Training checkpoint at step 3000
Browse files- trainer_state.json +366 -6
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
-
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_7x1024_mem32_bs64_hf_armt_dmem64/run_30/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -728,6 +728,366 @@
|
|
| 728 |
"eval_samples_per_second": 3.516,
|
| 729 |
"eval_steps_per_second": 1.773,
|
| 730 |
"step": 2000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 731 |
}
|
| 732 |
],
|
| 733 |
"logging_steps": 25,
|
|
@@ -747,7 +1107,7 @@
|
|
| 747 |
"attributes": {}
|
| 748 |
}
|
| 749 |
},
|
| 750 |
-
"total_flos":
|
| 751 |
"train_batch_size": 1,
|
| 752 |
"trial_name": null,
|
| 753 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 3000,
|
| 3 |
+
"best_metric": 2.436969041824341,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_7x1024_mem32_bs64_hf_armt_dmem64/run_30/checkpoint-3000",
|
| 5 |
+
"epoch": 0.06,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 3000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 728 |
"eval_samples_per_second": 3.516,
|
| 729 |
"eval_steps_per_second": 1.773,
|
| 730 |
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.0405,
|
| 734 |
+
"grad_norm": 0.034945541932647324,
|
| 735 |
+
"learning_rate": 4.048e-06,
|
| 736 |
+
"loss": 2.4357,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.041,
|
| 741 |
+
"grad_norm": 0.029322959861707003,
|
| 742 |
+
"learning_rate": 4.098e-06,
|
| 743 |
+
"loss": 2.4373,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.0415,
|
| 748 |
+
"grad_norm": 0.027365033479394632,
|
| 749 |
+
"learning_rate": 4.148000000000001e-06,
|
| 750 |
+
"loss": 2.442,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.042,
|
| 755 |
+
"grad_norm": 0.042214130565513416,
|
| 756 |
+
"learning_rate": 4.198e-06,
|
| 757 |
+
"loss": 2.4362,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.042,
|
| 762 |
+
"eval_loss": 2.448322296142578,
|
| 763 |
+
"eval_runtime": 33.466,
|
| 764 |
+
"eval_samples_per_second": 3.496,
|
| 765 |
+
"eval_steps_per_second": 1.763,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.0425,
|
| 770 |
+
"grad_norm": 0.028874346576168566,
|
| 771 |
+
"learning_rate": 4.248000000000001e-06,
|
| 772 |
+
"loss": 2.4428,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.043,
|
| 777 |
+
"grad_norm": 0.029771861998040296,
|
| 778 |
+
"learning_rate": 4.298e-06,
|
| 779 |
+
"loss": 2.4298,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.0435,
|
| 784 |
+
"grad_norm": 0.029668415484575914,
|
| 785 |
+
"learning_rate": 4.3480000000000006e-06,
|
| 786 |
+
"loss": 2.4352,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.044,
|
| 791 |
+
"grad_norm": 0.02564927582570633,
|
| 792 |
+
"learning_rate": 4.398000000000001e-06,
|
| 793 |
+
"loss": 2.4349,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.044,
|
| 798 |
+
"eval_loss": 2.4465889930725098,
|
| 799 |
+
"eval_runtime": 33.3555,
|
| 800 |
+
"eval_samples_per_second": 3.508,
|
| 801 |
+
"eval_steps_per_second": 1.769,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.0445,
|
| 806 |
+
"grad_norm": 0.024797235968250814,
|
| 807 |
+
"learning_rate": 4.4480000000000004e-06,
|
| 808 |
+
"loss": 2.4409,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.045,
|
| 813 |
+
"grad_norm": 0.02813189377877088,
|
| 814 |
+
"learning_rate": 4.498e-06,
|
| 815 |
+
"loss": 2.4367,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 0.0455,
|
| 820 |
+
"grad_norm": 0.02750903211389184,
|
| 821 |
+
"learning_rate": 4.548e-06,
|
| 822 |
+
"loss": 2.4326,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.046,
|
| 827 |
+
"grad_norm": 0.027737559952553607,
|
| 828 |
+
"learning_rate": 4.598e-06,
|
| 829 |
+
"loss": 2.4375,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.046,
|
| 834 |
+
"eval_loss": 2.4448626041412354,
|
| 835 |
+
"eval_runtime": 33.2658,
|
| 836 |
+
"eval_samples_per_second": 3.517,
|
| 837 |
+
"eval_steps_per_second": 1.774,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.0465,
|
| 842 |
+
"grad_norm": 0.02630663299301831,
|
| 843 |
+
"learning_rate": 4.648e-06,
|
| 844 |
+
"loss": 2.4392,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.047,
|
| 849 |
+
"grad_norm": 0.027929449055597393,
|
| 850 |
+
"learning_rate": 4.698000000000001e-06,
|
| 851 |
+
"loss": 2.4256,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.0475,
|
| 856 |
+
"grad_norm": 0.0283193243102273,
|
| 857 |
+
"learning_rate": 4.748e-06,
|
| 858 |
+
"loss": 2.429,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.048,
|
| 863 |
+
"grad_norm": 0.029295313451333963,
|
| 864 |
+
"learning_rate": 4.7980000000000005e-06,
|
| 865 |
+
"loss": 2.4393,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.048,
|
| 870 |
+
"eval_loss": 2.4432175159454346,
|
| 871 |
+
"eval_runtime": 33.3067,
|
| 872 |
+
"eval_samples_per_second": 3.513,
|
| 873 |
+
"eval_steps_per_second": 1.771,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.0485,
|
| 878 |
+
"grad_norm": 0.025382897552394503,
|
| 879 |
+
"learning_rate": 4.848000000000001e-06,
|
| 880 |
+
"loss": 2.4322,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.049,
|
| 885 |
+
"grad_norm": 0.02450548193909556,
|
| 886 |
+
"learning_rate": 4.898e-06,
|
| 887 |
+
"loss": 2.4314,
|
| 888 |
+
"step": 2450
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.0495,
|
| 892 |
+
"grad_norm": 0.033065483070063684,
|
| 893 |
+
"learning_rate": 4.948000000000001e-06,
|
| 894 |
+
"loss": 2.4338,
|
| 895 |
+
"step": 2475
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.05,
|
| 899 |
+
"grad_norm": 0.027543894857825314,
|
| 900 |
+
"learning_rate": 4.998e-06,
|
| 901 |
+
"loss": 2.4333,
|
| 902 |
+
"step": 2500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.05,
|
| 906 |
+
"eval_loss": 2.441807985305786,
|
| 907 |
+
"eval_runtime": 33.0379,
|
| 908 |
+
"eval_samples_per_second": 3.541,
|
| 909 |
+
"eval_steps_per_second": 1.786,
|
| 910 |
+
"step": 2500
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.0505,
|
| 914 |
+
"grad_norm": 0.027354239436717945,
|
| 915 |
+
"learning_rate": 5.048000000000001e-06,
|
| 916 |
+
"loss": 2.439,
|
| 917 |
+
"step": 2525
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.051,
|
| 921 |
+
"grad_norm": 0.022458884368301627,
|
| 922 |
+
"learning_rate": 5.098000000000001e-06,
|
| 923 |
+
"loss": 2.427,
|
| 924 |
+
"step": 2550
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.0515,
|
| 928 |
+
"grad_norm": 0.033350881745701555,
|
| 929 |
+
"learning_rate": 5.1480000000000005e-06,
|
| 930 |
+
"loss": 2.4275,
|
| 931 |
+
"step": 2575
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 0.052,
|
| 935 |
+
"grad_norm": 0.025032545530163004,
|
| 936 |
+
"learning_rate": 5.198000000000001e-06,
|
| 937 |
+
"loss": 2.4275,
|
| 938 |
+
"step": 2600
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.052,
|
| 942 |
+
"eval_loss": 2.440882444381714,
|
| 943 |
+
"eval_runtime": 33.1835,
|
| 944 |
+
"eval_samples_per_second": 3.526,
|
| 945 |
+
"eval_steps_per_second": 1.778,
|
| 946 |
+
"step": 2600
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.0525,
|
| 950 |
+
"grad_norm": 0.026294170044068685,
|
| 951 |
+
"learning_rate": 5.248000000000001e-06,
|
| 952 |
+
"loss": 2.4312,
|
| 953 |
+
"step": 2625
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.053,
|
| 957 |
+
"grad_norm": 0.03301155351988982,
|
| 958 |
+
"learning_rate": 5.298000000000001e-06,
|
| 959 |
+
"loss": 2.4203,
|
| 960 |
+
"step": 2650
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.0535,
|
| 964 |
+
"grad_norm": 0.02389586194961339,
|
| 965 |
+
"learning_rate": 5.348000000000001e-06,
|
| 966 |
+
"loss": 2.4332,
|
| 967 |
+
"step": 2675
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.054,
|
| 971 |
+
"grad_norm": 0.056862279743176244,
|
| 972 |
+
"learning_rate": 5.398e-06,
|
| 973 |
+
"loss": 2.4313,
|
| 974 |
+
"step": 2700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.054,
|
| 978 |
+
"eval_loss": 2.4402644634246826,
|
| 979 |
+
"eval_runtime": 33.2071,
|
| 980 |
+
"eval_samples_per_second": 3.523,
|
| 981 |
+
"eval_steps_per_second": 1.777,
|
| 982 |
+
"step": 2700
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.0545,
|
| 986 |
+
"grad_norm": 0.025636671246445756,
|
| 987 |
+
"learning_rate": 5.448e-06,
|
| 988 |
+
"loss": 2.4311,
|
| 989 |
+
"step": 2725
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.055,
|
| 993 |
+
"grad_norm": 0.022083605910153424,
|
| 994 |
+
"learning_rate": 5.498e-06,
|
| 995 |
+
"loss": 2.4357,
|
| 996 |
+
"step": 2750
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.0555,
|
| 1000 |
+
"grad_norm": 0.024223735712298522,
|
| 1001 |
+
"learning_rate": 5.548e-06,
|
| 1002 |
+
"loss": 2.4294,
|
| 1003 |
+
"step": 2775
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.056,
|
| 1007 |
+
"grad_norm": 0.029847698463432104,
|
| 1008 |
+
"learning_rate": 5.5980000000000004e-06,
|
| 1009 |
+
"loss": 2.4344,
|
| 1010 |
+
"step": 2800
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.056,
|
| 1014 |
+
"eval_loss": 2.4389007091522217,
|
| 1015 |
+
"eval_runtime": 33.2705,
|
| 1016 |
+
"eval_samples_per_second": 3.517,
|
| 1017 |
+
"eval_steps_per_second": 1.773,
|
| 1018 |
+
"step": 2800
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.0565,
|
| 1022 |
+
"grad_norm": 0.032144633236930065,
|
| 1023 |
+
"learning_rate": 5.648e-06,
|
| 1024 |
+
"loss": 2.4282,
|
| 1025 |
+
"step": 2825
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.057,
|
| 1029 |
+
"grad_norm": 0.02355863809037046,
|
| 1030 |
+
"learning_rate": 5.698e-06,
|
| 1031 |
+
"loss": 2.4322,
|
| 1032 |
+
"step": 2850
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.0575,
|
| 1036 |
+
"grad_norm": 0.023728744427970416,
|
| 1037 |
+
"learning_rate": 5.748e-06,
|
| 1038 |
+
"loss": 2.4286,
|
| 1039 |
+
"step": 2875
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.058,
|
| 1043 |
+
"grad_norm": 0.025539915034515293,
|
| 1044 |
+
"learning_rate": 5.798e-06,
|
| 1045 |
+
"loss": 2.4287,
|
| 1046 |
+
"step": 2900
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.058,
|
| 1050 |
+
"eval_loss": 2.4376914501190186,
|
| 1051 |
+
"eval_runtime": 33.3179,
|
| 1052 |
+
"eval_samples_per_second": 3.512,
|
| 1053 |
+
"eval_steps_per_second": 1.771,
|
| 1054 |
+
"step": 2900
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.0585,
|
| 1058 |
+
"grad_norm": 0.023457547558388747,
|
| 1059 |
+
"learning_rate": 5.848000000000001e-06,
|
| 1060 |
+
"loss": 2.4289,
|
| 1061 |
+
"step": 2925
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 0.059,
|
| 1065 |
+
"grad_norm": 0.025297710201421797,
|
| 1066 |
+
"learning_rate": 5.898e-06,
|
| 1067 |
+
"loss": 2.4274,
|
| 1068 |
+
"step": 2950
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.0595,
|
| 1072 |
+
"grad_norm": 0.024155176530161276,
|
| 1073 |
+
"learning_rate": 5.9480000000000005e-06,
|
| 1074 |
+
"loss": 2.4169,
|
| 1075 |
+
"step": 2975
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.06,
|
| 1079 |
+
"grad_norm": 0.023954841726960448,
|
| 1080 |
+
"learning_rate": 5.998000000000001e-06,
|
| 1081 |
+
"loss": 2.4244,
|
| 1082 |
+
"step": 3000
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.06,
|
| 1086 |
+
"eval_loss": 2.436969041824341,
|
| 1087 |
+
"eval_runtime": 33.2713,
|
| 1088 |
+
"eval_samples_per_second": 3.517,
|
| 1089 |
+
"eval_steps_per_second": 1.773,
|
| 1090 |
+
"step": 3000
|
| 1091 |
}
|
| 1092 |
],
|
| 1093 |
"logging_steps": 25,
|
|
|
|
| 1107 |
"attributes": {}
|
| 1108 |
}
|
| 1109 |
},
|
| 1110 |
+
"total_flos": 8.355905775410872e+18,
|
| 1111 |
"train_batch_size": 1,
|
| 1112 |
"trial_name": null,
|
| 1113 |
"trial_params": null
|