Training in progress, step 1418, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 50365768
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e5fe024b05282e9bb676645113cf1f85610eac73136fae5348ae36a04d2cbb6
|
| 3 |
size 50365768
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 25859412
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8e2743229c4d6b76d07092cc5a9aa6a62e6d544983c1da45b6da4ac3e145ebc
|
| 3 |
size 25859412
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:89e3388213cbf144349706e8461755d803ad7dcb4b2b866ed5ec4e122a355c8e
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05b6639f438a19725b9ad30fb7c4d3588581e257326137576a14039812b3c0fe
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 355,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7494,6 +7494,2477 @@
|
|
| 7494 |
"eval_samples_per_second": 34.005,
|
| 7495 |
"eval_steps_per_second": 17.002,
|
| 7496 |
"step": 1065
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7497 |
}
|
| 7498 |
],
|
| 7499 |
"logging_steps": 1,
|
|
@@ -7508,12 +9979,12 @@
|
|
| 7508 |
"should_evaluate": false,
|
| 7509 |
"should_log": false,
|
| 7510 |
"should_save": true,
|
| 7511 |
-
"should_training_stop":
|
| 7512 |
},
|
| 7513 |
"attributes": {}
|
| 7514 |
}
|
| 7515 |
},
|
| 7516 |
-
"total_flos":
|
| 7517 |
"train_batch_size": 2,
|
| 7518 |
"trial_name": null,
|
| 7519 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.16597413238134254,
|
| 5 |
"eval_steps": 355,
|
| 6 |
+
"global_step": 1418,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7494 |
"eval_samples_per_second": 34.005,
|
| 7495 |
"eval_steps_per_second": 17.002,
|
| 7496 |
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.1247732194065664,
|
| 7500 |
+
"grad_norm": 3.8553192615509033,
|
| 7501 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 7502 |
+
"loss": 6.4458,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.12489026745479019,
|
| 7507 |
+
"grad_norm": 4.623905658721924,
|
| 7508 |
+
"learning_rate": 2.913172519547589e-05,
|
| 7509 |
+
"loss": 6.282,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.125007315503014,
|
| 7514 |
+
"grad_norm": 3.767019748687744,
|
| 7515 |
+
"learning_rate": 2.897448132385525e-05,
|
| 7516 |
+
"loss": 4.6797,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.1251243635512378,
|
| 7521 |
+
"grad_norm": 3.8985183238983154,
|
| 7522 |
+
"learning_rate": 2.8817591049314264e-05,
|
| 7523 |
+
"loss": 5.8846,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.1252414115994616,
|
| 7528 |
+
"grad_norm": 4.024424076080322,
|
| 7529 |
+
"learning_rate": 2.8661055152923456e-05,
|
| 7530 |
+
"loss": 6.677,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.12535845964768538,
|
| 7535 |
+
"grad_norm": 4.238265037536621,
|
| 7536 |
+
"learning_rate": 2.8504874413989214e-05,
|
| 7537 |
+
"loss": 6.3787,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.12547550769590918,
|
| 7542 |
+
"grad_norm": 3.741295576095581,
|
| 7543 |
+
"learning_rate": 2.8349049610049683e-05,
|
| 7544 |
+
"loss": 5.0746,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.12559255574413297,
|
| 7549 |
+
"grad_norm": 3.197049856185913,
|
| 7550 |
+
"learning_rate": 2.819358151687107e-05,
|
| 7551 |
+
"loss": 4.0846,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.12570960379235677,
|
| 7556 |
+
"grad_norm": 3.4590532779693604,
|
| 7557 |
+
"learning_rate": 2.8038470908443714e-05,
|
| 7558 |
+
"loss": 5.8085,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.12582665184058056,
|
| 7563 |
+
"grad_norm": 3.989264488220215,
|
| 7564 |
+
"learning_rate": 2.7883718556978188e-05,
|
| 7565 |
+
"loss": 6.0178,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.12594369988880436,
|
| 7570 |
+
"grad_norm": 3.369285821914673,
|
| 7571 |
+
"learning_rate": 2.7729325232901494e-05,
|
| 7572 |
+
"loss": 4.4873,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.12606074793702815,
|
| 7577 |
+
"grad_norm": 4.070693016052246,
|
| 7578 |
+
"learning_rate": 2.7575291704853323e-05,
|
| 7579 |
+
"loss": 5.3489,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.12617779598525194,
|
| 7584 |
+
"grad_norm": 5.596129894256592,
|
| 7585 |
+
"learning_rate": 2.742161873968202e-05,
|
| 7586 |
+
"loss": 6.5073,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.12629484403347574,
|
| 7591 |
+
"grad_norm": 4.095518112182617,
|
| 7592 |
+
"learning_rate": 2.7268307102441005e-05,
|
| 7593 |
+
"loss": 6.0858,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.12641189208169953,
|
| 7598 |
+
"grad_norm": 4.547975063323975,
|
| 7599 |
+
"learning_rate": 2.7115357556384756e-05,
|
| 7600 |
+
"loss": 6.5706,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.12652894012992333,
|
| 7605 |
+
"grad_norm": 4.1257758140563965,
|
| 7606 |
+
"learning_rate": 2.696277086296509e-05,
|
| 7607 |
+
"loss": 6.7871,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.12664598817814712,
|
| 7612 |
+
"grad_norm": 4.423878192901611,
|
| 7613 |
+
"learning_rate": 2.681054778182748e-05,
|
| 7614 |
+
"loss": 6.0472,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.12676303622637092,
|
| 7619 |
+
"grad_norm": 3.5452682971954346,
|
| 7620 |
+
"learning_rate": 2.665868907080705e-05,
|
| 7621 |
+
"loss": 5.121,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.1268800842745947,
|
| 7626 |
+
"grad_norm": 3.882988691329956,
|
| 7627 |
+
"learning_rate": 2.650719548592502e-05,
|
| 7628 |
+
"loss": 6.0486,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.1269971323228185,
|
| 7633 |
+
"grad_norm": 4.451942443847656,
|
| 7634 |
+
"learning_rate": 2.635606778138483e-05,
|
| 7635 |
+
"loss": 4.5282,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.1271141803710423,
|
| 7640 |
+
"grad_norm": 6.146245956420898,
|
| 7641 |
+
"learning_rate": 2.6205306709568354e-05,
|
| 7642 |
+
"loss": 7.7677,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.1272312284192661,
|
| 7647 |
+
"grad_norm": 4.536795139312744,
|
| 7648 |
+
"learning_rate": 2.6054913021032224e-05,
|
| 7649 |
+
"loss": 6.4225,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.12734827646748992,
|
| 7654 |
+
"grad_norm": 3.937973737716675,
|
| 7655 |
+
"learning_rate": 2.5904887464504114e-05,
|
| 7656 |
+
"loss": 6.1264,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.1274653245157137,
|
| 7661 |
+
"grad_norm": 4.33305549621582,
|
| 7662 |
+
"learning_rate": 2.5755230786878882e-05,
|
| 7663 |
+
"loss": 5.1653,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.1275823725639375,
|
| 7668 |
+
"grad_norm": 3.740837335586548,
|
| 7669 |
+
"learning_rate": 2.5605943733215042e-05,
|
| 7670 |
+
"loss": 5.2321,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.1276994206121613,
|
| 7675 |
+
"grad_norm": 5.293520450592041,
|
| 7676 |
+
"learning_rate": 2.5457027046730964e-05,
|
| 7677 |
+
"loss": 7.359,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.1278164686603851,
|
| 7682 |
+
"grad_norm": 6.307898998260498,
|
| 7683 |
+
"learning_rate": 2.530848146880097e-05,
|
| 7684 |
+
"loss": 8.2541,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.1279335167086089,
|
| 7689 |
+
"grad_norm": 4.090138912200928,
|
| 7690 |
+
"learning_rate": 2.516030773895205e-05,
|
| 7691 |
+
"loss": 4.9293,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.12805056475683269,
|
| 7696 |
+
"grad_norm": 4.111153602600098,
|
| 7697 |
+
"learning_rate": 2.501250659485992e-05,
|
| 7698 |
+
"loss": 4.8477,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.12816761280505648,
|
| 7703 |
+
"grad_norm": 4.494771480560303,
|
| 7704 |
+
"learning_rate": 2.4865078772345285e-05,
|
| 7705 |
+
"loss": 7.0018,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.12828466085328027,
|
| 7710 |
+
"grad_norm": 3.813302755355835,
|
| 7711 |
+
"learning_rate": 2.4718025005370414e-05,
|
| 7712 |
+
"loss": 5.6723,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.12840170890150407,
|
| 7717 |
+
"grad_norm": 5.024593353271484,
|
| 7718 |
+
"learning_rate": 2.457134602603538e-05,
|
| 7719 |
+
"loss": 8.4544,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.12851875694972786,
|
| 7724 |
+
"grad_norm": 3.703540325164795,
|
| 7725 |
+
"learning_rate": 2.4425042564574184e-05,
|
| 7726 |
+
"loss": 4.9045,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.12863580499795166,
|
| 7731 |
+
"grad_norm": 3.296093702316284,
|
| 7732 |
+
"learning_rate": 2.4279115349351543e-05,
|
| 7733 |
+
"loss": 5.1584,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.12875285304617545,
|
| 7738 |
+
"grad_norm": 4.957206726074219,
|
| 7739 |
+
"learning_rate": 2.4133565106859003e-05,
|
| 7740 |
+
"loss": 6.2911,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.12886990109439925,
|
| 7745 |
+
"grad_norm": 3.4661080837249756,
|
| 7746 |
+
"learning_rate": 2.3988392561711292e-05,
|
| 7747 |
+
"loss": 6.6898,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.12898694914262304,
|
| 7752 |
+
"grad_norm": 4.296874523162842,
|
| 7753 |
+
"learning_rate": 2.38435984366429e-05,
|
| 7754 |
+
"loss": 6.5291,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.12910399719084684,
|
| 7759 |
+
"grad_norm": 5.050826549530029,
|
| 7760 |
+
"learning_rate": 2.369918345250428e-05,
|
| 7761 |
+
"loss": 9.5722,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.12922104523907063,
|
| 7766 |
+
"grad_norm": 4.0029191970825195,
|
| 7767 |
+
"learning_rate": 2.3555148328258368e-05,
|
| 7768 |
+
"loss": 4.2838,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.12933809328729443,
|
| 7773 |
+
"grad_norm": 3.974111318588257,
|
| 7774 |
+
"learning_rate": 2.3411493780977058e-05,
|
| 7775 |
+
"loss": 6.8469,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.12945514133551822,
|
| 7780 |
+
"grad_norm": 3.8164901733398438,
|
| 7781 |
+
"learning_rate": 2.3268220525837437e-05,
|
| 7782 |
+
"loss": 5.8446,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.12957218938374201,
|
| 7787 |
+
"grad_norm": 4.173857688903809,
|
| 7788 |
+
"learning_rate": 2.3125329276118446e-05,
|
| 7789 |
+
"loss": 6.362,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.1296892374319658,
|
| 7794 |
+
"grad_norm": 3.859119176864624,
|
| 7795 |
+
"learning_rate": 2.2982820743197243e-05,
|
| 7796 |
+
"loss": 5.4499,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.12980628548018963,
|
| 7801 |
+
"grad_norm": 3.6904869079589844,
|
| 7802 |
+
"learning_rate": 2.2840695636545574e-05,
|
| 7803 |
+
"loss": 6.5662,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.12992333352841343,
|
| 7808 |
+
"grad_norm": 4.197831630706787,
|
| 7809 |
+
"learning_rate": 2.26989546637263e-05,
|
| 7810 |
+
"loss": 6.6831,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.13004038157663722,
|
| 7815 |
+
"grad_norm": 5.4231157302856445,
|
| 7816 |
+
"learning_rate": 2.2557598530390022e-05,
|
| 7817 |
+
"loss": 6.2548,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.13015742962486102,
|
| 7822 |
+
"grad_norm": 4.809431552886963,
|
| 7823 |
+
"learning_rate": 2.2416627940271297e-05,
|
| 7824 |
+
"loss": 5.1483,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.1302744776730848,
|
| 7829 |
+
"grad_norm": 3.8242275714874268,
|
| 7830 |
+
"learning_rate": 2.227604359518536e-05,
|
| 7831 |
+
"loss": 7.5611,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.1303915257213086,
|
| 7836 |
+
"grad_norm": 3.2712440490722656,
|
| 7837 |
+
"learning_rate": 2.2135846195024513e-05,
|
| 7838 |
+
"loss": 4.8346,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.1305085737695324,
|
| 7843 |
+
"grad_norm": 3.6756529808044434,
|
| 7844 |
+
"learning_rate": 2.199603643775463e-05,
|
| 7845 |
+
"loss": 5.3905,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.1306256218177562,
|
| 7850 |
+
"grad_norm": 4.0744733810424805,
|
| 7851 |
+
"learning_rate": 2.1856615019411797e-05,
|
| 7852 |
+
"loss": 6.4944,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.13074266986598,
|
| 7857 |
+
"grad_norm": 5.272459983825684,
|
| 7858 |
+
"learning_rate": 2.1717582634098776e-05,
|
| 7859 |
+
"loss": 7.2233,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.13085971791420378,
|
| 7864 |
+
"grad_norm": 4.850452899932861,
|
| 7865 |
+
"learning_rate": 2.157893997398146e-05,
|
| 7866 |
+
"loss": 7.8259,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.13097676596242758,
|
| 7871 |
+
"grad_norm": 6.229705810546875,
|
| 7872 |
+
"learning_rate": 2.1440687729285614e-05,
|
| 7873 |
+
"loss": 8.0993,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.13109381401065137,
|
| 7878 |
+
"grad_norm": 4.742966175079346,
|
| 7879 |
+
"learning_rate": 2.130282658829328e-05,
|
| 7880 |
+
"loss": 6.4568,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.13121086205887517,
|
| 7885 |
+
"grad_norm": 4.198139190673828,
|
| 7886 |
+
"learning_rate": 2.116535723733938e-05,
|
| 7887 |
+
"loss": 7.4337,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.13132791010709896,
|
| 7892 |
+
"grad_norm": 3.4211654663085938,
|
| 7893 |
+
"learning_rate": 2.1028280360808407e-05,
|
| 7894 |
+
"loss": 5.6395,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.13144495815532276,
|
| 7899 |
+
"grad_norm": 4.0057148933410645,
|
| 7900 |
+
"learning_rate": 2.0891596641130928e-05,
|
| 7901 |
+
"loss": 6.4509,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.13156200620354655,
|
| 7906 |
+
"grad_norm": 6.410972595214844,
|
| 7907 |
+
"learning_rate": 2.075530675878016e-05,
|
| 7908 |
+
"loss": 9.8986,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.13167905425177034,
|
| 7913 |
+
"grad_norm": 4.581459045410156,
|
| 7914 |
+
"learning_rate": 2.0619411392268582e-05,
|
| 7915 |
+
"loss": 5.9499,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.13179610229999414,
|
| 7920 |
+
"grad_norm": 4.086128234863281,
|
| 7921 |
+
"learning_rate": 2.0483911218144715e-05,
|
| 7922 |
+
"loss": 5.0018,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.13191315034821793,
|
| 7927 |
+
"grad_norm": 4.77893590927124,
|
| 7928 |
+
"learning_rate": 2.0348806910989505e-05,
|
| 7929 |
+
"loss": 6.3637,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.13203019839644173,
|
| 7934 |
+
"grad_norm": 4.110125541687012,
|
| 7935 |
+
"learning_rate": 2.0214099143413212e-05,
|
| 7936 |
+
"loss": 5.6604,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.13214724644466552,
|
| 7941 |
+
"grad_norm": 3.951420545578003,
|
| 7942 |
+
"learning_rate": 2.0079788586051828e-05,
|
| 7943 |
+
"loss": 6.2952,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.13226429449288934,
|
| 7948 |
+
"grad_norm": 4.686469554901123,
|
| 7949 |
+
"learning_rate": 1.994587590756397e-05,
|
| 7950 |
+
"loss": 5.4759,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.13238134254111314,
|
| 7955 |
+
"grad_norm": 4.113170146942139,
|
| 7956 |
+
"learning_rate": 1.981236177462732e-05,
|
| 7957 |
+
"loss": 6.4982,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.13249839058933693,
|
| 7962 |
+
"grad_norm": 3.955838441848755,
|
| 7963 |
+
"learning_rate": 1.967924685193552e-05,
|
| 7964 |
+
"loss": 6.8908,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.13261543863756073,
|
| 7969 |
+
"grad_norm": 4.751936435699463,
|
| 7970 |
+
"learning_rate": 1.9546531802194667e-05,
|
| 7971 |
+
"loss": 5.8245,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.13273248668578452,
|
| 7976 |
+
"grad_norm": 3.9754884243011475,
|
| 7977 |
+
"learning_rate": 1.941421728612023e-05,
|
| 7978 |
+
"loss": 6.0481,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.13284953473400832,
|
| 7983 |
+
"grad_norm": 3.9083471298217773,
|
| 7984 |
+
"learning_rate": 1.9282303962433524e-05,
|
| 7985 |
+
"loss": 5.8108,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.1329665827822321,
|
| 7990 |
+
"grad_norm": 4.179833889007568,
|
| 7991 |
+
"learning_rate": 1.915079248785858e-05,
|
| 7992 |
+
"loss": 5.887,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.1330836308304559,
|
| 7997 |
+
"grad_norm": 5.102409839630127,
|
| 7998 |
+
"learning_rate": 1.9019683517118914e-05,
|
| 7999 |
+
"loss": 7.1645,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.1332006788786797,
|
| 8004 |
+
"grad_norm": 3.4902656078338623,
|
| 8005 |
+
"learning_rate": 1.8888977702934085e-05,
|
| 8006 |
+
"loss": 5.9588,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.1333177269269035,
|
| 8011 |
+
"grad_norm": 3.6271302700042725,
|
| 8012 |
+
"learning_rate": 1.8758675696016638e-05,
|
| 8013 |
+
"loss": 4.7721,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.1334347749751273,
|
| 8018 |
+
"grad_norm": 4.156149864196777,
|
| 8019 |
+
"learning_rate": 1.862877814506878e-05,
|
| 8020 |
+
"loss": 5.1536,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.13355182302335109,
|
| 8025 |
+
"grad_norm": 4.350010395050049,
|
| 8026 |
+
"learning_rate": 1.849928569677911e-05,
|
| 8027 |
+
"loss": 5.7833,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.13366887107157488,
|
| 8032 |
+
"grad_norm": 3.8963491916656494,
|
| 8033 |
+
"learning_rate": 1.837019899581943e-05,
|
| 8034 |
+
"loss": 6.7263,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.13378591911979867,
|
| 8039 |
+
"grad_norm": 4.305629253387451,
|
| 8040 |
+
"learning_rate": 1.824151868484164e-05,
|
| 8041 |
+
"loss": 6.9536,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.13390296716802247,
|
| 8046 |
+
"grad_norm": 5.229813098907471,
|
| 8047 |
+
"learning_rate": 1.8113245404474354e-05,
|
| 8048 |
+
"loss": 7.3727,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.13402001521624626,
|
| 8053 |
+
"grad_norm": 4.409402847290039,
|
| 8054 |
+
"learning_rate": 1.7985379793319835e-05,
|
| 8055 |
+
"loss": 7.0412,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.13413706326447006,
|
| 8060 |
+
"grad_norm": 4.176473140716553,
|
| 8061 |
+
"learning_rate": 1.7857922487950874e-05,
|
| 8062 |
+
"loss": 6.8747,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.13425411131269385,
|
| 8067 |
+
"grad_norm": 4.215778827667236,
|
| 8068 |
+
"learning_rate": 1.7730874122907315e-05,
|
| 8069 |
+
"loss": 5.9242,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.13437115936091765,
|
| 8074 |
+
"grad_norm": 3.710967540740967,
|
| 8075 |
+
"learning_rate": 1.760423533069332e-05,
|
| 8076 |
+
"loss": 5.7409,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.13448820740914144,
|
| 8081 |
+
"grad_norm": 3.2800509929656982,
|
| 8082 |
+
"learning_rate": 1.7478006741773946e-05,
|
| 8083 |
+
"loss": 5.1176,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.13460525545736524,
|
| 8088 |
+
"grad_norm": 3.88649845123291,
|
| 8089 |
+
"learning_rate": 1.7352188984572026e-05,
|
| 8090 |
+
"loss": 5.6042,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.13472230350558903,
|
| 8095 |
+
"grad_norm": 5.705807685852051,
|
| 8096 |
+
"learning_rate": 1.722678268546517e-05,
|
| 8097 |
+
"loss": 5.3125,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.13483935155381285,
|
| 8102 |
+
"grad_norm": 4.61563777923584,
|
| 8103 |
+
"learning_rate": 1.710178846878251e-05,
|
| 8104 |
+
"loss": 4.8456,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.13495639960203665,
|
| 8109 |
+
"grad_norm": 3.9143693447113037,
|
| 8110 |
+
"learning_rate": 1.6977206956801604e-05,
|
| 8111 |
+
"loss": 6.1729,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.13507344765026044,
|
| 8116 |
+
"grad_norm": 4.544604778289795,
|
| 8117 |
+
"learning_rate": 1.6853038769745467e-05,
|
| 8118 |
+
"loss": 7.4931,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.13519049569848424,
|
| 8123 |
+
"grad_norm": 4.558419227600098,
|
| 8124 |
+
"learning_rate": 1.6729284525779397e-05,
|
| 8125 |
+
"loss": 7.7546,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.13530754374670803,
|
| 8130 |
+
"grad_norm": 5.486094951629639,
|
| 8131 |
+
"learning_rate": 1.6605944841007802e-05,
|
| 8132 |
+
"loss": 5.386,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.13542459179493183,
|
| 8137 |
+
"grad_norm": 3.2692084312438965,
|
| 8138 |
+
"learning_rate": 1.6483020329471333e-05,
|
| 8139 |
+
"loss": 5.2034,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.13554163984315562,
|
| 8144 |
+
"grad_norm": 4.308760166168213,
|
| 8145 |
+
"learning_rate": 1.6360511603143648e-05,
|
| 8146 |
+
"loss": 7.1668,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.13565868789137941,
|
| 8151 |
+
"grad_norm": 3.5268983840942383,
|
| 8152 |
+
"learning_rate": 1.623841927192845e-05,
|
| 8153 |
+
"loss": 7.3562,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.1357757359396032,
|
| 8158 |
+
"grad_norm": 4.54243278503418,
|
| 8159 |
+
"learning_rate": 1.6116743943656488e-05,
|
| 8160 |
+
"loss": 6.7075,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.135892783987827,
|
| 8165 |
+
"grad_norm": 4.709784507751465,
|
| 8166 |
+
"learning_rate": 1.59954862240824e-05,
|
| 8167 |
+
"loss": 6.8764,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.1360098320360508,
|
| 8172 |
+
"grad_norm": 4.561254501342773,
|
| 8173 |
+
"learning_rate": 1.587464671688187e-05,
|
| 8174 |
+
"loss": 7.9346,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.1361268800842746,
|
| 8179 |
+
"grad_norm": 5.485746383666992,
|
| 8180 |
+
"learning_rate": 1.5754226023648523e-05,
|
| 8181 |
+
"loss": 6.0937,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.1362439281324984,
|
| 8186 |
+
"grad_norm": 3.9633190631866455,
|
| 8187 |
+
"learning_rate": 1.56342247438908e-05,
|
| 8188 |
+
"loss": 6.3625,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.13636097618072218,
|
| 8193 |
+
"grad_norm": 4.687920093536377,
|
| 8194 |
+
"learning_rate": 1.551464347502929e-05,
|
| 8195 |
+
"loss": 6.1423,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.13647802422894598,
|
| 8200 |
+
"grad_norm": 4.646004676818848,
|
| 8201 |
+
"learning_rate": 1.5395482812393514e-05,
|
| 8202 |
+
"loss": 7.1366,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.13659507227716977,
|
| 8207 |
+
"grad_norm": 6.156735897064209,
|
| 8208 |
+
"learning_rate": 1.527674334921897e-05,
|
| 8209 |
+
"loss": 6.2637,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.13671212032539357,
|
| 8214 |
+
"grad_norm": 3.9354870319366455,
|
| 8215 |
+
"learning_rate": 1.5158425676644329e-05,
|
| 8216 |
+
"loss": 5.6502,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.13682916837361736,
|
| 8221 |
+
"grad_norm": 5.1963582038879395,
|
| 8222 |
+
"learning_rate": 1.504053038370834e-05,
|
| 8223 |
+
"loss": 7.1623,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.13694621642184115,
|
| 8228 |
+
"grad_norm": 4.08088493347168,
|
| 8229 |
+
"learning_rate": 1.4923058057346929e-05,
|
| 8230 |
+
"loss": 6.3154,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.13706326447006495,
|
| 8235 |
+
"grad_norm": 3.8931407928466797,
|
| 8236 |
+
"learning_rate": 1.4806009282390376e-05,
|
| 8237 |
+
"loss": 4.262,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.13718031251828874,
|
| 8242 |
+
"grad_norm": 4.435689449310303,
|
| 8243 |
+
"learning_rate": 1.4689384641560311e-05,
|
| 8244 |
+
"loss": 7.5797,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.13729736056651257,
|
| 8249 |
+
"grad_norm": 3.5573792457580566,
|
| 8250 |
+
"learning_rate": 1.457318471546677e-05,
|
| 8251 |
+
"loss": 5.2003,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.13741440861473636,
|
| 8256 |
+
"grad_norm": 3.3824315071105957,
|
| 8257 |
+
"learning_rate": 1.4457410082605483e-05,
|
| 8258 |
+
"loss": 6.3396,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.13753145666296016,
|
| 8263 |
+
"grad_norm": 3.9588029384613037,
|
| 8264 |
+
"learning_rate": 1.434206131935476e-05,
|
| 8265 |
+
"loss": 4.7833,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.13764850471118395,
|
| 8270 |
+
"grad_norm": 4.222960948944092,
|
| 8271 |
+
"learning_rate": 1.42271389999728e-05,
|
| 8272 |
+
"loss": 6.784,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.13776555275940774,
|
| 8277 |
+
"grad_norm": 3.721405506134033,
|
| 8278 |
+
"learning_rate": 1.4112643696594773e-05,
|
| 8279 |
+
"loss": 6.2174,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.13788260080763154,
|
| 8284 |
+
"grad_norm": 4.587942123413086,
|
| 8285 |
+
"learning_rate": 1.3998575979229944e-05,
|
| 8286 |
+
"loss": 6.1981,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.13799964885585533,
|
| 8291 |
+
"grad_norm": 4.8053297996521,
|
| 8292 |
+
"learning_rate": 1.388493641575892e-05,
|
| 8293 |
+
"loss": 7.8202,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.13811669690407913,
|
| 8298 |
+
"grad_norm": 3.7762508392333984,
|
| 8299 |
+
"learning_rate": 1.3771725571930672e-05,
|
| 8300 |
+
"loss": 6.4063,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.13823374495230292,
|
| 8305 |
+
"grad_norm": 4.313928127288818,
|
| 8306 |
+
"learning_rate": 1.3658944011359897e-05,
|
| 8307 |
+
"loss": 6.9899,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.13835079300052672,
|
| 8312 |
+
"grad_norm": 3.4070425033569336,
|
| 8313 |
+
"learning_rate": 1.3546592295524074e-05,
|
| 8314 |
+
"loss": 4.5993,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.1384678410487505,
|
| 8319 |
+
"grad_norm": 3.8275327682495117,
|
| 8320 |
+
"learning_rate": 1.3434670983760755e-05,
|
| 8321 |
+
"loss": 5.5647,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.1385848890969743,
|
| 8326 |
+
"grad_norm": 3.8242149353027344,
|
| 8327 |
+
"learning_rate": 1.332318063326471e-05,
|
| 8328 |
+
"loss": 5.6159,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.1387019371451981,
|
| 8333 |
+
"grad_norm": 3.9725754261016846,
|
| 8334 |
+
"learning_rate": 1.3212121799085242e-05,
|
| 8335 |
+
"loss": 6.6851,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.1388189851934219,
|
| 8340 |
+
"grad_norm": 3.9829583168029785,
|
| 8341 |
+
"learning_rate": 1.3101495034123313e-05,
|
| 8342 |
+
"loss": 6.4324,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.1389360332416457,
|
| 8347 |
+
"grad_norm": 4.031448841094971,
|
| 8348 |
+
"learning_rate": 1.2991300889128866e-05,
|
| 8349 |
+
"loss": 6.7313,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.13905308128986948,
|
| 8354 |
+
"grad_norm": 3.654980421066284,
|
| 8355 |
+
"learning_rate": 1.2881539912698082e-05,
|
| 8356 |
+
"loss": 5.5801,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.13917012933809328,
|
| 8361 |
+
"grad_norm": 4.534533977508545,
|
| 8362 |
+
"learning_rate": 1.2772212651270666e-05,
|
| 8363 |
+
"loss": 6.5362,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.13928717738631707,
|
| 8368 |
+
"grad_norm": 4.087856769561768,
|
| 8369 |
+
"learning_rate": 1.2663319649127026e-05,
|
| 8370 |
+
"loss": 5.3049,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.13940422543454087,
|
| 8375 |
+
"grad_norm": 5.091198921203613,
|
| 8376 |
+
"learning_rate": 1.2554861448385636e-05,
|
| 8377 |
+
"loss": 6.9913,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.13952127348276466,
|
| 8382 |
+
"grad_norm": 4.729816436767578,
|
| 8383 |
+
"learning_rate": 1.2446838589000397e-05,
|
| 8384 |
+
"loss": 6.3906,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.13963832153098846,
|
| 8389 |
+
"grad_norm": 3.5689735412597656,
|
| 8390 |
+
"learning_rate": 1.2339251608757819e-05,
|
| 8391 |
+
"loss": 5.8104,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.13975536957921228,
|
| 8396 |
+
"grad_norm": 4.361255645751953,
|
| 8397 |
+
"learning_rate": 1.2232101043274436e-05,
|
| 8398 |
+
"loss": 6.1139,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.13987241762743607,
|
| 8403 |
+
"grad_norm": 5.06920862197876,
|
| 8404 |
+
"learning_rate": 1.2125387425994138e-05,
|
| 8405 |
+
"loss": 5.8007,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.13998946567565987,
|
| 8410 |
+
"grad_norm": 5.215285301208496,
|
| 8411 |
+
"learning_rate": 1.2019111288185437e-05,
|
| 8412 |
+
"loss": 8.4369,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.14010651372388366,
|
| 8417 |
+
"grad_norm": 4.381458282470703,
|
| 8418 |
+
"learning_rate": 1.1913273158938898e-05,
|
| 8419 |
+
"loss": 7.4393,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.14022356177210746,
|
| 8424 |
+
"grad_norm": 4.536776065826416,
|
| 8425 |
+
"learning_rate": 1.1807873565164506e-05,
|
| 8426 |
+
"loss": 5.11,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.14034060982033125,
|
| 8431 |
+
"grad_norm": 4.459112644195557,
|
| 8432 |
+
"learning_rate": 1.1702913031588969e-05,
|
| 8433 |
+
"loss": 5.7944,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.14045765786855505,
|
| 8438 |
+
"grad_norm": 4.268187999725342,
|
| 8439 |
+
"learning_rate": 1.1598392080753206e-05,
|
| 8440 |
+
"loss": 6.4448,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.14057470591677884,
|
| 8445 |
+
"grad_norm": 3.9678657054901123,
|
| 8446 |
+
"learning_rate": 1.149431123300967e-05,
|
| 8447 |
+
"loss": 5.8919,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.14069175396500264,
|
| 8452 |
+
"grad_norm": 5.267107009887695,
|
| 8453 |
+
"learning_rate": 1.139067100651976e-05,
|
| 8454 |
+
"loss": 6.3598,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.14080880201322643,
|
| 8459 |
+
"grad_norm": 3.5337772369384766,
|
| 8460 |
+
"learning_rate": 1.1287471917251291e-05,
|
| 8461 |
+
"loss": 5.4271,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.14092585006145023,
|
| 8466 |
+
"grad_norm": 3.8083269596099854,
|
| 8467 |
+
"learning_rate": 1.1184714478975934e-05,
|
| 8468 |
+
"loss": 6.5638,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.14104289810967402,
|
| 8473 |
+
"grad_norm": 4.413158893585205,
|
| 8474 |
+
"learning_rate": 1.1082399203266536e-05,
|
| 8475 |
+
"loss": 5.692,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.14115994615789781,
|
| 8480 |
+
"grad_norm": 3.8955037593841553,
|
| 8481 |
+
"learning_rate": 1.0980526599494733e-05,
|
| 8482 |
+
"loss": 6.7729,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.1412769942061216,
|
| 8487 |
+
"grad_norm": 3.710289478302002,
|
| 8488 |
+
"learning_rate": 1.0879097174828313e-05,
|
| 8489 |
+
"loss": 6.4334,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.1413940422543454,
|
| 8494 |
+
"grad_norm": 4.112654209136963,
|
| 8495 |
+
"learning_rate": 1.0778111434228689e-05,
|
| 8496 |
+
"loss": 5.5303,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.1415110903025692,
|
| 8501 |
+
"grad_norm": 4.588508129119873,
|
| 8502 |
+
"learning_rate": 1.067756988044848e-05,
|
| 8503 |
+
"loss": 7.4335,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.141628138350793,
|
| 8508 |
+
"grad_norm": 4.789179801940918,
|
| 8509 |
+
"learning_rate": 1.057747301402887e-05,
|
| 8510 |
+
"loss": 6.6236,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.1417451863990168,
|
| 8515 |
+
"grad_norm": 3.701345920562744,
|
| 8516 |
+
"learning_rate": 1.0477821333297255e-05,
|
| 8517 |
+
"loss": 5.2883,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.14186223444724058,
|
| 8522 |
+
"grad_norm": 4.159261703491211,
|
| 8523 |
+
"learning_rate": 1.0378615334364694e-05,
|
| 8524 |
+
"loss": 7.2449,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.14197928249546438,
|
| 8529 |
+
"grad_norm": 3.8805110454559326,
|
| 8530 |
+
"learning_rate": 1.0279855511123326e-05,
|
| 8531 |
+
"loss": 6.4649,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.14209633054368817,
|
| 8536 |
+
"grad_norm": 4.1621832847595215,
|
| 8537 |
+
"learning_rate": 1.0181542355244167e-05,
|
| 8538 |
+
"loss": 6.5887,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.142213378591912,
|
| 8543 |
+
"grad_norm": 4.077401161193848,
|
| 8544 |
+
"learning_rate": 1.0083676356174487e-05,
|
| 8545 |
+
"loss": 7.551,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.1423304266401358,
|
| 8550 |
+
"grad_norm": 4.078848838806152,
|
| 8551 |
+
"learning_rate": 9.98625800113534e-06,
|
| 8552 |
+
"loss": 7.5951,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.14244747468835958,
|
| 8557 |
+
"grad_norm": 3.4837894439697266,
|
| 8558 |
+
"learning_rate": 9.889287775119272e-06,
|
| 8559 |
+
"loss": 4.6695,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.14256452273658338,
|
| 8564 |
+
"grad_norm": 4.082892894744873,
|
| 8565 |
+
"learning_rate": 9.792766160887868e-06,
|
| 8566 |
+
"loss": 5.3314,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.14268157078480717,
|
| 8571 |
+
"grad_norm": 4.751600742340088,
|
| 8572 |
+
"learning_rate": 9.696693638969201e-06,
|
| 8573 |
+
"loss": 6.9273,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.14279861883303097,
|
| 8578 |
+
"grad_norm": 4.398174285888672,
|
| 8579 |
+
"learning_rate": 9.601070687655667e-06,
|
| 8580 |
+
"loss": 5.1706,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.14291566688125476,
|
| 8585 |
+
"grad_norm": 3.9142744541168213,
|
| 8586 |
+
"learning_rate": 9.505897783001483e-06,
|
| 8587 |
+
"loss": 6.5655,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.14303271492947855,
|
| 8592 |
+
"grad_norm": 3.7572720050811768,
|
| 8593 |
+
"learning_rate": 9.411175398820271e-06,
|
| 8594 |
+
"loss": 6.5502,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.14314976297770235,
|
| 8599 |
+
"grad_norm": 4.0552568435668945,
|
| 8600 |
+
"learning_rate": 9.316904006682847e-06,
|
| 8601 |
+
"loss": 6.2258,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.14326681102592614,
|
| 8606 |
+
"grad_norm": 4.32166862487793,
|
| 8607 |
+
"learning_rate": 9.223084075914712e-06,
|
| 8608 |
+
"loss": 4.6951,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.14338385907414994,
|
| 8613 |
+
"grad_norm": 3.618835926055908,
|
| 8614 |
+
"learning_rate": 9.129716073593797e-06,
|
| 8615 |
+
"loss": 5.3391,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.14350090712237373,
|
| 8620 |
+
"grad_norm": 3.269143581390381,
|
| 8621 |
+
"learning_rate": 9.036800464548157e-06,
|
| 8622 |
+
"loss": 5.5711,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.14361795517059753,
|
| 8627 |
+
"grad_norm": 4.386624813079834,
|
| 8628 |
+
"learning_rate": 8.944337711353646e-06,
|
| 8629 |
+
"loss": 5.6035,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.14373500321882132,
|
| 8634 |
+
"grad_norm": 3.8390941619873047,
|
| 8635 |
+
"learning_rate": 8.852328274331545e-06,
|
| 8636 |
+
"loss": 6.4548,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.14385205126704512,
|
| 8641 |
+
"grad_norm": 4.132387638092041,
|
| 8642 |
+
"learning_rate": 8.760772611546352e-06,
|
| 8643 |
+
"loss": 6.5087,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.1439690993152689,
|
| 8648 |
+
"grad_norm": 3.1986963748931885,
|
| 8649 |
+
"learning_rate": 8.669671178803485e-06,
|
| 8650 |
+
"loss": 4.2964,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.1440861473634927,
|
| 8655 |
+
"grad_norm": 3.56307053565979,
|
| 8656 |
+
"learning_rate": 8.579024429646932e-06,
|
| 8657 |
+
"loss": 5.8899,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.1442031954117165,
|
| 8662 |
+
"grad_norm": 3.848541736602783,
|
| 8663 |
+
"learning_rate": 8.488832815357173e-06,
|
| 8664 |
+
"loss": 5.4332,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.1443202434599403,
|
| 8669 |
+
"grad_norm": 3.926161766052246,
|
| 8670 |
+
"learning_rate": 8.399096784948712e-06,
|
| 8671 |
+
"loss": 6.3473,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.1444372915081641,
|
| 8676 |
+
"grad_norm": 3.7061004638671875,
|
| 8677 |
+
"learning_rate": 8.309816785168034e-06,
|
| 8678 |
+
"loss": 7.0913,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.14455433955638788,
|
| 8683 |
+
"grad_norm": 4.211513519287109,
|
| 8684 |
+
"learning_rate": 8.22099326049124e-06,
|
| 8685 |
+
"loss": 5.9397,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.1446713876046117,
|
| 8690 |
+
"grad_norm": 3.505610704421997,
|
| 8691 |
+
"learning_rate": 8.132626653121912e-06,
|
| 8692 |
+
"loss": 5.3207,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.1447884356528355,
|
| 8697 |
+
"grad_norm": 4.281852722167969,
|
| 8698 |
+
"learning_rate": 8.044717402988899e-06,
|
| 8699 |
+
"loss": 7.2471,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.1449054837010593,
|
| 8704 |
+
"grad_norm": 4.114862442016602,
|
| 8705 |
+
"learning_rate": 7.95726594774413e-06,
|
| 8706 |
+
"loss": 5.2312,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.1450225317492831,
|
| 8711 |
+
"grad_norm": 3.8156607151031494,
|
| 8712 |
+
"learning_rate": 7.870272722760386e-06,
|
| 8713 |
+
"loss": 6.8572,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.14513957979750688,
|
| 8718 |
+
"grad_norm": 4.314128398895264,
|
| 8719 |
+
"learning_rate": 7.783738161129206e-06,
|
| 8720 |
+
"loss": 6.6666,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.14525662784573068,
|
| 8725 |
+
"grad_norm": 4.673570156097412,
|
| 8726 |
+
"learning_rate": 7.697662693658691e-06,
|
| 8727 |
+
"loss": 5.9078,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.14537367589395447,
|
| 8732 |
+
"grad_norm": 3.7882399559020996,
|
| 8733 |
+
"learning_rate": 7.612046748871327e-06,
|
| 8734 |
+
"loss": 5.6484,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.14549072394217827,
|
| 8739 |
+
"grad_norm": 4.3639678955078125,
|
| 8740 |
+
"learning_rate": 7.526890753001936e-06,
|
| 8741 |
+
"loss": 6.0124,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.14560777199040206,
|
| 8746 |
+
"grad_norm": 4.610547065734863,
|
| 8747 |
+
"learning_rate": 7.44219512999551e-06,
|
| 8748 |
+
"loss": 5.6481,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.14572482003862586,
|
| 8753 |
+
"grad_norm": 4.014462471008301,
|
| 8754 |
+
"learning_rate": 7.3579603015050514e-06,
|
| 8755 |
+
"loss": 5.9941,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.14584186808684965,
|
| 8760 |
+
"grad_norm": 3.9123036861419678,
|
| 8761 |
+
"learning_rate": 7.2741866868895395e-06,
|
| 8762 |
+
"loss": 6.5066,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.14595891613507345,
|
| 8767 |
+
"grad_norm": 3.4274072647094727,
|
| 8768 |
+
"learning_rate": 7.1908747032118475e-06,
|
| 8769 |
+
"loss": 5.5077,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.14607596418329724,
|
| 8774 |
+
"grad_norm": 5.065500259399414,
|
| 8775 |
+
"learning_rate": 7.108024765236588e-06,
|
| 8776 |
+
"loss": 5.307,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.14619301223152104,
|
| 8781 |
+
"grad_norm": 4.586196422576904,
|
| 8782 |
+
"learning_rate": 7.0256372854281664e-06,
|
| 8783 |
+
"loss": 5.7997,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.14631006027974483,
|
| 8788 |
+
"grad_norm": 3.3022146224975586,
|
| 8789 |
+
"learning_rate": 6.943712673948644e-06,
|
| 8790 |
+
"loss": 5.0822,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.14642710832796862,
|
| 8795 |
+
"grad_norm": 3.9266810417175293,
|
| 8796 |
+
"learning_rate": 6.862251338655668e-06,
|
| 8797 |
+
"loss": 6.1271,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.14654415637619242,
|
| 8802 |
+
"grad_norm": 4.2883806228637695,
|
| 8803 |
+
"learning_rate": 6.78125368510053e-06,
|
| 8804 |
+
"loss": 7.4963,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.1466612044244162,
|
| 8809 |
+
"grad_norm": 4.894154071807861,
|
| 8810 |
+
"learning_rate": 6.700720116526116e-06,
|
| 8811 |
+
"loss": 6.4526,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.14677825247264,
|
| 8816 |
+
"grad_norm": 3.5264487266540527,
|
| 8817 |
+
"learning_rate": 6.620651033864844e-06,
|
| 8818 |
+
"loss": 5.0733,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.1468953005208638,
|
| 8823 |
+
"grad_norm": 3.764070749282837,
|
| 8824 |
+
"learning_rate": 6.541046835736753e-06,
|
| 8825 |
+
"loss": 4.8802,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.1470123485690876,
|
| 8830 |
+
"grad_norm": 2.602025032043457,
|
| 8831 |
+
"learning_rate": 6.461907918447463e-06,
|
| 8832 |
+
"loss": 3.7907,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.14712939661731142,
|
| 8837 |
+
"grad_norm": 4.767065525054932,
|
| 8838 |
+
"learning_rate": 6.3832346759861715e-06,
|
| 8839 |
+
"loss": 5.0684,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.14724644466553521,
|
| 8844 |
+
"grad_norm": 3.957653284072876,
|
| 8845 |
+
"learning_rate": 6.3050275000238414e-06,
|
| 8846 |
+
"loss": 5.9702,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.147363492713759,
|
| 8851 |
+
"grad_norm": 4.439967155456543,
|
| 8852 |
+
"learning_rate": 6.227286779911024e-06,
|
| 8853 |
+
"loss": 5.3836,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.1474805407619828,
|
| 8858 |
+
"grad_norm": 3.9382688999176025,
|
| 8859 |
+
"learning_rate": 6.150012902676139e-06,
|
| 8860 |
+
"loss": 6.1059,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.1475975888102066,
|
| 8865 |
+
"grad_norm": 4.807851314544678,
|
| 8866 |
+
"learning_rate": 6.073206253023411e-06,
|
| 8867 |
+
"loss": 8.5415,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.1477146368584304,
|
| 8872 |
+
"grad_norm": 4.171043872833252,
|
| 8873 |
+
"learning_rate": 5.996867213330992e-06,
|
| 8874 |
+
"loss": 6.8448,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.1478316849066542,
|
| 8879 |
+
"grad_norm": 3.0742545127868652,
|
| 8880 |
+
"learning_rate": 5.920996163649051e-06,
|
| 8881 |
+
"loss": 5.4045,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.14794873295487798,
|
| 8886 |
+
"grad_norm": 4.342021465301514,
|
| 8887 |
+
"learning_rate": 5.8455934816979305e-06,
|
| 8888 |
+
"loss": 8.3246,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.14806578100310178,
|
| 8893 |
+
"grad_norm": 4.4686784744262695,
|
| 8894 |
+
"learning_rate": 5.770659542866186e-06,
|
| 8895 |
+
"loss": 4.934,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.14818282905132557,
|
| 8900 |
+
"grad_norm": 4.780887126922607,
|
| 8901 |
+
"learning_rate": 5.696194720208792e-06,
|
| 8902 |
+
"loss": 6.1045,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.14829987709954937,
|
| 8907 |
+
"grad_norm": 3.9128787517547607,
|
| 8908 |
+
"learning_rate": 5.622199384445248e-06,
|
| 8909 |
+
"loss": 4.723,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.14841692514777316,
|
| 8914 |
+
"grad_norm": 4.773082256317139,
|
| 8915 |
+
"learning_rate": 5.54867390395768e-06,
|
| 8916 |
+
"loss": 6.9858,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.14853397319599695,
|
| 8921 |
+
"grad_norm": 5.843385219573975,
|
| 8922 |
+
"learning_rate": 5.475618644789126e-06,
|
| 8923 |
+
"loss": 8.4942,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.14865102124422075,
|
| 8928 |
+
"grad_norm": 4.275879383087158,
|
| 8929 |
+
"learning_rate": 5.403033970641647e-06,
|
| 8930 |
+
"loss": 7.2213,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.14876806929244454,
|
| 8935 |
+
"grad_norm": 4.2997612953186035,
|
| 8936 |
+
"learning_rate": 5.330920242874482e-06,
|
| 8937 |
+
"loss": 5.9771,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.14888511734066834,
|
| 8942 |
+
"grad_norm": 4.245179653167725,
|
| 8943 |
+
"learning_rate": 5.259277820502306e-06,
|
| 8944 |
+
"loss": 7.8182,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.14900216538889213,
|
| 8949 |
+
"grad_norm": 3.4496350288391113,
|
| 8950 |
+
"learning_rate": 5.188107060193459e-06,
|
| 8951 |
+
"loss": 5.7554,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.14911921343711593,
|
| 8956 |
+
"grad_norm": 3.8000648021698,
|
| 8957 |
+
"learning_rate": 5.1174083162680465e-06,
|
| 8958 |
+
"loss": 5.6367,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.14923626148533972,
|
| 8963 |
+
"grad_norm": 3.9714572429656982,
|
| 8964 |
+
"learning_rate": 5.047181940696333e-06,
|
| 8965 |
+
"loss": 6.7048,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.14935330953356352,
|
| 8970 |
+
"grad_norm": 3.9067649841308594,
|
| 8971 |
+
"learning_rate": 4.9774282830969296e-06,
|
| 8972 |
+
"loss": 5.4613,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.1494703575817873,
|
| 8977 |
+
"grad_norm": 4.62085485458374,
|
| 8978 |
+
"learning_rate": 4.908147690734977e-06,
|
| 8979 |
+
"loss": 5.9719,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.14958740563001113,
|
| 8984 |
+
"grad_norm": 3.595506429672241,
|
| 8985 |
+
"learning_rate": 4.839340508520562e-06,
|
| 8986 |
+
"loss": 6.4994,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.14970445367823493,
|
| 8991 |
+
"grad_norm": 3.4741034507751465,
|
| 8992 |
+
"learning_rate": 4.7710070790068725e-06,
|
| 8993 |
+
"loss": 5.2502,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.14982150172645872,
|
| 8998 |
+
"grad_norm": 3.96079158782959,
|
| 8999 |
+
"learning_rate": 4.703147742388536e-06,
|
| 9000 |
+
"loss": 5.5573,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.14993854977468252,
|
| 9005 |
+
"grad_norm": 6.466307640075684,
|
| 9006 |
+
"learning_rate": 4.635762836499968e-06,
|
| 9007 |
+
"loss": 6.8669,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.1500555978229063,
|
| 9012 |
+
"grad_norm": 5.081732273101807,
|
| 9013 |
+
"learning_rate": 4.568852696813619e-06,
|
| 9014 |
+
"loss": 7.5618,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.1501726458711301,
|
| 9019 |
+
"grad_norm": 4.093739032745361,
|
| 9020 |
+
"learning_rate": 4.502417656438362e-06,
|
| 9021 |
+
"loss": 6.4742,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.1502896939193539,
|
| 9026 |
+
"grad_norm": 4.444731712341309,
|
| 9027 |
+
"learning_rate": 4.436458046117831e-06,
|
| 9028 |
+
"loss": 6.6706,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.1504067419675777,
|
| 9033 |
+
"grad_norm": 4.529944896697998,
|
| 9034 |
+
"learning_rate": 4.370974194228694e-06,
|
| 9035 |
+
"loss": 7.2217,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.1505237900158015,
|
| 9040 |
+
"grad_norm": 4.988407611846924,
|
| 9041 |
+
"learning_rate": 4.305966426779118e-06,
|
| 9042 |
+
"loss": 7.1923,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.15064083806402528,
|
| 9047 |
+
"grad_norm": 4.239985942840576,
|
| 9048 |
+
"learning_rate": 4.241435067407107e-06,
|
| 9049 |
+
"loss": 5.4475,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.15075788611224908,
|
| 9054 |
+
"grad_norm": 3.7280960083007812,
|
| 9055 |
+
"learning_rate": 4.177380437378886e-06,
|
| 9056 |
+
"loss": 6.4699,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.15087493416047287,
|
| 9061 |
+
"grad_norm": 3.6265618801116943,
|
| 9062 |
+
"learning_rate": 4.113802855587312e-06,
|
| 9063 |
+
"loss": 6.6276,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.15099198220869667,
|
| 9068 |
+
"grad_norm": 3.895657777786255,
|
| 9069 |
+
"learning_rate": 4.050702638550275e-06,
|
| 9070 |
+
"loss": 7.2802,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.15110903025692046,
|
| 9075 |
+
"grad_norm": 3.437020778656006,
|
| 9076 |
+
"learning_rate": 3.988080100409098e-06,
|
| 9077 |
+
"loss": 6.7151,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.15122607830514426,
|
| 9082 |
+
"grad_norm": 4.021770000457764,
|
| 9083 |
+
"learning_rate": 3.9259355529270625e-06,
|
| 9084 |
+
"loss": 6.0971,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.15134312635336805,
|
| 9089 |
+
"grad_norm": 4.238542556762695,
|
| 9090 |
+
"learning_rate": 3.864269305487756e-06,
|
| 9091 |
+
"loss": 7.0094,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.15146017440159185,
|
| 9096 |
+
"grad_norm": 4.376027584075928,
|
| 9097 |
+
"learning_rate": 3.8030816650935776e-06,
|
| 9098 |
+
"loss": 7.0964,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.15157722244981564,
|
| 9103 |
+
"grad_norm": 4.593097686767578,
|
| 9104 |
+
"learning_rate": 3.7423729363642356e-06,
|
| 9105 |
+
"loss": 6.5553,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.15169427049803944,
|
| 9110 |
+
"grad_norm": 3.8831586837768555,
|
| 9111 |
+
"learning_rate": 3.682143421535156e-06,
|
| 9112 |
+
"loss": 5.7876,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.15181131854626323,
|
| 9117 |
+
"grad_norm": 3.351270914077759,
|
| 9118 |
+
"learning_rate": 3.622393420456016e-06,
|
| 9119 |
+
"loss": 4.9215,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.15192836659448702,
|
| 9124 |
+
"grad_norm": 3.97438383102417,
|
| 9125 |
+
"learning_rate": 3.5631232305893046e-06,
|
| 9126 |
+
"loss": 6.8098,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.15204541464271085,
|
| 9131 |
+
"grad_norm": 4.048185348510742,
|
| 9132 |
+
"learning_rate": 3.504333147008787e-06,
|
| 9133 |
+
"loss": 5.711,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.15216246269093464,
|
| 9138 |
+
"grad_norm": 3.498582124710083,
|
| 9139 |
+
"learning_rate": 3.446023462398018e-06,
|
| 9140 |
+
"loss": 5.8116,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.15227951073915844,
|
| 9145 |
+
"grad_norm": 3.9243156909942627,
|
| 9146 |
+
"learning_rate": 3.3881944670488776e-06,
|
| 9147 |
+
"loss": 6.6084,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.15239655878738223,
|
| 9152 |
+
"grad_norm": 3.5060412883758545,
|
| 9153 |
+
"learning_rate": 3.330846448860259e-06,
|
| 9154 |
+
"loss": 6.7144,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.15251360683560602,
|
| 9159 |
+
"grad_norm": 4.419942378997803,
|
| 9160 |
+
"learning_rate": 3.2739796933364155e-06,
|
| 9161 |
+
"loss": 6.0434,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.15263065488382982,
|
| 9166 |
+
"grad_norm": 3.8820512294769287,
|
| 9167 |
+
"learning_rate": 3.2175944835857617e-06,
|
| 9168 |
+
"loss": 5.3865,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.15274770293205361,
|
| 9173 |
+
"grad_norm": 3.651808500289917,
|
| 9174 |
+
"learning_rate": 3.161691100319286e-06,
|
| 9175 |
+
"loss": 4.0544,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.1528647509802774,
|
| 9180 |
+
"grad_norm": 4.261025428771973,
|
| 9181 |
+
"learning_rate": 3.1062698218492724e-06,
|
| 9182 |
+
"loss": 7.047,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.1529817990285012,
|
| 9187 |
+
"grad_norm": 4.000607967376709,
|
| 9188 |
+
"learning_rate": 3.0513309240878253e-06,
|
| 9189 |
+
"loss": 6.6384,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.153098847076725,
|
| 9194 |
+
"grad_norm": 5.608572959899902,
|
| 9195 |
+
"learning_rate": 2.996874680545603e-06,
|
| 9196 |
+
"loss": 7.3719,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.1532158951249488,
|
| 9201 |
+
"grad_norm": 4.026655673980713,
|
| 9202 |
+
"learning_rate": 2.9429013623303304e-06,
|
| 9203 |
+
"loss": 6.3124,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.1533329431731726,
|
| 9208 |
+
"grad_norm": 4.070751667022705,
|
| 9209 |
+
"learning_rate": 2.889411238145545e-06,
|
| 9210 |
+
"loss": 6.1873,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.15344999122139638,
|
| 9215 |
+
"grad_norm": 4.576844215393066,
|
| 9216 |
+
"learning_rate": 2.8364045742892308e-06,
|
| 9217 |
+
"loss": 7.2124,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.15356703926962018,
|
| 9222 |
+
"grad_norm": 4.491173267364502,
|
| 9223 |
+
"learning_rate": 2.7838816346524634e-06,
|
| 9224 |
+
"loss": 7.2297,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.15368408731784397,
|
| 9229 |
+
"grad_norm": 3.549685478210449,
|
| 9230 |
+
"learning_rate": 2.7318426807181442e-06,
|
| 9231 |
+
"loss": 5.0413,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.15380113536606776,
|
| 9236 |
+
"grad_norm": 4.685144901275635,
|
| 9237 |
+
"learning_rate": 2.6802879715596585e-06,
|
| 9238 |
+
"loss": 6.6046,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.15391818341429156,
|
| 9243 |
+
"grad_norm": 4.207958698272705,
|
| 9244 |
+
"learning_rate": 2.6292177638396086e-06,
|
| 9245 |
+
"loss": 5.6841,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.15403523146251535,
|
| 9250 |
+
"grad_norm": 3.4470295906066895,
|
| 9251 |
+
"learning_rate": 2.5786323118085596e-06,
|
| 9252 |
+
"loss": 6.3598,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.15415227951073915,
|
| 9257 |
+
"grad_norm": 3.9577460289001465,
|
| 9258 |
+
"learning_rate": 2.528531867303696e-06,
|
| 9259 |
+
"loss": 6.2133,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.15426932755896294,
|
| 9264 |
+
"grad_norm": 4.349520683288574,
|
| 9265 |
+
"learning_rate": 2.4789166797476228e-06,
|
| 9266 |
+
"loss": 6.48,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.15438637560718674,
|
| 9271 |
+
"grad_norm": 4.203663349151611,
|
| 9272 |
+
"learning_rate": 2.429786996147154e-06,
|
| 9273 |
+
"loss": 6.9134,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.15450342365541053,
|
| 9278 |
+
"grad_norm": 4.4230451583862305,
|
| 9279 |
+
"learning_rate": 2.381143061092006e-06,
|
| 9280 |
+
"loss": 6.4944,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.15462047170363435,
|
| 9285 |
+
"grad_norm": 4.428316116333008,
|
| 9286 |
+
"learning_rate": 2.332985116753639e-06,
|
| 9287 |
+
"loss": 6.1822,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.15473751975185815,
|
| 9292 |
+
"grad_norm": 4.547220230102539,
|
| 9293 |
+
"learning_rate": 2.2853134028840594e-06,
|
| 9294 |
+
"loss": 6.831,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.15485456780008194,
|
| 9299 |
+
"grad_norm": 3.5556628704071045,
|
| 9300 |
+
"learning_rate": 2.238128156814523e-06,
|
| 9301 |
+
"loss": 5.236,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.15497161584830574,
|
| 9306 |
+
"grad_norm": 4.256648063659668,
|
| 9307 |
+
"learning_rate": 2.191429613454499e-06,
|
| 9308 |
+
"loss": 4.1367,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.15508866389652953,
|
| 9313 |
+
"grad_norm": 3.9103307723999023,
|
| 9314 |
+
"learning_rate": 2.145218005290406e-06,
|
| 9315 |
+
"loss": 6.7366,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.15520571194475333,
|
| 9320 |
+
"grad_norm": 3.4093990325927734,
|
| 9321 |
+
"learning_rate": 2.0994935623844692e-06,
|
| 9322 |
+
"loss": 5.3234,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.15532275999297712,
|
| 9327 |
+
"grad_norm": 2.927978038787842,
|
| 9328 |
+
"learning_rate": 2.054256512373631e-06,
|
| 9329 |
+
"loss": 4.6302,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.15543980804120092,
|
| 9334 |
+
"grad_norm": 4.108157634735107,
|
| 9335 |
+
"learning_rate": 2.009507080468298e-06,
|
| 9336 |
+
"loss": 6.3839,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.1555568560894247,
|
| 9341 |
+
"grad_norm": 3.956204414367676,
|
| 9342 |
+
"learning_rate": 1.9652454894513504e-06,
|
| 9343 |
+
"loss": 5.6238,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.1556739041376485,
|
| 9348 |
+
"grad_norm": 5.535799503326416,
|
| 9349 |
+
"learning_rate": 1.921471959676957e-06,
|
| 9350 |
+
"loss": 7.456,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.1557909521858723,
|
| 9355 |
+
"grad_norm": 3.87038516998291,
|
| 9356 |
+
"learning_rate": 1.8781867090695071e-06,
|
| 9357 |
+
"loss": 5.5837,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.1559080002340961,
|
| 9362 |
+
"grad_norm": 4.128233909606934,
|
| 9363 |
+
"learning_rate": 1.8353899531224906e-06,
|
| 9364 |
+
"loss": 6.6661,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.1560250482823199,
|
| 9369 |
+
"grad_norm": 3.827026844024658,
|
| 9370 |
+
"learning_rate": 1.793081904897498e-06,
|
| 9371 |
+
"loss": 6.5843,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.15614209633054368,
|
| 9376 |
+
"grad_norm": 4.454151153564453,
|
| 9377 |
+
"learning_rate": 1.751262775023077e-06,
|
| 9378 |
+
"loss": 5.8657,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.15625914437876748,
|
| 9383 |
+
"grad_norm": 4.341606616973877,
|
| 9384 |
+
"learning_rate": 1.7099327716936896e-06,
|
| 9385 |
+
"loss": 6.8369,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.15637619242699127,
|
| 9390 |
+
"grad_norm": 3.8359739780426025,
|
| 9391 |
+
"learning_rate": 1.6690921006687899e-06,
|
| 9392 |
+
"loss": 6.4584,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.15649324047521507,
|
| 9397 |
+
"grad_norm": 3.7251949310302734,
|
| 9398 |
+
"learning_rate": 1.6287409652716356e-06,
|
| 9399 |
+
"loss": 5.3272,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.15661028852343886,
|
| 9404 |
+
"grad_norm": 3.5989632606506348,
|
| 9405 |
+
"learning_rate": 1.5888795663883904e-06,
|
| 9406 |
+
"loss": 6.3437,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.15672733657166266,
|
| 9411 |
+
"grad_norm": 3.4688527584075928,
|
| 9412 |
+
"learning_rate": 1.549508102467101e-06,
|
| 9413 |
+
"loss": 6.0845,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.15684438461988645,
|
| 9418 |
+
"grad_norm": 5.154304027557373,
|
| 9419 |
+
"learning_rate": 1.5106267695166543e-06,
|
| 9420 |
+
"loss": 7.1261,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.15696143266811025,
|
| 9425 |
+
"grad_norm": 4.6669921875,
|
| 9426 |
+
"learning_rate": 1.472235761105878e-06,
|
| 9427 |
+
"loss": 6.0136,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.15707848071633407,
|
| 9432 |
+
"grad_norm": 4.336490154266357,
|
| 9433 |
+
"learning_rate": 1.4343352683625411e-06,
|
| 9434 |
+
"loss": 6.6578,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.15719552876455786,
|
| 9439 |
+
"grad_norm": 6.8128814697265625,
|
| 9440 |
+
"learning_rate": 1.3969254799723997e-06,
|
| 9441 |
+
"loss": 5.6969,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.15731257681278166,
|
| 9446 |
+
"grad_norm": 3.7662675380706787,
|
| 9447 |
+
"learning_rate": 1.3600065821782526e-06,
|
| 9448 |
+
"loss": 6.8376,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.15742962486100545,
|
| 9453 |
+
"grad_norm": 4.1669840812683105,
|
| 9454 |
+
"learning_rate": 1.3235787587790426e-06,
|
| 9455 |
+
"loss": 5.9976,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.15754667290922925,
|
| 9460 |
+
"grad_norm": 3.880065441131592,
|
| 9461 |
+
"learning_rate": 1.2876421911288905e-06,
|
| 9462 |
+
"loss": 6.0656,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.15766372095745304,
|
| 9467 |
+
"grad_norm": 4.043405532836914,
|
| 9468 |
+
"learning_rate": 1.2521970581362507e-06,
|
| 9469 |
+
"loss": 5.9426,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.15778076900567684,
|
| 9474 |
+
"grad_norm": 4.752688407897949,
|
| 9475 |
+
"learning_rate": 1.217243536263013e-06,
|
| 9476 |
+
"loss": 7.8083,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.15789781705390063,
|
| 9481 |
+
"grad_norm": 3.737199544906616,
|
| 9482 |
+
"learning_rate": 1.182781799523547e-06,
|
| 9483 |
+
"loss": 5.9821,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.15801486510212442,
|
| 9488 |
+
"grad_norm": 4.315102577209473,
|
| 9489 |
+
"learning_rate": 1.148812019483958e-06,
|
| 9490 |
+
"loss": 6.5228,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.15813191315034822,
|
| 9495 |
+
"grad_norm": 3.739389657974243,
|
| 9496 |
+
"learning_rate": 1.1153343652611227e-06,
|
| 9497 |
+
"loss": 7.5063,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.158248961198572,
|
| 9502 |
+
"grad_norm": 4.729193687438965,
|
| 9503 |
+
"learning_rate": 1.0823490035218987e-06,
|
| 9504 |
+
"loss": 6.2379,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.1583660092467958,
|
| 9509 |
+
"grad_norm": 4.6792988777160645,
|
| 9510 |
+
"learning_rate": 1.049856098482327e-06,
|
| 9511 |
+
"loss": 6.5573,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.1584830572950196,
|
| 9516 |
+
"grad_norm": 4.178571701049805,
|
| 9517 |
+
"learning_rate": 1.0178558119067315e-06,
|
| 9518 |
+
"loss": 5.8899,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.1586001053432434,
|
| 9523 |
+
"grad_norm": 4.782796382904053,
|
| 9524 |
+
"learning_rate": 9.863483031069764e-07,
|
| 9525 |
+
"loss": 7.5387,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.1587171533914672,
|
| 9530 |
+
"grad_norm": 3.7984838485717773,
|
| 9531 |
+
"learning_rate": 9.553337289416653e-07,
|
| 9532 |
+
"loss": 4.3348,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.158834201439691,
|
| 9537 |
+
"grad_norm": 4.128501892089844,
|
| 9538 |
+
"learning_rate": 9.248122438153429e-07,
|
| 9539 |
+
"loss": 4.9796,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.15895124948791478,
|
| 9544 |
+
"grad_norm": 4.67678165435791,
|
| 9545 |
+
"learning_rate": 8.947839996777285e-07,
|
| 9546 |
+
"loss": 6.2732,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.15906829753613858,
|
| 9551 |
+
"grad_norm": 3.0954668521881104,
|
| 9552 |
+
"learning_rate": 8.652491460229728e-07,
|
| 9553 |
+
"loss": 5.2698,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.15918534558436237,
|
| 9558 |
+
"grad_norm": 3.4296183586120605,
|
| 9559 |
+
"learning_rate": 8.362078298889021e-07,
|
| 9560 |
+
"loss": 5.2506,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.15930239363258616,
|
| 9565 |
+
"grad_norm": 3.3314638137817383,
|
| 9566 |
+
"learning_rate": 8.076601958562968e-07,
|
| 9567 |
+
"loss": 5.7465,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.15941944168080996,
|
| 9572 |
+
"grad_norm": 4.422887802124023,
|
| 9573 |
+
"learning_rate": 7.796063860481595e-07,
|
| 9574 |
+
"loss": 6.7263,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.15953648972903378,
|
| 9579 |
+
"grad_norm": 3.5842325687408447,
|
| 9580 |
+
"learning_rate": 7.520465401290033e-07,
|
| 9581 |
+
"loss": 4.8379,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.15965353777725758,
|
| 9586 |
+
"grad_norm": 4.023311614990234,
|
| 9587 |
+
"learning_rate": 7.249807953041865e-07,
|
| 9588 |
+
"loss": 5.214,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.15977058582548137,
|
| 9593 |
+
"grad_norm": 4.647928714752197,
|
| 9594 |
+
"learning_rate": 6.984092863192016e-07,
|
| 9595 |
+
"loss": 5.8553,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.15988763387370516,
|
| 9600 |
+
"grad_norm": 4.061383247375488,
|
| 9601 |
+
"learning_rate": 6.723321454590092e-07,
|
| 9602 |
+
"loss": 6.4471,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.16000468192192896,
|
| 9607 |
+
"grad_norm": 4.1114068031311035,
|
| 9608 |
+
"learning_rate": 6.467495025473614e-07,
|
| 9609 |
+
"loss": 5.6854,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.16012172997015275,
|
| 9614 |
+
"grad_norm": 5.476286888122559,
|
| 9615 |
+
"learning_rate": 6.216614849462343e-07,
|
| 9616 |
+
"loss": 6.2115,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.16023877801837655,
|
| 9621 |
+
"grad_norm": 3.9494688510894775,
|
| 9622 |
+
"learning_rate": 5.970682175550745e-07,
|
| 9623 |
+
"loss": 5.5924,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.16035582606660034,
|
| 9628 |
+
"grad_norm": 5.803978443145752,
|
| 9629 |
+
"learning_rate": 5.729698228102653e-07,
|
| 9630 |
+
"loss": 7.363,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.16047287411482414,
|
| 9635 |
+
"grad_norm": 4.088033676147461,
|
| 9636 |
+
"learning_rate": 5.49366420684505e-07,
|
| 9637 |
+
"loss": 5.8067,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.16058992216304793,
|
| 9642 |
+
"grad_norm": 4.184268951416016,
|
| 9643 |
+
"learning_rate": 5.262581286861634e-07,
|
| 9644 |
+
"loss": 7.9438,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.16070697021127173,
|
| 9649 |
+
"grad_norm": 4.242025852203369,
|
| 9650 |
+
"learning_rate": 5.036450618587485e-07,
|
| 9651 |
+
"loss": 7.2965,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.16082401825949552,
|
| 9656 |
+
"grad_norm": 4.872281551361084,
|
| 9657 |
+
"learning_rate": 4.815273327803182e-07,
|
| 9658 |
+
"loss": 7.8095,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.16094106630771932,
|
| 9663 |
+
"grad_norm": 3.669004440307617,
|
| 9664 |
+
"learning_rate": 4.599050515628922e-07,
|
| 9665 |
+
"loss": 6.6182,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.1610581143559431,
|
| 9670 |
+
"grad_norm": 3.983104705810547,
|
| 9671 |
+
"learning_rate": 4.387783258519518e-07,
|
| 9672 |
+
"loss": 5.4207,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.1611751624041669,
|
| 9677 |
+
"grad_norm": 4.213310241699219,
|
| 9678 |
+
"learning_rate": 4.1814726082588516e-07,
|
| 9679 |
+
"loss": 5.407,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.1612922104523907,
|
| 9684 |
+
"grad_norm": 3.563631296157837,
|
| 9685 |
+
"learning_rate": 3.9801195919541014e-07,
|
| 9686 |
+
"loss": 6.6246,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.1614092585006145,
|
| 9691 |
+
"grad_norm": 3.3321008682250977,
|
| 9692 |
+
"learning_rate": 3.7837252120317415e-07,
|
| 9693 |
+
"loss": 5.045,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.1615263065488383,
|
| 9698 |
+
"grad_norm": 3.832080841064453,
|
| 9699 |
+
"learning_rate": 3.5922904462314387e-07,
|
| 9700 |
+
"loss": 5.2409,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.16164335459706208,
|
| 9705 |
+
"grad_norm": 3.959026575088501,
|
| 9706 |
+
"learning_rate": 3.405816247601834e-07,
|
| 9707 |
+
"loss": 7.1682,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.16176040264528588,
|
| 9712 |
+
"grad_norm": 4.033455848693848,
|
| 9713 |
+
"learning_rate": 3.224303544495766e-07,
|
| 9714 |
+
"loss": 5.2303,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.16187745069350967,
|
| 9719 |
+
"grad_norm": 3.818906545639038,
|
| 9720 |
+
"learning_rate": 3.0477532405652763e-07,
|
| 9721 |
+
"loss": 5.657,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.1619944987417335,
|
| 9726 |
+
"grad_norm": 6.2514142990112305,
|
| 9727 |
+
"learning_rate": 2.876166214757392e-07,
|
| 9728 |
+
"loss": 6.9008,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.1621115467899573,
|
| 9733 |
+
"grad_norm": 3.2104997634887695,
|
| 9734 |
+
"learning_rate": 2.709543321309793e-07,
|
| 9735 |
+
"loss": 4.9992,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.16222859483818108,
|
| 9740 |
+
"grad_norm": 3.438826084136963,
|
| 9741 |
+
"learning_rate": 2.547885389746485e-07,
|
| 9742 |
+
"loss": 4.4849,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.16234564288640488,
|
| 9747 |
+
"grad_norm": 4.253631591796875,
|
| 9748 |
+
"learning_rate": 2.391193224873467e-07,
|
| 9749 |
+
"loss": 6.8902,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.16246269093462867,
|
| 9754 |
+
"grad_norm": 3.2461020946502686,
|
| 9755 |
+
"learning_rate": 2.2394676067752918e-07,
|
| 9756 |
+
"loss": 5.3384,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.16257973898285247,
|
| 9761 |
+
"grad_norm": 5.023898601531982,
|
| 9762 |
+
"learning_rate": 2.0927092908101796e-07,
|
| 9763 |
+
"loss": 7.4382,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.16269678703107626,
|
| 9768 |
+
"grad_norm": 4.078502655029297,
|
| 9769 |
+
"learning_rate": 1.9509190076074657e-07,
|
| 9770 |
+
"loss": 5.3726,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.16281383507930006,
|
| 9775 |
+
"grad_norm": 4.447368621826172,
|
| 9776 |
+
"learning_rate": 1.814097463063269e-07,
|
| 9777 |
+
"loss": 6.0735,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.16293088312752385,
|
| 9782 |
+
"grad_norm": 4.023862361907959,
|
| 9783 |
+
"learning_rate": 1.6822453383367186e-07,
|
| 9784 |
+
"loss": 6.4528,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.16304793117574765,
|
| 9789 |
+
"grad_norm": 4.2901291847229,
|
| 9790 |
+
"learning_rate": 1.5553632898472892e-07,
|
| 9791 |
+
"loss": 8.4979,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.16316497922397144,
|
| 9796 |
+
"grad_norm": 3.606577157974243,
|
| 9797 |
+
"learning_rate": 1.4334519492711362e-07,
|
| 9798 |
+
"loss": 6.1818,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.16328202727219523,
|
| 9803 |
+
"grad_norm": 4.6511945724487305,
|
| 9804 |
+
"learning_rate": 1.3165119235377665e-07,
|
| 9805 |
+
"loss": 7.7793,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.16339907532041903,
|
| 9810 |
+
"grad_norm": 4.561491012573242,
|
| 9811 |
+
"learning_rate": 1.204543794827595e-07,
|
| 9812 |
+
"loss": 5.1814,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.16351612336864282,
|
| 9817 |
+
"grad_norm": 4.02864408493042,
|
| 9818 |
+
"learning_rate": 1.0975481205683924e-07,
|
| 9819 |
+
"loss": 5.8196,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.16363317141686662,
|
| 9824 |
+
"grad_norm": 3.305347442626953,
|
| 9825 |
+
"learning_rate": 9.955254334328423e-08,
|
| 9826 |
+
"loss": 4.7082,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.1637502194650904,
|
| 9831 |
+
"grad_norm": 4.880128860473633,
|
| 9832 |
+
"learning_rate": 8.984762413359881e-08,
|
| 9833 |
+
"loss": 5.2619,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.1638672675133142,
|
| 9838 |
+
"grad_norm": 4.935993671417236,
|
| 9839 |
+
"learning_rate": 8.064010274324573e-08,
|
| 9840 |
+
"loss": 6.816,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.163984315561538,
|
| 9845 |
+
"grad_norm": 3.9603235721588135,
|
| 9846 |
+
"learning_rate": 7.19300250114241e-08,
|
| 9847 |
+
"loss": 5.4697,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.1641013636097618,
|
| 9852 |
+
"grad_norm": 3.8613545894622803,
|
| 9853 |
+
"learning_rate": 6.37174343008251e-08,
|
| 9854 |
+
"loss": 6.3692,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.1642184116579856,
|
| 9859 |
+
"grad_norm": 3.9567618370056152,
|
| 9860 |
+
"learning_rate": 5.600237149744336e-08,
|
| 9861 |
+
"loss": 5.542,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.16433545970620939,
|
| 9866 |
+
"grad_norm": 4.240719795227051,
|
| 9867 |
+
"learning_rate": 4.878487501033258e-08,
|
| 9868 |
+
"loss": 7.4315,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.1644525077544332,
|
| 9873 |
+
"grad_norm": 3.8969204425811768,
|
| 9874 |
+
"learning_rate": 4.206498077145016e-08,
|
| 9875 |
+
"loss": 6.1796,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.164569555802657,
|
| 9880 |
+
"grad_norm": 3.9615015983581543,
|
| 9881 |
+
"learning_rate": 3.584272223546847e-08,
|
| 9882 |
+
"loss": 6.6647,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.1646866038508808,
|
| 9887 |
+
"grad_norm": 4.455677509307861,
|
| 9888 |
+
"learning_rate": 3.0118130379575005e-08,
|
| 9889 |
+
"loss": 6.8467,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.1648036518991046,
|
| 9894 |
+
"grad_norm": 6.2688398361206055,
|
| 9895 |
+
"learning_rate": 2.4891233703394634e-08,
|
| 9896 |
+
"loss": 8.9478,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.1649206999473284,
|
| 9901 |
+
"grad_norm": 4.4571003913879395,
|
| 9902 |
+
"learning_rate": 2.01620582287676e-08,
|
| 9903 |
+
"loss": 7.2163,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.16503774799555218,
|
| 9908 |
+
"grad_norm": 4.036981105804443,
|
| 9909 |
+
"learning_rate": 1.593062749967178e-08,
|
| 9910 |
+
"loss": 6.3315,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.16515479604377598,
|
| 9915 |
+
"grad_norm": 4.721566200256348,
|
| 9916 |
+
"learning_rate": 1.2196962582067262e-08,
|
| 9917 |
+
"loss": 6.8959,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.16527184409199977,
|
| 9922 |
+
"grad_norm": 4.5873212814331055,
|
| 9923 |
+
"learning_rate": 8.961082063829729e-09,
|
| 9924 |
+
"loss": 7.9275,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.16538889214022356,
|
| 9929 |
+
"grad_norm": 4.162585258483887,
|
| 9930 |
+
"learning_rate": 6.223002054617233e-09,
|
| 9931 |
+
"loss": 5.7238,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.16550594018844736,
|
| 9936 |
+
"grad_norm": 4.646335601806641,
|
| 9937 |
+
"learning_rate": 3.982736185859093e-09,
|
| 9938 |
+
"loss": 6.4431,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.16562298823667115,
|
| 9943 |
+
"grad_norm": 4.095743656158447,
|
| 9944 |
+
"learning_rate": 2.240295610578258e-09,
|
| 9945 |
+
"loss": 6.9833,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.16574003628489495,
|
| 9950 |
+
"grad_norm": 4.3485822677612305,
|
| 9951 |
+
"learning_rate": 9.956890034468202e-10,
|
| 9952 |
+
"loss": 6.0359,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.16585708433311874,
|
| 9957 |
+
"grad_norm": 5.006742477416992,
|
| 9958 |
+
"learning_rate": 2.489225606749912e-10,
|
| 9959 |
+
"loss": 7.3317,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.16597413238134254,
|
| 9964 |
+
"grad_norm": 3.7422544956207275,
|
| 9965 |
+
"learning_rate": 0.0,
|
| 9966 |
+
"loss": 4.7782,
|
| 9967 |
+
"step": 1418
|
| 9968 |
}
|
| 9969 |
],
|
| 9970 |
"logging_steps": 1,
|
|
|
|
| 9979 |
"should_evaluate": false,
|
| 9980 |
"should_log": false,
|
| 9981 |
"should_save": true,
|
| 9982 |
+
"should_training_stop": true
|
| 9983 |
},
|
| 9984 |
"attributes": {}
|
| 9985 |
}
|
| 9986 |
},
|
| 9987 |
+
"total_flos": 1.301657889621934e+17,
|
| 9988 |
"train_batch_size": 2,
|
| 9989 |
"trial_name": null,
|
| 9990 |
"trial_params": null
|