Commit
•
5d8ab3b
1
Parent(s):
cb30f37
Training in progress, step 19200
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +396 -3
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 745634697
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d041fc9e6d55c4a7915f0599d0972686813610cfd5a2d83bd76580f7087c5ca
|
| 3 |
size 745634697
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 372832803
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7a084a9b3d69be038d1f70310127204c769b6f31132335d3c43f2359a442b86
|
| 3 |
size 372832803
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15523
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac85050f1de5a3da93b15d68ec19d08f9c128973d47940d52332ce7a8a430098
|
| 3 |
size 15523
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12a6154fa53f0286557ec7a9b6bf6b9f5b2fb01f4345510fa7b96c5e44005857
|
| 3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -792,11 +792,404 @@
|
|
| 792 |
"eval_samples_per_second": 45.406,
|
| 793 |
"eval_steps_per_second": 2.838,
|
| 794 |
"step": 12800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 795 |
}
|
| 796 |
],
|
| 797 |
"max_steps": 64000,
|
| 798 |
"num_train_epochs": 9223372036854775807,
|
| 799 |
-
"total_flos":
|
| 800 |
"trial_name": null,
|
| 801 |
"trial_params": null
|
| 802 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.3,
|
| 5 |
+
"global_step": 19200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 792 |
"eval_samples_per_second": 45.406,
|
| 793 |
"eval_steps_per_second": 2.838,
|
| 794 |
"step": 12800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.2,
|
| 798 |
+
"learning_rate": 0.001,
|
| 799 |
+
"loss": 8.0474,
|
| 800 |
+
"step": 12900
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.2,
|
| 804 |
+
"learning_rate": 0.001,
|
| 805 |
+
"loss": 8.054,
|
| 806 |
+
"step": 13000
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.2,
|
| 810 |
+
"learning_rate": 0.001,
|
| 811 |
+
"loss": 8.0506,
|
| 812 |
+
"step": 13100
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.21,
|
| 816 |
+
"learning_rate": 0.001,
|
| 817 |
+
"loss": 8.0485,
|
| 818 |
+
"step": 13200
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.21,
|
| 822 |
+
"learning_rate": 0.001,
|
| 823 |
+
"loss": 8.0485,
|
| 824 |
+
"step": 13300
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.21,
|
| 828 |
+
"learning_rate": 0.001,
|
| 829 |
+
"loss": 8.0515,
|
| 830 |
+
"step": 13400
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.21,
|
| 834 |
+
"learning_rate": 0.001,
|
| 835 |
+
"loss": 8.0295,
|
| 836 |
+
"step": 13500
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.21,
|
| 840 |
+
"learning_rate": 0.001,
|
| 841 |
+
"loss": 8.0456,
|
| 842 |
+
"step": 13600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.21,
|
| 846 |
+
"learning_rate": 0.001,
|
| 847 |
+
"loss": 8.0278,
|
| 848 |
+
"step": 13700
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.22,
|
| 852 |
+
"learning_rate": 0.001,
|
| 853 |
+
"loss": 8.0358,
|
| 854 |
+
"step": 13800
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.22,
|
| 858 |
+
"learning_rate": 0.001,
|
| 859 |
+
"loss": 8.0513,
|
| 860 |
+
"step": 13900
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.22,
|
| 864 |
+
"learning_rate": 0.001,
|
| 865 |
+
"loss": 8.0443,
|
| 866 |
+
"step": 14000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.22,
|
| 870 |
+
"learning_rate": 0.001,
|
| 871 |
+
"loss": 8.0657,
|
| 872 |
+
"step": 14100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.22,
|
| 876 |
+
"learning_rate": 0.001,
|
| 877 |
+
"loss": 8.0381,
|
| 878 |
+
"step": 14200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.22,
|
| 882 |
+
"learning_rate": 0.001,
|
| 883 |
+
"loss": 8.047,
|
| 884 |
+
"step": 14300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.23,
|
| 888 |
+
"learning_rate": 0.001,
|
| 889 |
+
"loss": 8.0441,
|
| 890 |
+
"step": 14400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.23,
|
| 894 |
+
"learning_rate": 0.001,
|
| 895 |
+
"loss": 8.0293,
|
| 896 |
+
"step": 14500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.23,
|
| 900 |
+
"learning_rate": 0.001,
|
| 901 |
+
"loss": 8.0308,
|
| 902 |
+
"step": 14600
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.23,
|
| 906 |
+
"learning_rate": 0.001,
|
| 907 |
+
"loss": 8.0353,
|
| 908 |
+
"step": 14700
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.23,
|
| 912 |
+
"learning_rate": 0.001,
|
| 913 |
+
"loss": 8.0098,
|
| 914 |
+
"step": 14800
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.23,
|
| 918 |
+
"learning_rate": 0.001,
|
| 919 |
+
"loss": 8.0422,
|
| 920 |
+
"step": 14900
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.23,
|
| 924 |
+
"learning_rate": 0.001,
|
| 925 |
+
"loss": 8.0427,
|
| 926 |
+
"step": 15000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.24,
|
| 930 |
+
"learning_rate": 0.001,
|
| 931 |
+
"loss": 8.0323,
|
| 932 |
+
"step": 15100
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.24,
|
| 936 |
+
"learning_rate": 0.001,
|
| 937 |
+
"loss": 8.035,
|
| 938 |
+
"step": 15200
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.24,
|
| 942 |
+
"learning_rate": 0.001,
|
| 943 |
+
"loss": 8.0574,
|
| 944 |
+
"step": 15300
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.24,
|
| 948 |
+
"learning_rate": 0.001,
|
| 949 |
+
"loss": 8.0365,
|
| 950 |
+
"step": 15400
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.24,
|
| 954 |
+
"learning_rate": 0.001,
|
| 955 |
+
"loss": 8.0341,
|
| 956 |
+
"step": 15500
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.24,
|
| 960 |
+
"learning_rate": 0.001,
|
| 961 |
+
"loss": 8.0484,
|
| 962 |
+
"step": 15600
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.25,
|
| 966 |
+
"learning_rate": 0.001,
|
| 967 |
+
"loss": 8.0409,
|
| 968 |
+
"step": 15700
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.25,
|
| 972 |
+
"learning_rate": 0.001,
|
| 973 |
+
"loss": 8.0363,
|
| 974 |
+
"step": 15800
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.25,
|
| 978 |
+
"learning_rate": 0.001,
|
| 979 |
+
"loss": 8.0415,
|
| 980 |
+
"step": 15900
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.25,
|
| 984 |
+
"learning_rate": 0.001,
|
| 985 |
+
"loss": 8.043,
|
| 986 |
+
"step": 16000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.25,
|
| 990 |
+
"learning_rate": 0.001,
|
| 991 |
+
"loss": 8.0307,
|
| 992 |
+
"step": 16100
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.25,
|
| 996 |
+
"learning_rate": 0.001,
|
| 997 |
+
"loss": 8.0179,
|
| 998 |
+
"step": 16200
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.25,
|
| 1002 |
+
"learning_rate": 0.001,
|
| 1003 |
+
"loss": 8.0279,
|
| 1004 |
+
"step": 16300
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.26,
|
| 1008 |
+
"learning_rate": 0.001,
|
| 1009 |
+
"loss": 8.0552,
|
| 1010 |
+
"step": 16400
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.26,
|
| 1014 |
+
"learning_rate": 0.001,
|
| 1015 |
+
"loss": 8.0437,
|
| 1016 |
+
"step": 16500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.26,
|
| 1020 |
+
"learning_rate": 0.001,
|
| 1021 |
+
"loss": 8.0356,
|
| 1022 |
+
"step": 16600
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.26,
|
| 1026 |
+
"learning_rate": 0.001,
|
| 1027 |
+
"loss": 8.0427,
|
| 1028 |
+
"step": 16700
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.26,
|
| 1032 |
+
"learning_rate": 0.001,
|
| 1033 |
+
"loss": 8.0472,
|
| 1034 |
+
"step": 16800
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.26,
|
| 1038 |
+
"learning_rate": 0.001,
|
| 1039 |
+
"loss": 8.043,
|
| 1040 |
+
"step": 16900
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.27,
|
| 1044 |
+
"learning_rate": 0.001,
|
| 1045 |
+
"loss": 8.0567,
|
| 1046 |
+
"step": 17000
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.27,
|
| 1050 |
+
"learning_rate": 0.001,
|
| 1051 |
+
"loss": 8.0474,
|
| 1052 |
+
"step": 17100
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.27,
|
| 1056 |
+
"learning_rate": 0.001,
|
| 1057 |
+
"loss": 8.0295,
|
| 1058 |
+
"step": 17200
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.27,
|
| 1062 |
+
"learning_rate": 0.001,
|
| 1063 |
+
"loss": 8.0364,
|
| 1064 |
+
"step": 17300
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.27,
|
| 1068 |
+
"learning_rate": 0.001,
|
| 1069 |
+
"loss": 8.0474,
|
| 1070 |
+
"step": 17400
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.27,
|
| 1074 |
+
"learning_rate": 0.001,
|
| 1075 |
+
"loss": 8.0282,
|
| 1076 |
+
"step": 17500
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.28,
|
| 1080 |
+
"learning_rate": 0.001,
|
| 1081 |
+
"loss": 8.0525,
|
| 1082 |
+
"step": 17600
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.28,
|
| 1086 |
+
"learning_rate": 0.001,
|
| 1087 |
+
"loss": 8.0464,
|
| 1088 |
+
"step": 17700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.28,
|
| 1092 |
+
"learning_rate": 0.001,
|
| 1093 |
+
"loss": 8.0516,
|
| 1094 |
+
"step": 17800
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.28,
|
| 1098 |
+
"learning_rate": 0.001,
|
| 1099 |
+
"loss": 8.0322,
|
| 1100 |
+
"step": 17900
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.28,
|
| 1104 |
+
"learning_rate": 0.001,
|
| 1105 |
+
"loss": 8.0376,
|
| 1106 |
+
"step": 18000
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.28,
|
| 1110 |
+
"learning_rate": 0.001,
|
| 1111 |
+
"loss": 8.0235,
|
| 1112 |
+
"step": 18100
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.28,
|
| 1116 |
+
"learning_rate": 0.001,
|
| 1117 |
+
"loss": 8.0316,
|
| 1118 |
+
"step": 18200
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.29,
|
| 1122 |
+
"learning_rate": 0.001,
|
| 1123 |
+
"loss": 8.0319,
|
| 1124 |
+
"step": 18300
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.29,
|
| 1128 |
+
"learning_rate": 0.001,
|
| 1129 |
+
"loss": 8.0393,
|
| 1130 |
+
"step": 18400
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.29,
|
| 1134 |
+
"learning_rate": 0.001,
|
| 1135 |
+
"loss": 8.0248,
|
| 1136 |
+
"step": 18500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.29,
|
| 1140 |
+
"learning_rate": 0.001,
|
| 1141 |
+
"loss": 8.0463,
|
| 1142 |
+
"step": 18600
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.29,
|
| 1146 |
+
"learning_rate": 0.001,
|
| 1147 |
+
"loss": 8.024,
|
| 1148 |
+
"step": 18700
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.29,
|
| 1152 |
+
"learning_rate": 0.001,
|
| 1153 |
+
"loss": 8.0062,
|
| 1154 |
+
"step": 18800
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.3,
|
| 1158 |
+
"learning_rate": 0.001,
|
| 1159 |
+
"loss": 8.0302,
|
| 1160 |
+
"step": 18900
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.3,
|
| 1164 |
+
"learning_rate": 0.001,
|
| 1165 |
+
"loss": 8.054,
|
| 1166 |
+
"step": 19000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.3,
|
| 1170 |
+
"learning_rate": 0.001,
|
| 1171 |
+
"loss": 8.0448,
|
| 1172 |
+
"step": 19100
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.3,
|
| 1176 |
+
"learning_rate": 0.001,
|
| 1177 |
+
"loss": 8.0465,
|
| 1178 |
+
"step": 19200
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.3,
|
| 1182 |
+
"eval_accuracy": 0.03331597495163198,
|
| 1183 |
+
"eval_loss": 8.039088249206543,
|
| 1184 |
+
"eval_runtime": 7133.5004,
|
| 1185 |
+
"eval_samples_per_second": 45.96,
|
| 1186 |
+
"eval_steps_per_second": 2.873,
|
| 1187 |
+
"step": 19200
|
| 1188 |
}
|
| 1189 |
],
|
| 1190 |
"max_steps": 64000,
|
| 1191 |
"num_train_epochs": 9223372036854775807,
|
| 1192 |
+
"total_flos": 1.015475293126656e+17,
|
| 1193 |
"trial_name": null,
|
| 1194 |
"trial_params": null
|
| 1195 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 372832803
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7a084a9b3d69be038d1f70310127204c769b6f31132335d3c43f2359a442b86
|
| 3 |
size 372832803
|