Training in progress, step 983, 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 83945296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26571770744a2807e54634950fe1a844a9616c0d217722286318a72990944b11
|
| 3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 43123028
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13c62a54b0407f9aaac98a51a0df34b5186a5becbfede0057c6af56407d98e59
|
| 3 |
size 43123028
|
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 sha256:911f2aa53c74eda8eabf6de4806dfe65f05caea3cb99728abd8c624a00ae6106
|
| 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:b53ae7c84062c49d524fd48890ecb5e8a3503965f943a71edc375aa9a9c5b4a8
|
| 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":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5173,6 +5173,1729 @@
|
|
| 5173 |
"learning_rate": 1.470149183220748e-05,
|
| 5174 |
"loss": 2.0191,
|
| 5175 |
"step": 738
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5176 |
}
|
| 5177 |
],
|
| 5178 |
"logging_steps": 1,
|
|
@@ -5187,12 +6910,12 @@
|
|
| 5187 |
"should_evaluate": false,
|
| 5188 |
"should_log": false,
|
| 5189 |
"should_save": true,
|
| 5190 |
-
"should_training_stop":
|
| 5191 |
},
|
| 5192 |
"attributes": {}
|
| 5193 |
}
|
| 5194 |
},
|
| 5195 |
-
"total_flos":
|
| 5196 |
"train_batch_size": 4,
|
| 5197 |
"trial_name": null,
|
| 5198 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.000763553066938,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 983,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5173 |
"learning_rate": 1.470149183220748e-05,
|
| 5174 |
"loss": 2.0191,
|
| 5175 |
"step": 738
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.7523542886230593,
|
| 5179 |
+
"grad_norm": 6.6719584465026855,
|
| 5180 |
+
"learning_rate": 1.4587921253782849e-05,
|
| 5181 |
+
"loss": 1.6597,
|
| 5182 |
+
"step": 739
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.7533723593789768,
|
| 5186 |
+
"grad_norm": 8.974640846252441,
|
| 5187 |
+
"learning_rate": 1.447471607917854e-05,
|
| 5188 |
+
"loss": 1.9953,
|
| 5189 |
+
"step": 740
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.7543904301348944,
|
| 5193 |
+
"grad_norm": 7.379059314727783,
|
| 5194 |
+
"learning_rate": 1.4361877476515889e-05,
|
| 5195 |
+
"loss": 1.8422,
|
| 5196 |
+
"step": 741
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.7554085008908119,
|
| 5200 |
+
"grad_norm": 7.163296699523926,
|
| 5201 |
+
"learning_rate": 1.4249406610133686e-05,
|
| 5202 |
+
"loss": 1.8372,
|
| 5203 |
+
"step": 742
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.7564265716467294,
|
| 5207 |
+
"grad_norm": 10.26382064819336,
|
| 5208 |
+
"learning_rate": 1.413730464057616e-05,
|
| 5209 |
+
"loss": 2.2328,
|
| 5210 |
+
"step": 743
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.757444642402647,
|
| 5214 |
+
"grad_norm": 7.997495651245117,
|
| 5215 |
+
"learning_rate": 1.4025572724581038e-05,
|
| 5216 |
+
"loss": 2.083,
|
| 5217 |
+
"step": 744
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.7584627131585645,
|
| 5221 |
+
"grad_norm": 8.966462135314941,
|
| 5222 |
+
"learning_rate": 1.3914212015067651e-05,
|
| 5223 |
+
"loss": 2.0716,
|
| 5224 |
+
"step": 745
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.759480783914482,
|
| 5228 |
+
"grad_norm": 9.182121276855469,
|
| 5229 |
+
"learning_rate": 1.3803223661124936e-05,
|
| 5230 |
+
"loss": 2.406,
|
| 5231 |
+
"step": 746
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.7604988546703996,
|
| 5235 |
+
"grad_norm": 8.59941577911377,
|
| 5236 |
+
"learning_rate": 1.3692608807999652e-05,
|
| 5237 |
+
"loss": 2.1763,
|
| 5238 |
+
"step": 747
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.7615169254263171,
|
| 5242 |
+
"grad_norm": 11.990951538085938,
|
| 5243 |
+
"learning_rate": 1.3582368597084566e-05,
|
| 5244 |
+
"loss": 2.7291,
|
| 5245 |
+
"step": 748
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.7625349961822346,
|
| 5249 |
+
"grad_norm": 7.6986260414123535,
|
| 5250 |
+
"learning_rate": 1.3472504165906613e-05,
|
| 5251 |
+
"loss": 1.9128,
|
| 5252 |
+
"step": 749
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.7635530669381522,
|
| 5256 |
+
"grad_norm": 8.994124412536621,
|
| 5257 |
+
"learning_rate": 1.3363016648115245e-05,
|
| 5258 |
+
"loss": 1.7094,
|
| 5259 |
+
"step": 750
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.7645711376940697,
|
| 5263 |
+
"grad_norm": 8.2550630569458,
|
| 5264 |
+
"learning_rate": 1.3253907173470648e-05,
|
| 5265 |
+
"loss": 3.7822,
|
| 5266 |
+
"step": 751
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.7655892084499872,
|
| 5270 |
+
"grad_norm": 8.363167762756348,
|
| 5271 |
+
"learning_rate": 1.3145176867832165e-05,
|
| 5272 |
+
"loss": 3.1741,
|
| 5273 |
+
"step": 752
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.7666072792059048,
|
| 5277 |
+
"grad_norm": 9.87235164642334,
|
| 5278 |
+
"learning_rate": 1.30368268531466e-05,
|
| 5279 |
+
"loss": 3.5932,
|
| 5280 |
+
"step": 753
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.7676253499618223,
|
| 5284 |
+
"grad_norm": 7.78696346282959,
|
| 5285 |
+
"learning_rate": 1.292885824743667e-05,
|
| 5286 |
+
"loss": 2.4161,
|
| 5287 |
+
"step": 754
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.7686434207177398,
|
| 5291 |
+
"grad_norm": 10.35235595703125,
|
| 5292 |
+
"learning_rate": 1.2821272164789544e-05,
|
| 5293 |
+
"loss": 3.0389,
|
| 5294 |
+
"step": 755
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.7696614914736574,
|
| 5298 |
+
"grad_norm": 13.586828231811523,
|
| 5299 |
+
"learning_rate": 1.2714069715345195e-05,
|
| 5300 |
+
"loss": 3.6863,
|
| 5301 |
+
"step": 756
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.770679562229575,
|
| 5305 |
+
"grad_norm": 11.710920333862305,
|
| 5306 |
+
"learning_rate": 1.2607252005285109e-05,
|
| 5307 |
+
"loss": 4.0324,
|
| 5308 |
+
"step": 757
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.7716976329854925,
|
| 5312 |
+
"grad_norm": 12.497662544250488,
|
| 5313 |
+
"learning_rate": 1.2500820136820734e-05,
|
| 5314 |
+
"loss": 2.9699,
|
| 5315 |
+
"step": 758
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.7727157037414101,
|
| 5319 |
+
"grad_norm": 15.810281753540039,
|
| 5320 |
+
"learning_rate": 1.2394775208182174e-05,
|
| 5321 |
+
"loss": 3.1219,
|
| 5322 |
+
"step": 759
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.7737337744973276,
|
| 5326 |
+
"grad_norm": 10.045539855957031,
|
| 5327 |
+
"learning_rate": 1.2289118313606896e-05,
|
| 5328 |
+
"loss": 1.7459,
|
| 5329 |
+
"step": 760
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.7747518452532451,
|
| 5333 |
+
"grad_norm": 8.060089111328125,
|
| 5334 |
+
"learning_rate": 1.2183850543328312e-05,
|
| 5335 |
+
"loss": 1.4901,
|
| 5336 |
+
"step": 761
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.7757699160091627,
|
| 5340 |
+
"grad_norm": 5.897403240203857,
|
| 5341 |
+
"learning_rate": 1.2078972983564684e-05,
|
| 5342 |
+
"loss": 1.1615,
|
| 5343 |
+
"step": 762
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.7767879867650802,
|
| 5347 |
+
"grad_norm": 9.653463363647461,
|
| 5348 |
+
"learning_rate": 1.1974486716507783e-05,
|
| 5349 |
+
"loss": 2.1851,
|
| 5350 |
+
"step": 763
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.7778060575209977,
|
| 5354 |
+
"grad_norm": 8.81517505645752,
|
| 5355 |
+
"learning_rate": 1.1870392820311821e-05,
|
| 5356 |
+
"loss": 2.0521,
|
| 5357 |
+
"step": 764
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.7788241282769153,
|
| 5361 |
+
"grad_norm": 8.291658401489258,
|
| 5362 |
+
"learning_rate": 1.1766692369082255e-05,
|
| 5363 |
+
"loss": 2.0249,
|
| 5364 |
+
"step": 765
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.7798421990328328,
|
| 5368 |
+
"grad_norm": 8.102690696716309,
|
| 5369 |
+
"learning_rate": 1.1663386432864727e-05,
|
| 5370 |
+
"loss": 2.065,
|
| 5371 |
+
"step": 766
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.7808602697887503,
|
| 5375 |
+
"grad_norm": 4.781749248504639,
|
| 5376 |
+
"learning_rate": 1.156047607763407e-05,
|
| 5377 |
+
"loss": 1.2103,
|
| 5378 |
+
"step": 767
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.7818783405446679,
|
| 5382 |
+
"grad_norm": 6.986288547515869,
|
| 5383 |
+
"learning_rate": 1.145796236528322e-05,
|
| 5384 |
+
"loss": 1.5723,
|
| 5385 |
+
"step": 768
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.7828964113005854,
|
| 5389 |
+
"grad_norm": 6.896148681640625,
|
| 5390 |
+
"learning_rate": 1.135584635361232e-05,
|
| 5391 |
+
"loss": 1.7923,
|
| 5392 |
+
"step": 769
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.7839144820565029,
|
| 5396 |
+
"grad_norm": 6.917054176330566,
|
| 5397 |
+
"learning_rate": 1.1254129096317806e-05,
|
| 5398 |
+
"loss": 1.8785,
|
| 5399 |
+
"step": 770
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.7849325528124205,
|
| 5403 |
+
"grad_norm": 8.382417678833008,
|
| 5404 |
+
"learning_rate": 1.115281164298153e-05,
|
| 5405 |
+
"loss": 2.0591,
|
| 5406 |
+
"step": 771
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.785950623568338,
|
| 5410 |
+
"grad_norm": 6.412557601928711,
|
| 5411 |
+
"learning_rate": 1.105189503905985e-05,
|
| 5412 |
+
"loss": 1.3064,
|
| 5413 |
+
"step": 772
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.7869686943242555,
|
| 5417 |
+
"grad_norm": 8.5576753616333,
|
| 5418 |
+
"learning_rate": 1.0951380325872979e-05,
|
| 5419 |
+
"loss": 2.1621,
|
| 5420 |
+
"step": 773
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.7879867650801731,
|
| 5424 |
+
"grad_norm": 6.601977348327637,
|
| 5425 |
+
"learning_rate": 1.0851268540594167e-05,
|
| 5426 |
+
"loss": 1.9123,
|
| 5427 |
+
"step": 774
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.7890048358360906,
|
| 5431 |
+
"grad_norm": 9.037349700927734,
|
| 5432 |
+
"learning_rate": 1.0751560716238967e-05,
|
| 5433 |
+
"loss": 2.7441,
|
| 5434 |
+
"step": 775
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.7900229065920081,
|
| 5438 |
+
"grad_norm": 8.547028541564941,
|
| 5439 |
+
"learning_rate": 1.0652257881654627e-05,
|
| 5440 |
+
"loss": 2.4516,
|
| 5441 |
+
"step": 776
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.7910409773479257,
|
| 5445 |
+
"grad_norm": 6.397313117980957,
|
| 5446 |
+
"learning_rate": 1.055336106150948e-05,
|
| 5447 |
+
"loss": 1.7555,
|
| 5448 |
+
"step": 777
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.7920590481038432,
|
| 5452 |
+
"grad_norm": 8.946637153625488,
|
| 5453 |
+
"learning_rate": 1.0454871276282335e-05,
|
| 5454 |
+
"loss": 2.4554,
|
| 5455 |
+
"step": 778
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.7930771188597607,
|
| 5459 |
+
"grad_norm": 10.802556037902832,
|
| 5460 |
+
"learning_rate": 1.0356789542251938e-05,
|
| 5461 |
+
"loss": 3.2162,
|
| 5462 |
+
"step": 779
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.7940951896156783,
|
| 5466 |
+
"grad_norm": 6.947890281677246,
|
| 5467 |
+
"learning_rate": 1.0259116871486557e-05,
|
| 5468 |
+
"loss": 1.7408,
|
| 5469 |
+
"step": 780
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.7951132603715958,
|
| 5473 |
+
"grad_norm": 6.768093585968018,
|
| 5474 |
+
"learning_rate": 1.0161854271833443e-05,
|
| 5475 |
+
"loss": 1.8601,
|
| 5476 |
+
"step": 781
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.7961313311275133,
|
| 5480 |
+
"grad_norm": 8.398831367492676,
|
| 5481 |
+
"learning_rate": 1.006500274690853e-05,
|
| 5482 |
+
"loss": 2.2156,
|
| 5483 |
+
"step": 782
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.7971494018834309,
|
| 5487 |
+
"grad_norm": 7.810449123382568,
|
| 5488 |
+
"learning_rate": 9.96856329608597e-06,
|
| 5489 |
+
"loss": 2.1837,
|
| 5490 |
+
"step": 783
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.7981674726393484,
|
| 5494 |
+
"grad_norm": 8.77087116241455,
|
| 5495 |
+
"learning_rate": 9.87253691448794e-06,
|
| 5496 |
+
"loss": 2.6596,
|
| 5497 |
+
"step": 784
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.7991855433952659,
|
| 5501 |
+
"grad_norm": 11.717060089111328,
|
| 5502 |
+
"learning_rate": 9.776924592974256e-06,
|
| 5503 |
+
"loss": 3.5775,
|
| 5504 |
+
"step": 785
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.8002036141511835,
|
| 5508 |
+
"grad_norm": 7.535914897918701,
|
| 5509 |
+
"learning_rate": 9.681727318132227e-06,
|
| 5510 |
+
"loss": 2.2536,
|
| 5511 |
+
"step": 786
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.801221684907101,
|
| 5515 |
+
"grad_norm": 9.274803161621094,
|
| 5516 |
+
"learning_rate": 9.586946072266478e-06,
|
| 5517 |
+
"loss": 2.4405,
|
| 5518 |
+
"step": 787
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.8022397556630185,
|
| 5522 |
+
"grad_norm": 7.637203216552734,
|
| 5523 |
+
"learning_rate": 9.492581833388736e-06,
|
| 5524 |
+
"loss": 1.9418,
|
| 5525 |
+
"step": 788
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.8032578264189361,
|
| 5529 |
+
"grad_norm": 11.312345504760742,
|
| 5530 |
+
"learning_rate": 9.398635575207854e-06,
|
| 5531 |
+
"loss": 3.0226,
|
| 5532 |
+
"step": 789
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.8042758971748537,
|
| 5536 |
+
"grad_norm": 8.732460021972656,
|
| 5537 |
+
"learning_rate": 9.305108267119645e-06,
|
| 5538 |
+
"loss": 2.0292,
|
| 5539 |
+
"step": 790
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.8052939679307712,
|
| 5543 |
+
"grad_norm": 6.998504638671875,
|
| 5544 |
+
"learning_rate": 9.212000874196953e-06,
|
| 5545 |
+
"loss": 1.7993,
|
| 5546 |
+
"step": 791
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.8063120386866888,
|
| 5550 |
+
"grad_norm": 9.693340301513672,
|
| 5551 |
+
"learning_rate": 9.119314357179687e-06,
|
| 5552 |
+
"loss": 2.4902,
|
| 5553 |
+
"step": 792
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.8073301094426063,
|
| 5557 |
+
"grad_norm": 8.360791206359863,
|
| 5558 |
+
"learning_rate": 9.027049672464916e-06,
|
| 5559 |
+
"loss": 2.0688,
|
| 5560 |
+
"step": 793
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.8083481801985238,
|
| 5564 |
+
"grad_norm": 7.457218647003174,
|
| 5565 |
+
"learning_rate": 8.935207772096904e-06,
|
| 5566 |
+
"loss": 2.1543,
|
| 5567 |
+
"step": 794
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.8093662509544414,
|
| 5571 |
+
"grad_norm": 8.154823303222656,
|
| 5572 |
+
"learning_rate": 8.843789603757446e-06,
|
| 5573 |
+
"loss": 2.5219,
|
| 5574 |
+
"step": 795
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.8103843217103589,
|
| 5578 |
+
"grad_norm": 8.503774642944336,
|
| 5579 |
+
"learning_rate": 8.752796110755984e-06,
|
| 5580 |
+
"loss": 1.7771,
|
| 5581 |
+
"step": 796
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.8114023924662764,
|
| 5585 |
+
"grad_norm": 9.03532600402832,
|
| 5586 |
+
"learning_rate": 8.662228232019876e-06,
|
| 5587 |
+
"loss": 2.1927,
|
| 5588 |
+
"step": 797
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.812420463222194,
|
| 5592 |
+
"grad_norm": 7.620565891265869,
|
| 5593 |
+
"learning_rate": 8.572086902084731e-06,
|
| 5594 |
+
"loss": 1.5837,
|
| 5595 |
+
"step": 798
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.8134385339781115,
|
| 5599 |
+
"grad_norm": 8.183737754821777,
|
| 5600 |
+
"learning_rate": 8.48237305108479e-06,
|
| 5601 |
+
"loss": 2.0261,
|
| 5602 |
+
"step": 799
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.814456604734029,
|
| 5606 |
+
"grad_norm": 7.848052978515625,
|
| 5607 |
+
"learning_rate": 8.393087604743283e-06,
|
| 5608 |
+
"loss": 1.7399,
|
| 5609 |
+
"step": 800
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 0.8154746754899466,
|
| 5613 |
+
"grad_norm": 6.207403182983398,
|
| 5614 |
+
"learning_rate": 8.304231484362868e-06,
|
| 5615 |
+
"loss": 2.4818,
|
| 5616 |
+
"step": 801
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 0.8164927462458641,
|
| 5620 |
+
"grad_norm": 7.363234043121338,
|
| 5621 |
+
"learning_rate": 8.215805606816191e-06,
|
| 5622 |
+
"loss": 2.7651,
|
| 5623 |
+
"step": 802
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 0.8175108170017816,
|
| 5627 |
+
"grad_norm": 8.982840538024902,
|
| 5628 |
+
"learning_rate": 8.127810884536403e-06,
|
| 5629 |
+
"loss": 3.0555,
|
| 5630 |
+
"step": 803
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 0.8185288877576992,
|
| 5634 |
+
"grad_norm": 8.631933212280273,
|
| 5635 |
+
"learning_rate": 8.040248225507641e-06,
|
| 5636 |
+
"loss": 2.9924,
|
| 5637 |
+
"step": 804
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 0.8195469585136167,
|
| 5641 |
+
"grad_norm": 11.01415729522705,
|
| 5642 |
+
"learning_rate": 7.95311853325582e-06,
|
| 5643 |
+
"loss": 3.5641,
|
| 5644 |
+
"step": 805
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.8205650292695342,
|
| 5648 |
+
"grad_norm": 11.288020133972168,
|
| 5649 |
+
"learning_rate": 7.866422706839238e-06,
|
| 5650 |
+
"loss": 3.6431,
|
| 5651 |
+
"step": 806
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 0.8215831000254518,
|
| 5655 |
+
"grad_norm": 18.66059112548828,
|
| 5656 |
+
"learning_rate": 7.780161640839257e-06,
|
| 5657 |
+
"loss": 3.8684,
|
| 5658 |
+
"step": 807
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 0.8226011707813693,
|
| 5662 |
+
"grad_norm": 10.357224464416504,
|
| 5663 |
+
"learning_rate": 7.694336225351107e-06,
|
| 5664 |
+
"loss": 2.31,
|
| 5665 |
+
"step": 808
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 0.8236192415372868,
|
| 5669 |
+
"grad_norm": 10.739027976989746,
|
| 5670 |
+
"learning_rate": 7.60894734597476e-06,
|
| 5671 |
+
"loss": 2.3002,
|
| 5672 |
+
"step": 809
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 0.8246373122932044,
|
| 5676 |
+
"grad_norm": 4.031766414642334,
|
| 5677 |
+
"learning_rate": 7.523995883805679e-06,
|
| 5678 |
+
"loss": 0.8728,
|
| 5679 |
+
"step": 810
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 0.8256553830491219,
|
| 5683 |
+
"grad_norm": 5.490253448486328,
|
| 5684 |
+
"learning_rate": 7.439482715425805e-06,
|
| 5685 |
+
"loss": 1.0823,
|
| 5686 |
+
"step": 811
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.8266734538050394,
|
| 5690 |
+
"grad_norm": 6.157886981964111,
|
| 5691 |
+
"learning_rate": 7.355408712894507e-06,
|
| 5692 |
+
"loss": 1.4668,
|
| 5693 |
+
"step": 812
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 0.827691524560957,
|
| 5697 |
+
"grad_norm": 4.612349510192871,
|
| 5698 |
+
"learning_rate": 7.271774743739545e-06,
|
| 5699 |
+
"loss": 0.9666,
|
| 5700 |
+
"step": 813
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 0.8287095953168745,
|
| 5704 |
+
"grad_norm": 4.856873989105225,
|
| 5705 |
+
"learning_rate": 7.188581670948169e-06,
|
| 5706 |
+
"loss": 1.2653,
|
| 5707 |
+
"step": 814
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 0.829727666072792,
|
| 5711 |
+
"grad_norm": 7.248535633087158,
|
| 5712 |
+
"learning_rate": 7.105830352958142e-06,
|
| 5713 |
+
"loss": 1.8515,
|
| 5714 |
+
"step": 815
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 0.8307457368287096,
|
| 5718 |
+
"grad_norm": 7.241623878479004,
|
| 5719 |
+
"learning_rate": 7.0235216436489835e-06,
|
| 5720 |
+
"loss": 1.4111,
|
| 5721 |
+
"step": 816
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 0.8317638075846271,
|
| 5725 |
+
"grad_norm": 5.958316802978516,
|
| 5726 |
+
"learning_rate": 6.941656392333046e-06,
|
| 5727 |
+
"loss": 1.4647,
|
| 5728 |
+
"step": 817
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.8327818783405446,
|
| 5732 |
+
"grad_norm": 7.3134684562683105,
|
| 5733 |
+
"learning_rate": 6.860235443746859e-06,
|
| 5734 |
+
"loss": 1.4169,
|
| 5735 |
+
"step": 818
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 0.8337999490964622,
|
| 5739 |
+
"grad_norm": 6.82220458984375,
|
| 5740 |
+
"learning_rate": 6.779259638042318e-06,
|
| 5741 |
+
"loss": 1.6871,
|
| 5742 |
+
"step": 819
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 0.8348180198523797,
|
| 5746 |
+
"grad_norm": 8.294244766235352,
|
| 5747 |
+
"learning_rate": 6.698729810778065e-06,
|
| 5748 |
+
"loss": 2.0008,
|
| 5749 |
+
"step": 820
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 0.8358360906082972,
|
| 5753 |
+
"grad_norm": 7.062936305999756,
|
| 5754 |
+
"learning_rate": 6.618646792910893e-06,
|
| 5755 |
+
"loss": 1.5006,
|
| 5756 |
+
"step": 821
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 0.8368541613642149,
|
| 5760 |
+
"grad_norm": 8.914663314819336,
|
| 5761 |
+
"learning_rate": 6.539011410787105e-06,
|
| 5762 |
+
"loss": 2.8718,
|
| 5763 |
+
"step": 822
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 0.8378722321201324,
|
| 5767 |
+
"grad_norm": 6.904050827026367,
|
| 5768 |
+
"learning_rate": 6.459824486134014e-06,
|
| 5769 |
+
"loss": 1.6913,
|
| 5770 |
+
"step": 823
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.8388903028760499,
|
| 5774 |
+
"grad_norm": 9.867286682128906,
|
| 5775 |
+
"learning_rate": 6.381086836051498e-06,
|
| 5776 |
+
"loss": 2.3808,
|
| 5777 |
+
"step": 824
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.8399083736319675,
|
| 5781 |
+
"grad_norm": 7.717339992523193,
|
| 5782 |
+
"learning_rate": 6.302799273003546e-06,
|
| 5783 |
+
"loss": 2.3309,
|
| 5784 |
+
"step": 825
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.840926444387885,
|
| 5788 |
+
"grad_norm": 7.964992046356201,
|
| 5789 |
+
"learning_rate": 6.224962604809819e-06,
|
| 5790 |
+
"loss": 1.8556,
|
| 5791 |
+
"step": 826
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.8419445151438025,
|
| 5795 |
+
"grad_norm": 8.438617706298828,
|
| 5796 |
+
"learning_rate": 6.147577634637414e-06,
|
| 5797 |
+
"loss": 2.8175,
|
| 5798 |
+
"step": 827
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.8429625858997201,
|
| 5802 |
+
"grad_norm": 6.18842077255249,
|
| 5803 |
+
"learning_rate": 6.070645160992522e-06,
|
| 5804 |
+
"loss": 1.8594,
|
| 5805 |
+
"step": 828
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 0.8439806566556376,
|
| 5809 |
+
"grad_norm": 9.132290840148926,
|
| 5810 |
+
"learning_rate": 5.994165977712174e-06,
|
| 5811 |
+
"loss": 2.4016,
|
| 5812 |
+
"step": 829
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.8449987274115551,
|
| 5816 |
+
"grad_norm": 8.243701934814453,
|
| 5817 |
+
"learning_rate": 5.918140873956063e-06,
|
| 5818 |
+
"loss": 2.8481,
|
| 5819 |
+
"step": 830
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 0.8460167981674727,
|
| 5823 |
+
"grad_norm": 7.9838457107543945,
|
| 5824 |
+
"learning_rate": 5.842570634198452e-06,
|
| 5825 |
+
"loss": 2.2464,
|
| 5826 |
+
"step": 831
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 0.8470348689233902,
|
| 5830 |
+
"grad_norm": 9.669371604919434,
|
| 5831 |
+
"learning_rate": 5.767456038219987e-06,
|
| 5832 |
+
"loss": 3.0041,
|
| 5833 |
+
"step": 832
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 0.8480529396793077,
|
| 5837 |
+
"grad_norm": 6.894510746002197,
|
| 5838 |
+
"learning_rate": 5.692797861099719e-06,
|
| 5839 |
+
"loss": 1.9256,
|
| 5840 |
+
"step": 833
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 0.8490710104352253,
|
| 5844 |
+
"grad_norm": 9.749018669128418,
|
| 5845 |
+
"learning_rate": 5.6185968732070825e-06,
|
| 5846 |
+
"loss": 2.0606,
|
| 5847 |
+
"step": 834
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 0.8500890811911428,
|
| 5851 |
+
"grad_norm": 10.20246410369873,
|
| 5852 |
+
"learning_rate": 5.544853840193981e-06,
|
| 5853 |
+
"loss": 2.5006,
|
| 5854 |
+
"step": 835
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.8511071519470603,
|
| 5858 |
+
"grad_norm": 7.829645156860352,
|
| 5859 |
+
"learning_rate": 5.471569522986774e-06,
|
| 5860 |
+
"loss": 2.2776,
|
| 5861 |
+
"step": 836
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 0.8521252227029779,
|
| 5865 |
+
"grad_norm": 8.704474449157715,
|
| 5866 |
+
"learning_rate": 5.398744677778594e-06,
|
| 5867 |
+
"loss": 2.155,
|
| 5868 |
+
"step": 837
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 0.8531432934588954,
|
| 5872 |
+
"grad_norm": 9.067277908325195,
|
| 5873 |
+
"learning_rate": 5.326380056021418e-06,
|
| 5874 |
+
"loss": 1.9537,
|
| 5875 |
+
"step": 838
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 0.8541613642148129,
|
| 5879 |
+
"grad_norm": 7.175379276275635,
|
| 5880 |
+
"learning_rate": 5.25447640441834e-06,
|
| 5881 |
+
"loss": 1.8142,
|
| 5882 |
+
"step": 839
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 0.8551794349707305,
|
| 5886 |
+
"grad_norm": 9.313763618469238,
|
| 5887 |
+
"learning_rate": 5.183034464915898e-06,
|
| 5888 |
+
"loss": 2.6184,
|
| 5889 |
+
"step": 840
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 0.856197505726648,
|
| 5893 |
+
"grad_norm": 8.87330150604248,
|
| 5894 |
+
"learning_rate": 5.112054974696395e-06,
|
| 5895 |
+
"loss": 2.2619,
|
| 5896 |
+
"step": 841
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.8572155764825655,
|
| 5900 |
+
"grad_norm": 7.452190399169922,
|
| 5901 |
+
"learning_rate": 5.041538666170281e-06,
|
| 5902 |
+
"loss": 2.0532,
|
| 5903 |
+
"step": 842
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 0.858233647238483,
|
| 5907 |
+
"grad_norm": 8.609152793884277,
|
| 5908 |
+
"learning_rate": 4.9714862669686335e-06,
|
| 5909 |
+
"loss": 2.3708,
|
| 5910 |
+
"step": 843
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 0.8592517179944006,
|
| 5914 |
+
"grad_norm": 9.013458251953125,
|
| 5915 |
+
"learning_rate": 4.901898499935609e-06,
|
| 5916 |
+
"loss": 2.2493,
|
| 5917 |
+
"step": 844
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 0.8602697887503181,
|
| 5921 |
+
"grad_norm": 6.508463382720947,
|
| 5922 |
+
"learning_rate": 4.832776083120982e-06,
|
| 5923 |
+
"loss": 1.6531,
|
| 5924 |
+
"step": 845
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 0.8612878595062357,
|
| 5928 |
+
"grad_norm": 8.564003944396973,
|
| 5929 |
+
"learning_rate": 4.764119729772809e-06,
|
| 5930 |
+
"loss": 2.2661,
|
| 5931 |
+
"step": 846
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 0.8623059302621532,
|
| 5935 |
+
"grad_norm": 10.198071479797363,
|
| 5936 |
+
"learning_rate": 4.695930148329958e-06,
|
| 5937 |
+
"loss": 2.5887,
|
| 5938 |
+
"step": 847
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.8633240010180707,
|
| 5942 |
+
"grad_norm": 7.703155994415283,
|
| 5943 |
+
"learning_rate": 4.628208042414889e-06,
|
| 5944 |
+
"loss": 2.1529,
|
| 5945 |
+
"step": 848
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.8643420717739883,
|
| 5949 |
+
"grad_norm": 9.272164344787598,
|
| 5950 |
+
"learning_rate": 4.560954110826337e-06,
|
| 5951 |
+
"loss": 2.1878,
|
| 5952 |
+
"step": 849
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.8653601425299058,
|
| 5956 |
+
"grad_norm": 9.344987869262695,
|
| 5957 |
+
"learning_rate": 4.494169047532154e-06,
|
| 5958 |
+
"loss": 1.8306,
|
| 5959 |
+
"step": 850
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.8663782132858233,
|
| 5963 |
+
"grad_norm": 7.905543327331543,
|
| 5964 |
+
"learning_rate": 4.427853541662091e-06,
|
| 5965 |
+
"loss": 3.7738,
|
| 5966 |
+
"step": 851
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.8673962840417409,
|
| 5970 |
+
"grad_norm": 11.58995532989502,
|
| 5971 |
+
"learning_rate": 4.362008277500701e-06,
|
| 5972 |
+
"loss": 3.8687,
|
| 5973 |
+
"step": 852
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.8684143547976584,
|
| 5977 |
+
"grad_norm": 10.805427551269531,
|
| 5978 |
+
"learning_rate": 4.296633934480337e-06,
|
| 5979 |
+
"loss": 4.1339,
|
| 5980 |
+
"step": 853
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.869432425553576,
|
| 5984 |
+
"grad_norm": 9.845876693725586,
|
| 5985 |
+
"learning_rate": 4.231731187174065e-06,
|
| 5986 |
+
"loss": 3.6647,
|
| 5987 |
+
"step": 854
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.8704504963094936,
|
| 5991 |
+
"grad_norm": 11.320969581604004,
|
| 5992 |
+
"learning_rate": 4.167300705288718e-06,
|
| 5993 |
+
"loss": 3.8381,
|
| 5994 |
+
"step": 855
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.8714685670654111,
|
| 5998 |
+
"grad_norm": 12.798758506774902,
|
| 5999 |
+
"learning_rate": 4.10334315365804e-06,
|
| 6000 |
+
"loss": 3.4318,
|
| 6001 |
+
"step": 856
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.8724866378213286,
|
| 6005 |
+
"grad_norm": 11.61473274230957,
|
| 6006 |
+
"learning_rate": 4.039859192235779e-06,
|
| 6007 |
+
"loss": 2.8296,
|
| 6008 |
+
"step": 857
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.8735047085772462,
|
| 6012 |
+
"grad_norm": 14.96419620513916,
|
| 6013 |
+
"learning_rate": 3.976849476088845e-06,
|
| 6014 |
+
"loss": 3.5524,
|
| 6015 |
+
"step": 858
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.8745227793331637,
|
| 6019 |
+
"grad_norm": 7.406204700469971,
|
| 6020 |
+
"learning_rate": 3.914314655390633e-06,
|
| 6021 |
+
"loss": 1.9455,
|
| 6022 |
+
"step": 859
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.8755408500890812,
|
| 6026 |
+
"grad_norm": 6.518743991851807,
|
| 6027 |
+
"learning_rate": 3.852255375414271e-06,
|
| 6028 |
+
"loss": 1.2701,
|
| 6029 |
+
"step": 860
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.8765589208449988,
|
| 6033 |
+
"grad_norm": 6.446305751800537,
|
| 6034 |
+
"learning_rate": 3.790672276525936e-06,
|
| 6035 |
+
"loss": 1.4215,
|
| 6036 |
+
"step": 861
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.8775769916009163,
|
| 6040 |
+
"grad_norm": 8.741048812866211,
|
| 6041 |
+
"learning_rate": 3.7295659941782855e-06,
|
| 6042 |
+
"loss": 1.8906,
|
| 6043 |
+
"step": 862
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.8785950623568338,
|
| 6047 |
+
"grad_norm": 8.466268539428711,
|
| 6048 |
+
"learning_rate": 3.668937158903901e-06,
|
| 6049 |
+
"loss": 1.8339,
|
| 6050 |
+
"step": 863
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.8796131331127514,
|
| 6054 |
+
"grad_norm": 9.552635192871094,
|
| 6055 |
+
"learning_rate": 3.6087863963087497e-06,
|
| 6056 |
+
"loss": 2.2068,
|
| 6057 |
+
"step": 864
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.8806312038686689,
|
| 6061 |
+
"grad_norm": 6.880453586578369,
|
| 6062 |
+
"learning_rate": 3.5491143270657446e-06,
|
| 6063 |
+
"loss": 1.4963,
|
| 6064 |
+
"step": 865
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.8816492746245864,
|
| 6068 |
+
"grad_norm": 6.255483627319336,
|
| 6069 |
+
"learning_rate": 3.4899215669083716e-06,
|
| 6070 |
+
"loss": 1.3132,
|
| 6071 |
+
"step": 866
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.882667345380504,
|
| 6075 |
+
"grad_norm": 6.0426025390625,
|
| 6076 |
+
"learning_rate": 3.4312087266242963e-06,
|
| 6077 |
+
"loss": 1.5957,
|
| 6078 |
+
"step": 867
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.8836854161364215,
|
| 6082 |
+
"grad_norm": 6.473054885864258,
|
| 6083 |
+
"learning_rate": 3.3729764120490446e-06,
|
| 6084 |
+
"loss": 1.7223,
|
| 6085 |
+
"step": 868
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.884703486892339,
|
| 6089 |
+
"grad_norm": 6.081118106842041,
|
| 6090 |
+
"learning_rate": 3.315225224059809e-06,
|
| 6091 |
+
"loss": 1.4486,
|
| 6092 |
+
"step": 869
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.8857215576482566,
|
| 6096 |
+
"grad_norm": 8.436514854431152,
|
| 6097 |
+
"learning_rate": 3.25795575856922e-06,
|
| 6098 |
+
"loss": 2.5112,
|
| 6099 |
+
"step": 870
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.8867396284041741,
|
| 6103 |
+
"grad_norm": 7.625750541687012,
|
| 6104 |
+
"learning_rate": 3.2011686065191895e-06,
|
| 6105 |
+
"loss": 1.7229,
|
| 6106 |
+
"step": 871
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.8877576991600916,
|
| 6110 |
+
"grad_norm": 8.813794136047363,
|
| 6111 |
+
"learning_rate": 3.1448643538748045e-06,
|
| 6112 |
+
"loss": 2.4128,
|
| 6113 |
+
"step": 872
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.8887757699160092,
|
| 6117 |
+
"grad_norm": 5.511772155761719,
|
| 6118 |
+
"learning_rate": 3.0890435816183226e-06,
|
| 6119 |
+
"loss": 1.3973,
|
| 6120 |
+
"step": 873
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.8897938406719267,
|
| 6124 |
+
"grad_norm": 8.631103515625,
|
| 6125 |
+
"learning_rate": 3.03370686574313e-06,
|
| 6126 |
+
"loss": 2.015,
|
| 6127 |
+
"step": 874
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.8908119114278442,
|
| 6131 |
+
"grad_norm": 7.364684104919434,
|
| 6132 |
+
"learning_rate": 2.9788547772478416e-06,
|
| 6133 |
+
"loss": 2.2994,
|
| 6134 |
+
"step": 875
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.8918299821837617,
|
| 6138 |
+
"grad_norm": 6.241342067718506,
|
| 6139 |
+
"learning_rate": 2.924487882130356e-06,
|
| 6140 |
+
"loss": 1.8339,
|
| 6141 |
+
"step": 876
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.8928480529396793,
|
| 6145 |
+
"grad_norm": 7.564022064208984,
|
| 6146 |
+
"learning_rate": 2.870606741382059e-06,
|
| 6147 |
+
"loss": 1.7091,
|
| 6148 |
+
"step": 877
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.8938661236955968,
|
| 6152 |
+
"grad_norm": 7.5156378746032715,
|
| 6153 |
+
"learning_rate": 2.817211910982037e-06,
|
| 6154 |
+
"loss": 2.1743,
|
| 6155 |
+
"step": 878
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.8948841944515143,
|
| 6159 |
+
"grad_norm": 7.265412330627441,
|
| 6160 |
+
"learning_rate": 2.7643039418913e-06,
|
| 6161 |
+
"loss": 2.0643,
|
| 6162 |
+
"step": 879
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.8959022652074319,
|
| 6166 |
+
"grad_norm": 7.189180850982666,
|
| 6167 |
+
"learning_rate": 2.711883380047131e-06,
|
| 6168 |
+
"loss": 2.1941,
|
| 6169 |
+
"step": 880
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.8969203359633494,
|
| 6173 |
+
"grad_norm": 7.311581134796143,
|
| 6174 |
+
"learning_rate": 2.6599507663574384e-06,
|
| 6175 |
+
"loss": 2.1736,
|
| 6176 |
+
"step": 881
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.897938406719267,
|
| 6180 |
+
"grad_norm": 10.869359016418457,
|
| 6181 |
+
"learning_rate": 2.6085066366951905e-06,
|
| 6182 |
+
"loss": 3.2378,
|
| 6183 |
+
"step": 882
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.8989564774751845,
|
| 6187 |
+
"grad_norm": 7.649423122406006,
|
| 6188 |
+
"learning_rate": 2.5575515218928592e-06,
|
| 6189 |
+
"loss": 2.1515,
|
| 6190 |
+
"step": 883
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.899974548231102,
|
| 6194 |
+
"grad_norm": 7.814465045928955,
|
| 6195 |
+
"learning_rate": 2.5070859477369645e-06,
|
| 6196 |
+
"loss": 2.4541,
|
| 6197 |
+
"step": 884
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.9009926189870195,
|
| 6201 |
+
"grad_norm": 7.679671764373779,
|
| 6202 |
+
"learning_rate": 2.457110434962645e-06,
|
| 6203 |
+
"loss": 2.6637,
|
| 6204 |
+
"step": 885
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.9020106897429372,
|
| 6208 |
+
"grad_norm": 8.16401481628418,
|
| 6209 |
+
"learning_rate": 2.407625499248273e-06,
|
| 6210 |
+
"loss": 2.547,
|
| 6211 |
+
"step": 886
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.9030287604988547,
|
| 6215 |
+
"grad_norm": 9.379379272460938,
|
| 6216 |
+
"learning_rate": 2.3586316512101416e-06,
|
| 6217 |
+
"loss": 2.5642,
|
| 6218 |
+
"step": 887
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.9040468312547723,
|
| 6222 |
+
"grad_norm": 7.037562370300293,
|
| 6223 |
+
"learning_rate": 2.3101293963972094e-06,
|
| 6224 |
+
"loss": 1.9438,
|
| 6225 |
+
"step": 888
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.9050649020106898,
|
| 6229 |
+
"grad_norm": 7.9886674880981445,
|
| 6230 |
+
"learning_rate": 2.26211923528587e-06,
|
| 6231 |
+
"loss": 2.2604,
|
| 6232 |
+
"step": 889
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.9060829727666073,
|
| 6236 |
+
"grad_norm": 7.768427848815918,
|
| 6237 |
+
"learning_rate": 2.2146016632747624e-06,
|
| 6238 |
+
"loss": 1.8138,
|
| 6239 |
+
"step": 890
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.9071010435225249,
|
| 6243 |
+
"grad_norm": 7.540080547332764,
|
| 6244 |
+
"learning_rate": 2.1675771706797132e-06,
|
| 6245 |
+
"loss": 1.8673,
|
| 6246 |
+
"step": 891
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.9081191142784424,
|
| 6250 |
+
"grad_norm": 8.09107780456543,
|
| 6251 |
+
"learning_rate": 2.1210462427286524e-06,
|
| 6252 |
+
"loss": 1.971,
|
| 6253 |
+
"step": 892
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.9091371850343599,
|
| 6257 |
+
"grad_norm": 6.946146011352539,
|
| 6258 |
+
"learning_rate": 2.0750093595565733e-06,
|
| 6259 |
+
"loss": 1.7914,
|
| 6260 |
+
"step": 893
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.9101552557902775,
|
| 6264 |
+
"grad_norm": 10.480133056640625,
|
| 6265 |
+
"learning_rate": 2.0294669962006354e-06,
|
| 6266 |
+
"loss": 2.4089,
|
| 6267 |
+
"step": 894
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.911173326546195,
|
| 6271 |
+
"grad_norm": 7.2262797355651855,
|
| 6272 |
+
"learning_rate": 1.984419622595224e-06,
|
| 6273 |
+
"loss": 1.8403,
|
| 6274 |
+
"step": 895
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.9121913973021125,
|
| 6278 |
+
"grad_norm": 8.057347297668457,
|
| 6279 |
+
"learning_rate": 1.939867703567122e-06,
|
| 6280 |
+
"loss": 1.7492,
|
| 6281 |
+
"step": 896
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.91320946805803,
|
| 6285 |
+
"grad_norm": 8.859241485595703,
|
| 6286 |
+
"learning_rate": 1.895811698830685e-06,
|
| 6287 |
+
"loss": 2.3681,
|
| 6288 |
+
"step": 897
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.9142275388139476,
|
| 6292 |
+
"grad_norm": 7.5691328048706055,
|
| 6293 |
+
"learning_rate": 1.8522520629831397e-06,
|
| 6294 |
+
"loss": 2.1921,
|
| 6295 |
+
"step": 898
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.9152456095698651,
|
| 6299 |
+
"grad_norm": 9.58875846862793,
|
| 6300 |
+
"learning_rate": 1.8091892454998594e-06,
|
| 6301 |
+
"loss": 2.4974,
|
| 6302 |
+
"step": 899
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.9162636803257826,
|
| 6306 |
+
"grad_norm": 7.032078266143799,
|
| 6307 |
+
"learning_rate": 1.7666236907297406e-06,
|
| 6308 |
+
"loss": 1.3266,
|
| 6309 |
+
"step": 900
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.9172817510817002,
|
| 6313 |
+
"grad_norm": 7.034844875335693,
|
| 6314 |
+
"learning_rate": 1.7245558378906013e-06,
|
| 6315 |
+
"loss": 2.617,
|
| 6316 |
+
"step": 901
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.9182998218376177,
|
| 6320 |
+
"grad_norm": 9.052960395812988,
|
| 6321 |
+
"learning_rate": 1.6829861210646891e-06,
|
| 6322 |
+
"loss": 3.9865,
|
| 6323 |
+
"step": 902
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.9193178925935352,
|
| 6327 |
+
"grad_norm": 11.31360912322998,
|
| 6328 |
+
"learning_rate": 1.641914969194147e-06,
|
| 6329 |
+
"loss": 3.9044,
|
| 6330 |
+
"step": 903
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.9203359633494528,
|
| 6334 |
+
"grad_norm": 11.041589736938477,
|
| 6335 |
+
"learning_rate": 1.6013428060766168e-06,
|
| 6336 |
+
"loss": 4.3406,
|
| 6337 |
+
"step": 904
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.9213540341053703,
|
| 6341 |
+
"grad_norm": 10.366722106933594,
|
| 6342 |
+
"learning_rate": 1.5612700503608968e-06,
|
| 6343 |
+
"loss": 2.948,
|
| 6344 |
+
"step": 905
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.9223721048612878,
|
| 6348 |
+
"grad_norm": 13.091897964477539,
|
| 6349 |
+
"learning_rate": 1.5216971155425475e-06,
|
| 6350 |
+
"loss": 4.3409,
|
| 6351 |
+
"step": 906
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.9233901756172054,
|
| 6355 |
+
"grad_norm": 10.320602416992188,
|
| 6356 |
+
"learning_rate": 1.4826244099596986e-06,
|
| 6357 |
+
"loss": 2.7963,
|
| 6358 |
+
"step": 907
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.9244082463731229,
|
| 6362 |
+
"grad_norm": 9.795669555664062,
|
| 6363 |
+
"learning_rate": 1.4440523367887871e-06,
|
| 6364 |
+
"loss": 2.7568,
|
| 6365 |
+
"step": 908
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.9254263171290404,
|
| 6369 |
+
"grad_norm": 5.081234931945801,
|
| 6370 |
+
"learning_rate": 1.4059812940404093e-06,
|
| 6371 |
+
"loss": 1.1005,
|
| 6372 |
+
"step": 909
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.926444387884958,
|
| 6376 |
+
"grad_norm": 5.965490818023682,
|
| 6377 |
+
"learning_rate": 1.3684116745552423e-06,
|
| 6378 |
+
"loss": 1.6534,
|
| 6379 |
+
"step": 910
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.9274624586408755,
|
| 6383 |
+
"grad_norm": 8.321152687072754,
|
| 6384 |
+
"learning_rate": 1.33134386599994e-06,
|
| 6385 |
+
"loss": 1.6702,
|
| 6386 |
+
"step": 911
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.928480529396793,
|
| 6390 |
+
"grad_norm": 6.844864845275879,
|
| 6391 |
+
"learning_rate": 1.2947782508631822e-06,
|
| 6392 |
+
"loss": 1.5389,
|
| 6393 |
+
"step": 912
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.9294986001527106,
|
| 6397 |
+
"grad_norm": 7.344895362854004,
|
| 6398 |
+
"learning_rate": 1.2587152064516827e-06,
|
| 6399 |
+
"loss": 1.6019,
|
| 6400 |
+
"step": 913
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.9305166709086281,
|
| 6404 |
+
"grad_norm": 6.270571708679199,
|
| 6405 |
+
"learning_rate": 1.223155104886342e-06,
|
| 6406 |
+
"loss": 1.322,
|
| 6407 |
+
"step": 914
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.9315347416645456,
|
| 6411 |
+
"grad_norm": 7.117447853088379,
|
| 6412 |
+
"learning_rate": 1.1880983130983626e-06,
|
| 6413 |
+
"loss": 1.6418,
|
| 6414 |
+
"step": 915
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.9325528124204632,
|
| 6418 |
+
"grad_norm": 8.100106239318848,
|
| 6419 |
+
"learning_rate": 1.1535451928254947e-06,
|
| 6420 |
+
"loss": 2.1211,
|
| 6421 |
+
"step": 916
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.9335708831763807,
|
| 6425 |
+
"grad_norm": 9.206981658935547,
|
| 6426 |
+
"learning_rate": 1.1194961006082972e-06,
|
| 6427 |
+
"loss": 2.2021,
|
| 6428 |
+
"step": 917
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.9345889539322983,
|
| 6432 |
+
"grad_norm": 5.570037364959717,
|
| 6433 |
+
"learning_rate": 1.085951387786438e-06,
|
| 6434 |
+
"loss": 1.5467,
|
| 6435 |
+
"step": 918
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.9356070246882159,
|
| 6439 |
+
"grad_norm": 8.013311386108398,
|
| 6440 |
+
"learning_rate": 1.0529114004951047e-06,
|
| 6441 |
+
"loss": 1.9642,
|
| 6442 |
+
"step": 919
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.9366250954441334,
|
| 6446 |
+
"grad_norm": 6.999322891235352,
|
| 6447 |
+
"learning_rate": 1.0203764796614058e-06,
|
| 6448 |
+
"loss": 1.772,
|
| 6449 |
+
"step": 920
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.937643166200051,
|
| 6453 |
+
"grad_norm": 7.812559604644775,
|
| 6454 |
+
"learning_rate": 9.883469610008577e-07,
|
| 6455 |
+
"loss": 1.9127,
|
| 6456 |
+
"step": 921
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.9386612369559685,
|
| 6460 |
+
"grad_norm": 9.275822639465332,
|
| 6461 |
+
"learning_rate": 9.568231750139212e-07,
|
| 6462 |
+
"loss": 2.3705,
|
| 6463 |
+
"step": 922
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.939679307711886,
|
| 6467 |
+
"grad_norm": 7.6475372314453125,
|
| 6468 |
+
"learning_rate": 9.258054469825972e-07,
|
| 6469 |
+
"loss": 2.306,
|
| 6470 |
+
"step": 923
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.9406973784678035,
|
| 6474 |
+
"grad_norm": 7.288696765899658,
|
| 6475 |
+
"learning_rate": 8.952940969670809e-07,
|
| 6476 |
+
"loss": 1.6429,
|
| 6477 |
+
"step": 924
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.9417154492237211,
|
| 6481 |
+
"grad_norm": 7.795031547546387,
|
| 6482 |
+
"learning_rate": 8.652894398024136e-07,
|
| 6483 |
+
"loss": 1.9986,
|
| 6484 |
+
"step": 925
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.9427335199796386,
|
| 6488 |
+
"grad_norm": 7.860483169555664,
|
| 6489 |
+
"learning_rate": 8.357917850952802e-07,
|
| 6490 |
+
"loss": 2.1139,
|
| 6491 |
+
"step": 926
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.9437515907355561,
|
| 6495 |
+
"grad_norm": 7.814316749572754,
|
| 6496 |
+
"learning_rate": 8.06801437220811e-07,
|
| 6497 |
+
"loss": 2.1013,
|
| 6498 |
+
"step": 927
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.9447696614914737,
|
| 6502 |
+
"grad_norm": 8.413445472717285,
|
| 6503 |
+
"learning_rate": 7.783186953194189e-07,
|
| 6504 |
+
"loss": 2.7227,
|
| 6505 |
+
"step": 928
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.9457877322473912,
|
| 6509 |
+
"grad_norm": 7.9406328201293945,
|
| 6510 |
+
"learning_rate": 7.503438532937168e-07,
|
| 6511 |
+
"loss": 2.1969,
|
| 6512 |
+
"step": 929
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.9468058030033087,
|
| 6516 |
+
"grad_norm": 8.008191108703613,
|
| 6517 |
+
"learning_rate": 7.228771998054995e-07,
|
| 6518 |
+
"loss": 2.2026,
|
| 6519 |
+
"step": 930
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.9478238737592263,
|
| 6523 |
+
"grad_norm": 9.503705978393555,
|
| 6524 |
+
"learning_rate": 6.959190182727615e-07,
|
| 6525 |
+
"loss": 2.6732,
|
| 6526 |
+
"step": 931
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.9488419445151438,
|
| 6530 |
+
"grad_norm": 9.472963333129883,
|
| 6531 |
+
"learning_rate": 6.694695868667556e-07,
|
| 6532 |
+
"loss": 3.0515,
|
| 6533 |
+
"step": 932
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.9498600152710613,
|
| 6537 |
+
"grad_norm": 7.070324420928955,
|
| 6538 |
+
"learning_rate": 6.43529178509139e-07,
|
| 6539 |
+
"loss": 1.9813,
|
| 6540 |
+
"step": 933
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.9508780860269789,
|
| 6544 |
+
"grad_norm": 9.686485290527344,
|
| 6545 |
+
"learning_rate": 6.180980608691655e-07,
|
| 6546 |
+
"loss": 2.8315,
|
| 6547 |
+
"step": 934
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.9518961567828964,
|
| 6551 |
+
"grad_norm": 8.254791259765625,
|
| 6552 |
+
"learning_rate": 5.931764963608866e-07,
|
| 6553 |
+
"loss": 2.2097,
|
| 6554 |
+
"step": 935
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.9529142275388139,
|
| 6558 |
+
"grad_norm": 8.103293418884277,
|
| 6559 |
+
"learning_rate": 5.687647421404874e-07,
|
| 6560 |
+
"loss": 2.4556,
|
| 6561 |
+
"step": 936
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.9539322982947315,
|
| 6565 |
+
"grad_norm": 7.946092128753662,
|
| 6566 |
+
"learning_rate": 5.448630501036112e-07,
|
| 6567 |
+
"loss": 1.7899,
|
| 6568 |
+
"step": 937
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.954950369050649,
|
| 6572 |
+
"grad_norm": 7.618176460266113,
|
| 6573 |
+
"learning_rate": 5.214716668827557e-07,
|
| 6574 |
+
"loss": 2.3681,
|
| 6575 |
+
"step": 938
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.9559684398065665,
|
| 6579 |
+
"grad_norm": 5.380380153656006,
|
| 6580 |
+
"learning_rate": 4.985908338447476e-07,
|
| 6581 |
+
"loss": 1.7725,
|
| 6582 |
+
"step": 939
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.9569865105624841,
|
| 6586 |
+
"grad_norm": 7.778873920440674,
|
| 6587 |
+
"learning_rate": 4.762207870882218e-07,
|
| 6588 |
+
"loss": 1.8244,
|
| 6589 |
+
"step": 940
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.9580045813184016,
|
| 6593 |
+
"grad_norm": 9.022889137268066,
|
| 6594 |
+
"learning_rate": 4.543617574412184e-07,
|
| 6595 |
+
"loss": 2.5223,
|
| 6596 |
+
"step": 941
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.9590226520743191,
|
| 6600 |
+
"grad_norm": 6.915138244628906,
|
| 6601 |
+
"learning_rate": 4.3301397045877876e-07,
|
| 6602 |
+
"loss": 1.5739,
|
| 6603 |
+
"step": 942
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.9600407228302367,
|
| 6607 |
+
"grad_norm": 7.640623092651367,
|
| 6608 |
+
"learning_rate": 4.121776464206251e-07,
|
| 6609 |
+
"loss": 2.0568,
|
| 6610 |
+
"step": 943
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.9610587935861542,
|
| 6614 |
+
"grad_norm": 7.426476955413818,
|
| 6615 |
+
"learning_rate": 3.9185300032889006e-07,
|
| 6616 |
+
"loss": 1.8825,
|
| 6617 |
+
"step": 944
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.9620768643420717,
|
| 6621 |
+
"grad_norm": 8.864518165588379,
|
| 6622 |
+
"learning_rate": 3.720402419058966e-07,
|
| 6623 |
+
"loss": 2.373,
|
| 6624 |
+
"step": 945
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.9630949350979893,
|
| 6628 |
+
"grad_norm": 7.7088541984558105,
|
| 6629 |
+
"learning_rate": 3.5273957559199266e-07,
|
| 6630 |
+
"loss": 1.6487,
|
| 6631 |
+
"step": 946
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.9641130058539068,
|
| 6635 |
+
"grad_norm": 7.193671703338623,
|
| 6636 |
+
"learning_rate": 3.339512005434309e-07,
|
| 6637 |
+
"loss": 1.8298,
|
| 6638 |
+
"step": 947
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.9651310766098243,
|
| 6642 |
+
"grad_norm": 7.4829230308532715,
|
| 6643 |
+
"learning_rate": 3.1567531063033673e-07,
|
| 6644 |
+
"loss": 1.1332,
|
| 6645 |
+
"step": 948
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.9661491473657419,
|
| 6649 |
+
"grad_norm": 27.006391525268555,
|
| 6650 |
+
"learning_rate": 2.979120944346936e-07,
|
| 6651 |
+
"loss": 2.0494,
|
| 6652 |
+
"step": 949
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.9671672181216594,
|
| 6656 |
+
"grad_norm": 11.99468994140625,
|
| 6657 |
+
"learning_rate": 2.806617352483998e-07,
|
| 6658 |
+
"loss": 2.4674,
|
| 6659 |
+
"step": 950
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.968185288877577,
|
| 6663 |
+
"grad_norm": 7.580793380737305,
|
| 6664 |
+
"learning_rate": 2.639244110713701e-07,
|
| 6665 |
+
"loss": 3.3584,
|
| 6666 |
+
"step": 951
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.9692033596334946,
|
| 6670 |
+
"grad_norm": 9.658464431762695,
|
| 6671 |
+
"learning_rate": 2.4770029460970954e-07,
|
| 6672 |
+
"loss": 3.5755,
|
| 6673 |
+
"step": 952
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.9702214303894121,
|
| 6677 |
+
"grad_norm": 9.27920150756836,
|
| 6678 |
+
"learning_rate": 2.319895532739369e-07,
|
| 6679 |
+
"loss": 2.8816,
|
| 6680 |
+
"step": 953
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.9712395011453296,
|
| 6684 |
+
"grad_norm": 11.56139850616455,
|
| 6685 |
+
"learning_rate": 2.1679234917721946e-07,
|
| 6686 |
+
"loss": 3.1379,
|
| 6687 |
+
"step": 954
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.9722575719012472,
|
| 6691 |
+
"grad_norm": 13.082681655883789,
|
| 6692 |
+
"learning_rate": 2.0210883913376334e-07,
|
| 6693 |
+
"loss": 3.5519,
|
| 6694 |
+
"step": 955
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.9732756426571647,
|
| 6698 |
+
"grad_norm": 10.118450164794922,
|
| 6699 |
+
"learning_rate": 1.8793917465713684e-07,
|
| 6700 |
+
"loss": 2.1806,
|
| 6701 |
+
"step": 956
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.9742937134130822,
|
| 6705 |
+
"grad_norm": 7.3693928718566895,
|
| 6706 |
+
"learning_rate": 1.742835019587441e-07,
|
| 6707 |
+
"loss": 1.6516,
|
| 6708 |
+
"step": 957
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.9753117841689998,
|
| 6712 |
+
"grad_norm": 7.52052640914917,
|
| 6713 |
+
"learning_rate": 1.6114196194628172e-07,
|
| 6714 |
+
"loss": 1.8945,
|
| 6715 |
+
"step": 958
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.9763298549249173,
|
| 6719 |
+
"grad_norm": 6.575441837310791,
|
| 6720 |
+
"learning_rate": 1.4851469022234e-07,
|
| 6721 |
+
"loss": 1.5346,
|
| 6722 |
+
"step": 959
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.9773479256808348,
|
| 6726 |
+
"grad_norm": 5.139125823974609,
|
| 6727 |
+
"learning_rate": 1.3640181708293731e-07,
|
| 6728 |
+
"loss": 1.2349,
|
| 6729 |
+
"step": 960
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.9783659964367524,
|
| 6733 |
+
"grad_norm": 5.116672992706299,
|
| 6734 |
+
"learning_rate": 1.2480346751622686e-07,
|
| 6735 |
+
"loss": 1.3371,
|
| 6736 |
+
"step": 961
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.9793840671926699,
|
| 6740 |
+
"grad_norm": 6.816611289978027,
|
| 6741 |
+
"learning_rate": 1.1371976120118088e-07,
|
| 6742 |
+
"loss": 1.8166,
|
| 6743 |
+
"step": 962
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.9804021379485874,
|
| 6747 |
+
"grad_norm": 5.658526420593262,
|
| 6748 |
+
"learning_rate": 1.0315081250636405e-07,
|
| 6749 |
+
"loss": 1.435,
|
| 6750 |
+
"step": 963
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.981420208704505,
|
| 6754 |
+
"grad_norm": 5.805619239807129,
|
| 6755 |
+
"learning_rate": 9.309673048875089e-08,
|
| 6756 |
+
"loss": 1.473,
|
| 6757 |
+
"step": 964
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.9824382794604225,
|
| 6761 |
+
"grad_norm": 8.438018798828125,
|
| 6762 |
+
"learning_rate": 8.355761889260461e-08,
|
| 6763 |
+
"loss": 1.9403,
|
| 6764 |
+
"step": 965
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.98345635021634,
|
| 6768 |
+
"grad_norm": 6.625545501708984,
|
| 6769 |
+
"learning_rate": 7.453357614841117e-08,
|
| 6770 |
+
"loss": 1.7268,
|
| 6771 |
+
"step": 966
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.9844744209722576,
|
| 6775 |
+
"grad_norm": 7.497494220733643,
|
| 6776 |
+
"learning_rate": 6.602469537183021e-08,
|
| 6777 |
+
"loss": 2.0333,
|
| 6778 |
+
"step": 967
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.9854924917281751,
|
| 6782 |
+
"grad_norm": 7.167290210723877,
|
| 6783 |
+
"learning_rate": 5.8031064362795705e-08,
|
| 6784 |
+
"loss": 1.9103,
|
| 6785 |
+
"step": 968
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.9865105624840926,
|
| 6789 |
+
"grad_norm": 8.998686790466309,
|
| 6790 |
+
"learning_rate": 5.0552765604544584e-08,
|
| 6791 |
+
"loss": 2.8485,
|
| 6792 |
+
"step": 969
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.9875286332400102,
|
| 6796 |
+
"grad_norm": 7.887483596801758,
|
| 6797 |
+
"learning_rate": 4.358987626281175e-08,
|
| 6798 |
+
"loss": 2.5035,
|
| 6799 |
+
"step": 970
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.9885467039959277,
|
| 6803 |
+
"grad_norm": 8.251432418823242,
|
| 6804 |
+
"learning_rate": 3.7142468185014104e-08,
|
| 6805 |
+
"loss": 2.1646,
|
| 6806 |
+
"step": 971
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.9895647747518452,
|
| 6810 |
+
"grad_norm": 8.031188011169434,
|
| 6811 |
+
"learning_rate": 3.121060789951225e-08,
|
| 6812 |
+
"loss": 2.3445,
|
| 6813 |
+
"step": 972
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.9905828455077628,
|
| 6817 |
+
"grad_norm": 10.717124938964844,
|
| 6818 |
+
"learning_rate": 2.5794356614922134e-08,
|
| 6819 |
+
"loss": 2.4195,
|
| 6820 |
+
"step": 973
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.9916009162636803,
|
| 6824 |
+
"grad_norm": 7.7094950675964355,
|
| 6825 |
+
"learning_rate": 2.0893770219493346e-08,
|
| 6826 |
+
"loss": 2.2724,
|
| 6827 |
+
"step": 974
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.9926189870195978,
|
| 6831 |
+
"grad_norm": 9.21927547454834,
|
| 6832 |
+
"learning_rate": 1.6508899280515134e-08,
|
| 6833 |
+
"loss": 2.9697,
|
| 6834 |
+
"step": 975
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.9936370577755154,
|
| 6838 |
+
"grad_norm": 7.8263092041015625,
|
| 6839 |
+
"learning_rate": 1.2639789043805694e-08,
|
| 6840 |
+
"loss": 1.9429,
|
| 6841 |
+
"step": 976
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.9946551285314329,
|
| 6845 |
+
"grad_norm": 8.359663009643555,
|
| 6846 |
+
"learning_rate": 9.286479433257e-09,
|
| 6847 |
+
"loss": 2.078,
|
| 6848 |
+
"step": 977
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.9956731992873504,
|
| 6852 |
+
"grad_norm": 9.0786714553833,
|
| 6853 |
+
"learning_rate": 6.449005050390699e-09,
|
| 6854 |
+
"loss": 2.2142,
|
| 6855 |
+
"step": 978
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.996691270043268,
|
| 6859 |
+
"grad_norm": 8.358176231384277,
|
| 6860 |
+
"learning_rate": 4.127395174036153e-09,
|
| 6861 |
+
"loss": 1.7529,
|
| 6862 |
+
"step": 979
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.9977093407991855,
|
| 6866 |
+
"grad_norm": 8.492805480957031,
|
| 6867 |
+
"learning_rate": 2.321673760002918e-09,
|
| 6868 |
+
"loss": 2.3433,
|
| 6869 |
+
"step": 980
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.998727411555103,
|
| 6873 |
+
"grad_norm": 10.61440372467041,
|
| 6874 |
+
"learning_rate": 1.0318594408476045e-09,
|
| 6875 |
+
"loss": 2.466,
|
| 6876 |
+
"step": 981
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.9997454823110206,
|
| 6880 |
+
"grad_norm": 9.659395217895508,
|
| 6881 |
+
"learning_rate": 2.57965525674031e-10,
|
| 6882 |
+
"loss": 1.3249,
|
| 6883 |
+
"step": 982
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.9997454823110206,
|
| 6887 |
+
"eval_loss": 0.5453814268112183,
|
| 6888 |
+
"eval_runtime": 50.2489,
|
| 6889 |
+
"eval_samples_per_second": 16.478,
|
| 6890 |
+
"eval_steps_per_second": 4.119,
|
| 6891 |
+
"step": 982
|
| 6892 |
+
},
|
| 6893 |
+
{
|
| 6894 |
+
"epoch": 1.000763553066938,
|
| 6895 |
+
"grad_norm": 6.880834102630615,
|
| 6896 |
+
"learning_rate": 0.0,
|
| 6897 |
+
"loss": 1.8668,
|
| 6898 |
+
"step": 983
|
| 6899 |
}
|
| 6900 |
],
|
| 6901 |
"logging_steps": 1,
|
|
|
|
| 6910 |
"should_evaluate": false,
|
| 6911 |
"should_log": false,
|
| 6912 |
"should_save": true,
|
| 6913 |
+
"should_training_stop": true
|
| 6914 |
},
|
| 6915 |
"attributes": {}
|
| 6916 |
}
|
| 6917 |
},
|
| 6918 |
+
"total_flos": 8.031180351948718e+17,
|
| 6919 |
"train_batch_size": 4,
|
| 6920 |
"trial_name": null,
|
| 6921 |
"trial_params": null
|