Training in progress, step 732, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3582214344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b01e5c7ba58cc0a8ceb5c097cf2452678dc78f80db4ffafbdc7d1f12a7b93235
|
| 3 |
size 3582214344
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5116251049
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6750ba3b9a48a1c182fa61beb15d922aa4b2ddd1d732efddeb80b267d908c130
|
| 3 |
size 5116251049
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d569a9c96ff4e89d32dfe8ca0b97952cc2edcd76f89c14caa7e57238c4b46c8a
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c8d3de20b2c14860168a267cd05d5c71bfafe85ad743a4b90197aae066604e65
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 244,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -3440,6 +3440,1722 @@
|
|
| 3440 |
"eval_samples_per_second": 39.047,
|
| 3441 |
"eval_steps_per_second": 9.762,
|
| 3442 |
"step": 488
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3443 |
}
|
| 3444 |
],
|
| 3445 |
"logging_steps": 1,
|
|
@@ -3459,7 +5175,7 @@
|
|
| 3459 |
"attributes": {}
|
| 3460 |
}
|
| 3461 |
},
|
| 3462 |
-
"total_flos": 1.
|
| 3463 |
"train_batch_size": 4,
|
| 3464 |
"trial_name": null,
|
| 3465 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.6006153846153847,
|
| 6 |
"eval_steps": 244,
|
| 7 |
+
"global_step": 732,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3440 |
"eval_samples_per_second": 39.047,
|
| 3441 |
"eval_steps_per_second": 9.762,
|
| 3442 |
"step": 488
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 0.4012307692307692,
|
| 3446 |
+
"grad_norm": 1.5625,
|
| 3447 |
+
"learning_rate": 8.25700626376351e-06,
|
| 3448 |
+
"loss": 1.108311414718628,
|
| 3449 |
+
"step": 489
|
| 3450 |
+
},
|
| 3451 |
+
{
|
| 3452 |
+
"epoch": 0.40205128205128204,
|
| 3453 |
+
"grad_norm": 1.6484375,
|
| 3454 |
+
"learning_rate": 8.241725270353607e-06,
|
| 3455 |
+
"loss": 1.0573620796203613,
|
| 3456 |
+
"step": 490
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 0.40287179487179486,
|
| 3460 |
+
"grad_norm": 1.8828125,
|
| 3461 |
+
"learning_rate": 8.226430352894541e-06,
|
| 3462 |
+
"loss": 1.1554553508758545,
|
| 3463 |
+
"step": 491
|
| 3464 |
+
},
|
| 3465 |
+
{
|
| 3466 |
+
"epoch": 0.4036923076923077,
|
| 3467 |
+
"grad_norm": 1.5078125,
|
| 3468 |
+
"learning_rate": 8.211121615693778e-06,
|
| 3469 |
+
"loss": 1.0917716026306152,
|
| 3470 |
+
"step": 492
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 0.4045128205128205,
|
| 3474 |
+
"grad_norm": 1.65625,
|
| 3475 |
+
"learning_rate": 8.195799163153027e-06,
|
| 3476 |
+
"loss": 1.0735331773757935,
|
| 3477 |
+
"step": 493
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 0.4053333333333333,
|
| 3481 |
+
"grad_norm": 1.5546875,
|
| 3482 |
+
"learning_rate": 8.180463099767533e-06,
|
| 3483 |
+
"loss": 1.0706208944320679,
|
| 3484 |
+
"step": 494
|
| 3485 |
+
},
|
| 3486 |
+
{
|
| 3487 |
+
"epoch": 0.40615384615384614,
|
| 3488 |
+
"grad_norm": 1.5546875,
|
| 3489 |
+
"learning_rate": 8.165113530125363e-06,
|
| 3490 |
+
"loss": 1.0984644889831543,
|
| 3491 |
+
"step": 495
|
| 3492 |
+
},
|
| 3493 |
+
{
|
| 3494 |
+
"epoch": 0.40697435897435896,
|
| 3495 |
+
"grad_norm": 1.6171875,
|
| 3496 |
+
"learning_rate": 8.149750558906694e-06,
|
| 3497 |
+
"loss": 1.1961491107940674,
|
| 3498 |
+
"step": 496
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 0.4077948717948718,
|
| 3502 |
+
"grad_norm": 1.6328125,
|
| 3503 |
+
"learning_rate": 8.134374290883098e-06,
|
| 3504 |
+
"loss": 1.119896650314331,
|
| 3505 |
+
"step": 497
|
| 3506 |
+
},
|
| 3507 |
+
{
|
| 3508 |
+
"epoch": 0.4086153846153846,
|
| 3509 |
+
"grad_norm": 1.4609375,
|
| 3510 |
+
"learning_rate": 8.118984830916828e-06,
|
| 3511 |
+
"loss": 1.1157628297805786,
|
| 3512 |
+
"step": 498
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 0.4094358974358974,
|
| 3516 |
+
"grad_norm": 1.671875,
|
| 3517 |
+
"learning_rate": 8.103582283960105e-06,
|
| 3518 |
+
"loss": 1.0383669137954712,
|
| 3519 |
+
"step": 499
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 0.41025641025641024,
|
| 3523 |
+
"grad_norm": 1.703125,
|
| 3524 |
+
"learning_rate": 8.088166755054395e-06,
|
| 3525 |
+
"loss": 1.1565940380096436,
|
| 3526 |
+
"step": 500
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 0.41107692307692306,
|
| 3530 |
+
"grad_norm": 1.71875,
|
| 3531 |
+
"learning_rate": 8.072738349329702e-06,
|
| 3532 |
+
"loss": 1.1026580333709717,
|
| 3533 |
+
"step": 501
|
| 3534 |
+
},
|
| 3535 |
+
{
|
| 3536 |
+
"epoch": 0.4118974358974359,
|
| 3537 |
+
"grad_norm": 1.640625,
|
| 3538 |
+
"learning_rate": 8.057297172003842e-06,
|
| 3539 |
+
"loss": 1.140966534614563,
|
| 3540 |
+
"step": 502
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 0.4127179487179487,
|
| 3544 |
+
"grad_norm": 1.6640625,
|
| 3545 |
+
"learning_rate": 8.041843328381735e-06,
|
| 3546 |
+
"loss": 1.1219335794448853,
|
| 3547 |
+
"step": 503
|
| 3548 |
+
},
|
| 3549 |
+
{
|
| 3550 |
+
"epoch": 0.4135384615384615,
|
| 3551 |
+
"grad_norm": 1.515625,
|
| 3552 |
+
"learning_rate": 8.026376923854683e-06,
|
| 3553 |
+
"loss": 1.1024707555770874,
|
| 3554 |
+
"step": 504
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 0.41435897435897434,
|
| 3558 |
+
"grad_norm": 1.7265625,
|
| 3559 |
+
"learning_rate": 8.010898063899641e-06,
|
| 3560 |
+
"loss": 1.204986810684204,
|
| 3561 |
+
"step": 505
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 0.41517948717948716,
|
| 3565 |
+
"grad_norm": 1.75,
|
| 3566 |
+
"learning_rate": 7.995406854078518e-06,
|
| 3567 |
+
"loss": 1.1070940494537354,
|
| 3568 |
+
"step": 506
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 0.416,
|
| 3572 |
+
"grad_norm": 1.625,
|
| 3573 |
+
"learning_rate": 7.979903400037435e-06,
|
| 3574 |
+
"loss": 1.096226453781128,
|
| 3575 |
+
"step": 507
|
| 3576 |
+
},
|
| 3577 |
+
{
|
| 3578 |
+
"epoch": 0.4168205128205128,
|
| 3579 |
+
"grad_norm": 1.609375,
|
| 3580 |
+
"learning_rate": 7.96438780750603e-06,
|
| 3581 |
+
"loss": 1.1158950328826904,
|
| 3582 |
+
"step": 508
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 0.4176410256410256,
|
| 3586 |
+
"grad_norm": 1.640625,
|
| 3587 |
+
"learning_rate": 7.948860182296706e-06,
|
| 3588 |
+
"loss": 1.1123420000076294,
|
| 3589 |
+
"step": 509
|
| 3590 |
+
},
|
| 3591 |
+
{
|
| 3592 |
+
"epoch": 0.41846153846153844,
|
| 3593 |
+
"grad_norm": 1.71875,
|
| 3594 |
+
"learning_rate": 7.93332063030394e-06,
|
| 3595 |
+
"loss": 1.1149189472198486,
|
| 3596 |
+
"step": 510
|
| 3597 |
+
},
|
| 3598 |
+
{
|
| 3599 |
+
"epoch": 0.41928205128205126,
|
| 3600 |
+
"grad_norm": 1.7265625,
|
| 3601 |
+
"learning_rate": 7.917769257503536e-06,
|
| 3602 |
+
"loss": 1.1472665071487427,
|
| 3603 |
+
"step": 511
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 0.4201025641025641,
|
| 3607 |
+
"grad_norm": 1.6484375,
|
| 3608 |
+
"learning_rate": 7.90220616995192e-06,
|
| 3609 |
+
"loss": 1.0900911092758179,
|
| 3610 |
+
"step": 512
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 0.4209230769230769,
|
| 3614 |
+
"grad_norm": 1.8203125,
|
| 3615 |
+
"learning_rate": 7.886631473785402e-06,
|
| 3616 |
+
"loss": 1.1053787469863892,
|
| 3617 |
+
"step": 513
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 0.4217435897435897,
|
| 3621 |
+
"grad_norm": 1.78125,
|
| 3622 |
+
"learning_rate": 7.87104527521947e-06,
|
| 3623 |
+
"loss": 1.1553611755371094,
|
| 3624 |
+
"step": 514
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 0.42256410256410254,
|
| 3628 |
+
"grad_norm": 1.6015625,
|
| 3629 |
+
"learning_rate": 7.855447680548048e-06,
|
| 3630 |
+
"loss": 1.150394082069397,
|
| 3631 |
+
"step": 515
|
| 3632 |
+
},
|
| 3633 |
+
{
|
| 3634 |
+
"epoch": 0.42338461538461536,
|
| 3635 |
+
"grad_norm": 1.609375,
|
| 3636 |
+
"learning_rate": 7.83983879614278e-06,
|
| 3637 |
+
"loss": 1.1880446672439575,
|
| 3638 |
+
"step": 516
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 0.4242051282051282,
|
| 3642 |
+
"grad_norm": 1.546875,
|
| 3643 |
+
"learning_rate": 7.824218728452307e-06,
|
| 3644 |
+
"loss": 1.0694695711135864,
|
| 3645 |
+
"step": 517
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 0.425025641025641,
|
| 3649 |
+
"grad_norm": 1.75,
|
| 3650 |
+
"learning_rate": 7.80858758400153e-06,
|
| 3651 |
+
"loss": 1.1604297161102295,
|
| 3652 |
+
"step": 518
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 0.4258461538461538,
|
| 3656 |
+
"grad_norm": 1.5390625,
|
| 3657 |
+
"learning_rate": 7.792945469390896e-06,
|
| 3658 |
+
"loss": 1.0425302982330322,
|
| 3659 |
+
"step": 519
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 0.4266666666666667,
|
| 3663 |
+
"grad_norm": 1.5234375,
|
| 3664 |
+
"learning_rate": 7.777292491295666e-06,
|
| 3665 |
+
"loss": 1.0891900062561035,
|
| 3666 |
+
"step": 520
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 0.4274871794871795,
|
| 3670 |
+
"grad_norm": 1.90625,
|
| 3671 |
+
"learning_rate": 7.761628756465185e-06,
|
| 3672 |
+
"loss": 1.1330201625823975,
|
| 3673 |
+
"step": 521
|
| 3674 |
+
},
|
| 3675 |
+
{
|
| 3676 |
+
"epoch": 0.42830769230769233,
|
| 3677 |
+
"grad_norm": 1.7265625,
|
| 3678 |
+
"learning_rate": 7.745954371722153e-06,
|
| 3679 |
+
"loss": 1.0741163492202759,
|
| 3680 |
+
"step": 522
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 0.42912820512820515,
|
| 3684 |
+
"grad_norm": 1.71875,
|
| 3685 |
+
"learning_rate": 7.730269443961909e-06,
|
| 3686 |
+
"loss": 1.1834020614624023,
|
| 3687 |
+
"step": 523
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 0.429948717948718,
|
| 3691 |
+
"grad_norm": 1.65625,
|
| 3692 |
+
"learning_rate": 7.714574080151685e-06,
|
| 3693 |
+
"loss": 1.120877742767334,
|
| 3694 |
+
"step": 524
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 0.4307692307692308,
|
| 3698 |
+
"grad_norm": 1.7734375,
|
| 3699 |
+
"learning_rate": 7.698868387329887e-06,
|
| 3700 |
+
"loss": 0.9986427426338196,
|
| 3701 |
+
"step": 525
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 0.4315897435897436,
|
| 3705 |
+
"grad_norm": 1.5859375,
|
| 3706 |
+
"learning_rate": 7.683152472605357e-06,
|
| 3707 |
+
"loss": 1.1671761274337769,
|
| 3708 |
+
"step": 526
|
| 3709 |
+
},
|
| 3710 |
+
{
|
| 3711 |
+
"epoch": 0.43241025641025643,
|
| 3712 |
+
"grad_norm": 1.6875,
|
| 3713 |
+
"learning_rate": 7.66742644315666e-06,
|
| 3714 |
+
"loss": 1.0759464502334595,
|
| 3715 |
+
"step": 527
|
| 3716 |
+
},
|
| 3717 |
+
{
|
| 3718 |
+
"epoch": 0.43323076923076925,
|
| 3719 |
+
"grad_norm": 1.578125,
|
| 3720 |
+
"learning_rate": 7.65169040623133e-06,
|
| 3721 |
+
"loss": 1.0863559246063232,
|
| 3722 |
+
"step": 528
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 0.4340512820512821,
|
| 3726 |
+
"grad_norm": 1.890625,
|
| 3727 |
+
"learning_rate": 7.63594446914515e-06,
|
| 3728 |
+
"loss": 1.1177570819854736,
|
| 3729 |
+
"step": 529
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 0.4348717948717949,
|
| 3733 |
+
"grad_norm": 1.71875,
|
| 3734 |
+
"learning_rate": 7.620188739281428e-06,
|
| 3735 |
+
"loss": 1.1569801568984985,
|
| 3736 |
+
"step": 530
|
| 3737 |
+
},
|
| 3738 |
+
{
|
| 3739 |
+
"epoch": 0.4356923076923077,
|
| 3740 |
+
"grad_norm": 1.6171875,
|
| 3741 |
+
"learning_rate": 7.604423324090247e-06,
|
| 3742 |
+
"loss": 1.139686942100525,
|
| 3743 |
+
"step": 531
|
| 3744 |
+
},
|
| 3745 |
+
{
|
| 3746 |
+
"epoch": 0.43651282051282053,
|
| 3747 |
+
"grad_norm": 1.7109375,
|
| 3748 |
+
"learning_rate": 7.588648331087744e-06,
|
| 3749 |
+
"loss": 1.0584099292755127,
|
| 3750 |
+
"step": 532
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 0.43733333333333335,
|
| 3754 |
+
"grad_norm": 1.546875,
|
| 3755 |
+
"learning_rate": 7.572863867855377e-06,
|
| 3756 |
+
"loss": 1.1394144296646118,
|
| 3757 |
+
"step": 533
|
| 3758 |
+
},
|
| 3759 |
+
{
|
| 3760 |
+
"epoch": 0.43815384615384617,
|
| 3761 |
+
"grad_norm": 1.7890625,
|
| 3762 |
+
"learning_rate": 7.557070042039188e-06,
|
| 3763 |
+
"loss": 1.2172819375991821,
|
| 3764 |
+
"step": 534
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 0.438974358974359,
|
| 3768 |
+
"grad_norm": 1.6015625,
|
| 3769 |
+
"learning_rate": 7.54126696134907e-06,
|
| 3770 |
+
"loss": 1.1586464643478394,
|
| 3771 |
+
"step": 535
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 0.4397948717948718,
|
| 3775 |
+
"grad_norm": 1.6875,
|
| 3776 |
+
"learning_rate": 7.525454733558024e-06,
|
| 3777 |
+
"loss": 1.0779222249984741,
|
| 3778 |
+
"step": 536
|
| 3779 |
+
},
|
| 3780 |
+
{
|
| 3781 |
+
"epoch": 0.44061538461538463,
|
| 3782 |
+
"grad_norm": 1.75,
|
| 3783 |
+
"learning_rate": 7.509633466501443e-06,
|
| 3784 |
+
"loss": 1.0934256315231323,
|
| 3785 |
+
"step": 537
|
| 3786 |
+
},
|
| 3787 |
+
{
|
| 3788 |
+
"epoch": 0.44143589743589745,
|
| 3789 |
+
"grad_norm": 1.53125,
|
| 3790 |
+
"learning_rate": 7.49380326807636e-06,
|
| 3791 |
+
"loss": 1.0459587574005127,
|
| 3792 |
+
"step": 538
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 0.44225641025641027,
|
| 3796 |
+
"grad_norm": 1.75,
|
| 3797 |
+
"learning_rate": 7.477964246240717e-06,
|
| 3798 |
+
"loss": 1.1203221082687378,
|
| 3799 |
+
"step": 539
|
| 3800 |
+
},
|
| 3801 |
+
{
|
| 3802 |
+
"epoch": 0.4430769230769231,
|
| 3803 |
+
"grad_norm": 1.5,
|
| 3804 |
+
"learning_rate": 7.462116509012633e-06,
|
| 3805 |
+
"loss": 1.1050634384155273,
|
| 3806 |
+
"step": 540
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 0.4438974358974359,
|
| 3810 |
+
"grad_norm": 1.6796875,
|
| 3811 |
+
"learning_rate": 7.446260164469659e-06,
|
| 3812 |
+
"loss": 1.0807363986968994,
|
| 3813 |
+
"step": 541
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 0.44471794871794873,
|
| 3817 |
+
"grad_norm": 1.5703125,
|
| 3818 |
+
"learning_rate": 7.43039532074805e-06,
|
| 3819 |
+
"loss": 1.0424985885620117,
|
| 3820 |
+
"step": 542
|
| 3821 |
+
},
|
| 3822 |
+
{
|
| 3823 |
+
"epoch": 0.44553846153846155,
|
| 3824 |
+
"grad_norm": 1.6875,
|
| 3825 |
+
"learning_rate": 7.414522086042021e-06,
|
| 3826 |
+
"loss": 1.071103572845459,
|
| 3827 |
+
"step": 543
|
| 3828 |
+
},
|
| 3829 |
+
{
|
| 3830 |
+
"epoch": 0.44635897435897437,
|
| 3831 |
+
"grad_norm": 1.65625,
|
| 3832 |
+
"learning_rate": 7.3986405686030125e-06,
|
| 3833 |
+
"loss": 1.2091856002807617,
|
| 3834 |
+
"step": 544
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 0.4471794871794872,
|
| 3838 |
+
"grad_norm": 1.609375,
|
| 3839 |
+
"learning_rate": 7.38275087673895e-06,
|
| 3840 |
+
"loss": 1.0825698375701904,
|
| 3841 |
+
"step": 545
|
| 3842 |
+
},
|
| 3843 |
+
{
|
| 3844 |
+
"epoch": 0.448,
|
| 3845 |
+
"grad_norm": 1.6484375,
|
| 3846 |
+
"learning_rate": 7.366853118813507e-06,
|
| 3847 |
+
"loss": 1.1572418212890625,
|
| 3848 |
+
"step": 546
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 0.44882051282051283,
|
| 3852 |
+
"grad_norm": 1.625,
|
| 3853 |
+
"learning_rate": 7.350947403245365e-06,
|
| 3854 |
+
"loss": 1.0924804210662842,
|
| 3855 |
+
"step": 547
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 0.44964102564102565,
|
| 3859 |
+
"grad_norm": 1.71875,
|
| 3860 |
+
"learning_rate": 7.335033838507475e-06,
|
| 3861 |
+
"loss": 1.0889440774917603,
|
| 3862 |
+
"step": 548
|
| 3863 |
+
},
|
| 3864 |
+
{
|
| 3865 |
+
"epoch": 0.45046153846153847,
|
| 3866 |
+
"grad_norm": 1.5390625,
|
| 3867 |
+
"learning_rate": 7.3191125331263184e-06,
|
| 3868 |
+
"loss": 1.1945247650146484,
|
| 3869 |
+
"step": 549
|
| 3870 |
+
},
|
| 3871 |
+
{
|
| 3872 |
+
"epoch": 0.4512820512820513,
|
| 3873 |
+
"grad_norm": 1.5625,
|
| 3874 |
+
"learning_rate": 7.303183595681165e-06,
|
| 3875 |
+
"loss": 1.158640742301941,
|
| 3876 |
+
"step": 550
|
| 3877 |
+
},
|
| 3878 |
+
{
|
| 3879 |
+
"epoch": 0.4521025641025641,
|
| 3880 |
+
"grad_norm": 1.453125,
|
| 3881 |
+
"learning_rate": 7.287247134803335e-06,
|
| 3882 |
+
"loss": 1.0312163829803467,
|
| 3883 |
+
"step": 551
|
| 3884 |
+
},
|
| 3885 |
+
{
|
| 3886 |
+
"epoch": 0.45292307692307693,
|
| 3887 |
+
"grad_norm": 1.796875,
|
| 3888 |
+
"learning_rate": 7.2713032591754496e-06,
|
| 3889 |
+
"loss": 1.1686253547668457,
|
| 3890 |
+
"step": 552
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 0.45374358974358975,
|
| 3894 |
+
"grad_norm": 1.71875,
|
| 3895 |
+
"learning_rate": 7.2553520775307065e-06,
|
| 3896 |
+
"loss": 1.1563432216644287,
|
| 3897 |
+
"step": 553
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 0.45456410256410257,
|
| 3901 |
+
"grad_norm": 1.8359375,
|
| 3902 |
+
"learning_rate": 7.2393936986521225e-06,
|
| 3903 |
+
"loss": 1.1137615442276,
|
| 3904 |
+
"step": 554
|
| 3905 |
+
},
|
| 3906 |
+
{
|
| 3907 |
+
"epoch": 0.4553846153846154,
|
| 3908 |
+
"grad_norm": 1.484375,
|
| 3909 |
+
"learning_rate": 7.2234282313718e-06,
|
| 3910 |
+
"loss": 1.126206398010254,
|
| 3911 |
+
"step": 555
|
| 3912 |
+
},
|
| 3913 |
+
{
|
| 3914 |
+
"epoch": 0.4562051282051282,
|
| 3915 |
+
"grad_norm": 1.796875,
|
| 3916 |
+
"learning_rate": 7.207455784570181e-06,
|
| 3917 |
+
"loss": 1.1172479391098022,
|
| 3918 |
+
"step": 556
|
| 3919 |
+
},
|
| 3920 |
+
{
|
| 3921 |
+
"epoch": 0.457025641025641,
|
| 3922 |
+
"grad_norm": 1.578125,
|
| 3923 |
+
"learning_rate": 7.1914764671753044e-06,
|
| 3924 |
+
"loss": 1.0680476427078247,
|
| 3925 |
+
"step": 557
|
| 3926 |
+
},
|
| 3927 |
+
{
|
| 3928 |
+
"epoch": 0.45784615384615385,
|
| 3929 |
+
"grad_norm": 1.796875,
|
| 3930 |
+
"learning_rate": 7.175490388162072e-06,
|
| 3931 |
+
"loss": 1.1442821025848389,
|
| 3932 |
+
"step": 558
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 0.45866666666666667,
|
| 3936 |
+
"grad_norm": 1.6796875,
|
| 3937 |
+
"learning_rate": 7.159497656551489e-06,
|
| 3938 |
+
"loss": 1.1398534774780273,
|
| 3939 |
+
"step": 559
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 0.4594871794871795,
|
| 3943 |
+
"grad_norm": 1.59375,
|
| 3944 |
+
"learning_rate": 7.143498381409937e-06,
|
| 3945 |
+
"loss": 1.2173717021942139,
|
| 3946 |
+
"step": 560
|
| 3947 |
+
},
|
| 3948 |
+
{
|
| 3949 |
+
"epoch": 0.4603076923076923,
|
| 3950 |
+
"grad_norm": 1.5703125,
|
| 3951 |
+
"learning_rate": 7.127492671848417e-06,
|
| 3952 |
+
"loss": 1.1026802062988281,
|
| 3953 |
+
"step": 561
|
| 3954 |
+
},
|
| 3955 |
+
{
|
| 3956 |
+
"epoch": 0.4611282051282051,
|
| 3957 |
+
"grad_norm": 1.5625,
|
| 3958 |
+
"learning_rate": 7.111480637021816e-06,
|
| 3959 |
+
"loss": 1.122018575668335,
|
| 3960 |
+
"step": 562
|
| 3961 |
+
},
|
| 3962 |
+
{
|
| 3963 |
+
"epoch": 0.46194871794871795,
|
| 3964 |
+
"grad_norm": 1.4609375,
|
| 3965 |
+
"learning_rate": 7.095462386128156e-06,
|
| 3966 |
+
"loss": 1.099457025527954,
|
| 3967 |
+
"step": 563
|
| 3968 |
+
},
|
| 3969 |
+
{
|
| 3970 |
+
"epoch": 0.46276923076923077,
|
| 3971 |
+
"grad_norm": 1.5546875,
|
| 3972 |
+
"learning_rate": 7.079438028407847e-06,
|
| 3973 |
+
"loss": 1.0360870361328125,
|
| 3974 |
+
"step": 564
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 0.4635897435897436,
|
| 3978 |
+
"grad_norm": 1.65625,
|
| 3979 |
+
"learning_rate": 7.063407673142953e-06,
|
| 3980 |
+
"loss": 1.1035900115966797,
|
| 3981 |
+
"step": 565
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 0.4644102564102564,
|
| 3985 |
+
"grad_norm": 1.765625,
|
| 3986 |
+
"learning_rate": 7.047371429656435e-06,
|
| 3987 |
+
"loss": 1.0730888843536377,
|
| 3988 |
+
"step": 566
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 0.4652307692307692,
|
| 3992 |
+
"grad_norm": 1.5703125,
|
| 3993 |
+
"learning_rate": 7.031329407311412e-06,
|
| 3994 |
+
"loss": 1.134545087814331,
|
| 3995 |
+
"step": 567
|
| 3996 |
+
},
|
| 3997 |
+
{
|
| 3998 |
+
"epoch": 0.46605128205128205,
|
| 3999 |
+
"grad_norm": 1.59375,
|
| 4000 |
+
"learning_rate": 7.015281715510413e-06,
|
| 4001 |
+
"loss": 1.0366904735565186,
|
| 4002 |
+
"step": 568
|
| 4003 |
+
},
|
| 4004 |
+
{
|
| 4005 |
+
"epoch": 0.46687179487179487,
|
| 4006 |
+
"grad_norm": 1.765625,
|
| 4007 |
+
"learning_rate": 6.999228463694629e-06,
|
| 4008 |
+
"loss": 1.1620510816574097,
|
| 4009 |
+
"step": 569
|
| 4010 |
+
},
|
| 4011 |
+
{
|
| 4012 |
+
"epoch": 0.4676923076923077,
|
| 4013 |
+
"grad_norm": 1.6953125,
|
| 4014 |
+
"learning_rate": 6.983169761343176e-06,
|
| 4015 |
+
"loss": 1.0937600135803223,
|
| 4016 |
+
"step": 570
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 0.4685128205128205,
|
| 4020 |
+
"grad_norm": 1.6796875,
|
| 4021 |
+
"learning_rate": 6.9671057179723295e-06,
|
| 4022 |
+
"loss": 1.1628119945526123,
|
| 4023 |
+
"step": 571
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 0.4693333333333333,
|
| 4027 |
+
"grad_norm": 1.6484375,
|
| 4028 |
+
"learning_rate": 6.951036443134799e-06,
|
| 4029 |
+
"loss": 1.1623189449310303,
|
| 4030 |
+
"step": 572
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 0.47015384615384614,
|
| 4034 |
+
"grad_norm": 1.5625,
|
| 4035 |
+
"learning_rate": 6.934962046418972e-06,
|
| 4036 |
+
"loss": 1.1499994993209839,
|
| 4037 |
+
"step": 573
|
| 4038 |
+
},
|
| 4039 |
+
{
|
| 4040 |
+
"epoch": 0.47097435897435896,
|
| 4041 |
+
"grad_norm": 1.640625,
|
| 4042 |
+
"learning_rate": 6.918882637448158e-06,
|
| 4043 |
+
"loss": 1.1366188526153564,
|
| 4044 |
+
"step": 574
|
| 4045 |
+
},
|
| 4046 |
+
{
|
| 4047 |
+
"epoch": 0.4717948717948718,
|
| 4048 |
+
"grad_norm": 1.5390625,
|
| 4049 |
+
"learning_rate": 6.902798325879854e-06,
|
| 4050 |
+
"loss": 1.1301484107971191,
|
| 4051 |
+
"step": 575
|
| 4052 |
+
},
|
| 4053 |
+
{
|
| 4054 |
+
"epoch": 0.4726153846153846,
|
| 4055 |
+
"grad_norm": 1.65625,
|
| 4056 |
+
"learning_rate": 6.886709221404992e-06,
|
| 4057 |
+
"loss": 1.0446761846542358,
|
| 4058 |
+
"step": 576
|
| 4059 |
+
},
|
| 4060 |
+
{
|
| 4061 |
+
"epoch": 0.4734358974358974,
|
| 4062 |
+
"grad_norm": 1.5234375,
|
| 4063 |
+
"learning_rate": 6.870615433747187e-06,
|
| 4064 |
+
"loss": 1.1570371389389038,
|
| 4065 |
+
"step": 577
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 0.47425641025641024,
|
| 4069 |
+
"grad_norm": 1.640625,
|
| 4070 |
+
"learning_rate": 6.854517072661996e-06,
|
| 4071 |
+
"loss": 1.1273854970932007,
|
| 4072 |
+
"step": 578
|
| 4073 |
+
},
|
| 4074 |
+
{
|
| 4075 |
+
"epoch": 0.47507692307692306,
|
| 4076 |
+
"grad_norm": 1.7421875,
|
| 4077 |
+
"learning_rate": 6.838414247936163e-06,
|
| 4078 |
+
"loss": 1.1674871444702148,
|
| 4079 |
+
"step": 579
|
| 4080 |
+
},
|
| 4081 |
+
{
|
| 4082 |
+
"epoch": 0.4758974358974359,
|
| 4083 |
+
"grad_norm": 1.625,
|
| 4084 |
+
"learning_rate": 6.822307069386875e-06,
|
| 4085 |
+
"loss": 1.1848071813583374,
|
| 4086 |
+
"step": 580
|
| 4087 |
+
},
|
| 4088 |
+
{
|
| 4089 |
+
"epoch": 0.4767179487179487,
|
| 4090 |
+
"grad_norm": 1.453125,
|
| 4091 |
+
"learning_rate": 6.806195646861008e-06,
|
| 4092 |
+
"loss": 1.091614842414856,
|
| 4093 |
+
"step": 581
|
| 4094 |
+
},
|
| 4095 |
+
{
|
| 4096 |
+
"epoch": 0.4775384615384615,
|
| 4097 |
+
"grad_norm": 1.46875,
|
| 4098 |
+
"learning_rate": 6.790080090234381e-06,
|
| 4099 |
+
"loss": 1.1283140182495117,
|
| 4100 |
+
"step": 582
|
| 4101 |
+
},
|
| 4102 |
+
{
|
| 4103 |
+
"epoch": 0.47835897435897434,
|
| 4104 |
+
"grad_norm": 1.671875,
|
| 4105 |
+
"learning_rate": 6.7739605094110105e-06,
|
| 4106 |
+
"loss": 1.1362240314483643,
|
| 4107 |
+
"step": 583
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 0.47917948717948716,
|
| 4111 |
+
"grad_norm": 1.6484375,
|
| 4112 |
+
"learning_rate": 6.757837014322356e-06,
|
| 4113 |
+
"loss": 1.1810652017593384,
|
| 4114 |
+
"step": 584
|
| 4115 |
+
},
|
| 4116 |
+
{
|
| 4117 |
+
"epoch": 0.48,
|
| 4118 |
+
"grad_norm": 1.7421875,
|
| 4119 |
+
"learning_rate": 6.741709714926567e-06,
|
| 4120 |
+
"loss": 1.1747609376907349,
|
| 4121 |
+
"step": 585
|
| 4122 |
+
},
|
| 4123 |
+
{
|
| 4124 |
+
"epoch": 0.4808205128205128,
|
| 4125 |
+
"grad_norm": 1.625,
|
| 4126 |
+
"learning_rate": 6.725578721207739e-06,
|
| 4127 |
+
"loss": 1.1179296970367432,
|
| 4128 |
+
"step": 586
|
| 4129 |
+
},
|
| 4130 |
+
{
|
| 4131 |
+
"epoch": 0.4816410256410256,
|
| 4132 |
+
"grad_norm": 1.453125,
|
| 4133 |
+
"learning_rate": 6.709444143175165e-06,
|
| 4134 |
+
"loss": 1.0193846225738525,
|
| 4135 |
+
"step": 587
|
| 4136 |
+
},
|
| 4137 |
+
{
|
| 4138 |
+
"epoch": 0.48246153846153844,
|
| 4139 |
+
"grad_norm": 1.8671875,
|
| 4140 |
+
"learning_rate": 6.693306090862583e-06,
|
| 4141 |
+
"loss": 1.1453070640563965,
|
| 4142 |
+
"step": 588
|
| 4143 |
+
},
|
| 4144 |
+
{
|
| 4145 |
+
"epoch": 0.48328205128205126,
|
| 4146 |
+
"grad_norm": 1.8125,
|
| 4147 |
+
"learning_rate": 6.677164674327418e-06,
|
| 4148 |
+
"loss": 1.0982394218444824,
|
| 4149 |
+
"step": 589
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 0.4841025641025641,
|
| 4153 |
+
"grad_norm": 1.5234375,
|
| 4154 |
+
"learning_rate": 6.661020003650043e-06,
|
| 4155 |
+
"loss": 1.098716139793396,
|
| 4156 |
+
"step": 590
|
| 4157 |
+
},
|
| 4158 |
+
{
|
| 4159 |
+
"epoch": 0.4849230769230769,
|
| 4160 |
+
"grad_norm": 1.5859375,
|
| 4161 |
+
"learning_rate": 6.644872188933022e-06,
|
| 4162 |
+
"loss": 1.1306407451629639,
|
| 4163 |
+
"step": 591
|
| 4164 |
+
},
|
| 4165 |
+
{
|
| 4166 |
+
"epoch": 0.4857435897435897,
|
| 4167 |
+
"grad_norm": 1.5703125,
|
| 4168 |
+
"learning_rate": 6.628721340300363e-06,
|
| 4169 |
+
"loss": 1.170407772064209,
|
| 4170 |
+
"step": 592
|
| 4171 |
+
},
|
| 4172 |
+
{
|
| 4173 |
+
"epoch": 0.48656410256410254,
|
| 4174 |
+
"grad_norm": 1.6015625,
|
| 4175 |
+
"learning_rate": 6.6125675678967595e-06,
|
| 4176 |
+
"loss": 1.093729019165039,
|
| 4177 |
+
"step": 593
|
| 4178 |
+
},
|
| 4179 |
+
{
|
| 4180 |
+
"epoch": 0.48738461538461536,
|
| 4181 |
+
"grad_norm": 1.6015625,
|
| 4182 |
+
"learning_rate": 6.59641098188685e-06,
|
| 4183 |
+
"loss": 1.1293773651123047,
|
| 4184 |
+
"step": 594
|
| 4185 |
+
},
|
| 4186 |
+
{
|
| 4187 |
+
"epoch": 0.4882051282051282,
|
| 4188 |
+
"grad_norm": 1.65625,
|
| 4189 |
+
"learning_rate": 6.580251692454459e-06,
|
| 4190 |
+
"loss": 1.0571609735488892,
|
| 4191 |
+
"step": 595
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 0.489025641025641,
|
| 4195 |
+
"grad_norm": 1.703125,
|
| 4196 |
+
"learning_rate": 6.564089809801844e-06,
|
| 4197 |
+
"loss": 1.2081211805343628,
|
| 4198 |
+
"step": 596
|
| 4199 |
+
},
|
| 4200 |
+
{
|
| 4201 |
+
"epoch": 0.4898461538461538,
|
| 4202 |
+
"grad_norm": 1.78125,
|
| 4203 |
+
"learning_rate": 6.547925444148952e-06,
|
| 4204 |
+
"loss": 1.2445255517959595,
|
| 4205 |
+
"step": 597
|
| 4206 |
+
},
|
| 4207 |
+
{
|
| 4208 |
+
"epoch": 0.49066666666666664,
|
| 4209 |
+
"grad_norm": 1.6015625,
|
| 4210 |
+
"learning_rate": 6.5317587057326635e-06,
|
| 4211 |
+
"loss": 1.1090102195739746,
|
| 4212 |
+
"step": 598
|
| 4213 |
+
},
|
| 4214 |
+
{
|
| 4215 |
+
"epoch": 0.49148717948717946,
|
| 4216 |
+
"grad_norm": 1.671875,
|
| 4217 |
+
"learning_rate": 6.515589704806038e-06,
|
| 4218 |
+
"loss": 1.2049899101257324,
|
| 4219 |
+
"step": 599
|
| 4220 |
+
},
|
| 4221 |
+
{
|
| 4222 |
+
"epoch": 0.49230769230769234,
|
| 4223 |
+
"grad_norm": 1.8671875,
|
| 4224 |
+
"learning_rate": 6.49941855163757e-06,
|
| 4225 |
+
"loss": 1.1750158071517944,
|
| 4226 |
+
"step": 600
|
| 4227 |
+
},
|
| 4228 |
+
{
|
| 4229 |
+
"epoch": 0.49312820512820515,
|
| 4230 |
+
"grad_norm": 1.640625,
|
| 4231 |
+
"learning_rate": 6.483245356510421e-06,
|
| 4232 |
+
"loss": 1.118952989578247,
|
| 4233 |
+
"step": 601
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 0.493948717948718,
|
| 4237 |
+
"grad_norm": 1.6171875,
|
| 4238 |
+
"learning_rate": 6.46707022972169e-06,
|
| 4239 |
+
"loss": 1.1094834804534912,
|
| 4240 |
+
"step": 602
|
| 4241 |
+
},
|
| 4242 |
+
{
|
| 4243 |
+
"epoch": 0.4947692307692308,
|
| 4244 |
+
"grad_norm": 1.7109375,
|
| 4245 |
+
"learning_rate": 6.450893281581643e-06,
|
| 4246 |
+
"loss": 1.1568554639816284,
|
| 4247 |
+
"step": 603
|
| 4248 |
+
},
|
| 4249 |
+
{
|
| 4250 |
+
"epoch": 0.4955897435897436,
|
| 4251 |
+
"grad_norm": 1.7421875,
|
| 4252 |
+
"learning_rate": 6.434714622412969e-06,
|
| 4253 |
+
"loss": 1.0595741271972656,
|
| 4254 |
+
"step": 604
|
| 4255 |
+
},
|
| 4256 |
+
{
|
| 4257 |
+
"epoch": 0.49641025641025643,
|
| 4258 |
+
"grad_norm": 1.8125,
|
| 4259 |
+
"learning_rate": 6.418534362550025e-06,
|
| 4260 |
+
"loss": 1.1385390758514404,
|
| 4261 |
+
"step": 605
|
| 4262 |
+
},
|
| 4263 |
+
{
|
| 4264 |
+
"epoch": 0.49723076923076925,
|
| 4265 |
+
"grad_norm": 1.65625,
|
| 4266 |
+
"learning_rate": 6.402352612338083e-06,
|
| 4267 |
+
"loss": 1.1630690097808838,
|
| 4268 |
+
"step": 606
|
| 4269 |
+
},
|
| 4270 |
+
{
|
| 4271 |
+
"epoch": 0.4980512820512821,
|
| 4272 |
+
"grad_norm": 1.5234375,
|
| 4273 |
+
"learning_rate": 6.3861694821325835e-06,
|
| 4274 |
+
"loss": 1.1319586038589478,
|
| 4275 |
+
"step": 607
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 0.4988717948717949,
|
| 4279 |
+
"grad_norm": 1.7265625,
|
| 4280 |
+
"learning_rate": 6.369985082298375e-06,
|
| 4281 |
+
"loss": 1.1615095138549805,
|
| 4282 |
+
"step": 608
|
| 4283 |
+
},
|
| 4284 |
+
{
|
| 4285 |
+
"epoch": 0.4996923076923077,
|
| 4286 |
+
"grad_norm": 1.6328125,
|
| 4287 |
+
"learning_rate": 6.353799523208961e-06,
|
| 4288 |
+
"loss": 1.1865564584732056,
|
| 4289 |
+
"step": 609
|
| 4290 |
+
},
|
| 4291 |
+
{
|
| 4292 |
+
"epoch": 0.5005128205128205,
|
| 4293 |
+
"grad_norm": 1.765625,
|
| 4294 |
+
"learning_rate": 6.3376129152457595e-06,
|
| 4295 |
+
"loss": 1.2127357721328735,
|
| 4296 |
+
"step": 610
|
| 4297 |
+
},
|
| 4298 |
+
{
|
| 4299 |
+
"epoch": 0.5013333333333333,
|
| 4300 |
+
"grad_norm": 1.8046875,
|
| 4301 |
+
"learning_rate": 6.321425368797333e-06,
|
| 4302 |
+
"loss": 1.1961997747421265,
|
| 4303 |
+
"step": 611
|
| 4304 |
+
},
|
| 4305 |
+
{
|
| 4306 |
+
"epoch": 0.5021538461538462,
|
| 4307 |
+
"grad_norm": 1.5625,
|
| 4308 |
+
"learning_rate": 6.305236994258648e-06,
|
| 4309 |
+
"loss": 1.070225715637207,
|
| 4310 |
+
"step": 612
|
| 4311 |
+
},
|
| 4312 |
+
{
|
| 4313 |
+
"epoch": 0.5029743589743589,
|
| 4314 |
+
"grad_norm": 1.5859375,
|
| 4315 |
+
"learning_rate": 6.289047902030322e-06,
|
| 4316 |
+
"loss": 1.1783050298690796,
|
| 4317 |
+
"step": 613
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 0.5037948717948718,
|
| 4321 |
+
"grad_norm": 1.4921875,
|
| 4322 |
+
"learning_rate": 6.2728582025178585e-06,
|
| 4323 |
+
"loss": 1.181092619895935,
|
| 4324 |
+
"step": 614
|
| 4325 |
+
},
|
| 4326 |
+
{
|
| 4327 |
+
"epoch": 0.5046153846153846,
|
| 4328 |
+
"grad_norm": 1.765625,
|
| 4329 |
+
"learning_rate": 6.25666800613091e-06,
|
| 4330 |
+
"loss": 1.1656084060668945,
|
| 4331 |
+
"step": 615
|
| 4332 |
+
},
|
| 4333 |
+
{
|
| 4334 |
+
"epoch": 0.5054358974358975,
|
| 4335 |
+
"grad_norm": 1.6796875,
|
| 4336 |
+
"learning_rate": 6.240477423282515e-06,
|
| 4337 |
+
"loss": 1.087341070175171,
|
| 4338 |
+
"step": 616
|
| 4339 |
+
},
|
| 4340 |
+
{
|
| 4341 |
+
"epoch": 0.5062564102564102,
|
| 4342 |
+
"grad_norm": 1.6484375,
|
| 4343 |
+
"learning_rate": 6.224286564388345e-06,
|
| 4344 |
+
"loss": 1.1174447536468506,
|
| 4345 |
+
"step": 617
|
| 4346 |
+
},
|
| 4347 |
+
{
|
| 4348 |
+
"epoch": 0.5070769230769231,
|
| 4349 |
+
"grad_norm": 1.578125,
|
| 4350 |
+
"learning_rate": 6.208095539865958e-06,
|
| 4351 |
+
"loss": 1.229354739189148,
|
| 4352 |
+
"step": 618
|
| 4353 |
+
},
|
| 4354 |
+
{
|
| 4355 |
+
"epoch": 0.5078974358974359,
|
| 4356 |
+
"grad_norm": 1.796875,
|
| 4357 |
+
"learning_rate": 6.191904460134042e-06,
|
| 4358 |
+
"loss": 1.1044552326202393,
|
| 4359 |
+
"step": 619
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 0.5087179487179487,
|
| 4363 |
+
"grad_norm": 1.6875,
|
| 4364 |
+
"learning_rate": 6.175713435611656e-06,
|
| 4365 |
+
"loss": 1.0737800598144531,
|
| 4366 |
+
"step": 620
|
| 4367 |
+
},
|
| 4368 |
+
{
|
| 4369 |
+
"epoch": 0.5095384615384615,
|
| 4370 |
+
"grad_norm": 1.6796875,
|
| 4371 |
+
"learning_rate": 6.159522576717486e-06,
|
| 4372 |
+
"loss": 1.077164888381958,
|
| 4373 |
+
"step": 621
|
| 4374 |
+
},
|
| 4375 |
+
{
|
| 4376 |
+
"epoch": 0.5103589743589744,
|
| 4377 |
+
"grad_norm": 1.6796875,
|
| 4378 |
+
"learning_rate": 6.14333199386909e-06,
|
| 4379 |
+
"loss": 1.1477549076080322,
|
| 4380 |
+
"step": 622
|
| 4381 |
+
},
|
| 4382 |
+
{
|
| 4383 |
+
"epoch": 0.5111794871794871,
|
| 4384 |
+
"grad_norm": 1.65625,
|
| 4385 |
+
"learning_rate": 6.127141797482141e-06,
|
| 4386 |
+
"loss": 1.101395845413208,
|
| 4387 |
+
"step": 623
|
| 4388 |
+
},
|
| 4389 |
+
{
|
| 4390 |
+
"epoch": 0.512,
|
| 4391 |
+
"grad_norm": 1.53125,
|
| 4392 |
+
"learning_rate": 6.1109520979696806e-06,
|
| 4393 |
+
"loss": 1.0559899806976318,
|
| 4394 |
+
"step": 624
|
| 4395 |
+
},
|
| 4396 |
+
{
|
| 4397 |
+
"epoch": 0.5128205128205128,
|
| 4398 |
+
"grad_norm": 1.6640625,
|
| 4399 |
+
"learning_rate": 6.094763005741352e-06,
|
| 4400 |
+
"loss": 1.1552640199661255,
|
| 4401 |
+
"step": 625
|
| 4402 |
+
},
|
| 4403 |
+
{
|
| 4404 |
+
"epoch": 0.5136410256410257,
|
| 4405 |
+
"grad_norm": 1.5,
|
| 4406 |
+
"learning_rate": 6.0785746312026676e-06,
|
| 4407 |
+
"loss": 1.0766817331314087,
|
| 4408 |
+
"step": 626
|
| 4409 |
+
},
|
| 4410 |
+
{
|
| 4411 |
+
"epoch": 0.5144615384615384,
|
| 4412 |
+
"grad_norm": 1.75,
|
| 4413 |
+
"learning_rate": 6.062387084754242e-06,
|
| 4414 |
+
"loss": 1.1779124736785889,
|
| 4415 |
+
"step": 627
|
| 4416 |
+
},
|
| 4417 |
+
{
|
| 4418 |
+
"epoch": 0.5152820512820513,
|
| 4419 |
+
"grad_norm": 1.609375,
|
| 4420 |
+
"learning_rate": 6.046200476791038e-06,
|
| 4421 |
+
"loss": 1.0955864191055298,
|
| 4422 |
+
"step": 628
|
| 4423 |
+
},
|
| 4424 |
+
{
|
| 4425 |
+
"epoch": 0.516102564102564,
|
| 4426 |
+
"grad_norm": 1.5625,
|
| 4427 |
+
"learning_rate": 6.0300149177016275e-06,
|
| 4428 |
+
"loss": 1.0860978364944458,
|
| 4429 |
+
"step": 629
|
| 4430 |
+
},
|
| 4431 |
+
{
|
| 4432 |
+
"epoch": 0.5169230769230769,
|
| 4433 |
+
"grad_norm": 1.6015625,
|
| 4434 |
+
"learning_rate": 6.013830517867416e-06,
|
| 4435 |
+
"loss": 1.1335543394088745,
|
| 4436 |
+
"step": 630
|
| 4437 |
+
},
|
| 4438 |
+
{
|
| 4439 |
+
"epoch": 0.5177435897435897,
|
| 4440 |
+
"grad_norm": 1.859375,
|
| 4441 |
+
"learning_rate": 5.997647387661917e-06,
|
| 4442 |
+
"loss": 1.2005040645599365,
|
| 4443 |
+
"step": 631
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 0.5185641025641026,
|
| 4447 |
+
"grad_norm": 1.6875,
|
| 4448 |
+
"learning_rate": 5.981465637449975e-06,
|
| 4449 |
+
"loss": 1.151053547859192,
|
| 4450 |
+
"step": 632
|
| 4451 |
+
},
|
| 4452 |
+
{
|
| 4453 |
+
"epoch": 0.5193846153846153,
|
| 4454 |
+
"grad_norm": 1.78125,
|
| 4455 |
+
"learning_rate": 5.96528537758703e-06,
|
| 4456 |
+
"loss": 1.087674856185913,
|
| 4457 |
+
"step": 633
|
| 4458 |
+
},
|
| 4459 |
+
{
|
| 4460 |
+
"epoch": 0.5202051282051282,
|
| 4461 |
+
"grad_norm": 1.6171875,
|
| 4462 |
+
"learning_rate": 5.9491067184183576e-06,
|
| 4463 |
+
"loss": 1.107380747795105,
|
| 4464 |
+
"step": 634
|
| 4465 |
+
},
|
| 4466 |
+
{
|
| 4467 |
+
"epoch": 0.521025641025641,
|
| 4468 |
+
"grad_norm": 1.6328125,
|
| 4469 |
+
"learning_rate": 5.932929770278311e-06,
|
| 4470 |
+
"loss": 1.1303094625473022,
|
| 4471 |
+
"step": 635
|
| 4472 |
+
},
|
| 4473 |
+
{
|
| 4474 |
+
"epoch": 0.5218461538461538,
|
| 4475 |
+
"grad_norm": 1.640625,
|
| 4476 |
+
"learning_rate": 5.916754643489579e-06,
|
| 4477 |
+
"loss": 1.2045444250106812,
|
| 4478 |
+
"step": 636
|
| 4479 |
+
},
|
| 4480 |
+
{
|
| 4481 |
+
"epoch": 0.5226666666666666,
|
| 4482 |
+
"grad_norm": 1.671875,
|
| 4483 |
+
"learning_rate": 5.9005814483624305e-06,
|
| 4484 |
+
"loss": 1.0858845710754395,
|
| 4485 |
+
"step": 637
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 0.5234871794871795,
|
| 4489 |
+
"grad_norm": 1.6484375,
|
| 4490 |
+
"learning_rate": 5.884410295193961e-06,
|
| 4491 |
+
"loss": 1.117233157157898,
|
| 4492 |
+
"step": 638
|
| 4493 |
+
},
|
| 4494 |
+
{
|
| 4495 |
+
"epoch": 0.5243076923076923,
|
| 4496 |
+
"grad_norm": 1.6640625,
|
| 4497 |
+
"learning_rate": 5.868241294267338e-06,
|
| 4498 |
+
"loss": 1.1313074827194214,
|
| 4499 |
+
"step": 639
|
| 4500 |
+
},
|
| 4501 |
+
{
|
| 4502 |
+
"epoch": 0.5251282051282051,
|
| 4503 |
+
"grad_norm": 1.6796875,
|
| 4504 |
+
"learning_rate": 5.852074555851048e-06,
|
| 4505 |
+
"loss": 1.0843769311904907,
|
| 4506 |
+
"step": 640
|
| 4507 |
+
},
|
| 4508 |
+
{
|
| 4509 |
+
"epoch": 0.525948717948718,
|
| 4510 |
+
"grad_norm": 1.765625,
|
| 4511 |
+
"learning_rate": 5.835910190198157e-06,
|
| 4512 |
+
"loss": 1.164339303970337,
|
| 4513 |
+
"step": 641
|
| 4514 |
+
},
|
| 4515 |
+
{
|
| 4516 |
+
"epoch": 0.5267692307692308,
|
| 4517 |
+
"grad_norm": 1.7421875,
|
| 4518 |
+
"learning_rate": 5.819748307545543e-06,
|
| 4519 |
+
"loss": 1.1073044538497925,
|
| 4520 |
+
"step": 642
|
| 4521 |
+
},
|
| 4522 |
+
{
|
| 4523 |
+
"epoch": 0.5275897435897436,
|
| 4524 |
+
"grad_norm": 1.7109375,
|
| 4525 |
+
"learning_rate": 5.80358901811315e-06,
|
| 4526 |
+
"loss": 1.089107632637024,
|
| 4527 |
+
"step": 643
|
| 4528 |
+
},
|
| 4529 |
+
{
|
| 4530 |
+
"epoch": 0.5284102564102564,
|
| 4531 |
+
"grad_norm": 1.640625,
|
| 4532 |
+
"learning_rate": 5.787432432103241e-06,
|
| 4533 |
+
"loss": 1.147425889968872,
|
| 4534 |
+
"step": 644
|
| 4535 |
+
},
|
| 4536 |
+
{
|
| 4537 |
+
"epoch": 0.5292307692307693,
|
| 4538 |
+
"grad_norm": 1.53125,
|
| 4539 |
+
"learning_rate": 5.771278659699638e-06,
|
| 4540 |
+
"loss": 1.107999324798584,
|
| 4541 |
+
"step": 645
|
| 4542 |
+
},
|
| 4543 |
+
{
|
| 4544 |
+
"epoch": 0.530051282051282,
|
| 4545 |
+
"grad_norm": 1.7109375,
|
| 4546 |
+
"learning_rate": 5.7551278110669795e-06,
|
| 4547 |
+
"loss": 1.1329853534698486,
|
| 4548 |
+
"step": 646
|
| 4549 |
+
},
|
| 4550 |
+
{
|
| 4551 |
+
"epoch": 0.5308717948717949,
|
| 4552 |
+
"grad_norm": 1.5625,
|
| 4553 |
+
"learning_rate": 5.738979996349958e-06,
|
| 4554 |
+
"loss": 1.0885168313980103,
|
| 4555 |
+
"step": 647
|
| 4556 |
+
},
|
| 4557 |
+
{
|
| 4558 |
+
"epoch": 0.5316923076923077,
|
| 4559 |
+
"grad_norm": 1.734375,
|
| 4560 |
+
"learning_rate": 5.722835325672581e-06,
|
| 4561 |
+
"loss": 1.063646912574768,
|
| 4562 |
+
"step": 648
|
| 4563 |
+
},
|
| 4564 |
+
{
|
| 4565 |
+
"epoch": 0.5325128205128206,
|
| 4566 |
+
"grad_norm": 1.7109375,
|
| 4567 |
+
"learning_rate": 5.706693909137418e-06,
|
| 4568 |
+
"loss": 1.2062866687774658,
|
| 4569 |
+
"step": 649
|
| 4570 |
+
},
|
| 4571 |
+
{
|
| 4572 |
+
"epoch": 0.5333333333333333,
|
| 4573 |
+
"grad_norm": 1.609375,
|
| 4574 |
+
"learning_rate": 5.690555856824835e-06,
|
| 4575 |
+
"loss": 1.0824732780456543,
|
| 4576 |
+
"step": 650
|
| 4577 |
+
},
|
| 4578 |
+
{
|
| 4579 |
+
"epoch": 0.5341538461538462,
|
| 4580 |
+
"grad_norm": 1.7890625,
|
| 4581 |
+
"learning_rate": 5.674421278792262e-06,
|
| 4582 |
+
"loss": 1.1828659772872925,
|
| 4583 |
+
"step": 651
|
| 4584 |
+
},
|
| 4585 |
+
{
|
| 4586 |
+
"epoch": 0.534974358974359,
|
| 4587 |
+
"grad_norm": 1.78125,
|
| 4588 |
+
"learning_rate": 5.658290285073434e-06,
|
| 4589 |
+
"loss": 1.0778831243515015,
|
| 4590 |
+
"step": 652
|
| 4591 |
+
},
|
| 4592 |
+
{
|
| 4593 |
+
"epoch": 0.5357948717948718,
|
| 4594 |
+
"grad_norm": 1.5859375,
|
| 4595 |
+
"learning_rate": 5.642162985677645e-06,
|
| 4596 |
+
"loss": 1.1457768678665161,
|
| 4597 |
+
"step": 653
|
| 4598 |
+
},
|
| 4599 |
+
{
|
| 4600 |
+
"epoch": 0.5366153846153846,
|
| 4601 |
+
"grad_norm": 1.5625,
|
| 4602 |
+
"learning_rate": 5.62603949058899e-06,
|
| 4603 |
+
"loss": 1.167357325553894,
|
| 4604 |
+
"step": 654
|
| 4605 |
+
},
|
| 4606 |
+
{
|
| 4607 |
+
"epoch": 0.5374358974358975,
|
| 4608 |
+
"grad_norm": 1.8515625,
|
| 4609 |
+
"learning_rate": 5.60991990976562e-06,
|
| 4610 |
+
"loss": 1.0818358659744263,
|
| 4611 |
+
"step": 655
|
| 4612 |
+
},
|
| 4613 |
+
{
|
| 4614 |
+
"epoch": 0.5382564102564102,
|
| 4615 |
+
"grad_norm": 1.546875,
|
| 4616 |
+
"learning_rate": 5.593804353138993e-06,
|
| 4617 |
+
"loss": 1.0808149576187134,
|
| 4618 |
+
"step": 656
|
| 4619 |
+
},
|
| 4620 |
+
{
|
| 4621 |
+
"epoch": 0.5390769230769231,
|
| 4622 |
+
"grad_norm": 1.7265625,
|
| 4623 |
+
"learning_rate": 5.577692930613126e-06,
|
| 4624 |
+
"loss": 1.1308292150497437,
|
| 4625 |
+
"step": 657
|
| 4626 |
+
},
|
| 4627 |
+
{
|
| 4628 |
+
"epoch": 0.5398974358974359,
|
| 4629 |
+
"grad_norm": 1.8046875,
|
| 4630 |
+
"learning_rate": 5.561585752063836e-06,
|
| 4631 |
+
"loss": 1.047197699546814,
|
| 4632 |
+
"step": 658
|
| 4633 |
+
},
|
| 4634 |
+
{
|
| 4635 |
+
"epoch": 0.5407179487179488,
|
| 4636 |
+
"grad_norm": 1.640625,
|
| 4637 |
+
"learning_rate": 5.545482927338005e-06,
|
| 4638 |
+
"loss": 1.1586151123046875,
|
| 4639 |
+
"step": 659
|
| 4640 |
+
},
|
| 4641 |
+
{
|
| 4642 |
+
"epoch": 0.5415384615384615,
|
| 4643 |
+
"grad_norm": 1.5859375,
|
| 4644 |
+
"learning_rate": 5.529384566252813e-06,
|
| 4645 |
+
"loss": 1.0410411357879639,
|
| 4646 |
+
"step": 660
|
| 4647 |
+
},
|
| 4648 |
+
{
|
| 4649 |
+
"epoch": 0.5423589743589744,
|
| 4650 |
+
"grad_norm": 1.65625,
|
| 4651 |
+
"learning_rate": 5.51329077859501e-06,
|
| 4652 |
+
"loss": 1.136472225189209,
|
| 4653 |
+
"step": 661
|
| 4654 |
+
},
|
| 4655 |
+
{
|
| 4656 |
+
"epoch": 0.5431794871794872,
|
| 4657 |
+
"grad_norm": 1.671875,
|
| 4658 |
+
"learning_rate": 5.497201674120146e-06,
|
| 4659 |
+
"loss": 1.1162110567092896,
|
| 4660 |
+
"step": 662
|
| 4661 |
+
},
|
| 4662 |
+
{
|
| 4663 |
+
"epoch": 0.544,
|
| 4664 |
+
"grad_norm": 1.71875,
|
| 4665 |
+
"learning_rate": 5.481117362551841e-06,
|
| 4666 |
+
"loss": 1.211832046508789,
|
| 4667 |
+
"step": 663
|
| 4668 |
+
},
|
| 4669 |
+
{
|
| 4670 |
+
"epoch": 0.5448205128205128,
|
| 4671 |
+
"grad_norm": 1.6796875,
|
| 4672 |
+
"learning_rate": 5.465037953581029e-06,
|
| 4673 |
+
"loss": 1.1091201305389404,
|
| 4674 |
+
"step": 664
|
| 4675 |
+
},
|
| 4676 |
+
{
|
| 4677 |
+
"epoch": 0.5456410256410257,
|
| 4678 |
+
"grad_norm": 1.6953125,
|
| 4679 |
+
"learning_rate": 5.448963556865201e-06,
|
| 4680 |
+
"loss": 1.0828438997268677,
|
| 4681 |
+
"step": 665
|
| 4682 |
+
},
|
| 4683 |
+
{
|
| 4684 |
+
"epoch": 0.5464615384615384,
|
| 4685 |
+
"grad_norm": 1.5625,
|
| 4686 |
+
"learning_rate": 5.432894282027672e-06,
|
| 4687 |
+
"loss": 1.0858161449432373,
|
| 4688 |
+
"step": 666
|
| 4689 |
+
},
|
| 4690 |
+
{
|
| 4691 |
+
"epoch": 0.5472820512820513,
|
| 4692 |
+
"grad_norm": 1.5625,
|
| 4693 |
+
"learning_rate": 5.416830238656826e-06,
|
| 4694 |
+
"loss": 1.1406219005584717,
|
| 4695 |
+
"step": 667
|
| 4696 |
+
},
|
| 4697 |
+
{
|
| 4698 |
+
"epoch": 0.5481025641025641,
|
| 4699 |
+
"grad_norm": 1.65625,
|
| 4700 |
+
"learning_rate": 5.400771536305371e-06,
|
| 4701 |
+
"loss": 1.091049075126648,
|
| 4702 |
+
"step": 668
|
| 4703 |
+
},
|
| 4704 |
+
{
|
| 4705 |
+
"epoch": 0.548923076923077,
|
| 4706 |
+
"grad_norm": 1.8671875,
|
| 4707 |
+
"learning_rate": 5.384718284489589e-06,
|
| 4708 |
+
"loss": 1.1222573518753052,
|
| 4709 |
+
"step": 669
|
| 4710 |
+
},
|
| 4711 |
+
{
|
| 4712 |
+
"epoch": 0.5497435897435897,
|
| 4713 |
+
"grad_norm": 1.5625,
|
| 4714 |
+
"learning_rate": 5.3686705926885895e-06,
|
| 4715 |
+
"loss": 1.2079274654388428,
|
| 4716 |
+
"step": 670
|
| 4717 |
+
},
|
| 4718 |
+
{
|
| 4719 |
+
"epoch": 0.5505641025641026,
|
| 4720 |
+
"grad_norm": 1.671875,
|
| 4721 |
+
"learning_rate": 5.352628570343565e-06,
|
| 4722 |
+
"loss": 1.1412111520767212,
|
| 4723 |
+
"step": 671
|
| 4724 |
+
},
|
| 4725 |
+
{
|
| 4726 |
+
"epoch": 0.5513846153846154,
|
| 4727 |
+
"grad_norm": 1.671875,
|
| 4728 |
+
"learning_rate": 5.3365923268570485e-06,
|
| 4729 |
+
"loss": 1.151916742324829,
|
| 4730 |
+
"step": 672
|
| 4731 |
+
},
|
| 4732 |
+
{
|
| 4733 |
+
"epoch": 0.5522051282051282,
|
| 4734 |
+
"grad_norm": 1.6953125,
|
| 4735 |
+
"learning_rate": 5.320561971592153e-06,
|
| 4736 |
+
"loss": 1.16988205909729,
|
| 4737 |
+
"step": 673
|
| 4738 |
+
},
|
| 4739 |
+
{
|
| 4740 |
+
"epoch": 0.553025641025641,
|
| 4741 |
+
"grad_norm": 1.609375,
|
| 4742 |
+
"learning_rate": 5.3045376138718465e-06,
|
| 4743 |
+
"loss": 1.2555769681930542,
|
| 4744 |
+
"step": 674
|
| 4745 |
+
},
|
| 4746 |
+
{
|
| 4747 |
+
"epoch": 0.5538461538461539,
|
| 4748 |
+
"grad_norm": 1.6640625,
|
| 4749 |
+
"learning_rate": 5.288519362978184e-06,
|
| 4750 |
+
"loss": 1.0746052265167236,
|
| 4751 |
+
"step": 675
|
| 4752 |
+
},
|
| 4753 |
+
{
|
| 4754 |
+
"epoch": 0.5546666666666666,
|
| 4755 |
+
"grad_norm": 1.671875,
|
| 4756 |
+
"learning_rate": 5.2725073281515836e-06,
|
| 4757 |
+
"loss": 1.0875591039657593,
|
| 4758 |
+
"step": 676
|
| 4759 |
+
},
|
| 4760 |
+
{
|
| 4761 |
+
"epoch": 0.5554871794871795,
|
| 4762 |
+
"grad_norm": 1.7265625,
|
| 4763 |
+
"learning_rate": 5.256501618590064e-06,
|
| 4764 |
+
"loss": 1.1699628829956055,
|
| 4765 |
+
"step": 677
|
| 4766 |
+
},
|
| 4767 |
+
{
|
| 4768 |
+
"epoch": 0.5563076923076923,
|
| 4769 |
+
"grad_norm": 1.640625,
|
| 4770 |
+
"learning_rate": 5.2405023434485105e-06,
|
| 4771 |
+
"loss": 1.0946143865585327,
|
| 4772 |
+
"step": 678
|
| 4773 |
+
},
|
| 4774 |
+
{
|
| 4775 |
+
"epoch": 0.5571282051282052,
|
| 4776 |
+
"grad_norm": 1.6015625,
|
| 4777 |
+
"learning_rate": 5.224509611837929e-06,
|
| 4778 |
+
"loss": 1.1036039590835571,
|
| 4779 |
+
"step": 679
|
| 4780 |
+
},
|
| 4781 |
+
{
|
| 4782 |
+
"epoch": 0.5579487179487179,
|
| 4783 |
+
"grad_norm": 1.53125,
|
| 4784 |
+
"learning_rate": 5.208523532824697e-06,
|
| 4785 |
+
"loss": 1.1801561117172241,
|
| 4786 |
+
"step": 680
|
| 4787 |
+
},
|
| 4788 |
+
{
|
| 4789 |
+
"epoch": 0.5587692307692308,
|
| 4790 |
+
"grad_norm": 1.6171875,
|
| 4791 |
+
"learning_rate": 5.192544215429821e-06,
|
| 4792 |
+
"loss": 1.1462246179580688,
|
| 4793 |
+
"step": 681
|
| 4794 |
+
},
|
| 4795 |
+
{
|
| 4796 |
+
"epoch": 0.5595897435897436,
|
| 4797 |
+
"grad_norm": 1.640625,
|
| 4798 |
+
"learning_rate": 5.1765717686282e-06,
|
| 4799 |
+
"loss": 1.218364953994751,
|
| 4800 |
+
"step": 682
|
| 4801 |
+
},
|
| 4802 |
+
{
|
| 4803 |
+
"epoch": 0.5604102564102564,
|
| 4804 |
+
"grad_norm": 1.71875,
|
| 4805 |
+
"learning_rate": 5.160606301347877e-06,
|
| 4806 |
+
"loss": 1.160405158996582,
|
| 4807 |
+
"step": 683
|
| 4808 |
+
},
|
| 4809 |
+
{
|
| 4810 |
+
"epoch": 0.5612307692307692,
|
| 4811 |
+
"grad_norm": 1.671875,
|
| 4812 |
+
"learning_rate": 5.144647922469293e-06,
|
| 4813 |
+
"loss": 1.1982057094573975,
|
| 4814 |
+
"step": 684
|
| 4815 |
+
},
|
| 4816 |
+
{
|
| 4817 |
+
"epoch": 0.5620512820512821,
|
| 4818 |
+
"grad_norm": 1.796875,
|
| 4819 |
+
"learning_rate": 5.128696740824551e-06,
|
| 4820 |
+
"loss": 1.056734561920166,
|
| 4821 |
+
"step": 685
|
| 4822 |
+
},
|
| 4823 |
+
{
|
| 4824 |
+
"epoch": 0.5628717948717948,
|
| 4825 |
+
"grad_norm": 1.7265625,
|
| 4826 |
+
"learning_rate": 5.112752865196667e-06,
|
| 4827 |
+
"loss": 1.1484451293945312,
|
| 4828 |
+
"step": 686
|
| 4829 |
+
},
|
| 4830 |
+
{
|
| 4831 |
+
"epoch": 0.5636923076923077,
|
| 4832 |
+
"grad_norm": 1.7890625,
|
| 4833 |
+
"learning_rate": 5.096816404318835e-06,
|
| 4834 |
+
"loss": 1.1355385780334473,
|
| 4835 |
+
"step": 687
|
| 4836 |
+
},
|
| 4837 |
+
{
|
| 4838 |
+
"epoch": 0.5645128205128205,
|
| 4839 |
+
"grad_norm": 1.6875,
|
| 4840 |
+
"learning_rate": 5.0808874668736814e-06,
|
| 4841 |
+
"loss": 1.089058756828308,
|
| 4842 |
+
"step": 688
|
| 4843 |
+
},
|
| 4844 |
+
{
|
| 4845 |
+
"epoch": 0.5653333333333334,
|
| 4846 |
+
"grad_norm": 1.5859375,
|
| 4847 |
+
"learning_rate": 5.064966161492527e-06,
|
| 4848 |
+
"loss": 1.023891568183899,
|
| 4849 |
+
"step": 689
|
| 4850 |
+
},
|
| 4851 |
+
{
|
| 4852 |
+
"epoch": 0.5661538461538461,
|
| 4853 |
+
"grad_norm": 1.578125,
|
| 4854 |
+
"learning_rate": 5.049052596754637e-06,
|
| 4855 |
+
"loss": 1.1688581705093384,
|
| 4856 |
+
"step": 690
|
| 4857 |
+
},
|
| 4858 |
+
{
|
| 4859 |
+
"epoch": 0.566974358974359,
|
| 4860 |
+
"grad_norm": 1.546875,
|
| 4861 |
+
"learning_rate": 5.033146881186494e-06,
|
| 4862 |
+
"loss": 1.139955997467041,
|
| 4863 |
+
"step": 691
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 0.5677948717948718,
|
| 4867 |
+
"grad_norm": 1.5859375,
|
| 4868 |
+
"learning_rate": 5.017249123261052e-06,
|
| 4869 |
+
"loss": 1.0958822965621948,
|
| 4870 |
+
"step": 692
|
| 4871 |
+
},
|
| 4872 |
+
{
|
| 4873 |
+
"epoch": 0.5686153846153846,
|
| 4874 |
+
"grad_norm": 1.5234375,
|
| 4875 |
+
"learning_rate": 5.001359431396987e-06,
|
| 4876 |
+
"loss": 1.128438949584961,
|
| 4877 |
+
"step": 693
|
| 4878 |
+
},
|
| 4879 |
+
{
|
| 4880 |
+
"epoch": 0.5694358974358974,
|
| 4881 |
+
"grad_norm": 1.625,
|
| 4882 |
+
"learning_rate": 4.9854779139579805e-06,
|
| 4883 |
+
"loss": 1.146626353263855,
|
| 4884 |
+
"step": 694
|
| 4885 |
+
},
|
| 4886 |
+
{
|
| 4887 |
+
"epoch": 0.5702564102564103,
|
| 4888 |
+
"grad_norm": 1.53125,
|
| 4889 |
+
"learning_rate": 4.96960467925195e-06,
|
| 4890 |
+
"loss": 1.1052286624908447,
|
| 4891 |
+
"step": 695
|
| 4892 |
+
},
|
| 4893 |
+
{
|
| 4894 |
+
"epoch": 0.571076923076923,
|
| 4895 |
+
"grad_norm": 1.75,
|
| 4896 |
+
"learning_rate": 4.953739835530342e-06,
|
| 4897 |
+
"loss": 1.1179757118225098,
|
| 4898 |
+
"step": 696
|
| 4899 |
+
},
|
| 4900 |
+
{
|
| 4901 |
+
"epoch": 0.5718974358974359,
|
| 4902 |
+
"grad_norm": 1.5078125,
|
| 4903 |
+
"learning_rate": 4.9378834909873675e-06,
|
| 4904 |
+
"loss": 1.0641415119171143,
|
| 4905 |
+
"step": 697
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 0.5727179487179487,
|
| 4909 |
+
"grad_norm": 1.6015625,
|
| 4910 |
+
"learning_rate": 4.922035753759282e-06,
|
| 4911 |
+
"loss": 1.0473688840866089,
|
| 4912 |
+
"step": 698
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.5735384615384616,
|
| 4916 |
+
"grad_norm": 1.796875,
|
| 4917 |
+
"learning_rate": 4.906196731923642e-06,
|
| 4918 |
+
"loss": 1.2188773155212402,
|
| 4919 |
+
"step": 699
|
| 4920 |
+
},
|
| 4921 |
+
{
|
| 4922 |
+
"epoch": 0.5743589743589743,
|
| 4923 |
+
"grad_norm": 1.6484375,
|
| 4924 |
+
"learning_rate": 4.890366533498559e-06,
|
| 4925 |
+
"loss": 1.1033180952072144,
|
| 4926 |
+
"step": 700
|
| 4927 |
+
},
|
| 4928 |
+
{
|
| 4929 |
+
"epoch": 0.5751794871794872,
|
| 4930 |
+
"grad_norm": 1.609375,
|
| 4931 |
+
"learning_rate": 4.874545266441977e-06,
|
| 4932 |
+
"loss": 1.1321452856063843,
|
| 4933 |
+
"step": 701
|
| 4934 |
+
},
|
| 4935 |
+
{
|
| 4936 |
+
"epoch": 0.576,
|
| 4937 |
+
"grad_norm": 1.640625,
|
| 4938 |
+
"learning_rate": 4.858733038650932e-06,
|
| 4939 |
+
"loss": 1.2013392448425293,
|
| 4940 |
+
"step": 702
|
| 4941 |
+
},
|
| 4942 |
+
{
|
| 4943 |
+
"epoch": 0.5768205128205128,
|
| 4944 |
+
"grad_norm": 1.578125,
|
| 4945 |
+
"learning_rate": 4.842929957960811e-06,
|
| 4946 |
+
"loss": 1.0100051164627075,
|
| 4947 |
+
"step": 703
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 0.5776410256410256,
|
| 4951 |
+
"grad_norm": 1.5859375,
|
| 4952 |
+
"learning_rate": 4.827136132144623e-06,
|
| 4953 |
+
"loss": 1.1534571647644043,
|
| 4954 |
+
"step": 704
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.5784615384615385,
|
| 4958 |
+
"grad_norm": 1.6171875,
|
| 4959 |
+
"learning_rate": 4.811351668912257e-06,
|
| 4960 |
+
"loss": 1.1213208436965942,
|
| 4961 |
+
"step": 705
|
| 4962 |
+
},
|
| 4963 |
+
{
|
| 4964 |
+
"epoch": 0.5792820512820512,
|
| 4965 |
+
"grad_norm": 1.671875,
|
| 4966 |
+
"learning_rate": 4.795576675909754e-06,
|
| 4967 |
+
"loss": 1.1385544538497925,
|
| 4968 |
+
"step": 706
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 0.5801025641025641,
|
| 4972 |
+
"grad_norm": 1.65625,
|
| 4973 |
+
"learning_rate": 4.779811260718572e-06,
|
| 4974 |
+
"loss": 1.1246585845947266,
|
| 4975 |
+
"step": 707
|
| 4976 |
+
},
|
| 4977 |
+
{
|
| 4978 |
+
"epoch": 0.5809230769230769,
|
| 4979 |
+
"grad_norm": 1.875,
|
| 4980 |
+
"learning_rate": 4.764055530854849e-06,
|
| 4981 |
+
"loss": 1.1996458768844604,
|
| 4982 |
+
"step": 708
|
| 4983 |
+
},
|
| 4984 |
+
{
|
| 4985 |
+
"epoch": 0.5817435897435898,
|
| 4986 |
+
"grad_norm": 1.625,
|
| 4987 |
+
"learning_rate": 4.7483095937686725e-06,
|
| 4988 |
+
"loss": 1.1832023859024048,
|
| 4989 |
+
"step": 709
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 0.5825641025641025,
|
| 4993 |
+
"grad_norm": 1.640625,
|
| 4994 |
+
"learning_rate": 4.73257355684334e-06,
|
| 4995 |
+
"loss": 1.125613808631897,
|
| 4996 |
+
"step": 710
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.5833846153846154,
|
| 5000 |
+
"grad_norm": 1.6875,
|
| 5001 |
+
"learning_rate": 4.716847527394642e-06,
|
| 5002 |
+
"loss": 1.1033669710159302,
|
| 5003 |
+
"step": 711
|
| 5004 |
+
},
|
| 5005 |
+
{
|
| 5006 |
+
"epoch": 0.5842051282051282,
|
| 5007 |
+
"grad_norm": 1.6640625,
|
| 5008 |
+
"learning_rate": 4.7011316126701145e-06,
|
| 5009 |
+
"loss": 1.1974871158599854,
|
| 5010 |
+
"step": 712
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 0.585025641025641,
|
| 5014 |
+
"grad_norm": 1.6796875,
|
| 5015 |
+
"learning_rate": 4.685425919848314e-06,
|
| 5016 |
+
"loss": 1.1260145902633667,
|
| 5017 |
+
"step": 713
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.5858461538461538,
|
| 5021 |
+
"grad_norm": 1.5859375,
|
| 5022 |
+
"learning_rate": 4.669730556038092e-06,
|
| 5023 |
+
"loss": 1.1053581237792969,
|
| 5024 |
+
"step": 714
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.5866666666666667,
|
| 5028 |
+
"grad_norm": 1.6796875,
|
| 5029 |
+
"learning_rate": 4.654045628277848e-06,
|
| 5030 |
+
"loss": 1.1527366638183594,
|
| 5031 |
+
"step": 715
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.5874871794871794,
|
| 5035 |
+
"grad_norm": 1.765625,
|
| 5036 |
+
"learning_rate": 4.638371243534817e-06,
|
| 5037 |
+
"loss": 1.0591764450073242,
|
| 5038 |
+
"step": 716
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.5883076923076923,
|
| 5042 |
+
"grad_norm": 1.5234375,
|
| 5043 |
+
"learning_rate": 4.622707508704334e-06,
|
| 5044 |
+
"loss": 1.0199596881866455,
|
| 5045 |
+
"step": 717
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.5891282051282051,
|
| 5049 |
+
"grad_norm": 1.6875,
|
| 5050 |
+
"learning_rate": 4.6070545306091034e-06,
|
| 5051 |
+
"loss": 1.0806047916412354,
|
| 5052 |
+
"step": 718
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.589948717948718,
|
| 5056 |
+
"grad_norm": 1.5625,
|
| 5057 |
+
"learning_rate": 4.591412415998471e-06,
|
| 5058 |
+
"loss": 1.161590814590454,
|
| 5059 |
+
"step": 719
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.5907692307692308,
|
| 5063 |
+
"grad_norm": 1.625,
|
| 5064 |
+
"learning_rate": 4.575781271547695e-06,
|
| 5065 |
+
"loss": 1.1730376482009888,
|
| 5066 |
+
"step": 720
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.5915897435897436,
|
| 5070 |
+
"grad_norm": 1.8515625,
|
| 5071 |
+
"learning_rate": 4.5601612038572185e-06,
|
| 5072 |
+
"loss": 1.159124493598938,
|
| 5073 |
+
"step": 721
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.5924102564102565,
|
| 5077 |
+
"grad_norm": 1.75,
|
| 5078 |
+
"learning_rate": 4.544552319451952e-06,
|
| 5079 |
+
"loss": 1.035894751548767,
|
| 5080 |
+
"step": 722
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.5932307692307692,
|
| 5084 |
+
"grad_norm": 1.609375,
|
| 5085 |
+
"learning_rate": 4.528954724780529e-06,
|
| 5086 |
+
"loss": 1.0906904935836792,
|
| 5087 |
+
"step": 723
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.5940512820512821,
|
| 5091 |
+
"grad_norm": 1.6015625,
|
| 5092 |
+
"learning_rate": 4.5133685262145994e-06,
|
| 5093 |
+
"loss": 1.1019519567489624,
|
| 5094 |
+
"step": 724
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.5948717948717949,
|
| 5098 |
+
"grad_norm": 1.5234375,
|
| 5099 |
+
"learning_rate": 4.497793830048082e-06,
|
| 5100 |
+
"loss": 1.0816446542739868,
|
| 5101 |
+
"step": 725
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.5956923076923077,
|
| 5105 |
+
"grad_norm": 1.5859375,
|
| 5106 |
+
"learning_rate": 4.4822307424964655e-06,
|
| 5107 |
+
"loss": 1.1538878679275513,
|
| 5108 |
+
"step": 726
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.5965128205128205,
|
| 5112 |
+
"grad_norm": 1.6796875,
|
| 5113 |
+
"learning_rate": 4.46667936969606e-06,
|
| 5114 |
+
"loss": 1.162054419517517,
|
| 5115 |
+
"step": 727
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.5973333333333334,
|
| 5119 |
+
"grad_norm": 1.6484375,
|
| 5120 |
+
"learning_rate": 4.451139817703292e-06,
|
| 5121 |
+
"loss": 1.0661221742630005,
|
| 5122 |
+
"step": 728
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.5981538461538461,
|
| 5126 |
+
"grad_norm": 1.6875,
|
| 5127 |
+
"learning_rate": 4.4356121924939715e-06,
|
| 5128 |
+
"loss": 1.1707451343536377,
|
| 5129 |
+
"step": 729
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.598974358974359,
|
| 5133 |
+
"grad_norm": 1.6484375,
|
| 5134 |
+
"learning_rate": 4.420096599962565e-06,
|
| 5135 |
+
"loss": 1.138479232788086,
|
| 5136 |
+
"step": 730
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.5997948717948718,
|
| 5140 |
+
"grad_norm": 1.5859375,
|
| 5141 |
+
"learning_rate": 4.404593145921484e-06,
|
| 5142 |
+
"loss": 1.0460493564605713,
|
| 5143 |
+
"step": 731
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.6006153846153847,
|
| 5147 |
+
"grad_norm": 1.578125,
|
| 5148 |
+
"learning_rate": 4.38910193610036e-06,
|
| 5149 |
+
"loss": 1.148463249206543,
|
| 5150 |
+
"step": 732
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.6006153846153847,
|
| 5154 |
+
"eval_loss": 1.1107702255249023,
|
| 5155 |
+
"eval_runtime": 25.6154,
|
| 5156 |
+
"eval_samples_per_second": 39.039,
|
| 5157 |
+
"eval_steps_per_second": 9.76,
|
| 5158 |
+
"step": 732
|
| 5159 |
}
|
| 5160 |
],
|
| 5161 |
"logging_steps": 1,
|
|
|
|
| 5175 |
"attributes": {}
|
| 5176 |
}
|
| 5177 |
},
|
| 5178 |
+
"total_flos": 1.9085921112644813e+17,
|
| 5179 |
"train_batch_size": 4,
|
| 5180 |
"trial_name": null,
|
| 5181 |
"trial_params": null
|