Commit
β’
73f4306
1
Parent(s):
747b2be
Training in progress, step 2100
Browse files- checkpoint-1700/latest +0 -1
- {checkpoint-1700 β checkpoint-2100}/config.json +0 -0
- {checkpoint-1700 β checkpoint-2100}/generation_config.json +0 -0
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-2100/latest +1 -0
- {checkpoint-1700 β checkpoint-2100}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-1700 β checkpoint-2100}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-1700 β checkpoint-2100}/model.safetensors.index.json +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_0.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_1.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_2.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_3.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_4.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_5.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_6.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/rng_state_7.pth +0 -0
- {checkpoint-1700 β checkpoint-2100}/special_tokens_map.json +0 -0
- {checkpoint-1700 β checkpoint-2100}/tokenizer.json +0 -0
- {checkpoint-1700 β checkpoint-2100}/tokenizer.model +0 -0
- {checkpoint-1700 β checkpoint-2100}/tokenizer_config.json +0 -0
- {checkpoint-1700 β checkpoint-2100}/trainer_state.json +243 -3
- {checkpoint-1700 β checkpoint-2100}/training_args.bin +0 -0
- {checkpoint-1700 β checkpoint-2100}/zero_to_fp32.py +0 -0
- runs/Aug22_18-42-03_ip-26-0-150-12/events.out.tfevents.1692729850.ip-26-0-150-12.2895584.0 +2 -2
checkpoint-1700/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step1700
|
|
|
|
{checkpoint-1700 β checkpoint-2100}/config.json
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/generation_config.json
RENAMED
File without changes
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b5e17adae64d158ed7b80018b62d4f64f76a717739f3667ab868bd08b5546e36
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:90103633ebee5c418c077e4b0d6fc4c2936161788ab990a5b63b49058bf980cd
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ac402f4e9dfd0a9cdeedc219939c8d2854a425000535a845a391b5a61be5fa21
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0587a206f02896afd81610085e44cc69b3fd43d70cc64144dda25c7471d6c721
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d431dee2fe9c6c5124ad116f289b7ae04ae192823760eb75d95eb02f500ec778
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:de5485088c9b41156808d8346e4617be253fea5de5cd5d4b91f71a26e2e24ea1
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5eed860602ba38c80b647d3f2412ee1d6b4778dbfce7131878373bbfbacb187c
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2a484f7976af13384e667021a7316ea630924e973e29a67120db984ff15076c2
|
3 |
size 10107626487
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_0_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:11aac4363e618fa55c62977722ae406ff69bb54c12e53e2fdeffa5f1632855c8
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_1_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4dfa5bde2a410bd133271a27c7be77ad69ce3015fcdc26b7f58ff280eea1d359
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_2_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2acccdd6e28633f3c199afe57b8d9b80e0d128bae92176d2309e4102e3de50e0
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_3_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2ce5abe052a95d79d00ceac0272d90efa7b18a68a220d8345df706b98c600cf1
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_4_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59a2ce519ef75a9502353294f95984f83b8c1473ae7ea64a1828949d6d3e0e41
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_5_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:851aceef9941342286cbd797e3b61362e8d088deabe520a249a37e2be273cc80
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_6_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cbc4e5fa62b4b71b517895a983df2fc1b618aef52984e9caa8fe7898198a2e10
|
3 |
size 168086
|
{checkpoint-1700/global_step1700 β checkpoint-2100/global_step2100}/zero_pp_rank_7_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e40d7b508c8727eadc3ef7115d9b998796e9f503c498fdea49536a01996484fe
|
3 |
size 168086
|
checkpoint-2100/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step2100
|
{checkpoint-1700 β checkpoint-2100}/model-00001-of-00002.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 9976576392
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6adc6b1233e38d5ffa6e2dd4bc49283f9ba67f0e6796f4957275ba8eac5007fa
|
3 |
size 9976576392
|
{checkpoint-1700 β checkpoint-2100}/model-00002-of-00002.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3500296504
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fb05fce5d6454ad204985cd2264d190114a5a59a137bd596e9372925ffd56913
|
3 |
size 3500296504
|
{checkpoint-1700 β checkpoint-2100}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/trainer_state.json
RENAMED
@@ -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,
|
@@ -1026,11 +1026,251 @@
|
|
1026 |
"learning_rate": 0.0003,
|
1027 |
"loss": 0.0904,
|
1028 |
"step": 1700
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1029 |
}
|
1030 |
],
|
1031 |
"max_steps": 3000,
|
1032 |
"num_train_epochs": 9223372036854775807,
|
1033 |
-
"total_flos":
|
1034 |
"trial_name": null,
|
1035 |
"trial_params": null
|
1036 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.7,
|
5 |
+
"global_step": 2100,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
1026 |
"learning_rate": 0.0003,
|
1027 |
"loss": 0.0904,
|
1028 |
"step": 1700
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 0.57,
|
1032 |
+
"learning_rate": 0.0003,
|
1033 |
+
"loss": 0.0785,
|
1034 |
+
"step": 1710
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 0.57,
|
1038 |
+
"learning_rate": 0.0003,
|
1039 |
+
"loss": 0.0747,
|
1040 |
+
"step": 1720
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 0.58,
|
1044 |
+
"learning_rate": 0.0003,
|
1045 |
+
"loss": 0.0704,
|
1046 |
+
"step": 1730
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 0.58,
|
1050 |
+
"learning_rate": 0.0003,
|
1051 |
+
"loss": 0.0634,
|
1052 |
+
"step": 1740
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 0.58,
|
1056 |
+
"learning_rate": 0.0003,
|
1057 |
+
"loss": 0.0629,
|
1058 |
+
"step": 1750
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 0.59,
|
1062 |
+
"learning_rate": 0.0003,
|
1063 |
+
"loss": 0.057,
|
1064 |
+
"step": 1760
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 0.59,
|
1068 |
+
"learning_rate": 0.0003,
|
1069 |
+
"loss": 0.0563,
|
1070 |
+
"step": 1770
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 0.59,
|
1074 |
+
"learning_rate": 0.0003,
|
1075 |
+
"loss": 0.054,
|
1076 |
+
"step": 1780
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 0.6,
|
1080 |
+
"learning_rate": 0.0003,
|
1081 |
+
"loss": 0.0532,
|
1082 |
+
"step": 1790
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 0.6,
|
1086 |
+
"learning_rate": 0.0003,
|
1087 |
+
"loss": 0.0525,
|
1088 |
+
"step": 1800
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 0.6,
|
1092 |
+
"learning_rate": 0.0003,
|
1093 |
+
"loss": 0.0519,
|
1094 |
+
"step": 1810
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 0.61,
|
1098 |
+
"learning_rate": 0.0003,
|
1099 |
+
"loss": 0.0478,
|
1100 |
+
"step": 1820
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 0.61,
|
1104 |
+
"learning_rate": 0.0003,
|
1105 |
+
"loss": 0.0473,
|
1106 |
+
"step": 1830
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 0.61,
|
1110 |
+
"learning_rate": 0.0003,
|
1111 |
+
"loss": 0.0473,
|
1112 |
+
"step": 1840
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 0.62,
|
1116 |
+
"learning_rate": 0.0003,
|
1117 |
+
"loss": 0.0492,
|
1118 |
+
"step": 1850
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 0.62,
|
1122 |
+
"learning_rate": 0.0003,
|
1123 |
+
"loss": 0.0451,
|
1124 |
+
"step": 1860
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 0.62,
|
1128 |
+
"learning_rate": 0.0003,
|
1129 |
+
"loss": 0.0453,
|
1130 |
+
"step": 1870
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 0.63,
|
1134 |
+
"learning_rate": 0.0003,
|
1135 |
+
"loss": 0.0439,
|
1136 |
+
"step": 1880
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 0.63,
|
1140 |
+
"learning_rate": 0.0003,
|
1141 |
+
"loss": 0.0435,
|
1142 |
+
"step": 1890
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 0.63,
|
1146 |
+
"learning_rate": 0.0003,
|
1147 |
+
"loss": 0.0435,
|
1148 |
+
"step": 1900
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 0.64,
|
1152 |
+
"learning_rate": 0.0003,
|
1153 |
+
"loss": 0.0412,
|
1154 |
+
"step": 1910
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 0.64,
|
1158 |
+
"learning_rate": 0.0003,
|
1159 |
+
"loss": 0.0441,
|
1160 |
+
"step": 1920
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 0.64,
|
1164 |
+
"learning_rate": 0.0003,
|
1165 |
+
"loss": 0.04,
|
1166 |
+
"step": 1930
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 0.65,
|
1170 |
+
"learning_rate": 0.0003,
|
1171 |
+
"loss": 0.041,
|
1172 |
+
"step": 1940
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 0.65,
|
1176 |
+
"learning_rate": 0.0003,
|
1177 |
+
"loss": 0.1391,
|
1178 |
+
"step": 1950
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 0.65,
|
1182 |
+
"learning_rate": 0.0003,
|
1183 |
+
"loss": 0.1502,
|
1184 |
+
"step": 1960
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 0.66,
|
1188 |
+
"learning_rate": 0.0003,
|
1189 |
+
"loss": 0.0767,
|
1190 |
+
"step": 1970
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 0.66,
|
1194 |
+
"learning_rate": 0.0003,
|
1195 |
+
"loss": 0.0563,
|
1196 |
+
"step": 1980
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 0.66,
|
1200 |
+
"learning_rate": 0.0003,
|
1201 |
+
"loss": 0.0487,
|
1202 |
+
"step": 1990
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 0.67,
|
1206 |
+
"learning_rate": 0.0003,
|
1207 |
+
"loss": 0.042,
|
1208 |
+
"step": 2000
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 0.67,
|
1212 |
+
"learning_rate": 0.0003,
|
1213 |
+
"loss": 0.0388,
|
1214 |
+
"step": 2010
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 0.67,
|
1218 |
+
"learning_rate": 0.0003,
|
1219 |
+
"loss": 0.0424,
|
1220 |
+
"step": 2020
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 0.68,
|
1224 |
+
"learning_rate": 0.0003,
|
1225 |
+
"loss": 0.0413,
|
1226 |
+
"step": 2030
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 0.68,
|
1230 |
+
"learning_rate": 0.0003,
|
1231 |
+
"loss": 0.036,
|
1232 |
+
"step": 2040
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 0.68,
|
1236 |
+
"learning_rate": 0.0003,
|
1237 |
+
"loss": 0.0333,
|
1238 |
+
"step": 2050
|
1239 |
+
},
|
1240 |
+
{
|
1241 |
+
"epoch": 0.69,
|
1242 |
+
"learning_rate": 0.0003,
|
1243 |
+
"loss": 0.0334,
|
1244 |
+
"step": 2060
|
1245 |
+
},
|
1246 |
+
{
|
1247 |
+
"epoch": 0.69,
|
1248 |
+
"learning_rate": 0.0003,
|
1249 |
+
"loss": 0.0298,
|
1250 |
+
"step": 2070
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 0.69,
|
1254 |
+
"learning_rate": 0.0003,
|
1255 |
+
"loss": 0.0302,
|
1256 |
+
"step": 2080
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 0.7,
|
1260 |
+
"learning_rate": 0.0003,
|
1261 |
+
"loss": 0.0302,
|
1262 |
+
"step": 2090
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 0.7,
|
1266 |
+
"learning_rate": 0.0003,
|
1267 |
+
"loss": 0.0293,
|
1268 |
+
"step": 2100
|
1269 |
}
|
1270 |
],
|
1271 |
"max_steps": 3000,
|
1272 |
"num_train_epochs": 9223372036854775807,
|
1273 |
+
"total_flos": 879394553856000.0,
|
1274 |
"trial_name": null,
|
1275 |
"trial_params": null
|
1276 |
}
|
{checkpoint-1700 β checkpoint-2100}/training_args.bin
RENAMED
File without changes
|
{checkpoint-1700 β checkpoint-2100}/zero_to_fp32.py
RENAMED
File without changes
|
runs/Aug22_18-42-03_ip-26-0-150-12/events.out.tfevents.1692729850.ip-26-0-150-12.2895584.0
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:b3476f29a411313ed060c01b35f1409f573b4dc1eb8ce083996033c57780434b
|
3 |
+
size 37121
|