Commit
β’
a2a6b42
1
Parent(s):
5b652bd
Training in progress, step 1100
Browse files- {checkpoint-700 β checkpoint-1100}/config.json +0 -0
- {checkpoint-700 β checkpoint-1100}/generation_config.json +0 -0
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-1100/latest +1 -0
- {checkpoint-700 β checkpoint-1100}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-700 β checkpoint-1100}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-700 β checkpoint-1100}/model.safetensors.index.json +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_0.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_1.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_2.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_3.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_4.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_5.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_6.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/rng_state_7.pth +0 -0
- {checkpoint-700 β checkpoint-1100}/special_tokens_map.json +0 -0
- {checkpoint-700 β checkpoint-1100}/tokenizer.json +0 -0
- {checkpoint-700 β checkpoint-1100}/tokenizer.model +0 -0
- {checkpoint-700 β checkpoint-1100}/tokenizer_config.json +0 -0
- {checkpoint-700 β checkpoint-1100}/trainer_state.json +243 -3
- {checkpoint-700 β checkpoint-1100}/training_args.bin +0 -0
- {checkpoint-700 β checkpoint-1100}/zero_to_fp32.py +0 -0
- checkpoint-700/latest +0 -1
- runs/Aug22_18-42-03_ip-26-0-150-12/events.out.tfevents.1692729850.ip-26-0-150-12.2895584.0 +2 -2
{checkpoint-700 β checkpoint-1100}/config.json
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/generation_config.json
RENAMED
File without changes
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:910192e566724dee1fc1dc6987f1c77385476d324f0f81bd9af23f2113e080a7
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:61f1e1b41f8262e168d6cbabaee5ff906517a3ee8a2ead7bd5be13778b3301f7
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:67ac954f2ed6ed05539aa95ae3231713b4fbe699591b84cc94c82f2aede2dd49
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:c9e49f162f388cf270931b7aa65b0b9c9aaf6cdb5d40302d91dd9f0e1ba295c2
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:1161a763f360f4a98ce6277167582207ee6fc9fd9c8b5abdfd5a209c4c49634a
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:f85ec6277b7860a4f2a0d7f64f8fe82fe641343c7c41b1c42929dd3015697314
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:67ec4562e79680723dc7ee7b0ac2cabacdb1efa6607963f2a994942598838e21
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:3bb6032fd72cafe86e8fcd99c359f85d4fd4f85e7374f13d9fc1a24aff2df188
|
3 |
size 10107626487
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:a022971e5382ee392191270920eba4792f2d4ec5585037c1fe9fa9ff54df8095
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:a21a722f8a14282dbc56a326a8a759992cd5dd28f453e399cf89d81dcb339ea6
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:46635e643d7c019b78e3565364f7bee809ef8e7197b5b9d0dea7d0c710b12bcb
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:0e4598ed421dbc4c3cb011d8de423da9683ea355a7f6e75d7dacfee0e89fe11f
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:5026560a0268883cf2ae7d43e32c2fcefca6cb057d4c730a8be77136302d3373
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:e86d87f88e26bd1ce11f9d7be3e75e20e283fe0db345548afaaa99dceed96052
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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:db3f7202cfede94534d5498f63e012ca1900f8ac02b249550b40fb836d5e9be8
|
3 |
size 168086
|
{checkpoint-700/global_step700 β checkpoint-1100/global_step1100}/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 sha256:4c59f381ff19b5347b25e41a80384718e30708c99755dbe2c66f53e3d7dbef70
|
3 |
size 168086
|
checkpoint-1100/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step1100
|
{checkpoint-700 β checkpoint-1100}/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:e9e589f280307fec92df9299851e14587c937700c97acd438840b30e79b1f025
|
3 |
size 9976576392
|
{checkpoint-700 β checkpoint-1100}/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 sha256:2561123382fff632d42a7bd0b76899385470af4cf48b4c423bfcccdec12b6205
|
3 |
size 3500296504
|
{checkpoint-700 β checkpoint-1100}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/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,
|
@@ -426,11 +426,251 @@
|
|
426 |
"learning_rate": 0.0003,
|
427 |
"loss": 1.379,
|
428 |
"step": 700
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
429 |
}
|
430 |
],
|
431 |
"max_steps": 3000,
|
432 |
"num_train_epochs": 9223372036854775807,
|
433 |
-
"total_flos":
|
434 |
"trial_name": null,
|
435 |
"trial_params": null
|
436 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.36666666666666664,
|
5 |
+
"global_step": 1100,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
426 |
"learning_rate": 0.0003,
|
427 |
"loss": 1.379,
|
428 |
"step": 700
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 0.24,
|
432 |
+
"learning_rate": 0.0003,
|
433 |
+
"loss": 1.3609,
|
434 |
+
"step": 710
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 0.24,
|
438 |
+
"learning_rate": 0.0003,
|
439 |
+
"loss": 1.3224,
|
440 |
+
"step": 720
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"epoch": 0.24,
|
444 |
+
"learning_rate": 0.0003,
|
445 |
+
"loss": 1.2853,
|
446 |
+
"step": 730
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 0.25,
|
450 |
+
"learning_rate": 0.0003,
|
451 |
+
"loss": 1.2762,
|
452 |
+
"step": 740
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 0.25,
|
456 |
+
"learning_rate": 0.0003,
|
457 |
+
"loss": 1.2265,
|
458 |
+
"step": 750
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"epoch": 0.25,
|
462 |
+
"learning_rate": 0.0003,
|
463 |
+
"loss": 1.2232,
|
464 |
+
"step": 760
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.26,
|
468 |
+
"learning_rate": 0.0003,
|
469 |
+
"loss": 1.2075,
|
470 |
+
"step": 770
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 0.26,
|
474 |
+
"learning_rate": 0.0003,
|
475 |
+
"loss": 1.168,
|
476 |
+
"step": 780
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 0.26,
|
480 |
+
"learning_rate": 0.0003,
|
481 |
+
"loss": 1.1472,
|
482 |
+
"step": 790
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 0.27,
|
486 |
+
"learning_rate": 0.0003,
|
487 |
+
"loss": 1.118,
|
488 |
+
"step": 800
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 0.27,
|
492 |
+
"learning_rate": 0.0003,
|
493 |
+
"loss": 1.1228,
|
494 |
+
"step": 810
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 0.27,
|
498 |
+
"learning_rate": 0.0003,
|
499 |
+
"loss": 1.1339,
|
500 |
+
"step": 820
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"epoch": 0.28,
|
504 |
+
"learning_rate": 0.0003,
|
505 |
+
"loss": 1.0853,
|
506 |
+
"step": 830
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.28,
|
510 |
+
"learning_rate": 0.0003,
|
511 |
+
"loss": 1.0676,
|
512 |
+
"step": 840
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 0.28,
|
516 |
+
"learning_rate": 0.0003,
|
517 |
+
"loss": 1.0905,
|
518 |
+
"step": 850
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 0.29,
|
522 |
+
"learning_rate": 0.0003,
|
523 |
+
"loss": 1.076,
|
524 |
+
"step": 860
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.29,
|
528 |
+
"learning_rate": 0.0003,
|
529 |
+
"loss": 1.0202,
|
530 |
+
"step": 870
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.29,
|
534 |
+
"learning_rate": 0.0003,
|
535 |
+
"loss": 1.0123,
|
536 |
+
"step": 880
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.3,
|
540 |
+
"learning_rate": 0.0003,
|
541 |
+
"loss": 0.9863,
|
542 |
+
"step": 890
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.3,
|
546 |
+
"learning_rate": 0.0003,
|
547 |
+
"loss": 0.9347,
|
548 |
+
"step": 900
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.3,
|
552 |
+
"learning_rate": 0.0003,
|
553 |
+
"loss": 0.9416,
|
554 |
+
"step": 910
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.31,
|
558 |
+
"learning_rate": 0.0003,
|
559 |
+
"loss": 0.9165,
|
560 |
+
"step": 920
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.31,
|
564 |
+
"learning_rate": 0.0003,
|
565 |
+
"loss": 0.8996,
|
566 |
+
"step": 930
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.31,
|
570 |
+
"learning_rate": 0.0003,
|
571 |
+
"loss": 0.8673,
|
572 |
+
"step": 940
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.32,
|
576 |
+
"learning_rate": 0.0003,
|
577 |
+
"loss": 0.8449,
|
578 |
+
"step": 950
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.32,
|
582 |
+
"learning_rate": 0.0003,
|
583 |
+
"loss": 0.8468,
|
584 |
+
"step": 960
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.32,
|
588 |
+
"learning_rate": 0.0003,
|
589 |
+
"loss": 0.817,
|
590 |
+
"step": 970
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.33,
|
594 |
+
"learning_rate": 0.0003,
|
595 |
+
"loss": 0.7947,
|
596 |
+
"step": 980
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.33,
|
600 |
+
"learning_rate": 0.0003,
|
601 |
+
"loss": 0.7706,
|
602 |
+
"step": 990
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.33,
|
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 |
"max_steps": 3000,
|
672 |
"num_train_epochs": 9223372036854775807,
|
673 |
+
"total_flos": 460635242496000.0,
|
674 |
"trial_name": null,
|
675 |
"trial_params": null
|
676 |
}
|
{checkpoint-700 β checkpoint-1100}/training_args.bin
RENAMED
File without changes
|
{checkpoint-700 β checkpoint-1100}/zero_to_fp32.py
RENAMED
File without changes
|
checkpoint-700/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step700
|
|
|
|
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 sha256:9195eb0ab07f76500bcac0b0eec3ca51997f05cae782877112fd61a0f25f510c
|
3 |
+
size 21421
|