Upload 17 files
Browse files- optimizer.pt +2 -2
- pytorch_model.bin +2 -2
- rng_state.pth +1 -1
- scaler.pt +1 -1
- scheduler.pt +1 -1
- trainer_state.json +3 -252
optimizer.pt
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 sha256:47ea4092ddde682d366002e26f501e8f2c43fa7a07916494a0a33eef1122e075
|
| 3 |
+
size 6111428695
|
pytorch_model.bin
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:f2685daba978c2b8d65cb43ad777cda963664eba2acb444794a1480b6a6afd6b
|
| 3 |
+
size 3055754841
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14575
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12d7e6b7df767721053ff9f6323a231cf0852ba9410e5b7d74ee3079d462119a
|
| 3 |
size 14575
|
scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 557
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e769351f21ac92d172aed90d39bd27fe1d1e537736f1318d4c2465433f1bbfa
|
| 3 |
size 557
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d8eb51cc9c26967d17a84e7b6820a0d26198ba86110325f72bafd766eeb13bb
|
| 3 |
size 627
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 100.0,
|
| 3 |
"best_model_checkpoint": "./Malasar_ASR_DICtandLuke/checkpoint-1000",
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -753,260 +753,11 @@
|
|
| 753 |
"eval_steps_per_second": 0.204,
|
| 754 |
"eval_wer": 100.0,
|
| 755 |
"step": 3000
|
| 756 |
-
},
|
| 757 |
-
{
|
| 758 |
-
"epoch": 5.74,
|
| 759 |
-
"learning_rate": 2.802857142857143e-06,
|
| 760 |
-
"loss": 0.0062,
|
| 761 |
-
"step": 3025
|
| 762 |
-
},
|
| 763 |
-
{
|
| 764 |
-
"epoch": 5.79,
|
| 765 |
-
"learning_rate": 2.7314285714285714e-06,
|
| 766 |
-
"loss": 0.0063,
|
| 767 |
-
"step": 3050
|
| 768 |
-
},
|
| 769 |
-
{
|
| 770 |
-
"epoch": 5.83,
|
| 771 |
-
"learning_rate": 2.6600000000000004e-06,
|
| 772 |
-
"loss": 0.0072,
|
| 773 |
-
"step": 3075
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 5.88,
|
| 777 |
-
"learning_rate": 2.5885714285714285e-06,
|
| 778 |
-
"loss": 0.0047,
|
| 779 |
-
"step": 3100
|
| 780 |
-
},
|
| 781 |
-
{
|
| 782 |
-
"epoch": 5.93,
|
| 783 |
-
"learning_rate": 2.5171428571428575e-06,
|
| 784 |
-
"loss": 0.0078,
|
| 785 |
-
"step": 3125
|
| 786 |
-
},
|
| 787 |
-
{
|
| 788 |
-
"epoch": 5.98,
|
| 789 |
-
"learning_rate": 2.445714285714286e-06,
|
| 790 |
-
"loss": 0.0069,
|
| 791 |
-
"step": 3150
|
| 792 |
-
},
|
| 793 |
-
{
|
| 794 |
-
"epoch": 6.02,
|
| 795 |
-
"learning_rate": 2.3742857142857147e-06,
|
| 796 |
-
"loss": 0.0039,
|
| 797 |
-
"step": 3175
|
| 798 |
-
},
|
| 799 |
-
{
|
| 800 |
-
"epoch": 6.07,
|
| 801 |
-
"learning_rate": 2.302857142857143e-06,
|
| 802 |
-
"loss": 0.0035,
|
| 803 |
-
"step": 3200
|
| 804 |
-
},
|
| 805 |
-
{
|
| 806 |
-
"epoch": 6.12,
|
| 807 |
-
"learning_rate": 2.2314285714285715e-06,
|
| 808 |
-
"loss": 0.0043,
|
| 809 |
-
"step": 3225
|
| 810 |
-
},
|
| 811 |
-
{
|
| 812 |
-
"epoch": 6.17,
|
| 813 |
-
"learning_rate": 2.16e-06,
|
| 814 |
-
"loss": 0.0045,
|
| 815 |
-
"step": 3250
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 6.21,
|
| 819 |
-
"learning_rate": 2.0885714285714287e-06,
|
| 820 |
-
"loss": 0.0023,
|
| 821 |
-
"step": 3275
|
| 822 |
-
},
|
| 823 |
-
{
|
| 824 |
-
"epoch": 6.26,
|
| 825 |
-
"learning_rate": 2.0171428571428573e-06,
|
| 826 |
-
"loss": 0.0031,
|
| 827 |
-
"step": 3300
|
| 828 |
-
},
|
| 829 |
-
{
|
| 830 |
-
"epoch": 6.31,
|
| 831 |
-
"learning_rate": 1.945714285714286e-06,
|
| 832 |
-
"loss": 0.0028,
|
| 833 |
-
"step": 3325
|
| 834 |
-
},
|
| 835 |
-
{
|
| 836 |
-
"epoch": 6.36,
|
| 837 |
-
"learning_rate": 1.8742857142857142e-06,
|
| 838 |
-
"loss": 0.0053,
|
| 839 |
-
"step": 3350
|
| 840 |
-
},
|
| 841 |
-
{
|
| 842 |
-
"epoch": 6.4,
|
| 843 |
-
"learning_rate": 1.8028571428571432e-06,
|
| 844 |
-
"loss": 0.0016,
|
| 845 |
-
"step": 3375
|
| 846 |
-
},
|
| 847 |
-
{
|
| 848 |
-
"epoch": 6.45,
|
| 849 |
-
"learning_rate": 1.7314285714285716e-06,
|
| 850 |
-
"loss": 0.0019,
|
| 851 |
-
"step": 3400
|
| 852 |
-
},
|
| 853 |
-
{
|
| 854 |
-
"epoch": 6.5,
|
| 855 |
-
"learning_rate": 1.6600000000000002e-06,
|
| 856 |
-
"loss": 0.0038,
|
| 857 |
-
"step": 3425
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 6.55,
|
| 861 |
-
"learning_rate": 1.5885714285714288e-06,
|
| 862 |
-
"loss": 0.0028,
|
| 863 |
-
"step": 3450
|
| 864 |
-
},
|
| 865 |
-
{
|
| 866 |
-
"epoch": 6.59,
|
| 867 |
-
"learning_rate": 1.5171428571428574e-06,
|
| 868 |
-
"loss": 0.0047,
|
| 869 |
-
"step": 3475
|
| 870 |
-
},
|
| 871 |
-
{
|
| 872 |
-
"epoch": 6.64,
|
| 873 |
-
"learning_rate": 1.4457142857142858e-06,
|
| 874 |
-
"loss": 0.0041,
|
| 875 |
-
"step": 3500
|
| 876 |
-
},
|
| 877 |
-
{
|
| 878 |
-
"epoch": 6.69,
|
| 879 |
-
"learning_rate": 1.3742857142857143e-06,
|
| 880 |
-
"loss": 0.0015,
|
| 881 |
-
"step": 3525
|
| 882 |
-
},
|
| 883 |
-
{
|
| 884 |
-
"epoch": 6.74,
|
| 885 |
-
"learning_rate": 1.302857142857143e-06,
|
| 886 |
-
"loss": 0.002,
|
| 887 |
-
"step": 3550
|
| 888 |
-
},
|
| 889 |
-
{
|
| 890 |
-
"epoch": 6.78,
|
| 891 |
-
"learning_rate": 1.2314285714285715e-06,
|
| 892 |
-
"loss": 0.007,
|
| 893 |
-
"step": 3575
|
| 894 |
-
},
|
| 895 |
-
{
|
| 896 |
-
"epoch": 6.83,
|
| 897 |
-
"learning_rate": 1.1600000000000001e-06,
|
| 898 |
-
"loss": 0.0023,
|
| 899 |
-
"step": 3600
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 6.88,
|
| 903 |
-
"learning_rate": 1.0885714285714287e-06,
|
| 904 |
-
"loss": 0.0023,
|
| 905 |
-
"step": 3625
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 6.93,
|
| 909 |
-
"learning_rate": 1.0171428571428573e-06,
|
| 910 |
-
"loss": 0.0022,
|
| 911 |
-
"step": 3650
|
| 912 |
-
},
|
| 913 |
-
{
|
| 914 |
-
"epoch": 6.97,
|
| 915 |
-
"learning_rate": 9.457142857142858e-07,
|
| 916 |
-
"loss": 0.0034,
|
| 917 |
-
"step": 3675
|
| 918 |
-
},
|
| 919 |
-
{
|
| 920 |
-
"epoch": 7.02,
|
| 921 |
-
"learning_rate": 8.742857142857144e-07,
|
| 922 |
-
"loss": 0.0057,
|
| 923 |
-
"step": 3700
|
| 924 |
-
},
|
| 925 |
-
{
|
| 926 |
-
"epoch": 7.07,
|
| 927 |
-
"learning_rate": 8.028571428571429e-07,
|
| 928 |
-
"loss": 0.0009,
|
| 929 |
-
"step": 3725
|
| 930 |
-
},
|
| 931 |
-
{
|
| 932 |
-
"epoch": 7.12,
|
| 933 |
-
"learning_rate": 7.314285714285715e-07,
|
| 934 |
-
"loss": 0.003,
|
| 935 |
-
"step": 3750
|
| 936 |
-
},
|
| 937 |
-
{
|
| 938 |
-
"epoch": 7.16,
|
| 939 |
-
"learning_rate": 6.6e-07,
|
| 940 |
-
"loss": 0.0037,
|
| 941 |
-
"step": 3775
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 7.21,
|
| 945 |
-
"learning_rate": 5.885714285714286e-07,
|
| 946 |
-
"loss": 0.0007,
|
| 947 |
-
"step": 3800
|
| 948 |
-
},
|
| 949 |
-
{
|
| 950 |
-
"epoch": 7.26,
|
| 951 |
-
"learning_rate": 5.171428571428572e-07,
|
| 952 |
-
"loss": 0.0009,
|
| 953 |
-
"step": 3825
|
| 954 |
-
},
|
| 955 |
-
{
|
| 956 |
-
"epoch": 7.31,
|
| 957 |
-
"learning_rate": 4.457142857142858e-07,
|
| 958 |
-
"loss": 0.0008,
|
| 959 |
-
"step": 3850
|
| 960 |
-
},
|
| 961 |
-
{
|
| 962 |
-
"epoch": 7.35,
|
| 963 |
-
"learning_rate": 3.7428571428571434e-07,
|
| 964 |
-
"loss": 0.0019,
|
| 965 |
-
"step": 3875
|
| 966 |
-
},
|
| 967 |
-
{
|
| 968 |
-
"epoch": 7.4,
|
| 969 |
-
"learning_rate": 3.028571428571429e-07,
|
| 970 |
-
"loss": 0.001,
|
| 971 |
-
"step": 3900
|
| 972 |
-
},
|
| 973 |
-
{
|
| 974 |
-
"epoch": 7.45,
|
| 975 |
-
"learning_rate": 2.3142857142857144e-07,
|
| 976 |
-
"loss": 0.0011,
|
| 977 |
-
"step": 3925
|
| 978 |
-
},
|
| 979 |
-
{
|
| 980 |
-
"epoch": 7.5,
|
| 981 |
-
"learning_rate": 1.6e-07,
|
| 982 |
-
"loss": 0.0031,
|
| 983 |
-
"step": 3950
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 7.54,
|
| 987 |
-
"learning_rate": 8.857142857142858e-08,
|
| 988 |
-
"loss": 0.002,
|
| 989 |
-
"step": 3975
|
| 990 |
-
},
|
| 991 |
-
{
|
| 992 |
-
"epoch": 7.59,
|
| 993 |
-
"learning_rate": 1.7142857142857143e-08,
|
| 994 |
-
"loss": 0.002,
|
| 995 |
-
"step": 4000
|
| 996 |
-
},
|
| 997 |
-
{
|
| 998 |
-
"epoch": 7.59,
|
| 999 |
-
"eval_loss": 0.09421070665121078,
|
| 1000 |
-
"eval_runtime": 575.1216,
|
| 1001 |
-
"eval_samples_per_second": 3.257,
|
| 1002 |
-
"eval_steps_per_second": 0.205,
|
| 1003 |
-
"eval_wer": 100.0,
|
| 1004 |
-
"step": 4000
|
| 1005 |
}
|
| 1006 |
],
|
| 1007 |
"max_steps": 4000,
|
| 1008 |
"num_train_epochs": 8,
|
| 1009 |
-
"total_flos":
|
| 1010 |
"trial_name": null,
|
| 1011 |
"trial_params": null
|
| 1012 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 100.0,
|
| 3 |
"best_model_checkpoint": "./Malasar_ASR_DICtandLuke/checkpoint-1000",
|
| 4 |
+
"epoch": 5.692599620493358,
|
| 5 |
+
"global_step": 3000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 753 |
"eval_steps_per_second": 0.204,
|
| 754 |
"eval_wer": 100.0,
|
| 755 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 756 |
}
|
| 757 |
],
|
| 758 |
"max_steps": 4000,
|
| 759 |
"num_train_epochs": 8,
|
| 760 |
+
"total_flos": 9.79066287783936e+19,
|
| 761 |
"trial_name": null,
|
| 762 |
"trial_params": null
|
| 763 |
}
|