Training in progress, step 20500
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +24 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 484085807
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:19b95781ca03c9e86d040b5c404b78a437638a3f0a1696e47d41a82934a10f8e
|
3 |
size 484085807
|
last-checkpoint/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242031355
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:05efcf5aababb852c7eccd6c53e27386a0693babf30041a64250374290be0a51
|
3 |
size 242031355
|
last-checkpoint/rng_state.pth
CHANGED
@@ -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:f8dea7ac2bda4e15808ad608ef7298e75b2d8fe2785da156ac3e1f364bc08ea1
|
3 |
size 14503
|
last-checkpoint/scaler.pt
CHANGED
@@ -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:cd51c2fa4070296203cb870e417cf342a24d6f4daf52267069468a9392c81b1e
|
3 |
size 559
|
last-checkpoint/scheduler.pt
CHANGED
@@ -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:d873908bf85daf85ad33f60109f938f0ba1dae30f111d7bece506678ebf5a435
|
3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
-
"best_model_checkpoint": "/home2/s5431786/jb-internship/results/t5-small-CodeXGLUE-CONCODE-faster/checkpoint-
|
4 |
-
"epoch": 6.
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -766,11 +766,30 @@
|
|
766 |
"eval_samples_per_second": 76.957,
|
767 |
"eval_steps_per_second": 2.424,
|
768 |
"step": 20000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
769 |
}
|
770 |
],
|
771 |
"max_steps": 62500,
|
772 |
"num_train_epochs": 20,
|
773 |
-
"total_flos": 8.
|
774 |
"trial_name": null,
|
775 |
"trial_params": null
|
776 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.6715394258499146,
|
3 |
+
"best_model_checkpoint": "/home2/s5431786/jb-internship/results/t5-small-CodeXGLUE-CONCODE-faster/checkpoint-20500",
|
4 |
+
"epoch": 6.5600000000000005,
|
5 |
+
"global_step": 20500,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
766 |
"eval_samples_per_second": 76.957,
|
767 |
"eval_steps_per_second": 2.424,
|
768 |
"step": 20000
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 6.56,
|
772 |
+
"learning_rate": 0.00035374315789473684,
|
773 |
+
"loss": 0.4563,
|
774 |
+
"step": 20500
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 6.56,
|
778 |
+
"eval_bleu": 0.06740545701870117,
|
779 |
+
"eval_exact_match": 0.1015,
|
780 |
+
"eval_loss": 0.6715394258499146,
|
781 |
+
"eval_rouge1": 0.5120845038766327,
|
782 |
+
"eval_rouge2": 0.35703709540753703,
|
783 |
+
"eval_rougeL": 0.49983280781141204,
|
784 |
+
"eval_runtime": 26.1007,
|
785 |
+
"eval_samples_per_second": 76.626,
|
786 |
+
"eval_steps_per_second": 2.414,
|
787 |
+
"step": 20500
|
788 |
}
|
789 |
],
|
790 |
"max_steps": 62500,
|
791 |
"num_train_epochs": 20,
|
792 |
+
"total_flos": 8.572045557025997e+16,
|
793 |
"trial_name": null,
|
794 |
"trial_params": null
|
795 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242031355
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:05efcf5aababb852c7eccd6c53e27386a0693babf30041a64250374290be0a51
|
3 |
size 242031355
|