445 |
-
"grad_norm": 213.62936401367188,
|
4446 |
-
"learning_rate": 1.9781622234826997e-07,
|
4447 |
-
"loss": 3.9892,
|
4448 |
-
"step": 15825
|
4449 |
-
},
|
4450 |
-
{
|
4451 |
-
"epoch": 1.254451919271864,
|
4452 |
-
"grad_norm": 120.2110595703125,
|
4453 |
-
"learning_rate": 1.9773377831128228e-07,
|
4454 |
-
"loss": 3.6972,
|
4455 |
-
"step": 15850
|
4456 |
-
},
|
4457 |
-
{
|
4458 |
-
"epoch": 1.2564305500593589,
|
4459 |
-
"grad_norm": 115.94564819335938,
|
4460 |
-
"learning_rate": 1.976513342742946e-07,
|
4461 |
-
"loss": 3.6659,
|
4462 |
-
"step": 15875
|
4463 |
-
},
|
4464 |
-
{
|
4465 |
-
"epoch": 1.258409180846854,
|
4466 |
-
"grad_norm": 100.15674591064453,
|
4467 |
-
"learning_rate": 1.975688902373069e-07,
|
4468 |
-
"loss": 3.6585,
|
4469 |
-
"step": 15900
|
4470 |
-
},
|
4471 |
-
{
|
4472 |
-
"epoch": 1.260387811634349,
|
4473 |
-
"grad_norm": 126.27003479003906,
|
4474 |
-
"learning_rate": 1.9748644620031922e-07,
|
4475 |
-
"loss": 3.793,
|
4476 |
-
"step": 15925
|
4477 |
-
},
|
4478 |
-
{
|
4479 |
-
"epoch": 1.2623664424218441,
|
4480 |
-
"grad_norm": 122.43645477294922,
|
4481 |
-
"learning_rate": 1.9740400216333153e-07,
|
4482 |
-
"loss": 3.7562,
|
4483 |
-
"step": 15950
|
4484 |
-
},
|
4485 |
-
{
|
4486 |
-
"epoch": 1.264345073209339,
|
4487 |
-
"grad_norm": 121.08145904541016,
|
4488 |
-
"learning_rate": 1.9732155812634381e-07,
|
4489 |
-
"loss": 3.759,
|
4490 |
-
"step": 15975
|
4491 |
-
},
|
4492 |
-
{
|
4493 |
-
"epoch": 1.2663237039968342,
|
4494 |
-
"grad_norm": 116.81194305419922,
|
4495 |
-
"learning_rate": 1.9723911408935615e-07,
|
4496 |
-
"loss": 3.6303,
|
4497 |
-
"step": 16000
|
4498 |
-
},
|
4499 |
-
{
|
4500 |
-
"epoch": 1.2683023347843292,
|
4501 |
-
"grad_norm": 107.8984146118164,
|
4502 |
-
"learning_rate": 1.9715667005236844e-07,
|
4503 |
-
"loss": 3.7525,
|
4504 |
-
"step": 16025
|
4505 |
-
},
|
4506 |
-
{
|
4507 |
-
"epoch": 1.2702809655718243,
|
4508 |
-
"grad_norm": 114.40152740478516,
|
4509 |
-
"learning_rate": 1.9707422601538075e-07,
|
4510 |
-
"loss": 3.6044,
|
4511 |
-
"step": 16050
|
4512 |
-
},
|
4513 |
-
{
|
4514 |
-
"epoch": 1.2722595963593193,
|
4515 |
-
"grad_norm": 128.336181640625,
|
4516 |
-
"learning_rate": 1.9699178197839306e-07,
|
4517 |
-
"loss": 3.6883,
|
4518 |
-
"step": 16075
|
4519 |
-
},
|
4520 |
-
{
|
4521 |
-
"epoch": 1.2742382271468145,
|
4522 |
-
"grad_norm": 177.9373321533203,
|
4523 |
-
"learning_rate": 1.9690933794140535e-07,
|
4524 |
-
"loss": 3.7305,
|
4525 |
-
"step": 16100
|
4526 |
-
},
|
4527 |
-
{
|
4528 |
-
"epoch": 1.2762168579343094,
|
4529 |
-
"grad_norm": 137.6163787841797,
|
4530 |
-
"learning_rate": 1.9682689390441768e-07,
|
4531 |
-
"loss": 3.4276,
|
4532 |
-
"step": 16125
|
4533 |
-
},
|
4534 |
-
{
|
4535 |
-
"epoch": 1.2781954887218046,
|
4536 |
-
"grad_norm": 93.84378814697266,
|
4537 |
-
"learning_rate": 1.9674444986742997e-07,
|
4538 |
-
"loss": 3.954,
|
4539 |
-
"step": 16150
|
4540 |
-
},
|
4541 |
-
{
|
4542 |
-
"epoch": 1.2801741195092995,
|
4543 |
-
"grad_norm": 114.7217788696289,
|
4544 |
-
"learning_rate": 1.9666200583044228e-07,
|
4545 |
-
"loss": 3.9032,
|
4546 |
-
"step": 16175
|
4547 |
-
},
|
4548 |
-
{
|
4549 |
-
"epoch": 1.2821527502967947,
|
4550 |
-
"grad_norm": 127.64576721191406,
|
4551 |
-
"learning_rate": 1.965795617934546e-07,
|
4552 |
-
"loss": 3.9561,
|
4553 |
-
"step": 16200
|
4554 |
-
},
|
4555 |
-
{
|
4556 |
-
"epoch": 1.2841313810842896,
|
4557 |
-
"grad_norm": 147.38531494140625,
|
4558 |
-
"learning_rate": 1.964971177564669e-07,
|
4559 |
-
"loss": 3.7639,
|
4560 |
-
"step": 16225
|
4561 |
-
},
|
4562 |
-
{
|
4563 |
-
"epoch": 1.2861100118717848,
|
4564 |
-
"grad_norm": 106.52879333496094,
|
4565 |
-
"learning_rate": 1.964146737194792e-07,
|
4566 |
-
"loss": 3.6835,
|
4567 |
-
"step": 16250
|
4568 |
-
},
|
4569 |
-
{
|
4570 |
-
"epoch": 1.2880886426592797,
|
4571 |
-
"grad_norm": 117.00553131103516,
|
4572 |
-
"learning_rate": 1.9633222968249153e-07,
|
4573 |
-
"loss": 3.5749,
|
4574 |
-
"step": 16275
|
4575 |
-
},
|
4576 |
-
{
|
4577 |
-
"epoch": 1.2900672734467749,
|
4578 |
-
"grad_norm": 145.50201416015625,
|
4579 |
-
"learning_rate": 1.962497856455038e-07,
|
4580 |
-
"loss": 3.8452,
|
4581 |
-
"step": 16300
|
4582 |
-
},
|
4583 |
-
{
|
4584 |
-
"epoch": 1.29204590423427,
|
4585 |
-
"grad_norm": 109.88056945800781,
|
4586 |
-
"learning_rate": 1.9616734160851615e-07,
|
4587 |
-
"loss": 3.7797,
|
4588 |
-
"step": 16325
|
4589 |
-
},
|
4590 |
-
{
|
4591 |
-
"epoch": 1.294024535021765,
|
4592 |
-
"grad_norm": 98.50968170166016,
|
4593 |
-
"learning_rate": 1.9608489757152844e-07,
|
4594 |
-
"loss": 3.9423,
|
4595 |
-
"step": 16350
|
4596 |
-
},
|
4597 |
-
{
|
4598 |
-
"epoch": 1.29600316580926,
|
4599 |
-
"grad_norm": 137.3463134765625,
|
4600 |
-
"learning_rate": 1.9600245353454075e-07,
|
4601 |
-
"loss": 3.8175,
|
4602 |
-
"step": 16375
|
4603 |
-
},
|
4604 |
-
{
|
4605 |
-
"epoch": 1.297981796596755,
|
4606 |
-
"grad_norm": 104.44812774658203,
|
4607 |
-
"learning_rate": 1.9592000949755306e-07,
|
4608 |
-
"loss": 3.8284,
|
4609 |
-
"step": 16400
|
4610 |
-
},
|
4611 |
-
{
|
4612 |
-
"epoch": 1.2999604273842502,
|
4613 |
-
"grad_norm": 134.79293823242188,
|
4614 |
-
"learning_rate": 1.9583756546056534e-07,
|
4615 |
-
"loss": 3.8295,
|
4616 |
-
"step": 16425
|
4617 |
-
},
|
4618 |
-
{
|
4619 |
-
"epoch": 1.3019390581717452,
|
4620 |
-
"grad_norm": 136.05784606933594,
|
4621 |
-
"learning_rate": 1.9575512142357766e-07,
|
4622 |
-
"loss": 3.9577,
|
4623 |
-
"step": 16450
|
4624 |
-
},
|
4625 |
-
{
|
4626 |
-
"epoch": 1.30391768895924,
|
4627 |
-
"grad_norm": 162.8216094970703,
|
4628 |
-
"learning_rate": 1.9567267738658997e-07,
|
4629 |
-
"loss": 3.9843,
|
4630 |
-
"step": 16475
|
4631 |
-
},
|
4632 |
-
{
|
4633 |
-
"epoch": 1.3058963197467353,
|
4634 |
-
"grad_norm": 132.25741577148438,
|
4635 |
-
"learning_rate": 1.9559023334960228e-07,
|
4636 |
-
"loss": 3.8045,
|
4637 |
-
"step": 16500
|
4638 |
-
},
|
4639 |
-
{
|
4640 |
-
"epoch": 1.3078749505342304,
|
4641 |
-
"grad_norm": 140.72642517089844,
|
4642 |
-
"learning_rate": 1.955077893126146e-07,
|
4643 |
-
"loss": 3.8217,
|
4644 |
-
"step": 16525
|
4645 |
-
},
|
4646 |
-
{
|
4647 |
-
"epoch": 1.3098535813217254,
|
4648 |
-
"grad_norm": 119.49310302734375,
|
4649 |
-
"learning_rate": 1.954253452756269e-07,
|
4650 |
-
"loss": 3.6075,
|
4651 |
-
"step": 16550
|
4652 |
-
},
|
4653 |
-
{
|
4654 |
-
"epoch": 1.3118322121092203,
|
4655 |
-
"grad_norm": 189.51280212402344,
|
4656 |
-
"learning_rate": 1.953429012386392e-07,
|
4657 |
-
"loss": 3.9516,
|
4658 |
-
"step": 16575
|
4659 |
-
},
|
4660 |
-
{
|
4661 |
-
"epoch": 1.3138108428967155,
|
4662 |
-
"grad_norm": 106.88180541992188,
|
4663 |
-
"learning_rate": 1.9526045720165153e-07,
|
4664 |
-
"loss": 3.9774,
|
4665 |
-
"step": 16600
|
4666 |
-
},
|
4667 |
-
{
|
4668 |
-
"epoch": 1.3157894736842106,
|
4669 |
-
"grad_norm": 136.42471313476562,
|
4670 |
-
"learning_rate": 1.951780131646638e-07,
|
4671 |
-
"loss": 3.7752,
|
4672 |
-
"step": 16625
|
4673 |
-
},
|
4674 |
-
{
|
4675 |
-
"epoch": 1.3177681044717056,
|
4676 |
-
"grad_norm": 134.68948364257812,
|
4677 |
-
"learning_rate": 1.9509556912767612e-07,
|
4678 |
-
"loss": 3.6695,
|
4679 |
-
"step": 16650
|
4680 |
-
},
|
4681 |
-
{
|
4682 |
-
"epoch": 1.3197467352592005,
|
4683 |
-
"grad_norm": 88.88003540039062,
|
4684 |
-
"learning_rate": 1.9501312509068843e-07,
|
4685 |
-
"loss": 3.9039,
|
4686 |
-
"step": 16675
|
4687 |
-
},
|
4688 |
-
{
|
4689 |
-
"epoch": 1.3217253660466957,
|
4690 |
-
"grad_norm": 106.34395599365234,
|
4691 |
-
"learning_rate": 1.9493068105370075e-07,
|
4692 |
-
"loss": 3.6726,
|
4693 |
-
"step": 16700
|
4694 |
-
},
|
4695 |
-
{
|
4696 |
-
"epoch": 1.3237039968341908,
|
4697 |
-
"grad_norm": 94.92339324951172,
|
4698 |
-
"learning_rate": 1.9484823701671306e-07,
|
4699 |
-
"loss": 3.7362,
|
4700 |
-
"step": 16725
|
4701 |
-
},
|
4702 |
-
{
|
4703 |
-
"epoch": 1.3256826276216858,
|
4704 |
-
"grad_norm": 131.17221069335938,
|
4705 |
-
"learning_rate": 1.9476579297972534e-07,
|
4706 |
-
"loss": 3.6668,
|
4707 |
-
"step": 16750
|
4708 |
-
},
|
4709 |
-
{
|
4710 |
-
"epoch": 1.327661258409181,
|
4711 |
-
"grad_norm": 115.50166320800781,
|
4712 |
-
"learning_rate": 1.9468334894273765e-07,
|
4713 |
-
"loss": 3.9881,
|
4714 |
-
"step": 16775
|
4715 |
-
},
|
4716 |
-
{
|
4717 |
-
"epoch": 1.3296398891966759,
|
4718 |
-
"grad_norm": 146.72767639160156,
|
4719 |
-
"learning_rate": 1.9460090490574997e-07,
|
4720 |
-
"loss": 3.8375,
|
4721 |
-
"step": 16800
|
4722 |
-
},
|
4723 |
-
{
|
4724 |
-
"epoch": 1.331618519984171,
|
4725 |
-
"grad_norm": 125.41651153564453,
|
4726 |
-
"learning_rate": 1.9451846086876228e-07,
|
4727 |
-
"loss": 3.7523,
|
4728 |
-
"step": 16825
|
4729 |
-
},
|
4730 |
-
{
|
4731 |
-
"epoch": 1.333597150771666,
|
4732 |
-
"grad_norm": 148.9279327392578,
|
4733 |
-
"learning_rate": 1.9443601683177456e-07,
|
4734 |
-
"loss": 3.7209,
|
4735 |
-
"step": 16850
|
4736 |
-
},
|
4737 |
-
{
|
4738 |
-
"epoch": 1.3355757815591611,
|
4739 |
-
"grad_norm": 132.95213317871094,
|
4740 |
-
"learning_rate": 1.943535727947869e-07,
|
4741 |
-
"loss": 3.8043,
|
4742 |
-
"step": 16875
|
4743 |
-
},
|
4744 |
-
{
|
4745 |
-
"epoch": 1.337554412346656,
|
4746 |
-
"grad_norm": 121.32286834716797,
|
4747 |
-
"learning_rate": 1.9427112875779919e-07,
|
4748 |
-
"loss": 3.8463,
|
4749 |
-
"step": 16900
|
4750 |
-
},
|
4751 |
-
{
|
4752 |
-
"epoch": 1.3395330431341512,
|
4753 |
-
"grad_norm": 104.13847351074219,
|
4754 |
-
"learning_rate": 1.9418868472081152e-07,
|
4755 |
-
"loss": 3.6028,
|
4756 |
-
"step": 16925
|
4757 |
-
},
|
4758 |
-
{
|
4759 |
-
"epoch": 1.3415116739216462,
|
4760 |
-
"grad_norm": 116.32896423339844,
|
4761 |
-
"learning_rate": 1.941062406838238e-07,
|
4762 |
-
"loss": 3.6065,
|
4763 |
-
"step": 16950
|
4764 |
-
},
|
4765 |
-
{
|
4766 |
-
"epoch": 1.3434903047091413,
|
4767 |
-
"grad_norm": 123.75446319580078,
|
4768 |
-
"learning_rate": 1.9402379664683612e-07,
|
4769 |
-
"loss": 3.8383,
|
4770 |
-
"step": 16975
|
4771 |
-
},
|
4772 |
-
{
|
4773 |
-
"epoch": 1.3454689354966363,
|
4774 |
-
"grad_norm": 146.36181640625,
|
4775 |
-
"learning_rate": 1.9394135260984843e-07,
|
4776 |
-
"loss": 3.7784,
|
4777 |
-
"step": 17000
|
4778 |
-
},
|
4779 |
-
{
|
4780 |
-
"epoch": 1.3474475662841314,
|
4781 |
-
"grad_norm": 153.75225830078125,
|
4782 |
-
"learning_rate": 1.9385890857286074e-07,
|
4783 |
-
"loss": 3.8691,
|
4784 |
-
"step": 17025
|
4785 |
-
},
|
4786 |
-
{
|
4787 |
-
"epoch": 1.3494261970716264,
|
4788 |
-
"grad_norm": 144.79983520507812,
|
4789 |
-
"learning_rate": 1.9377646453587303e-07,
|
4790 |
-
"loss": 3.7902,
|
4791 |
-
"step": 17050
|
4792 |
-
},
|
4793 |
-
{
|
4794 |
-
"epoch": 1.3514048278591215,
|
4795 |
-
"grad_norm": 107.72118377685547,
|
4796 |
-
"learning_rate": 1.9369402049888534e-07,
|
4797 |
-
"loss": 3.7424,
|
4798 |
-
"step": 17075
|
4799 |
-
},
|
4800 |
-
{
|
4801 |
-
"epoch": 1.3533834586466165,
|
4802 |
-
"grad_norm": 107.33843994140625,
|
4803 |
-
"learning_rate": 1.9361157646189765e-07,
|
4804 |
-
"loss": 3.7423,
|
4805 |
-
"step": 17100
|
4806 |
-
},
|
4807 |
-
{
|
4808 |
-
"epoch": 1.3553620894341116,
|
4809 |
-
"grad_norm": 154.79566955566406,
|
4810 |
-
"learning_rate": 1.9352913242490996e-07,
|
4811 |
-
"loss": 3.5984,
|
4812 |
-
"step": 17125
|
4813 |
-
},
|
4814 |
-
{
|
4815 |
-
"epoch": 1.3573407202216066,
|
4816 |
-
"grad_norm": 118.29646301269531,
|
4817 |
-
"learning_rate": 1.9344668838792227e-07,
|
4818 |
-
"loss": 3.6184,
|
4819 |
-
"step": 17150
|
4820 |
-
},
|
4821 |
-
{
|
4822 |
-
"epoch": 1.3593193510091017,
|
4823 |
-
"grad_norm": 194.39981079101562,
|
4824 |
-
"learning_rate": 1.9336424435093456e-07,
|
4825 |
-
"loss": 3.7566,
|
4826 |
-
"step": 17175
|
4827 |
-
},
|
4828 |
-
{
|
4829 |
-
"epoch": 1.361297981796597,
|
4830 |
-
"grad_norm": 126.4017333984375,
|
4831 |
-
"learning_rate": 1.932818003139469e-07,
|
4832 |
-
"loss": 3.7378,
|
4833 |
-
"step": 17200
|
4834 |
-
},
|
4835 |
-
{
|
4836 |
-
"epoch": 1.3632766125840918,
|
4837 |
-
"grad_norm": 154.46499633789062,
|
4838 |
-
"learning_rate": 1.9319935627695918e-07,
|
4839 |
-
"loss": 3.575,
|
4840 |
-
"step": 17225
|
4841 |
-
},
|
4842 |
-
{
|
4843 |
-
"epoch": 1.3652552433715868,
|
4844 |
-
"grad_norm": 121.02112579345703,
|
4845 |
-
"learning_rate": 1.9311691223997152e-07,
|
4846 |
-
"loss": 3.5395,
|
4847 |
-
"step": 17250
|
4848 |
-
},
|
4849 |
-
{
|
4850 |
-
"epoch": 1.367233874159082,
|
4851 |
-
"grad_norm": 112.81676483154297,
|
4852 |
-
"learning_rate": 1.930344682029838e-07,
|
4853 |
-
"loss": 3.7668,
|
4854 |
-
"step": 17275
|
4855 |
-
},
|
4856 |
-
{
|
4857 |
-
"epoch": 1.369212504946577,
|
4858 |
-
"grad_norm": 128.57958984375,
|
4859 |
-
"learning_rate": 1.9295202416599612e-07,
|
4860 |
-
"loss": 3.6508,
|
4861 |
-
"step": 17300
|
4862 |
-
},
|
4863 |
-
{
|
4864 |
-
"epoch": 1.371191135734072,
|
4865 |
-
"grad_norm": 100.09276580810547,
|
4866 |
-
"learning_rate": 1.9286958012900843e-07,
|
4867 |
-
"loss": 3.568,
|
4868 |
-
"step": 17325
|
4869 |
-
},
|
4870 |
-
{
|
4871 |
-
"epoch": 1.373169766521567,
|
4872 |
-
"grad_norm": 102.0973892211914,
|
4873 |
-
"learning_rate": 1.9278713609202074e-07,
|
4874 |
-
"loss": 3.6707,
|
4875 |
-
"step": 17350
|
4876 |
-
},
|
4877 |
-
{
|
4878 |
-
"epoch": 1.3751483973090621,
|
4879 |
-
"grad_norm": 180.7505645751953,
|
4880 |
-
"learning_rate": 1.9270469205503303e-07,
|
4881 |
-
"loss": 3.6053,
|
4882 |
-
"step": 17375
|
4883 |
-
},
|
4884 |
-
{
|
4885 |
-
"epoch": 1.3771270280965573,
|
4886 |
-
"grad_norm": 141.00286865234375,
|
4887 |
-
"learning_rate": 1.9262224801804534e-07,
|
4888 |
-
"loss": 3.802,
|
4889 |
-
"step": 17400
|
4890 |
-
},
|
4891 |
-
{
|
4892 |
-
"epoch": 1.3791056588840522,
|
4893 |
-
"grad_norm": 105.14612579345703,
|
4894 |
-
"learning_rate": 1.9253980398105765e-07,
|
4895 |
-
"loss": 3.8125,
|
4896 |
-
"step": 17425
|
4897 |
-
},
|
4898 |
-
{
|
4899 |
-
"epoch": 1.3810842896715472,
|
4900 |
-
"grad_norm": 134.7199249267578,
|
4901 |
-
"learning_rate": 1.9245735994406993e-07,
|
4902 |
-
"loss": 3.6837,
|
4903 |
-
"step": 17450
|
4904 |
-
},
|
4905 |
-
{
|
4906 |
-
"epoch": 1.3830629204590423,
|
4907 |
-
"grad_norm": 161.79367065429688,
|
4908 |
-
"learning_rate": 1.9237491590708227e-07,
|
4909 |
-
"loss": 3.4173,
|
4910 |
-
"step": 17475
|
4911 |
-
},
|
4912 |
-
{
|
4913 |
-
"epoch": 1.3850415512465375,
|
4914 |
-
"grad_norm": 90.2686996459961,
|
4915 |
-
"learning_rate": 1.9229247187009456e-07,
|
4916 |
-
"loss": 3.7035,
|
4917 |
-
"step": 17500
|
4918 |
-
},
|
4919 |
-
{
|
4920 |
-
"epoch": 1.3870201820340324,
|
4921 |
-
"grad_norm": 138.9261016845703,
|
4922 |
-
"learning_rate": 1.922100278331069e-07,
|
4923 |
-
"loss": 3.7375,
|
4924 |
-
"step": 17525
|
4925 |
-
},
|
4926 |
-
{
|
4927 |
-
"epoch": 1.3889988128215274,
|
4928 |
-
"grad_norm": 99.18257141113281,
|
4929 |
-
"learning_rate": 1.9212758379611918e-07,
|
4930 |
-
"loss": 3.7228,
|
4931 |
-
"step": 17550
|
4932 |
-
},
|
4933 |
-
{
|
4934 |
-
"epoch": 1.3909774436090225,
|
4935 |
-
"grad_norm": 108.61868286132812,
|
4936 |
-
"learning_rate": 1.920451397591315e-07,
|
4937 |
-
"loss": 3.8505,
|
4938 |
-
"step": 17575
|
4939 |
-
},
|
4940 |
-
{
|
4941 |
-
"epoch": 1.3929560743965177,
|
4942 |
-
"grad_norm": 89.02510070800781,
|
4943 |
-
"learning_rate": 1.919626957221438e-07,
|
4944 |
-
"loss": 3.7366,
|
4945 |
-
"step": 17600
|
4946 |
-
},
|
4947 |
-
{
|
4948 |
-
"epoch": 1.3949347051840126,
|
4949 |
-
"grad_norm": 183.8499755859375,
|
4950 |
-
"learning_rate": 1.9188025168515612e-07,
|
4951 |
-
"loss": 3.7842,
|
4952 |
-
"step": 17625
|
4953 |
-
},
|
4954 |
-
{
|
4955 |
-
"epoch": 1.3969133359715078,
|
4956 |
-
"grad_norm": 111.4132080078125,
|
4957 |
-
"learning_rate": 1.917978076481684e-07,
|
4958 |
-
"loss": 4.0444,
|
4959 |
-
"step": 17650
|
4960 |
-
},
|
4961 |
-
{
|
4962 |
-
"epoch": 1.3988919667590027,
|
4963 |
-
"grad_norm": 114.52892303466797,
|
4964 |
-
"learning_rate": 1.917153636111807e-07,
|
4965 |
-
"loss": 3.6568,
|
4966 |
-
"step": 17675
|
4967 |
-
},
|
4968 |
-
{
|
4969 |
-
"epoch": 1.400870597546498,
|
4970 |
-
"grad_norm": 108.65158081054688,
|
4971 |
-
"learning_rate": 1.9163291957419302e-07,
|
4972 |
-
"loss": 3.7598,
|
4973 |
-
"step": 17700
|
4974 |
-
},
|
4975 |
-
{
|
4976 |
-
"epoch": 1.4028492283339928,
|
4977 |
-
"grad_norm": 118.24126434326172,
|
4978 |
-
"learning_rate": 1.9155047553720534e-07,
|
4979 |
-
"loss": 3.5768,
|
4980 |
-
"step": 17725
|
4981 |
-
},
|
4982 |
-
{
|
4983 |
-
"epoch": 1.404827859121488,
|
4984 |
-
"grad_norm": 116.43769073486328,
|
4985 |
-
"learning_rate": 1.9146803150021765e-07,
|
4986 |
-
"loss": 3.6195,
|
4987 |
-
"step": 17750
|
4988 |
-
},
|
4989 |
-
{
|
4990 |
-
"epoch": 1.406806489908983,
|
4991 |
-
"grad_norm": 96.20083618164062,
|
4992 |
-
"learning_rate": 1.9138558746322993e-07,
|
4993 |
-
"loss": 3.7732,
|
4994 |
-
"step": 17775
|
4995 |
-
},
|
4996 |
-
{
|
4997 |
-
"epoch": 1.408785120696478,
|
4998 |
-
"grad_norm": 106.93498229980469,
|
4999 |
-
"learning_rate": 1.9130314342624227e-07,
|
5000 |
-
"loss": 3.5733,
|
5001 |
-
"step": 17800
|
5002 |
-
},
|
5003 |
-
{
|
5004 |
-
"epoch": 1.410763751483973,
|
5005 |
-
"grad_norm": 126.78841400146484,
|
5006 |
-
"learning_rate": 1.9122069938925456e-07,
|
5007 |
-
"loss": 3.4865,
|
5008 |
-
"step": 17825
|
5009 |
-
},
|
5010 |
-
{
|
5011 |
-
"epoch": 1.4127423822714682,
|
5012 |
-
"grad_norm": 161.895263671875,
|
5013 |
-
"learning_rate": 1.911382553522669e-07,
|
5014 |
-
"loss": 3.7985,
|
5015 |
-
"step": 17850
|
5016 |
-
},
|
5017 |
-
{
|
5018 |
-
"epoch": 1.4147210130589631,
|
5019 |
-
"grad_norm": 137.6743927001953,
|
5020 |
-
"learning_rate": 1.9105581131527918e-07,
|
5021 |
-
"loss": 3.5315,
|
5022 |
-
"step": 17875
|
5023 |
-
},
|
5024 |
-
{
|
5025 |
-
"epoch": 1.4166996438464583,
|
5026 |
-
"grad_norm": 81.24407196044922,
|
5027 |
-
"learning_rate": 1.909733672782915e-07,
|
5028 |
-
"loss": 3.6007,
|
5029 |
-
"step": 17900
|
5030 |
-
},
|
5031 |
-
{
|
5032 |
-
"epoch": 1.4186782746339532,
|
5033 |
-
"grad_norm": 109.42854309082031,
|
5034 |
-
"learning_rate": 1.908909232413038e-07,
|
5035 |
-
"loss": 3.5264,
|
5036 |
-
"step": 17925
|
5037 |
-
},
|
5038 |
-
{
|
5039 |
-
"epoch": 1.4206569054214484,
|
5040 |
-
"grad_norm": 113.56609344482422,
|
5041 |
-
"learning_rate": 1.9080847920431611e-07,
|
5042 |
-
"loss": 3.7304,
|
5043 |
-
"step": 17950
|
5044 |
-
},
|
5045 |
-
{
|
5046 |
-
"epoch": 1.4226355362089433,
|
5047 |
-
"grad_norm": 148.14793395996094,
|
5048 |
-
"learning_rate": 1.907260351673284e-07,
|
5049 |
-
"loss": 3.6923,
|
5050 |
-
"step": 17975
|
5051 |
-
},
|
5052 |
-
{
|
5053 |
-
"epoch": 1.4246141669964385,
|
5054 |
-
"grad_norm": 141.70896911621094,
|
5055 |
-
"learning_rate": 1.906435911303407e-07,
|
5056 |
-
"loss": 3.813,
|
5057 |
-
"step": 18000
|
5058 |
-
},
|
5059 |
-
{
|
5060 |
-
"epoch": 1.4265927977839334,
|
5061 |
-
"grad_norm": 107.8005142211914,
|
5062 |
-
"learning_rate": 1.9056114709335302e-07,
|
5063 |
-
"loss": 3.8025,
|
5064 |
-
"step": 18025
|
5065 |
-
},
|
5066 |
-
{
|
5067 |
-
"epoch": 1.4285714285714286,
|
5068 |
-
"grad_norm": 119.71678924560547,
|
5069 |
-
"learning_rate": 1.904787030563653e-07,
|
5070 |
-
"loss": 3.6632,
|
5071 |
-
"step": 18050
|
5072 |
-
},
|
5073 |
-
{
|
5074 |
-
"epoch": 1.4305500593589235,
|
5075 |
-
"grad_norm": 89.46141815185547,
|
5076 |
-
"learning_rate": 1.9039625901937765e-07,
|
5077 |
-
"loss": 3.5954,
|
5078 |
-
"step": 18075
|
5079 |
-
},
|
5080 |
-
{
|
5081 |
-
"epoch": 1.4325286901464187,
|
5082 |
-
"grad_norm": 94.6341552734375,
|
5083 |
-
"learning_rate": 1.9031381498238993e-07,
|
5084 |
-
"loss": 3.7922,
|
5085 |
-
"step": 18100
|
5086 |
-
},
|
5087 |
-
{
|
5088 |
-
"epoch": 1.4345073209339136,
|
5089 |
-
"grad_norm": 119.52442932128906,
|
5090 |
-
"learning_rate": 1.9023137094540227e-07,
|
5091 |
-
"loss": 3.8677,
|
5092 |
-
"step": 18125
|
5093 |
-
},
|
5094 |
-
{
|
5095 |
-
"epoch": 1.4364859517214088,
|
5096 |
-
"grad_norm": 85.59278869628906,
|
5097 |
-
"learning_rate": 1.9014892690841455e-07,
|
5098 |
-
"loss": 3.7679,
|
5099 |
-
"step": 18150
|
5100 |
-
},
|
5101 |
-
{
|
5102 |
-
"epoch": 1.438464582508904,
|
5103 |
-
"grad_norm": 140.89190673828125,
|
5104 |
-
"learning_rate": 1.9006648287142687e-07,
|
5105 |
-
"loss": 3.7617,
|
5106 |
-
"step": 18175
|
5107 |
-
},
|
5108 |
-
{
|
5109 |
-
"epoch": 1.440443213296399,
|
5110 |
-
"grad_norm": 114.29212188720703,
|
5111 |
-
"learning_rate": 1.8998403883443918e-07,
|
5112 |
-
"loss": 3.9376,
|
5113 |
-
"step": 18200
|
5114 |
-
},
|
5115 |
-
{
|
5116 |
-
"epoch": 1.4424218440838938,
|
5117 |
-
"grad_norm": 118.00497436523438,
|
5118 |
-
"learning_rate": 1.899015947974515e-07,
|
5119 |
-
"loss": 3.7841,
|
5120 |
-
"step": 18225
|
5121 |
-
},
|
5122 |
-
{
|
5123 |
-
"epoch": 1.444400474871389,
|
5124 |
-
"grad_norm": 140.94200134277344,
|
5125 |
-
"learning_rate": 1.8981915076046377e-07,
|
5126 |
-
"loss": 3.694,
|
5127 |
-
"step": 18250
|
5128 |
-
},
|
5129 |
-
{
|
5130 |
-
"epoch": 1.4463791056588842,
|
5131 |
-
"grad_norm": 153.66397094726562,
|
5132 |
-
"learning_rate": 1.897367067234761e-07,
|
5133 |
-
"loss": 3.7603,
|
5134 |
-
"step": 18275
|
5135 |
-
},
|
5136 |
-
{
|
5137 |
-
"epoch": 1.448357736446379,
|
5138 |
-
"grad_norm": 128.06732177734375,
|
5139 |
-
"learning_rate": 1.896542626864884e-07,
|
5140 |
-
"loss": 3.7799,
|
5141 |
-
"step": 18300
|
5142 |
-
},
|
5143 |
-
{
|
5144 |
-
"epoch": 1.450336367233874,
|
5145 |
-
"grad_norm": 112.23291778564453,
|
5146 |
-
"learning_rate": 1.895718186495007e-07,
|
5147 |
-
"loss": 3.7655,
|
5148 |
-
"step": 18325
|
5149 |
-
},
|
5150 |
-
{
|
5151 |
-
"epoch": 1.4523149980213692,
|
5152 |
-
"grad_norm": 91.07350158691406,
|
5153 |
-
"learning_rate": 1.8948937461251302e-07,
|
5154 |
-
"loss": 3.643,
|
5155 |
-
"step": 18350
|
5156 |
-
},
|
5157 |
-
{
|
5158 |
-
"epoch": 1.4542936288088644,
|
5159 |
-
"grad_norm": 98.65179443359375,
|
5160 |
-
"learning_rate": 1.894069305755253e-07,
|
5161 |
-
"loss": 3.7346,
|
5162 |
-
"step": 18375
|
5163 |
-
},
|
5164 |
-
{
|
5165 |
-
"epoch": 1.4562722595963593,
|
5166 |
-
"grad_norm": 118.8671875,
|
5167 |
-
"learning_rate": 1.8932448653853764e-07,
|
5168 |
-
"loss": 3.7361,
|
5169 |
-
"step": 18400
|
5170 |
-
},
|
5171 |
-
{
|
5172 |
-
"epoch": 1.4582508903838542,
|
5173 |
-
"grad_norm": 121.17200469970703,
|
5174 |
-
"learning_rate": 1.8924204250154993e-07,
|
5175 |
-
"loss": 3.7823,
|
5176 |
-
"step": 18425
|
5177 |
-
},
|
5178 |
-
{
|
5179 |
-
"epoch": 1.4602295211713494,
|
5180 |
-
"grad_norm": 91.53225708007812,
|
5181 |
-
"learning_rate": 1.8915959846456227e-07,
|
5182 |
-
"loss": 3.659,
|
5183 |
-
"step": 18450
|
5184 |
-
},
|
5185 |
-
{
|
5186 |
-
"epoch": 1.4622081519588446,
|
5187 |
-
"grad_norm": 140.78236389160156,
|
5188 |
-
"learning_rate": 1.8907715442757455e-07,
|
5189 |
-
"loss": 3.8495,
|
5190 |
-
"step": 18475
|
5191 |
-
},
|
5192 |
-
{
|
5193 |
-
"epoch": 1.4641867827463395,
|
5194 |
-
"grad_norm": 136.43206787109375,
|
5195 |
-
"learning_rate": 1.8899471039058686e-07,
|
5196 |
-
"loss": 3.8468,
|
5197 |
-
"step": 18500
|
5198 |
-
},
|
5199 |
-
{
|
5200 |
-
"epoch": 1.4661654135338344,
|
5201 |
-
"grad_norm": 106.83531951904297,
|
5202 |
-
"learning_rate": 1.8891226635359917e-07,
|
5203 |
-
"loss": 3.6799,
|
5204 |
-
"step": 18525
|
5205 |
-
},
|
5206 |
-
{
|
5207 |
-
"epoch": 1.4681440443213296,
|
5208 |
-
"grad_norm": 132.8029327392578,
|
5209 |
-
"learning_rate": 1.8882982231661149e-07,
|
5210 |
-
"loss": 3.7441,
|
5211 |
-
"step": 18550
|
5212 |
-
},
|
5213 |
-
{
|
5214 |
-
"epoch": 1.4701226751088248,
|
5215 |
-
"grad_norm": 125.74601745605469,
|
5216 |
-
"learning_rate": 1.8874737827962377e-07,
|
5217 |
-
"loss": 3.586,
|
5218 |
-
"step": 18575
|
5219 |
-
},
|
5220 |
-
{
|
5221 |
-
"epoch": 1.4721013058963197,
|
5222 |
-
"grad_norm": 126.61151123046875,
|
5223 |
-
"learning_rate": 1.886649342426361e-07,
|
5224 |
-
"loss": 3.7686,
|
5225 |
-
"step": 18600
|
5226 |
-
},
|
5227 |
-
{
|
5228 |
-
"epoch": 1.4740799366838149,
|
5229 |
-
"grad_norm": 127.62877655029297,
|
5230 |
-
"learning_rate": 1.885824902056484e-07,
|
5231 |
-
"loss": 3.74,
|
5232 |
-
"step": 18625
|
5233 |
-
},
|
5234 |
-
{
|
5235 |
-
"epoch": 1.4760585674713098,
|
5236 |
-
"grad_norm": 134.3148193359375,
|
5237 |
-
"learning_rate": 1.8850004616866068e-07,
|
5238 |
-
"loss": 3.8815,
|
5239 |
-
"step": 18650
|
5240 |
-
},
|
5241 |
-
{
|
5242 |
-
"epoch": 1.478037198258805,
|
5243 |
-
"grad_norm": 167.0948028564453,
|
5244 |
-
"learning_rate": 1.8841760213167302e-07,
|
5245 |
-
"loss": 3.7258,
|
5246 |
-
"step": 18675
|
5247 |
-
},
|
5248 |
-
{
|
5249 |
-
"epoch": 1.4800158290463,
|
5250 |
-
"grad_norm": 103.17122650146484,
|
5251 |
-
"learning_rate": 1.883351580946853e-07,
|
5252 |
-
"loss": 3.7211,
|
5253 |
-
"step": 18700
|
5254 |
-
},
|
5255 |
-
{
|
5256 |
-
"epoch": 1.481994459833795,
|
5257 |
-
"grad_norm": 98.9283218383789,
|
5258 |
-
"learning_rate": 1.8825271405769764e-07,
|
5259 |
-
"loss": 3.6401,
|
5260 |
-
"step": 18725
|
5261 |
-
},
|
5262 |
-
{
|
5263 |
-
"epoch": 1.48397309062129,
|
5264 |
-
"grad_norm": 110.53363800048828,
|
5265 |
-
"learning_rate": 1.8817027002070993e-07,
|
5266 |
-
"loss": 3.84,
|
5267 |
-
"step": 18750
|
5268 |
-
},
|
5269 |
-
{
|
5270 |
-
"epoch": 1.4859517214087852,
|
5271 |
-
"grad_norm": 115.60739135742188,
|
5272 |
-
"learning_rate": 1.8808782598372224e-07,
|
5273 |
-
"loss": 3.9043,
|
5274 |
-
"step": 18775
|
5275 |
-
},
|
5276 |
-
{
|
5277 |
-
"epoch": 1.4879303521962801,
|
5278 |
-
"grad_norm": 130.4183807373047,
|
5279 |
-
"learning_rate": 1.8800538194673455e-07,
|
5280 |
-
"loss": 3.6778,
|
5281 |
-
"step": 18800
|
5282 |
-
},
|
5283 |
-
{
|
5284 |
-
"epoch": 1.4899089829837753,
|
5285 |
-
"grad_norm": 94.30120086669922,
|
5286 |
-
"learning_rate": 1.8792293790974686e-07,
|
5287 |
-
"loss": 3.5853,
|
5288 |
-
"step": 18825
|
5289 |
-
},
|
5290 |
-
{
|
5291 |
-
"epoch": 1.4918876137712702,
|
5292 |
-
"grad_norm": 102.75646209716797,
|
5293 |
-
"learning_rate": 1.8784049387275917e-07,
|
5294 |
-
"loss": 4.0326,
|
5295 |
-
"step": 18850
|
5296 |
-
},
|
5297 |
-
{
|
5298 |
-
"epoch": 1.4938662445587654,
|
5299 |
-
"grad_norm": 153.1327362060547,
|
5300 |
-
"learning_rate": 1.8775804983577148e-07,
|
5301 |
-
"loss": 3.8785,
|
5302 |
-
"step": 18875
|
5303 |
-
},
|
5304 |
-
{
|
5305 |
-
"epoch": 1.4958448753462603,
|
5306 |
-
"grad_norm": 100.03012084960938,
|
5307 |
-
"learning_rate": 1.8767560579878377e-07,
|
5308 |
-
"loss": 3.692,
|
5309 |
-
"step": 18900
|
5310 |
-
},
|
5311 |
-
{
|
5312 |
-
"epoch": 1.4978235061337555,
|
5313 |
-
"grad_norm": 128.64657592773438,
|
5314 |
-
"learning_rate": 1.8759316176179608e-07,
|
5315 |
-
"loss": 3.7799,
|
5316 |
-
"step": 18925
|
5317 |
-
},
|
5318 |
-
{
|
5319 |
-
"epoch": 1.4998021369212504,
|
5320 |
-
"grad_norm": 99.55155944824219,
|
5321 |
-
"learning_rate": 1.875107177248084e-07,
|
5322 |
-
"loss": 3.7112,
|
5323 |
-
"step": 18950
|
5324 |
-
},
|
5325 |
-
{
|
5326 |
-
"epoch": 1.5017807677087456,
|
5327 |
-
"grad_norm": 171.94178771972656,
|
5328 |
-
"learning_rate": 1.8742827368782068e-07,
|
5329 |
-
"loss": 3.7663,
|
5330 |
-
"step": 18975
|
5331 |
-
},
|
5332 |
-
{
|
5333 |
-
"epoch": 1.5037593984962405,
|
5334 |
-
"grad_norm": 107.23114013671875,
|
5335 |
-
"learning_rate": 1.8734582965083302e-07,
|
5336 |
-
"loss": 3.6127,
|
5337 |
-
"step": 19000
|
5338 |
-
},
|
5339 |
-
{
|
5340 |
-
"epoch": 1.5057380292837357,
|
5341 |
-
"grad_norm": 101.01669311523438,
|
5342 |
-
"learning_rate": 1.872633856138453e-07,
|
5343 |
-
"loss": 3.5948,
|
5344 |
-
"step": 19025
|
5345 |
-
},
|
5346 |
-
{
|
5347 |
-
"epoch": 1.5077166600712308,
|
5348 |
-
"grad_norm": 158.38211059570312,
|
5349 |
-
"learning_rate": 1.8718094157685764e-07,
|
5350 |
-
"loss": 3.5552,
|
5351 |
-
"step": 19050
|
5352 |
-
},
|
5353 |
-
{
|
5354 |
-
"epoch": 1.5096952908587258,
|
5355 |
-
"grad_norm": 117.05270385742188,
|
5356 |
-
"learning_rate": 1.8709849753986992e-07,
|
5357 |
-
"loss": 3.7912,
|
5358 |
-
"step": 19075
|
5359 |
-
},
|
5360 |
-
{
|
5361 |
-
"epoch": 1.5116739216462207,
|
5362 |
-
"grad_norm": 117.87310028076172,
|
5363 |
-
"learning_rate": 1.8701605350288224e-07,
|
5364 |
-
"loss": 3.7512,
|
5365 |
-
"step": 19100
|
5366 |
-
},
|
5367 |
-
{
|
5368 |
-
"epoch": 1.5136525524337159,
|
5369 |
-
"grad_norm": 106.46678161621094,
|
5370 |
-
"learning_rate": 1.8693360946589455e-07,
|
5371 |
-
"loss": 3.7715,
|
5372 |
-
"step": 19125
|
5373 |
-
},
|
5374 |
-
{
|
5375 |
-
"epoch": 1.515631183221211,
|
5376 |
-
"grad_norm": 107.43925476074219,
|
5377 |
-
"learning_rate": 1.8685116542890686e-07,
|
5378 |
-
"loss": 3.7395,
|
5379 |
-
"step": 19150
|
5380 |
-
},
|
5381 |
-
{
|
5382 |
-
"epoch": 1.517609814008706,
|
5383 |
-
"grad_norm": 101.0313491821289,
|
5384 |
-
"learning_rate": 1.8676872139191914e-07,
|
5385 |
-
"loss": 3.6996,
|
5386 |
-
"step": 19175
|
5387 |
-
},
|
5388 |
-
{
|
5389 |
-
"epoch": 1.519588444796201,
|
5390 |
-
"grad_norm": 179.96051025390625,
|
5391 |
-
"learning_rate": 1.8668627735493148e-07,
|
5392 |
-
"loss": 3.6133,
|
5393 |
-
"step": 19200
|
5394 |
-
},
|
5395 |
-
{
|
5396 |
-
"epoch": 1.521567075583696,
|
5397 |
-
"grad_norm": 90.7691879272461,
|
5398 |
-
"learning_rate": 1.8660383331794377e-07,
|
5399 |
-
"loss": 3.6053,
|
5400 |
-
"step": 19225
|
5401 |
-
},
|
5402 |
-
{
|
5403 |
-
"epoch": 1.5235457063711912,
|
5404 |
-
"grad_norm": 137.12637329101562,
|
5405 |
-
"learning_rate": 1.8652138928095605e-07,
|
5406 |
-
"loss": 3.6739,
|
5407 |
-
"step": 19250
|
5408 |
-
},
|
5409 |
-
{
|
5410 |
-
"epoch": 1.5255243371586862,
|
5411 |
-
"grad_norm": 138.26014709472656,
|
5412 |
-
"learning_rate": 1.864389452439684e-07,
|
5413 |
-
"loss": 3.8328,
|
5414 |
-
"step": 19275
|
5415 |
-
},
|
5416 |
-
{
|
5417 |
-
"epoch": 1.5275029679461811,
|
5418 |
-
"grad_norm": 94.83419799804688,
|
5419 |
-
"learning_rate": 1.8635650120698068e-07,
|
5420 |
-
"loss": 3.7555,
|
5421 |
-
"step": 19300
|
5422 |
-
},
|
5423 |
-
{
|
5424 |
-
"epoch": 1.5294815987336763,
|
5425 |
-
"grad_norm": 105.08883666992188,
|
5426 |
-
"learning_rate": 1.8627405716999301e-07,
|
5427 |
-
"loss": 3.8474,
|
5428 |
-
"step": 19325
|
5429 |
-
},
|
5430 |
-
{
|
5431 |
-
"epoch": 1.5314602295211714,
|
5432 |
-
"grad_norm": 124.04432678222656,
|
5433 |
-
"learning_rate": 1.861916131330053e-07,
|
5434 |
-
"loss": 3.5361,
|
5435 |
-
"step": 19350
|
5436 |
-
},
|
5437 |
-
{
|
5438 |
-
"epoch": 1.5334388603086664,
|
5439 |
-
"grad_norm": 97.83289337158203,
|
5440 |
-
"learning_rate": 1.861091690960176e-07,
|
5441 |
-
"loss": 3.6283,
|
5442 |
-
"step": 19375
|
5443 |
-
},
|
5444 |
-
{
|
5445 |
-
"epoch": 1.5354174910961613,
|
5446 |
-
"grad_norm": 110.26960754394531,
|
5447 |
-
"learning_rate": 1.8602672505902992e-07,
|
5448 |
-
"loss": 3.7872,
|
5449 |
-
"step": 19400
|
5450 |
-
},
|
5451 |
-
{
|
5452 |
-
"epoch": 1.5373961218836565,
|
5453 |
-
"grad_norm": 115.99392700195312,
|
5454 |
-
"learning_rate": 1.8594428102204223e-07,
|
5455 |
-
"loss": 3.6191,
|
5456 |
-
"step": 19425
|
5457 |
-
},
|
5458 |
-
{
|
5459 |
-
"epoch": 1.5393747526711516,
|
5460 |
-
"grad_norm": 105.29808044433594,
|
5461 |
-
"learning_rate": 1.8586183698505454e-07,
|
5462 |
-
"loss": 3.6837,
|
5463 |
-
"step": 19450
|
5464 |
-
},
|
5465 |
-
{
|
5466 |
-
"epoch": 1.5413533834586466,
|
5467 |
-
"grad_norm": 100.55680847167969,
|
5468 |
-
"learning_rate": 1.8577939294806686e-07,
|
5469 |
-
"loss": 3.7439,
|
5470 |
-
"step": 19475
|
5471 |
-
},
|
5472 |
-
{
|
5473 |
-
"epoch": 1.5433320142461415,
|
5474 |
-
"grad_norm": 117.83526611328125,
|
5475 |
-
"learning_rate": 1.8569694891107914e-07,
|
5476 |
-
"loss": 3.5938,
|
5477 |
-
"step": 19500
|
5478 |
-
},
|
5479 |
-
{
|
5480 |
-
"epoch": 1.5453106450336367,
|
5481 |
-
"grad_norm": 84.95865631103516,
|
5482 |
-
"learning_rate": 1.8561450487409148e-07,
|
5483 |
-
"loss": 3.7425,
|
5484 |
-
"step": 19525
|
5485 |
-
},
|
5486 |
-
{
|
5487 |
-
"epoch": 1.5472892758211318,
|
5488 |
-
"grad_norm": 128.07754516601562,
|
5489 |
-
"learning_rate": 1.8553206083710376e-07,
|
5490 |
-
"loss": 3.7379,
|
5491 |
-
"step": 19550
|
5492 |
-
},
|
5493 |
-
{
|
5494 |
-
"epoch": 1.5492679066086268,
|
5495 |
-
"grad_norm": 103.81727600097656,
|
5496 |
-
"learning_rate": 1.8544961680011605e-07,
|
5497 |
-
"loss": 3.8121,
|
5498 |
-
"step": 19575
|
5499 |
-
},
|
5500 |
-
{
|
5501 |
-
"epoch": 1.5512465373961217,
|
5502 |
-
"grad_norm": 113.71387481689453,
|
5503 |
-
"learning_rate": 1.853671727631284e-07,
|
5504 |
-
"loss": 3.5854,
|
5505 |
-
"step": 19600
|
5506 |
-
},
|
5507 |
-
{
|
5508 |
-
"epoch": 1.5532251681836169,
|
5509 |
-
"grad_norm": 106.94216918945312,
|
5510 |
-
"learning_rate": 1.8528472872614067e-07,
|
5511 |
-
"loss": 3.7723,
|
5512 |
-
"step": 19625
|
5513 |
-
},
|
5514 |
-
{
|
5515 |
-
"epoch": 1.555203798971112,
|
5516 |
-
"grad_norm": 113.77094268798828,
|
5517 |
-
"learning_rate": 1.85202284689153e-07,
|
5518 |
-
"loss": 3.7218,
|
5519 |
-
"step": 19650
|
5520 |
-
},
|
5521 |
-
{
|
5522 |
-
"epoch": 1.557182429758607,
|
5523 |
-
"grad_norm": 102.69156646728516,
|
5524 |
-
"learning_rate": 1.851198406521653e-07,
|
5525 |
-
"loss": 3.5478,
|
5526 |
-
"step": 19675
|
5527 |
-
},
|
5528 |
-
{
|
5529 |
-
"epoch": 1.5591610605461022,
|
5530 |
-
"grad_norm": 146.14686584472656,
|
5531 |
-
"learning_rate": 1.850373966151776e-07,
|
5532 |
-
"loss": 3.5888,
|
5533 |
-
"step": 19700
|
5534 |
-
},
|
5535 |
-
{
|
5536 |
-
"epoch": 1.5611396913335973,
|
5537 |
-
"grad_norm": 117.55293273925781,
|
5538 |
-
"learning_rate": 1.8495495257818992e-07,
|
5539 |
-
"loss": 3.5802,
|
5540 |
-
"step": 19725
|
5541 |
-
},
|
5542 |
-
{
|
5543 |
-
"epoch": 1.5631183221210923,
|
5544 |
-
"grad_norm": 116.557861328125,
|
5545 |
-
"learning_rate": 1.8487250854120223e-07,
|
5546 |
-
"loss": 3.6588,
|
5547 |
-
"step": 19750
|
5548 |
-
},
|
5549 |
-
{
|
5550 |
-
"epoch": 1.5650969529085872,
|
5551 |
-
"grad_norm": 143.797607421875,
|
5552 |
-
"learning_rate": 1.8479006450421452e-07,
|
5553 |
-
"loss": 3.7888,
|
5554 |
-
"step": 19775
|
5555 |
-
},
|
5556 |
-
{
|
5557 |
-
"epoch": 1.5670755836960824,
|
5558 |
-
"grad_norm": 90.89989471435547,
|
5559 |
-
"learning_rate": 1.8470762046722685e-07,
|
5560 |
-
"loss": 3.7565,
|
5561 |
-
"step": 19800
|
5562 |
-
},
|
5563 |
-
{
|
5564 |
-
"epoch": 1.5690542144835775,
|
5565 |
-
"grad_norm": 93.96920776367188,
|
5566 |
-
"learning_rate": 1.8462517643023914e-07,
|
5567 |
-
"loss": 3.7576,
|
5568 |
-
"step": 19825
|
5569 |
-
},
|
5570 |
-
{
|
5571 |
-
"epoch": 1.5710328452710725,
|
5572 |
-
"grad_norm": 146.9014129638672,
|
5573 |
-
"learning_rate": 1.8454273239325148e-07,
|
5574 |
-
"loss": 3.7507,
|
5575 |
-
"step": 19850
|
5576 |
-
},
|
5577 |
-
{
|
5578 |
-
"epoch": 1.5730114760585674,
|
5579 |
-
"grad_norm": 119.38358306884766,
|
5580 |
-
"learning_rate": 1.8446028835626376e-07,
|
5581 |
-
"loss": 3.6588,
|
5582 |
-
"step": 19875
|
5583 |
-
},
|
5584 |
-
{
|
5585 |
-
"epoch": 1.5749901068460626,
|
5586 |
-
"grad_norm": 113.38998413085938,
|
5587 |
-
"learning_rate": 1.8437784431927605e-07,
|
5588 |
-
"loss": 3.7866,
|
5589 |
-
"step": 19900
|
5590 |
-
},
|
5591 |
-
{
|
5592 |
-
"epoch": 1.5769687376335577,
|
5593 |
-
"grad_norm": 103.051513671875,
|
5594 |
-
"learning_rate": 1.8429540028228839e-07,
|
5595 |
-
"loss": 3.6443,
|
5596 |
-
"step": 19925
|
5597 |
-
},
|
5598 |
-
{
|
5599 |
-
"epoch": 1.5789473684210527,
|
5600 |
-
"grad_norm": 94.77304077148438,
|
5601 |
-
"learning_rate": 1.8421295624530067e-07,
|
5602 |
-
"loss": 3.6966,
|
5603 |
-
"step": 19950
|
5604 |
-
},
|
5605 |
-
{
|
5606 |
-
"epoch": 1.5809259992085476,
|
5607 |
-
"grad_norm": 167.68310546875,
|
5608 |
-
"learning_rate": 1.8413051220831298e-07,
|
5609 |
-
"loss": 3.684,
|
5610 |
-
"step": 19975
|
5611 |
-
},
|
5612 |
-
{
|
5613 |
-
"epoch": 1.5829046299960428,
|
5614 |
-
"grad_norm": 96.78482055664062,
|
5615 |
-
"learning_rate": 1.840480681713253e-07,
|
5616 |
-
"loss": 3.733,
|
5617 |
-
"step": 20000
|
5618 |
-
},
|
5619 |
-
{
|
5620 |
-
"epoch": 1.5829046299960428,
|
5621 |
-
"eval_loss": 3.6955573558807373,
|
5622 |
-
"eval_runtime": 9.5434,
|
5623 |
-
"eval_samples_per_second": 264.894,
|
5624 |
-
"eval_steps_per_second": 33.112,
|
5625 |
-
"step": 20000
|
5626 |
}
|
5627 |
],
|
5628 |
"logging_steps": 25,
|
@@ -5642,7 +2834,7 @@
|
|
5642 |
"attributes": {}
|
5643 |
}
|
5644 |
},
|
5645 |
-
"total_flos":
|
5646 |
"train_batch_size": 2,
|
5647 |
"trial_name": null,
|
5648 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 3.8681728839874268,
|
3 |
+
"best_model_checkpoint": "checkpoints/test_1M_1-2025-02-12-12-32/checkpoint-10000",
|
4 |
+
"epoch": 0.7914523149980214,
|
5 |
"eval_steps": 10000,
|
6 |
+
"global_step": 10000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
2815 |
"eval_samples_per_second": 264.165,
|
2816 |
"eval_steps_per_second": 33.021,
|
2817 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2818 |
}
|
2819 |
],
|
2820 |
"logging_steps": 25,
|
|
|
2834 |
"attributes": {}
|
2835 |
}
|
2836 |
},
|
2837 |
+
"total_flos": 35767296000000.0,
|
2838 |
"train_batch_size": 2,
|
2839 |
"trial_name": null,
|
2840 |
"trial_params": null
|