Commit
β’
d508a96
1
Parent(s):
6744131
Training in progress, step 1400
Browse files- checkpoint-1000/latest +0 -1
- {checkpoint-1000 β checkpoint-1400}/config.json +0 -0
- {checkpoint-1000 β checkpoint-1400}/generation_config.json +0 -0
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-1400/latest +1 -0
- {checkpoint-1000 β checkpoint-1400}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-1000 β checkpoint-1400}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-1000 β checkpoint-1400}/model.safetensors.index.json +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_0.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_1.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_2.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_3.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_4.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_5.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_6.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/rng_state_7.pth +0 -0
- {checkpoint-1000 β checkpoint-1400}/special_tokens_map.json +0 -0
- {checkpoint-1000 β checkpoint-1400}/tokenizer.json +0 -0
- {checkpoint-1000 β checkpoint-1400}/tokenizer.model +0 -0
- {checkpoint-1000 β checkpoint-1400}/tokenizer_config.json +0 -0
- {checkpoint-1000 β checkpoint-1400}/trainer_state.json +243 -3
- {checkpoint-1000 β checkpoint-1400}/training_args.bin +0 -0
- {checkpoint-1000 β checkpoint-1400}/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-1000/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step1000
|
|
|
|
{checkpoint-1000 β checkpoint-1400}/config.json
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/generation_config.json
RENAMED
File without changes
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:d48928da7236975b7d563675900115bd40723d9bb946a5e12aeb3802cda65a29
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:60d0660e0465b031017aa8ac057dadaad3fefbbe9a853e985e3fbcc3b9e3d14b
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:fc3b54faf82a144d566ec1e13391ced58d82ba950098490d54bc7e6c12b2f665
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:8730071eb5352476076fecea2c16332b76414b1587e9170294cd8be64d658ae4
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:e1ecb36573935c513134e905740b8404414b0756f654afb39ce67f9eed0f089c
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:2977e65540cd59b6e490d355dee2caa45edce69df4ccad7d08cd12e64717c58d
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:a91e4e9224ed4db284c54c1c60b66d8d8db36f96ef264a6d0724d17d6d4159e2
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:88a2ea44cd053990b0f804df2960b863b7ee45e2d96229c63aeb1b4f124b33f1
|
3 |
size 10107626487
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:5b427970d78d97a2990a387d0111436e57780839b349873130d3acb3276e6590
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:25cd95fea28ed0b382448d3984a3b9c57253251be9b58ca7a14bdb60e0044800
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:a12da95cea77c0e4bf6c95cf547c52358a1a3dcc47d277ebc6a798c9caa01814
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:883c364ce237d2eaae2ccaffd1020d9bf975a0709445f76ec935f476a9e0355c
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:1f7dee3b066702da3217434a6fdf3be471ea2359e4a7f550407604a5b0eac806
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:144d70ba74905a12764aa691b542b267d9b7798360cf8174acfb33c6916e7df6
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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 sha256:1c9ff5c63b0581bc3bf5296b9188a1bf794ee1128efd17617a1b6d6c5bd71ef6
|
3 |
size 168086
|
{checkpoint-1000/global_step1000 β checkpoint-1400/global_step1400}/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:ddbafdcfd970d545721c1221e491df7ab9c50322e6312b7ec1b036b6481b3dcb
|
3 |
size 168086
|
checkpoint-1400/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step1400
|
{checkpoint-1000 β checkpoint-1400}/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:d81d64070bf6ada487dbdc2aaec06e52c932926918d08c4298cb20772e593337
|
3 |
size 9976576392
|
{checkpoint-1000 β checkpoint-1400}/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:df84e357f53f857462e754af1e9a7f50bdbf8395868f7a7403b6b72c5d94b5a7
|
3 |
size 3500296504
|
{checkpoint-1000 β checkpoint-1400}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/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,
|
@@ -606,11 +606,251 @@
|
|
606 |
"learning_rate": 0.0003,
|
607 |
"loss": 0.7357,
|
608 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
609 |
}
|
610 |
],
|
611 |
"max_steps": 3000,
|
612 |
"num_train_epochs": 9223372036854775807,
|
613 |
-
"total_flos":
|
614 |
"trial_name": null,
|
615 |
"trial_params": null
|
616 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.4666666666666667,
|
5 |
+
"global_step": 1400,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
606 |
"learning_rate": 0.0003,
|
607 |
"loss": 0.7357,
|
608 |
"step": 1000
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.34,
|
612 |
+
"learning_rate": 0.0003,
|
613 |
+
"loss": 0.6983,
|
614 |
+
"step": 1010
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.34,
|
618 |
+
"learning_rate": 0.0003,
|
619 |
+
"loss": 0.6989,
|
620 |
+
"step": 1020
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.34,
|
624 |
+
"learning_rate": 0.0003,
|
625 |
+
"loss": 0.6692,
|
626 |
+
"step": 1030
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 0.35,
|
630 |
+
"learning_rate": 0.0003,
|
631 |
+
"loss": 0.6553,
|
632 |
+
"step": 1040
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.35,
|
636 |
+
"learning_rate": 0.0003,
|
637 |
+
"loss": 0.6272,
|
638 |
+
"step": 1050
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 0.35,
|
642 |
+
"learning_rate": 0.0003,
|
643 |
+
"loss": 0.6092,
|
644 |
+
"step": 1060
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 0.36,
|
648 |
+
"learning_rate": 0.0003,
|
649 |
+
"loss": 0.604,
|
650 |
+
"step": 1070
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 0.36,
|
654 |
+
"learning_rate": 0.0003,
|
655 |
+
"loss": 0.6494,
|
656 |
+
"step": 1080
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 0.36,
|
660 |
+
"learning_rate": 0.0003,
|
661 |
+
"loss": 0.6155,
|
662 |
+
"step": 1090
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 0.37,
|
666 |
+
"learning_rate": 0.0003,
|
667 |
+
"loss": 0.5713,
|
668 |
+
"step": 1100
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 0.37,
|
672 |
+
"learning_rate": 0.0003,
|
673 |
+
"loss": 0.5358,
|
674 |
+
"step": 1110
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.37,
|
678 |
+
"learning_rate": 0.0003,
|
679 |
+
"loss": 0.5005,
|
680 |
+
"step": 1120
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 0.38,
|
684 |
+
"learning_rate": 0.0003,
|
685 |
+
"loss": 0.483,
|
686 |
+
"step": 1130
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 0.38,
|
690 |
+
"learning_rate": 0.0003,
|
691 |
+
"loss": 0.4736,
|
692 |
+
"step": 1140
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 0.38,
|
696 |
+
"learning_rate": 0.0003,
|
697 |
+
"loss": 0.4585,
|
698 |
+
"step": 1150
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 0.39,
|
702 |
+
"learning_rate": 0.0003,
|
703 |
+
"loss": 0.4442,
|
704 |
+
"step": 1160
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 0.39,
|
708 |
+
"learning_rate": 0.0003,
|
709 |
+
"loss": 0.4392,
|
710 |
+
"step": 1170
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 0.39,
|
714 |
+
"learning_rate": 0.0003,
|
715 |
+
"loss": 0.4084,
|
716 |
+
"step": 1180
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.4,
|
720 |
+
"learning_rate": 0.0003,
|
721 |
+
"loss": 0.4104,
|
722 |
+
"step": 1190
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 0.4,
|
726 |
+
"learning_rate": 0.0003,
|
727 |
+
"loss": 0.3855,
|
728 |
+
"step": 1200
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 0.4,
|
732 |
+
"learning_rate": 0.0003,
|
733 |
+
"loss": 0.3662,
|
734 |
+
"step": 1210
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 0.41,
|
738 |
+
"learning_rate": 0.0003,
|
739 |
+
"loss": 0.3727,
|
740 |
+
"step": 1220
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 0.41,
|
744 |
+
"learning_rate": 0.0003,
|
745 |
+
"loss": 0.3563,
|
746 |
+
"step": 1230
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 0.41,
|
750 |
+
"learning_rate": 0.0003,
|
751 |
+
"loss": 0.3297,
|
752 |
+
"step": 1240
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 0.42,
|
756 |
+
"learning_rate": 0.0003,
|
757 |
+
"loss": 0.314,
|
758 |
+
"step": 1250
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 0.42,
|
762 |
+
"learning_rate": 0.0003,
|
763 |
+
"loss": 0.2996,
|
764 |
+
"step": 1260
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 0.42,
|
768 |
+
"learning_rate": 0.0003,
|
769 |
+
"loss": 0.3,
|
770 |
+
"step": 1270
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 0.43,
|
774 |
+
"learning_rate": 0.0003,
|
775 |
+
"loss": 0.2773,
|
776 |
+
"step": 1280
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 0.43,
|
780 |
+
"learning_rate": 0.0003,
|
781 |
+
"loss": 0.2665,
|
782 |
+
"step": 1290
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 0.43,
|
786 |
+
"learning_rate": 0.0003,
|
787 |
+
"loss": 0.2646,
|
788 |
+
"step": 1300
|
789 |
+
},
|
790 |
+
{
|
791 |
+
"epoch": 0.44,
|
792 |
+
"learning_rate": 0.0003,
|
793 |
+
"loss": 0.2406,
|
794 |
+
"step": 1310
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"epoch": 0.44,
|
798 |
+
"learning_rate": 0.0003,
|
799 |
+
"loss": 0.2206,
|
800 |
+
"step": 1320
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 0.44,
|
804 |
+
"learning_rate": 0.0003,
|
805 |
+
"loss": 0.2325,
|
806 |
+
"step": 1330
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 0.45,
|
810 |
+
"learning_rate": 0.0003,
|
811 |
+
"loss": 0.2152,
|
812 |
+
"step": 1340
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 0.45,
|
816 |
+
"learning_rate": 0.0003,
|
817 |
+
"loss": 0.2011,
|
818 |
+
"step": 1350
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 0.45,
|
822 |
+
"learning_rate": 0.0003,
|
823 |
+
"loss": 0.1953,
|
824 |
+
"step": 1360
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 0.46,
|
828 |
+
"learning_rate": 0.0003,
|
829 |
+
"loss": 0.1862,
|
830 |
+
"step": 1370
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 0.46,
|
834 |
+
"learning_rate": 0.0003,
|
835 |
+
"loss": 0.1814,
|
836 |
+
"step": 1380
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 0.46,
|
840 |
+
"learning_rate": 0.0003,
|
841 |
+
"loss": 0.1677,
|
842 |
+
"step": 1390
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 0.47,
|
846 |
+
"learning_rate": 0.0003,
|
847 |
+
"loss": 0.1657,
|
848 |
+
"step": 1400
|
849 |
}
|
850 |
],
|
851 |
"max_steps": 3000,
|
852 |
"num_train_epochs": 9223372036854775807,
|
853 |
+
"total_flos": 586263035904000.0,
|
854 |
"trial_name": null,
|
855 |
"trial_params": null
|
856 |
}
|
{checkpoint-1000 β checkpoint-1400}/training_args.bin
RENAMED
File without changes
|
{checkpoint-1000 β checkpoint-1400}/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 sha256:4b5454467270806d8ae2ed14569e1e08a427c329f3a4cebd109224d30bd23d58
|
3 |
+
size 26131
|