Commit
β’
5b652bd
1
Parent(s):
42b7ff7
Training in progress, step 1000
Browse files- {checkpoint-600 β checkpoint-1000}/config.json +0 -0
- {checkpoint-600 β checkpoint-1000}/generation_config.json +0 -0
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-1000/latest +1 -0
- {checkpoint-600 β checkpoint-1000}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-600 β checkpoint-1000}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-600 β checkpoint-1000}/model.safetensors.index.json +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_0.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_1.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_2.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_3.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_4.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_5.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_6.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/rng_state_7.pth +0 -0
- {checkpoint-600 β checkpoint-1000}/special_tokens_map.json +0 -0
- {checkpoint-600 β checkpoint-1000}/tokenizer.json +0 -0
- {checkpoint-600 β checkpoint-1000}/tokenizer.model +0 -0
- {checkpoint-600 β checkpoint-1000}/tokenizer_config.json +0 -0
- {checkpoint-600 β checkpoint-1000}/trainer_state.json +243 -3
- {checkpoint-600 β checkpoint-1000}/training_args.bin +0 -0
- {checkpoint-600 β checkpoint-1000}/zero_to_fp32.py +0 -0
- checkpoint-600/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-600 β checkpoint-1000}/config.json
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/generation_config.json
RENAMED
File without changes
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:5703ed353a444bcb887496e437dcc7b16f8bdaae831e4e6d4f7ae38a220a5df3
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:65d6937e778c81679ecadc56c59bb90e942cd63ffc72959111931f27b22bea57
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:7078c4a9cb10247701dadff610cea1a671df2436f25c85ce7ec7f830d5864cff
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:632c044c424e36e9ac4d04c3db4d08e0453493e42ad88a6bbfdceadec83fd685
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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:c31e28791a65e07de46a9f174419774d315db15b86b1c34705af7a4bc7d99a66
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:07b03db6da0fdf58a5abbc86c3e32353bc4ea225df4814dd9b0c87094cba0a38
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:1621abeb507c8e04e54646297d7dfc104a12de8750d5ef2e450703166e2e6a23
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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:eb1c8f3f16701d2f9fe90ecb59b247f72b3a0bde6d3db2267a5c44807cf77842
|
3 |
size 10107626487
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:4658b6b3c7f7a48dd296534d1ec3dc6d546541ef01a0a3c11c403c23380b9050
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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:fc1c982bfd73913563caf1b20156081f202141ef2b4942de7b04a60a2735599a
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:3c42c53335aafb877cfbba5749485ee3cd61082a493faf56f57f8709eb5e6252
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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:ccd7e1a57d35ed384ee88e36e9a684ec710295b14dacde4ea9e97a2e0dd5e2f4
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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:aa0a5cc62566f10c81232feb782efb1f9f3ef3a38928981584b06095d904ad93
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:8585a4986217ec7db11028c99ffd2fb8f753ad9973841c9b2be8b872fa0f5af7
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:689aff8732618feb10643e53dc24544b9e68c3fe79059dd39c4324f792c95f8e
|
3 |
size 168086
|
{checkpoint-600/global_step600 β checkpoint-1000/global_step1000}/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 sha256:200527511d1203250da233f27a50a276bf203ca991380ba5c7c096f1955a2752
|
3 |
size 168086
|
checkpoint-1000/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step1000
|
{checkpoint-600 β checkpoint-1000}/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 sha256:11913c7f1753ca039f36c4c2b6bbefb982404ad4203fe8ba96a2f892793f3922
|
3 |
size 9976576392
|
{checkpoint-600 β checkpoint-1000}/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:b9877b1573d48279b0f52a3d16935fdb3bf67e40baa64bc8774bbb8005dd3519
|
3 |
size 3500296504
|
{checkpoint-600 β checkpoint-1000}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/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,
|
@@ -366,11 +366,251 @@
|
|
366 |
"learning_rate": 0.0003,
|
367 |
"loss": 1.706,
|
368 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
369 |
}
|
370 |
],
|
371 |
"max_steps": 3000,
|
372 |
"num_train_epochs": 9223372036854775807,
|
373 |
-
"total_flos":
|
374 |
"trial_name": null,
|
375 |
"trial_params": null
|
376 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.3333333333333333,
|
5 |
+
"global_step": 1000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
366 |
"learning_rate": 0.0003,
|
367 |
"loss": 1.706,
|
368 |
"step": 600
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.2,
|
372 |
+
"learning_rate": 0.0003,
|
373 |
+
"loss": 1.6328,
|
374 |
+
"step": 610
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.21,
|
378 |
+
"learning_rate": 0.0003,
|
379 |
+
"loss": 1.6197,
|
380 |
+
"step": 620
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.21,
|
384 |
+
"learning_rate": 0.0003,
|
385 |
+
"loss": 1.6106,
|
386 |
+
"step": 630
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.21,
|
390 |
+
"learning_rate": 0.0003,
|
391 |
+
"loss": 1.5649,
|
392 |
+
"step": 640
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 0.22,
|
396 |
+
"learning_rate": 0.0003,
|
397 |
+
"loss": 1.5001,
|
398 |
+
"step": 650
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 0.22,
|
402 |
+
"learning_rate": 0.0003,
|
403 |
+
"loss": 1.4926,
|
404 |
+
"step": 660
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 0.22,
|
408 |
+
"learning_rate": 0.0003,
|
409 |
+
"loss": 1.4682,
|
410 |
+
"step": 670
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 0.23,
|
414 |
+
"learning_rate": 0.0003,
|
415 |
+
"loss": 1.442,
|
416 |
+
"step": 680
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 0.23,
|
420 |
+
"learning_rate": 0.0003,
|
421 |
+
"loss": 1.4117,
|
422 |
+
"step": 690
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.23,
|
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 |
"max_steps": 3000,
|
612 |
"num_train_epochs": 9223372036854775807,
|
613 |
+
"total_flos": 418759311360000.0,
|
614 |
"trial_name": null,
|
615 |
"trial_params": null
|
616 |
}
|
{checkpoint-600 β checkpoint-1000}/training_args.bin
RENAMED
File without changes
|
{checkpoint-600 β checkpoint-1000}/zero_to_fp32.py
RENAMED
File without changes
|
checkpoint-600/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step600
|
|
|
|
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 sha256:1fef6847951e7a7a5d7472c77aeb6d8f614b223d630dfc7f7950ed07e82dfab5
|
3 |
+
size 19851
|