"auto-commit"
Browse files- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/config.json +0 -0
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/optimizer.pt +1 -1
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/preprocessor_config.json +0 -0
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/pytorch_model.bin +1 -1
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/rng_state.pth +1 -1
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/scaler.pt +1 -1
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/scheduler.pt +1 -1
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/trainer_state.json +799 -4
- model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/training_args.bin +0 -0
- model-bin/finetune/base/log/1630140557.4831562/events.out.tfevents.1630140557.86bb0ddabf9b.4092.91 +3 -0
- model-bin/finetune/base/log/1630140950.614884/events.out.tfevents.1630140950.86bb0ddabf9b.4092.93 +3 -0
- model-bin/finetune/base/log/1630141343.6640568/events.out.tfevents.1630141343.86bb0ddabf9b.4092.95 +3 -0
- model-bin/finetune/base/log/1630141737.0182173/events.out.tfevents.1630141737.86bb0ddabf9b.4092.97 +3 -0
- model-bin/finetune/base/log/1630142244.9424796/events.out.tfevents.1630142244.86bb0ddabf9b.4092.99 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1630140557.86bb0ddabf9b.4092.90 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1630140950.86bb0ddabf9b.4092.92 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1630141343.86bb0ddabf9b.4092.94 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1630141737.86bb0ddabf9b.4092.96 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1630142244.86bb0ddabf9b.4092.98 +3 -0
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 722165393
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:444e2c1f4325b4b11162c9a02d9ac033783454f04b0ebac59f29f3f36dacd94e
|
| 3 |
size 722165393
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/preprocessor_config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/pytorch_model.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 377909911
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c99be2cb9f6bfd0d130822c68453cf6d1ebfc7d18545903ae81217c345b550ed
|
| 3 |
size 377909911
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14503
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:588ff1d2630a26b0738ac1935a67831bad436083ccce50a276fe4bac0f0ef39e
|
| 3 |
size 14503
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 559
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45ec721615ea8aa2c0f6eb839aa7fff0833136cc34604aa9f2ae17982ad6c3dd
|
| 3 |
size 559
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4693177c98f9aa50f7adffdb4243c309abaede1dcf51469b33e334fd32bdf34
|
| 3 |
size 623
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/trainer_state.json
RENAMED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.1735723097017633,
|
| 3 |
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-123203",
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -256593,11 +256593,806 @@
|
|
| 256593 |
"eval_steps_per_second": 0.738,
|
| 256594 |
"eval_wer": 0.17904090475141368,
|
| 256595 |
"step": 123577
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 256596 |
}
|
| 256597 |
],
|
| 256598 |
-
"max_steps":
|
| 256599 |
"num_train_epochs": 5000,
|
| 256600 |
-
"total_flos": 3.
|
| 256601 |
"trial_name": null,
|
| 256602 |
"trial_params": null
|
| 256603 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.1735723097017633,
|
| 3 |
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-123203",
|
| 4 |
+
"epoch": 1000.995983935743,
|
| 5 |
+
"global_step": 124200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 256593 |
"eval_steps_per_second": 0.738,
|
| 256594 |
"eval_wer": 0.17904090475141368,
|
| 256595 |
"step": 123577
|
| 256596 |
+
},
|
| 256597 |
+
{
|
| 256598 |
+
"epoch": 996.02,
|
| 256599 |
+
"learning_rate": 8.03636217948718e-06,
|
| 256600 |
+
"loss": 0.2697,
|
| 256601 |
+
"step": 123580
|
| 256602 |
+
},
|
| 256603 |
+
{
|
| 256604 |
+
"epoch": 996.06,
|
| 256605 |
+
"learning_rate": 8.036282051282052e-06,
|
| 256606 |
+
"loss": 0.2864,
|
| 256607 |
+
"step": 123585
|
| 256608 |
+
},
|
| 256609 |
+
{
|
| 256610 |
+
"epoch": 996.1,
|
| 256611 |
+
"learning_rate": 8.036201923076925e-06,
|
| 256612 |
+
"loss": 0.2807,
|
| 256613 |
+
"step": 123590
|
| 256614 |
+
},
|
| 256615 |
+
{
|
| 256616 |
+
"epoch": 996.15,
|
| 256617 |
+
"learning_rate": 8.036121794871794e-06,
|
| 256618 |
+
"loss": 0.3349,
|
| 256619 |
+
"step": 123595
|
| 256620 |
+
},
|
| 256621 |
+
{
|
| 256622 |
+
"epoch": 996.19,
|
| 256623 |
+
"learning_rate": 8.036041666666667e-06,
|
| 256624 |
+
"loss": 0.7529,
|
| 256625 |
+
"step": 123600
|
| 256626 |
+
},
|
| 256627 |
+
{
|
| 256628 |
+
"epoch": 996.23,
|
| 256629 |
+
"learning_rate": 8.035961538461539e-06,
|
| 256630 |
+
"loss": 0.8701,
|
| 256631 |
+
"step": 123605
|
| 256632 |
+
},
|
| 256633 |
+
{
|
| 256634 |
+
"epoch": 996.27,
|
| 256635 |
+
"learning_rate": 8.03588141025641e-06,
|
| 256636 |
+
"loss": 0.2396,
|
| 256637 |
+
"step": 123610
|
| 256638 |
+
},
|
| 256639 |
+
{
|
| 256640 |
+
"epoch": 996.31,
|
| 256641 |
+
"learning_rate": 8.035801282051283e-06,
|
| 256642 |
+
"loss": 0.3098,
|
| 256643 |
+
"step": 123615
|
| 256644 |
+
},
|
| 256645 |
+
{
|
| 256646 |
+
"epoch": 996.35,
|
| 256647 |
+
"learning_rate": 8.035721153846155e-06,
|
| 256648 |
+
"loss": 0.3529,
|
| 256649 |
+
"step": 123620
|
| 256650 |
+
},
|
| 256651 |
+
{
|
| 256652 |
+
"epoch": 996.39,
|
| 256653 |
+
"learning_rate": 8.035641025641026e-06,
|
| 256654 |
+
"loss": 0.7107,
|
| 256655 |
+
"step": 123625
|
| 256656 |
+
},
|
| 256657 |
+
{
|
| 256658 |
+
"epoch": 996.43,
|
| 256659 |
+
"learning_rate": 8.035560897435897e-06,
|
| 256660 |
+
"loss": 0.8661,
|
| 256661 |
+
"step": 123630
|
| 256662 |
+
},
|
| 256663 |
+
{
|
| 256664 |
+
"epoch": 996.47,
|
| 256665 |
+
"learning_rate": 8.03548076923077e-06,
|
| 256666 |
+
"loss": 0.2525,
|
| 256667 |
+
"step": 123635
|
| 256668 |
+
},
|
| 256669 |
+
{
|
| 256670 |
+
"epoch": 996.51,
|
| 256671 |
+
"learning_rate": 8.035400641025642e-06,
|
| 256672 |
+
"loss": 0.277,
|
| 256673 |
+
"step": 123640
|
| 256674 |
+
},
|
| 256675 |
+
{
|
| 256676 |
+
"epoch": 996.55,
|
| 256677 |
+
"learning_rate": 8.035320512820513e-06,
|
| 256678 |
+
"loss": 0.3795,
|
| 256679 |
+
"step": 123645
|
| 256680 |
+
},
|
| 256681 |
+
{
|
| 256682 |
+
"epoch": 996.59,
|
| 256683 |
+
"learning_rate": 8.035240384615384e-06,
|
| 256684 |
+
"loss": 0.6803,
|
| 256685 |
+
"step": 123650
|
| 256686 |
+
},
|
| 256687 |
+
{
|
| 256688 |
+
"epoch": 996.63,
|
| 256689 |
+
"learning_rate": 8.035160256410257e-06,
|
| 256690 |
+
"loss": 0.8381,
|
| 256691 |
+
"step": 123655
|
| 256692 |
+
},
|
| 256693 |
+
{
|
| 256694 |
+
"epoch": 996.67,
|
| 256695 |
+
"learning_rate": 8.035080128205129e-06,
|
| 256696 |
+
"loss": 0.2627,
|
| 256697 |
+
"step": 123660
|
| 256698 |
+
},
|
| 256699 |
+
{
|
| 256700 |
+
"epoch": 996.71,
|
| 256701 |
+
"learning_rate": 8.035e-06,
|
| 256702 |
+
"loss": 0.2651,
|
| 256703 |
+
"step": 123665
|
| 256704 |
+
},
|
| 256705 |
+
{
|
| 256706 |
+
"epoch": 996.75,
|
| 256707 |
+
"learning_rate": 8.034919871794873e-06,
|
| 256708 |
+
"loss": 0.3103,
|
| 256709 |
+
"step": 123670
|
| 256710 |
+
},
|
| 256711 |
+
{
|
| 256712 |
+
"epoch": 996.79,
|
| 256713 |
+
"learning_rate": 8.034839743589745e-06,
|
| 256714 |
+
"loss": 0.7412,
|
| 256715 |
+
"step": 123675
|
| 256716 |
+
},
|
| 256717 |
+
{
|
| 256718 |
+
"epoch": 996.83,
|
| 256719 |
+
"learning_rate": 8.034759615384616e-06,
|
| 256720 |
+
"loss": 0.816,
|
| 256721 |
+
"step": 123680
|
| 256722 |
+
},
|
| 256723 |
+
{
|
| 256724 |
+
"epoch": 996.87,
|
| 256725 |
+
"learning_rate": 8.034679487179487e-06,
|
| 256726 |
+
"loss": 0.2746,
|
| 256727 |
+
"step": 123685
|
| 256728 |
+
},
|
| 256729 |
+
{
|
| 256730 |
+
"epoch": 996.91,
|
| 256731 |
+
"learning_rate": 8.03459935897436e-06,
|
| 256732 |
+
"loss": 0.3075,
|
| 256733 |
+
"step": 123690
|
| 256734 |
+
},
|
| 256735 |
+
{
|
| 256736 |
+
"epoch": 996.95,
|
| 256737 |
+
"learning_rate": 8.034519230769232e-06,
|
| 256738 |
+
"loss": 0.4634,
|
| 256739 |
+
"step": 123695
|
| 256740 |
+
},
|
| 256741 |
+
{
|
| 256742 |
+
"epoch": 996.99,
|
| 256743 |
+
"learning_rate": 8.034439102564103e-06,
|
| 256744 |
+
"loss": 0.9143,
|
| 256745 |
+
"step": 123700
|
| 256746 |
+
},
|
| 256747 |
+
{
|
| 256748 |
+
"epoch": 997.0,
|
| 256749 |
+
"eval_loss": 0.35271042585372925,
|
| 256750 |
+
"eval_runtime": 37.4395,
|
| 256751 |
+
"eval_samples_per_second": 22.49,
|
| 256752 |
+
"eval_steps_per_second": 0.721,
|
| 256753 |
+
"eval_wer": 0.18502449265220433,
|
| 256754 |
+
"step": 123701
|
| 256755 |
+
},
|
| 256756 |
+
{
|
| 256757 |
+
"epoch": 989.03,
|
| 256758 |
+
"learning_rate": 8.034358974358974e-06,
|
| 256759 |
+
"loss": 0.3226,
|
| 256760 |
+
"step": 123705
|
| 256761 |
+
},
|
| 256762 |
+
{
|
| 256763 |
+
"epoch": 989.07,
|
| 256764 |
+
"learning_rate": 8.034278846153847e-06,
|
| 256765 |
+
"loss": 0.3369,
|
| 256766 |
+
"step": 123710
|
| 256767 |
+
},
|
| 256768 |
+
{
|
| 256769 |
+
"epoch": 989.11,
|
| 256770 |
+
"learning_rate": 8.034198717948719e-06,
|
| 256771 |
+
"loss": 0.2795,
|
| 256772 |
+
"step": 123715
|
| 256773 |
+
},
|
| 256774 |
+
{
|
| 256775 |
+
"epoch": 989.15,
|
| 256776 |
+
"learning_rate": 8.03411858974359e-06,
|
| 256777 |
+
"loss": 0.407,
|
| 256778 |
+
"step": 123720
|
| 256779 |
+
},
|
| 256780 |
+
{
|
| 256781 |
+
"epoch": 989.19,
|
| 256782 |
+
"learning_rate": 8.034038461538463e-06,
|
| 256783 |
+
"loss": 0.9145,
|
| 256784 |
+
"step": 123725
|
| 256785 |
+
},
|
| 256786 |
+
{
|
| 256787 |
+
"epoch": 989.23,
|
| 256788 |
+
"learning_rate": 8.033958333333335e-06,
|
| 256789 |
+
"loss": 0.6577,
|
| 256790 |
+
"step": 123730
|
| 256791 |
+
},
|
| 256792 |
+
{
|
| 256793 |
+
"epoch": 989.27,
|
| 256794 |
+
"learning_rate": 8.033878205128206e-06,
|
| 256795 |
+
"loss": 0.2938,
|
| 256796 |
+
"step": 123735
|
| 256797 |
+
},
|
| 256798 |
+
{
|
| 256799 |
+
"epoch": 989.31,
|
| 256800 |
+
"learning_rate": 8.033798076923077e-06,
|
| 256801 |
+
"loss": 0.3101,
|
| 256802 |
+
"step": 123740
|
| 256803 |
+
},
|
| 256804 |
+
{
|
| 256805 |
+
"epoch": 989.35,
|
| 256806 |
+
"learning_rate": 8.03371794871795e-06,
|
| 256807 |
+
"loss": 0.3811,
|
| 256808 |
+
"step": 123745
|
| 256809 |
+
},
|
| 256810 |
+
{
|
| 256811 |
+
"epoch": 989.39,
|
| 256812 |
+
"learning_rate": 8.03363782051282e-06,
|
| 256813 |
+
"loss": 0.8656,
|
| 256814 |
+
"step": 123750
|
| 256815 |
+
},
|
| 256816 |
+
{
|
| 256817 |
+
"epoch": 989.43,
|
| 256818 |
+
"learning_rate": 8.033557692307693e-06,
|
| 256819 |
+
"loss": 0.5815,
|
| 256820 |
+
"step": 123755
|
| 256821 |
+
},
|
| 256822 |
+
{
|
| 256823 |
+
"epoch": 989.47,
|
| 256824 |
+
"learning_rate": 8.033477564102566e-06,
|
| 256825 |
+
"loss": 0.2756,
|
| 256826 |
+
"step": 123760
|
| 256827 |
+
},
|
| 256828 |
+
{
|
| 256829 |
+
"epoch": 989.51,
|
| 256830 |
+
"learning_rate": 8.033397435897436e-06,
|
| 256831 |
+
"loss": 0.3436,
|
| 256832 |
+
"step": 123765
|
| 256833 |
+
},
|
| 256834 |
+
{
|
| 256835 |
+
"epoch": 989.55,
|
| 256836 |
+
"learning_rate": 8.033317307692309e-06,
|
| 256837 |
+
"loss": 0.444,
|
| 256838 |
+
"step": 123770
|
| 256839 |
+
},
|
| 256840 |
+
{
|
| 256841 |
+
"epoch": 989.59,
|
| 256842 |
+
"learning_rate": 8.03323717948718e-06,
|
| 256843 |
+
"loss": 0.8274,
|
| 256844 |
+
"step": 123775
|
| 256845 |
+
},
|
| 256846 |
+
{
|
| 256847 |
+
"epoch": 989.63,
|
| 256848 |
+
"learning_rate": 8.033157051282052e-06,
|
| 256849 |
+
"loss": 0.6641,
|
| 256850 |
+
"step": 123780
|
| 256851 |
+
},
|
| 256852 |
+
{
|
| 256853 |
+
"epoch": 989.67,
|
| 256854 |
+
"learning_rate": 8.033076923076923e-06,
|
| 256855 |
+
"loss": 0.2923,
|
| 256856 |
+
"step": 123785
|
| 256857 |
+
},
|
| 256858 |
+
{
|
| 256859 |
+
"epoch": 989.71,
|
| 256860 |
+
"learning_rate": 8.032996794871796e-06,
|
| 256861 |
+
"loss": 0.3045,
|
| 256862 |
+
"step": 123790
|
| 256863 |
+
},
|
| 256864 |
+
{
|
| 256865 |
+
"epoch": 989.75,
|
| 256866 |
+
"learning_rate": 8.032916666666667e-06,
|
| 256867 |
+
"loss": 0.5925,
|
| 256868 |
+
"step": 123795
|
| 256869 |
+
},
|
| 256870 |
+
{
|
| 256871 |
+
"epoch": 989.79,
|
| 256872 |
+
"learning_rate": 8.032836538461539e-06,
|
| 256873 |
+
"loss": 0.9411,
|
| 256874 |
+
"step": 123800
|
| 256875 |
+
},
|
| 256876 |
+
{
|
| 256877 |
+
"epoch": 989.83,
|
| 256878 |
+
"learning_rate": 8.03275641025641e-06,
|
| 256879 |
+
"loss": 0.6954,
|
| 256880 |
+
"step": 123805
|
| 256881 |
+
},
|
| 256882 |
+
{
|
| 256883 |
+
"epoch": 989.87,
|
| 256884 |
+
"learning_rate": 8.032676282051283e-06,
|
| 256885 |
+
"loss": 0.2363,
|
| 256886 |
+
"step": 123810
|
| 256887 |
+
},
|
| 256888 |
+
{
|
| 256889 |
+
"epoch": 989.91,
|
| 256890 |
+
"learning_rate": 8.032596153846154e-06,
|
| 256891 |
+
"loss": 0.3087,
|
| 256892 |
+
"step": 123815
|
| 256893 |
+
},
|
| 256894 |
+
{
|
| 256895 |
+
"epoch": 989.95,
|
| 256896 |
+
"learning_rate": 8.032516025641026e-06,
|
| 256897 |
+
"loss": 0.4986,
|
| 256898 |
+
"step": 123820
|
| 256899 |
+
},
|
| 256900 |
+
{
|
| 256901 |
+
"epoch": 989.99,
|
| 256902 |
+
"learning_rate": 8.032435897435899e-06,
|
| 256903 |
+
"loss": 0.9063,
|
| 256904 |
+
"step": 123825
|
| 256905 |
+
},
|
| 256906 |
+
{
|
| 256907 |
+
"epoch": 990.0,
|
| 256908 |
+
"eval_loss": 0.37780308723449707,
|
| 256909 |
+
"eval_runtime": 36.1129,
|
| 256910 |
+
"eval_samples_per_second": 23.316,
|
| 256911 |
+
"eval_steps_per_second": 0.748,
|
| 256912 |
+
"eval_wer": 0.18303048065650646,
|
| 256913 |
+
"step": 123826
|
| 256914 |
+
},
|
| 256915 |
+
{
|
| 256916 |
+
"epoch": 990.03,
|
| 256917 |
+
"learning_rate": 8.03235576923077e-06,
|
| 256918 |
+
"loss": 0.3242,
|
| 256919 |
+
"step": 123830
|
| 256920 |
+
},
|
| 256921 |
+
{
|
| 256922 |
+
"epoch": 990.07,
|
| 256923 |
+
"learning_rate": 8.032275641025642e-06,
|
| 256924 |
+
"loss": 0.2932,
|
| 256925 |
+
"step": 123835
|
| 256926 |
+
},
|
| 256927 |
+
{
|
| 256928 |
+
"epoch": 990.11,
|
| 256929 |
+
"learning_rate": 8.032195512820513e-06,
|
| 256930 |
+
"loss": 0.334,
|
| 256931 |
+
"step": 123840
|
| 256932 |
+
},
|
| 256933 |
+
{
|
| 256934 |
+
"epoch": 990.15,
|
| 256935 |
+
"learning_rate": 8.032115384615386e-06,
|
| 256936 |
+
"loss": 0.3641,
|
| 256937 |
+
"step": 123845
|
| 256938 |
+
},
|
| 256939 |
+
{
|
| 256940 |
+
"epoch": 990.19,
|
| 256941 |
+
"learning_rate": 8.032035256410257e-06,
|
| 256942 |
+
"loss": 0.9662,
|
| 256943 |
+
"step": 123850
|
| 256944 |
+
},
|
| 256945 |
+
{
|
| 256946 |
+
"epoch": 990.23,
|
| 256947 |
+
"learning_rate": 8.031955128205129e-06,
|
| 256948 |
+
"loss": 0.7018,
|
| 256949 |
+
"step": 123855
|
| 256950 |
+
},
|
| 256951 |
+
{
|
| 256952 |
+
"epoch": 990.27,
|
| 256953 |
+
"learning_rate": 8.031875000000002e-06,
|
| 256954 |
+
"loss": 0.2948,
|
| 256955 |
+
"step": 123860
|
| 256956 |
+
},
|
| 256957 |
+
{
|
| 256958 |
+
"epoch": 990.31,
|
| 256959 |
+
"learning_rate": 8.031794871794873e-06,
|
| 256960 |
+
"loss": 0.3453,
|
| 256961 |
+
"step": 123865
|
| 256962 |
+
},
|
| 256963 |
+
{
|
| 256964 |
+
"epoch": 990.35,
|
| 256965 |
+
"learning_rate": 8.031714743589744e-06,
|
| 256966 |
+
"loss": 0.439,
|
| 256967 |
+
"step": 123870
|
| 256968 |
+
},
|
| 256969 |
+
{
|
| 256970 |
+
"epoch": 990.39,
|
| 256971 |
+
"learning_rate": 8.031634615384616e-06,
|
| 256972 |
+
"loss": 1.012,
|
| 256973 |
+
"step": 123875
|
| 256974 |
+
},
|
| 256975 |
+
{
|
| 256976 |
+
"epoch": 990.43,
|
| 256977 |
+
"learning_rate": 8.031554487179489e-06,
|
| 256978 |
+
"loss": 0.6446,
|
| 256979 |
+
"step": 123880
|
| 256980 |
+
},
|
| 256981 |
+
{
|
| 256982 |
+
"epoch": 990.47,
|
| 256983 |
+
"learning_rate": 8.031474358974359e-06,
|
| 256984 |
+
"loss": 0.2602,
|
| 256985 |
+
"step": 123885
|
| 256986 |
+
},
|
| 256987 |
+
{
|
| 256988 |
+
"epoch": 990.51,
|
| 256989 |
+
"learning_rate": 8.031394230769232e-06,
|
| 256990 |
+
"loss": 0.3188,
|
| 256991 |
+
"step": 123890
|
| 256992 |
+
},
|
| 256993 |
+
{
|
| 256994 |
+
"epoch": 990.55,
|
| 256995 |
+
"learning_rate": 8.031314102564103e-06,
|
| 256996 |
+
"loss": 0.4144,
|
| 256997 |
+
"step": 123895
|
| 256998 |
+
},
|
| 256999 |
+
{
|
| 257000 |
+
"epoch": 990.59,
|
| 257001 |
+
"learning_rate": 8.031233974358974e-06,
|
| 257002 |
+
"loss": 0.7588,
|
| 257003 |
+
"step": 123900
|
| 257004 |
+
},
|
| 257005 |
+
{
|
| 257006 |
+
"epoch": 990.63,
|
| 257007 |
+
"learning_rate": 8.031153846153846e-06,
|
| 257008 |
+
"loss": 0.7133,
|
| 257009 |
+
"step": 123905
|
| 257010 |
+
},
|
| 257011 |
+
{
|
| 257012 |
+
"epoch": 990.67,
|
| 257013 |
+
"learning_rate": 8.031073717948719e-06,
|
| 257014 |
+
"loss": 0.2741,
|
| 257015 |
+
"step": 123910
|
| 257016 |
+
},
|
| 257017 |
+
{
|
| 257018 |
+
"epoch": 990.71,
|
| 257019 |
+
"learning_rate": 8.03099358974359e-06,
|
| 257020 |
+
"loss": 0.385,
|
| 257021 |
+
"step": 123915
|
| 257022 |
+
},
|
| 257023 |
+
{
|
| 257024 |
+
"epoch": 990.75,
|
| 257025 |
+
"learning_rate": 8.030913461538461e-06,
|
| 257026 |
+
"loss": 0.4035,
|
| 257027 |
+
"step": 123920
|
| 257028 |
+
},
|
| 257029 |
+
{
|
| 257030 |
+
"epoch": 990.79,
|
| 257031 |
+
"learning_rate": 8.030833333333335e-06,
|
| 257032 |
+
"loss": 0.9382,
|
| 257033 |
+
"step": 123925
|
| 257034 |
+
},
|
| 257035 |
+
{
|
| 257036 |
+
"epoch": 990.83,
|
| 257037 |
+
"learning_rate": 8.030753205128206e-06,
|
| 257038 |
+
"loss": 0.6219,
|
| 257039 |
+
"step": 123930
|
| 257040 |
+
},
|
| 257041 |
+
{
|
| 257042 |
+
"epoch": 990.87,
|
| 257043 |
+
"learning_rate": 8.030673076923077e-06,
|
| 257044 |
+
"loss": 0.2894,
|
| 257045 |
+
"step": 123935
|
| 257046 |
+
},
|
| 257047 |
+
{
|
| 257048 |
+
"epoch": 990.91,
|
| 257049 |
+
"learning_rate": 8.030592948717949e-06,
|
| 257050 |
+
"loss": 0.3362,
|
| 257051 |
+
"step": 123940
|
| 257052 |
+
},
|
| 257053 |
+
{
|
| 257054 |
+
"epoch": 990.95,
|
| 257055 |
+
"learning_rate": 8.030512820512822e-06,
|
| 257056 |
+
"loss": 0.4499,
|
| 257057 |
+
"step": 123945
|
| 257058 |
+
},
|
| 257059 |
+
{
|
| 257060 |
+
"epoch": 990.99,
|
| 257061 |
+
"learning_rate": 8.030432692307693e-06,
|
| 257062 |
+
"loss": 0.9499,
|
| 257063 |
+
"step": 123950
|
| 257064 |
+
},
|
| 257065 |
+
{
|
| 257066 |
+
"epoch": 991.0,
|
| 257067 |
+
"eval_loss": 0.38879507780075073,
|
| 257068 |
+
"eval_runtime": 36.3541,
|
| 257069 |
+
"eval_samples_per_second": 23.134,
|
| 257070 |
+
"eval_steps_per_second": 0.743,
|
| 257071 |
+
"eval_wer": 0.17872121256714002,
|
| 257072 |
+
"step": 123951
|
| 257073 |
+
},
|
| 257074 |
+
{
|
| 257075 |
+
"epoch": 991.03,
|
| 257076 |
+
"learning_rate": 8.030352564102564e-06,
|
| 257077 |
+
"loss": 0.3227,
|
| 257078 |
+
"step": 123955
|
| 257079 |
+
},
|
| 257080 |
+
{
|
| 257081 |
+
"epoch": 991.07,
|
| 257082 |
+
"learning_rate": 8.030272435897437e-06,
|
| 257083 |
+
"loss": 0.275,
|
| 257084 |
+
"step": 123960
|
| 257085 |
+
},
|
| 257086 |
+
{
|
| 257087 |
+
"epoch": 991.11,
|
| 257088 |
+
"learning_rate": 8.030192307692309e-06,
|
| 257089 |
+
"loss": 0.3181,
|
| 257090 |
+
"step": 123965
|
| 257091 |
+
},
|
| 257092 |
+
{
|
| 257093 |
+
"epoch": 991.15,
|
| 257094 |
+
"learning_rate": 8.03011217948718e-06,
|
| 257095 |
+
"loss": 0.3956,
|
| 257096 |
+
"step": 123970
|
| 257097 |
+
},
|
| 257098 |
+
{
|
| 257099 |
+
"epoch": 991.19,
|
| 257100 |
+
"learning_rate": 8.030032051282051e-06,
|
| 257101 |
+
"loss": 0.9103,
|
| 257102 |
+
"step": 123975
|
| 257103 |
+
},
|
| 257104 |
+
{
|
| 257105 |
+
"epoch": 991.23,
|
| 257106 |
+
"learning_rate": 8.029951923076925e-06,
|
| 257107 |
+
"loss": 0.6729,
|
| 257108 |
+
"step": 123980
|
| 257109 |
+
},
|
| 257110 |
+
{
|
| 257111 |
+
"epoch": 991.27,
|
| 257112 |
+
"learning_rate": 8.029871794871796e-06,
|
| 257113 |
+
"loss": 0.2789,
|
| 257114 |
+
"step": 123985
|
| 257115 |
+
},
|
| 257116 |
+
{
|
| 257117 |
+
"epoch": 991.31,
|
| 257118 |
+
"learning_rate": 8.029791666666667e-06,
|
| 257119 |
+
"loss": 0.3114,
|
| 257120 |
+
"step": 123990
|
| 257121 |
+
},
|
| 257122 |
+
{
|
| 257123 |
+
"epoch": 991.35,
|
| 257124 |
+
"learning_rate": 8.029711538461539e-06,
|
| 257125 |
+
"loss": 0.3959,
|
| 257126 |
+
"step": 123995
|
| 257127 |
+
},
|
| 257128 |
+
{
|
| 257129 |
+
"epoch": 991.39,
|
| 257130 |
+
"learning_rate": 8.029631410256412e-06,
|
| 257131 |
+
"loss": 0.9762,
|
| 257132 |
+
"step": 124000
|
| 257133 |
+
},
|
| 257134 |
+
{
|
| 257135 |
+
"epoch": 991.43,
|
| 257136 |
+
"learning_rate": 8.029551282051283e-06,
|
| 257137 |
+
"loss": 0.6035,
|
| 257138 |
+
"step": 124005
|
| 257139 |
+
},
|
| 257140 |
+
{
|
| 257141 |
+
"epoch": 991.47,
|
| 257142 |
+
"learning_rate": 8.029471153846154e-06,
|
| 257143 |
+
"loss": 0.3083,
|
| 257144 |
+
"step": 124010
|
| 257145 |
+
},
|
| 257146 |
+
{
|
| 257147 |
+
"epoch": 991.51,
|
| 257148 |
+
"learning_rate": 8.029391025641027e-06,
|
| 257149 |
+
"loss": 0.3193,
|
| 257150 |
+
"step": 124015
|
| 257151 |
+
},
|
| 257152 |
+
{
|
| 257153 |
+
"epoch": 991.55,
|
| 257154 |
+
"learning_rate": 8.029310897435899e-06,
|
| 257155 |
+
"loss": 0.3466,
|
| 257156 |
+
"step": 124020
|
| 257157 |
+
},
|
| 257158 |
+
{
|
| 257159 |
+
"epoch": 991.59,
|
| 257160 |
+
"learning_rate": 8.02923076923077e-06,
|
| 257161 |
+
"loss": 0.9612,
|
| 257162 |
+
"step": 124025
|
| 257163 |
+
},
|
| 257164 |
+
{
|
| 257165 |
+
"epoch": 991.63,
|
| 257166 |
+
"learning_rate": 8.029150641025642e-06,
|
| 257167 |
+
"loss": 0.6535,
|
| 257168 |
+
"step": 124030
|
| 257169 |
+
},
|
| 257170 |
+
{
|
| 257171 |
+
"epoch": 991.67,
|
| 257172 |
+
"learning_rate": 8.029070512820515e-06,
|
| 257173 |
+
"loss": 0.2786,
|
| 257174 |
+
"step": 124035
|
| 257175 |
+
},
|
| 257176 |
+
{
|
| 257177 |
+
"epoch": 991.71,
|
| 257178 |
+
"learning_rate": 8.028990384615384e-06,
|
| 257179 |
+
"loss": 0.2491,
|
| 257180 |
+
"step": 124040
|
| 257181 |
+
},
|
| 257182 |
+
{
|
| 257183 |
+
"epoch": 991.75,
|
| 257184 |
+
"learning_rate": 8.028910256410257e-06,
|
| 257185 |
+
"loss": 0.4216,
|
| 257186 |
+
"step": 124045
|
| 257187 |
+
},
|
| 257188 |
+
{
|
| 257189 |
+
"epoch": 991.79,
|
| 257190 |
+
"learning_rate": 8.028830128205129e-06,
|
| 257191 |
+
"loss": 0.9788,
|
| 257192 |
+
"step": 124050
|
| 257193 |
+
},
|
| 257194 |
+
{
|
| 257195 |
+
"epoch": 991.83,
|
| 257196 |
+
"learning_rate": 8.02875e-06,
|
| 257197 |
+
"loss": 0.5756,
|
| 257198 |
+
"step": 124055
|
| 257199 |
+
},
|
| 257200 |
+
{
|
| 257201 |
+
"epoch": 991.87,
|
| 257202 |
+
"learning_rate": 8.028669871794873e-06,
|
| 257203 |
+
"loss": 0.2539,
|
| 257204 |
+
"step": 124060
|
| 257205 |
+
},
|
| 257206 |
+
{
|
| 257207 |
+
"epoch": 991.91,
|
| 257208 |
+
"learning_rate": 8.028589743589744e-06,
|
| 257209 |
+
"loss": 0.3347,
|
| 257210 |
+
"step": 124065
|
| 257211 |
+
},
|
| 257212 |
+
{
|
| 257213 |
+
"epoch": 991.95,
|
| 257214 |
+
"learning_rate": 8.028509615384616e-06,
|
| 257215 |
+
"loss": 0.4351,
|
| 257216 |
+
"step": 124070
|
| 257217 |
+
},
|
| 257218 |
+
{
|
| 257219 |
+
"epoch": 991.99,
|
| 257220 |
+
"learning_rate": 8.028429487179487e-06,
|
| 257221 |
+
"loss": 0.8815,
|
| 257222 |
+
"step": 124075
|
| 257223 |
+
},
|
| 257224 |
+
{
|
| 257225 |
+
"epoch": 992.0,
|
| 257226 |
+
"eval_loss": 0.39232033491134644,
|
| 257227 |
+
"eval_runtime": 47.6402,
|
| 257228 |
+
"eval_samples_per_second": 17.653,
|
| 257229 |
+
"eval_steps_per_second": 0.567,
|
| 257230 |
+
"eval_wer": 0.18814771566194716,
|
| 257231 |
+
"step": 124076
|
| 257232 |
+
},
|
| 257233 |
+
{
|
| 257234 |
+
"epoch": 1000.03,
|
| 257235 |
+
"learning_rate": 8.02834935897436e-06,
|
| 257236 |
+
"loss": 0.4001,
|
| 257237 |
+
"step": 124080
|
| 257238 |
+
},
|
| 257239 |
+
{
|
| 257240 |
+
"epoch": 1000.07,
|
| 257241 |
+
"learning_rate": 8.028269230769232e-06,
|
| 257242 |
+
"loss": 0.2938,
|
| 257243 |
+
"step": 124085
|
| 257244 |
+
},
|
| 257245 |
+
{
|
| 257246 |
+
"epoch": 1000.11,
|
| 257247 |
+
"learning_rate": 8.028189102564103e-06,
|
| 257248 |
+
"loss": 0.3341,
|
| 257249 |
+
"step": 124090
|
| 257250 |
+
},
|
| 257251 |
+
{
|
| 257252 |
+
"epoch": 1000.15,
|
| 257253 |
+
"learning_rate": 8.028108974358974e-06,
|
| 257254 |
+
"loss": 0.3691,
|
| 257255 |
+
"step": 124095
|
| 257256 |
+
},
|
| 257257 |
+
{
|
| 257258 |
+
"epoch": 1000.19,
|
| 257259 |
+
"learning_rate": 8.028028846153847e-06,
|
| 257260 |
+
"loss": 0.7611,
|
| 257261 |
+
"step": 124100
|
| 257262 |
+
},
|
| 257263 |
+
{
|
| 257264 |
+
"epoch": 1000.23,
|
| 257265 |
+
"learning_rate": 8.027948717948719e-06,
|
| 257266 |
+
"loss": 0.8048,
|
| 257267 |
+
"step": 124105
|
| 257268 |
+
},
|
| 257269 |
+
{
|
| 257270 |
+
"epoch": 1000.27,
|
| 257271 |
+
"learning_rate": 8.02786858974359e-06,
|
| 257272 |
+
"loss": 0.2299,
|
| 257273 |
+
"step": 124110
|
| 257274 |
+
},
|
| 257275 |
+
{
|
| 257276 |
+
"epoch": 1000.31,
|
| 257277 |
+
"learning_rate": 8.027788461538463e-06,
|
| 257278 |
+
"loss": 0.2911,
|
| 257279 |
+
"step": 124115
|
| 257280 |
+
},
|
| 257281 |
+
{
|
| 257282 |
+
"epoch": 1000.35,
|
| 257283 |
+
"learning_rate": 8.027708333333334e-06,
|
| 257284 |
+
"loss": 0.367,
|
| 257285 |
+
"step": 124120
|
| 257286 |
+
},
|
| 257287 |
+
{
|
| 257288 |
+
"epoch": 1000.39,
|
| 257289 |
+
"learning_rate": 8.027628205128206e-06,
|
| 257290 |
+
"loss": 1.2227,
|
| 257291 |
+
"step": 124125
|
| 257292 |
+
},
|
| 257293 |
+
{
|
| 257294 |
+
"epoch": 1000.43,
|
| 257295 |
+
"learning_rate": 8.027548076923077e-06,
|
| 257296 |
+
"loss": 0.6543,
|
| 257297 |
+
"step": 124130
|
| 257298 |
+
},
|
| 257299 |
+
{
|
| 257300 |
+
"epoch": 1000.47,
|
| 257301 |
+
"learning_rate": 8.02746794871795e-06,
|
| 257302 |
+
"loss": 0.2603,
|
| 257303 |
+
"step": 124135
|
| 257304 |
+
},
|
| 257305 |
+
{
|
| 257306 |
+
"epoch": 1000.51,
|
| 257307 |
+
"learning_rate": 8.027387820512822e-06,
|
| 257308 |
+
"loss": 0.3135,
|
| 257309 |
+
"step": 124140
|
| 257310 |
+
},
|
| 257311 |
+
{
|
| 257312 |
+
"epoch": 1000.55,
|
| 257313 |
+
"learning_rate": 8.027307692307693e-06,
|
| 257314 |
+
"loss": 0.4096,
|
| 257315 |
+
"step": 124145
|
| 257316 |
+
},
|
| 257317 |
+
{
|
| 257318 |
+
"epoch": 1000.59,
|
| 257319 |
+
"learning_rate": 8.027227564102564e-06,
|
| 257320 |
+
"loss": 0.8186,
|
| 257321 |
+
"step": 124150
|
| 257322 |
+
},
|
| 257323 |
+
{
|
| 257324 |
+
"epoch": 1000.63,
|
| 257325 |
+
"learning_rate": 8.027147435897437e-06,
|
| 257326 |
+
"loss": 0.5708,
|
| 257327 |
+
"step": 124155
|
| 257328 |
+
},
|
| 257329 |
+
{
|
| 257330 |
+
"epoch": 1000.67,
|
| 257331 |
+
"learning_rate": 8.027067307692307e-06,
|
| 257332 |
+
"loss": 0.2524,
|
| 257333 |
+
"step": 124160
|
| 257334 |
+
},
|
| 257335 |
+
{
|
| 257336 |
+
"epoch": 1000.71,
|
| 257337 |
+
"learning_rate": 8.02698717948718e-06,
|
| 257338 |
+
"loss": 0.3512,
|
| 257339 |
+
"step": 124165
|
| 257340 |
+
},
|
| 257341 |
+
{
|
| 257342 |
+
"epoch": 1000.76,
|
| 257343 |
+
"learning_rate": 8.026907051282053e-06,
|
| 257344 |
+
"loss": 0.3847,
|
| 257345 |
+
"step": 124170
|
| 257346 |
+
},
|
| 257347 |
+
{
|
| 257348 |
+
"epoch": 1000.8,
|
| 257349 |
+
"learning_rate": 8.026826923076923e-06,
|
| 257350 |
+
"loss": 0.8454,
|
| 257351 |
+
"step": 124175
|
| 257352 |
+
},
|
| 257353 |
+
{
|
| 257354 |
+
"epoch": 1000.84,
|
| 257355 |
+
"learning_rate": 8.026746794871796e-06,
|
| 257356 |
+
"loss": 0.6479,
|
| 257357 |
+
"step": 124180
|
| 257358 |
+
},
|
| 257359 |
+
{
|
| 257360 |
+
"epoch": 1000.88,
|
| 257361 |
+
"learning_rate": 8.026666666666667e-06,
|
| 257362 |
+
"loss": 0.3131,
|
| 257363 |
+
"step": 124185
|
| 257364 |
+
},
|
| 257365 |
+
{
|
| 257366 |
+
"epoch": 1000.92,
|
| 257367 |
+
"learning_rate": 8.026586538461539e-06,
|
| 257368 |
+
"loss": 0.3086,
|
| 257369 |
+
"step": 124190
|
| 257370 |
+
},
|
| 257371 |
+
{
|
| 257372 |
+
"epoch": 1000.96,
|
| 257373 |
+
"learning_rate": 8.02650641025641e-06,
|
| 257374 |
+
"loss": 0.4573,
|
| 257375 |
+
"step": 124195
|
| 257376 |
+
},
|
| 257377 |
+
{
|
| 257378 |
+
"epoch": 1001.0,
|
| 257379 |
+
"learning_rate": 8.026426282051283e-06,
|
| 257380 |
+
"loss": 0.9595,
|
| 257381 |
+
"step": 124200
|
| 257382 |
+
},
|
| 257383 |
+
{
|
| 257384 |
+
"epoch": 1001.0,
|
| 257385 |
+
"eval_loss": 0.3651779294013977,
|
| 257386 |
+
"eval_runtime": 38.1351,
|
| 257387 |
+
"eval_samples_per_second": 22.027,
|
| 257388 |
+
"eval_steps_per_second": 0.708,
|
| 257389 |
+
"eval_wer": 0.17809599194070663,
|
| 257390 |
+
"step": 124200
|
| 257391 |
}
|
| 257392 |
],
|
| 257393 |
+
"max_steps": 620000,
|
| 257394 |
"num_train_epochs": 5000,
|
| 257395 |
+
"total_flos": 3.495167016128345e+20,
|
| 257396 |
"trial_name": null,
|
| 257397 |
"trial_params": null
|
| 257398 |
}
|
model-bin/finetune/base/{checkpoint-123577 → checkpoint-124200}/training_args.bin
RENAMED
|
File without changes
|
model-bin/finetune/base/log/1630140557.4831562/events.out.tfevents.1630140557.86bb0ddabf9b.4092.91
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d73248dcc2b7910acfc0a677dde4c151d7899c2286efd1aac34ee6d974a413b
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1630140950.614884/events.out.tfevents.1630140950.86bb0ddabf9b.4092.93
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ffa9ec9c1181f602da573c19d0a0493f728604b421e88c7131d0109bcb7b0ce4
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1630141343.6640568/events.out.tfevents.1630141343.86bb0ddabf9b.4092.95
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f13ddb00d1f9568e3ae74524e018c20610803841c833f78e5775d7a5b7fc713
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1630141737.0182173/events.out.tfevents.1630141737.86bb0ddabf9b.4092.97
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5147694ce0091023478759fcfb13569e248993596d086c758a5ee1343ca93f9
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1630142244.9424796/events.out.tfevents.1630142244.86bb0ddabf9b.4092.99
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1e885954ef1e20bce9cfe8d13c1518ff5c67664c98f7fe4d14eaa71e4708ad6
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/events.out.tfevents.1630140557.86bb0ddabf9b.4092.90
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7203831ac8ca7cc0616e15630080885e158dd0d69e7f87cd2a67ae3afed5d97c
|
| 3 |
+
size 8622
|
model-bin/finetune/base/log/events.out.tfevents.1630140950.86bb0ddabf9b.4092.92
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e571256d6df9f671e427227f6da6533abb61847f2c28a568e74ca57fa2f34976
|
| 3 |
+
size 8622
|
model-bin/finetune/base/log/events.out.tfevents.1630141343.86bb0ddabf9b.4092.94
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:889f2ddb22b58feb268eb865b8883ea8ed638ad0c711a484e4cda48112694e78
|
| 3 |
+
size 8622
|
model-bin/finetune/base/log/events.out.tfevents.1630141737.86bb0ddabf9b.4092.96
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02726508ab3fb5220414baabc518871a21c605f9170023db3ba530c086f7f352
|
| 3 |
+
size 8622
|
model-bin/finetune/base/log/events.out.tfevents.1630142244.86bb0ddabf9b.4092.98
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:849522741e53cdb2c7c0a09411191446657a52978759b04b9dac7701a37019fe
|
| 3 |
+
size 8622
|