Commit
β’
42b7ff7
1
Parent(s):
70719e2
Training in progress, step 900
Browse files- checkpoint-500/latest +0 -1
- {checkpoint-500 β checkpoint-900}/config.json +0 -0
- {checkpoint-500 β checkpoint-900}/generation_config.json +0 -0
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-500/global_step500 β checkpoint-900/global_step900}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-900/latest +1 -0
- {checkpoint-500 β checkpoint-900}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-500 β checkpoint-900}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-500 β checkpoint-900}/model.safetensors.index.json +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_0.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_1.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_2.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_3.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_4.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_5.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_6.pth +0 -0
- {checkpoint-500 β checkpoint-900}/rng_state_7.pth +0 -0
- {checkpoint-500 β checkpoint-900}/special_tokens_map.json +0 -0
- {checkpoint-500 β checkpoint-900}/tokenizer.json +0 -0
- {checkpoint-500 β checkpoint-900}/tokenizer.model +0 -0
- {checkpoint-500 β checkpoint-900}/tokenizer_config.json +0 -0
- {checkpoint-500 β checkpoint-900}/trainer_state.json +243 -3
- {checkpoint-500 β checkpoint-900}/training_args.bin +0 -0
- {checkpoint-500 β checkpoint-900}/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-500/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step500
|
|
|
|
{checkpoint-500 β checkpoint-900}/config.json
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/generation_config.json
RENAMED
File without changes
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:c836b489002b4f2bc5a29c06d9ac1ff0ff9ee30d6de5c19ef31aa606835ed938
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:4638b20ef605dce6f96e75096cbfbe87a7877b1abd3a9b5afa3af8cc52517378
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:dfa69d6467f9bab3fe28d1cafc0998fc3f9c75a2e03996a04ce6fba11aeb676b
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:d12fa247d8406d10eafbdbb9e572469bb46469fb29b3ac3c995838172c4d47bf
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:27053d51c07b64de70a3af83bd9016c36eb79b84a0dd2fb87265c9b049096f2d
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:22b618f502d939e19b4a622a3fef3afa25bf90479b8a9d95f720c791005f095f
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:02425bbc1465b95b6560156504de32ec52855dd852eaf031c47a5568fe26e41d
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:c6b125eda1defc963593bac1d3086653fad3dd4e8dc70af33e84bc47f2edb5c7
|
3 |
size 10107626487
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:ecaf7b76d8592c78c8d76cedb0651a24329208a93db287965af7bdfc9587e4cb
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:fe3a2cfd541e47e579feddb0c9c70ece50f40de01bea95d993706f57708b0f10
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:e6c7b547ced115282b43a097bec3858ae4be3f9d8eacbfe01abbe9c124852b28
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:7be2ade1bb19c43fa3475e66a5cac9425535095126eacd7834f3a050e3615189
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:46d8ecc23ea617dfdf47c0cf31104948b3d98ae33c56c21b8064ff14b2b7b4e4
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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:b541819e7f1bd67ada94b000d3d6ae85f17e1992c76dab6be7f76e14f3fd99ff
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:752a19e2ff2dce2c503cd3c5d3a9a87205f3073b5dff8606dc8ceae6c5635e86
|
3 |
size 168086
|
{checkpoint-500/global_step500 β checkpoint-900/global_step900}/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 sha256:6d0a0f4b76828975e030fcb18822398f75abb2aa403baaf7b4b0837b13cd85d3
|
3 |
size 168086
|
checkpoint-900/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step900
|
{checkpoint-500 β checkpoint-900}/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 sha256:99a737b5b9f70216ba3b2df80ced780c4425ff708bbf429fde9862ab588ca6c4
|
3 |
size 9976576392
|
{checkpoint-500 β checkpoint-900}/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:fb8020484c2edfec4b31c0e0297c66b7bcc1add6fb17a98cf8c8de7f9aeef9bd
|
3 |
size 3500296504
|
{checkpoint-500 β checkpoint-900}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/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,
|
@@ -306,11 +306,251 @@
|
|
306 |
"learning_rate": 0.0003,
|
307 |
"loss": 2.1092,
|
308 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
309 |
}
|
310 |
],
|
311 |
"max_steps": 3000,
|
312 |
"num_train_epochs": 9223372036854775807,
|
313 |
-
"total_flos":
|
314 |
"trial_name": null,
|
315 |
"trial_params": null
|
316 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.3,
|
5 |
+
"global_step": 900,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
306 |
"learning_rate": 0.0003,
|
307 |
"loss": 2.1092,
|
308 |
"step": 500
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.17,
|
312 |
+
"learning_rate": 0.0003,
|
313 |
+
"loss": 2.0715,
|
314 |
+
"step": 510
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.17,
|
318 |
+
"learning_rate": 0.0003,
|
319 |
+
"loss": 2.0473,
|
320 |
+
"step": 520
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.18,
|
324 |
+
"learning_rate": 0.0003,
|
325 |
+
"loss": 2.0566,
|
326 |
+
"step": 530
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.18,
|
330 |
+
"learning_rate": 0.0003,
|
331 |
+
"loss": 1.9818,
|
332 |
+
"step": 540
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.18,
|
336 |
+
"learning_rate": 0.0003,
|
337 |
+
"loss": 1.9592,
|
338 |
+
"step": 550
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.19,
|
342 |
+
"learning_rate": 0.0003,
|
343 |
+
"loss": 1.8939,
|
344 |
+
"step": 560
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.19,
|
348 |
+
"learning_rate": 0.0003,
|
349 |
+
"loss": 1.8441,
|
350 |
+
"step": 570
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.19,
|
354 |
+
"learning_rate": 0.0003,
|
355 |
+
"loss": 1.808,
|
356 |
+
"step": 580
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.2,
|
360 |
+
"learning_rate": 0.0003,
|
361 |
+
"loss": 1.765,
|
362 |
+
"step": 590
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.2,
|
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 |
"max_steps": 3000,
|
552 |
"num_train_epochs": 9223372036854775807,
|
553 |
+
"total_flos": 376883380224000.0,
|
554 |
"trial_name": null,
|
555 |
"trial_params": null
|
556 |
}
|
{checkpoint-500 β checkpoint-900}/training_args.bin
RENAMED
File without changes
|
{checkpoint-500 β checkpoint-900}/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:bf0edb971b8a90918c57c5b33c853a049ca331754786d7d176c54919449c9aaa
|
3 |
+
size 18281
|