Training in progress, step 1080, 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 100059752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:caf1dfc663d54875b2858a0d751629545f6e776794355547863522a659a0d194
|
| 3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 51244404
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5b137c8eb22b98e3f5361ff4d5bf05eeb7b1d2b6e251d4374a83036ee6cf458
|
| 3 |
size 51244404
|
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 sha256:2db36a653237a36900e264b9145238e7d817d66288441b8c45d83b48afa65679
|
| 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 sha256:57b5b8603397b8d79216ebee27d8b8db17a05e79b402ad6bde186043bff2a03a
|
| 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": 360,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5071,6 +5071,2534 @@
|
|
| 5071 |
"eval_samples_per_second": 28.162,
|
| 5072 |
"eval_steps_per_second": 14.081,
|
| 5073 |
"step": 720
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5074 |
}
|
| 5075 |
],
|
| 5076 |
"logging_steps": 1,
|
|
@@ -5090,7 +7618,7 @@
|
|
| 5090 |
"attributes": {}
|
| 5091 |
}
|
| 5092 |
},
|
| 5093 |
-
"total_flos":
|
| 5094 |
"train_batch_size": 2,
|
| 5095 |
"trial_name": null,
|
| 5096 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.09112962767640544,
|
| 5 |
"eval_steps": 360,
|
| 6 |
+
"global_step": 1080,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5071 |
"eval_samples_per_second": 28.162,
|
| 5072 |
"eval_steps_per_second": 14.081,
|
| 5073 |
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.06083746440248919,
|
| 5077 |
+
"grad_norm": 3.229191780090332,
|
| 5078 |
+
"learning_rate": 0.00010055038135820225,
|
| 5079 |
+
"loss": 1.4377,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.06092184368737475,
|
| 5084 |
+
"grad_norm": 3.5044431686401367,
|
| 5085 |
+
"learning_rate": 0.00010033022988195178,
|
| 5086 |
+
"loss": 1.4915,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.06100622297226031,
|
| 5091 |
+
"grad_norm": 3.6912529468536377,
|
| 5092 |
+
"learning_rate": 0.00010011007680515593,
|
| 5093 |
+
"loss": 1.4665,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.06109060225714587,
|
| 5098 |
+
"grad_norm": 3.602473735809326,
|
| 5099 |
+
"learning_rate": 9.988992319484408e-05,
|
| 5100 |
+
"loss": 1.6845,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.06117498154203143,
|
| 5105 |
+
"grad_norm": 4.103678226470947,
|
| 5106 |
+
"learning_rate": 9.966977011804823e-05,
|
| 5107 |
+
"loss": 2.0198,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.06125936082691699,
|
| 5112 |
+
"grad_norm": 3.193321943283081,
|
| 5113 |
+
"learning_rate": 9.944961864179774e-05,
|
| 5114 |
+
"loss": 1.5488,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.061343740111802554,
|
| 5119 |
+
"grad_norm": 3.6511833667755127,
|
| 5120 |
+
"learning_rate": 9.92294698331143e-05,
|
| 5121 |
+
"loss": 1.3478,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.061428119396688116,
|
| 5126 |
+
"grad_norm": 3.9025187492370605,
|
| 5127 |
+
"learning_rate": 9.90093247590066e-05,
|
| 5128 |
+
"loss": 1.5672,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.06151249868157367,
|
| 5133 |
+
"grad_norm": 3.2794833183288574,
|
| 5134 |
+
"learning_rate": 9.87891844864652e-05,
|
| 5135 |
+
"loss": 1.4662,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.06159687796645923,
|
| 5140 |
+
"grad_norm": 3.1737897396087646,
|
| 5141 |
+
"learning_rate": 9.856905008245747e-05,
|
| 5142 |
+
"loss": 1.5617,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.061681257251344795,
|
| 5147 |
+
"grad_norm": 3.4398791790008545,
|
| 5148 |
+
"learning_rate": 9.83489226139223e-05,
|
| 5149 |
+
"loss": 1.6734,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.06176563653623036,
|
| 5154 |
+
"grad_norm": 3.4238171577453613,
|
| 5155 |
+
"learning_rate": 9.812880314776496e-05,
|
| 5156 |
+
"loss": 1.532,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.06185001582111591,
|
| 5161 |
+
"grad_norm": 4.550687313079834,
|
| 5162 |
+
"learning_rate": 9.790869275085196e-05,
|
| 5163 |
+
"loss": 1.681,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.061934395106001475,
|
| 5168 |
+
"grad_norm": 3.855168581008911,
|
| 5169 |
+
"learning_rate": 9.76885924900058e-05,
|
| 5170 |
+
"loss": 1.8827,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.06201877439088704,
|
| 5175 |
+
"grad_norm": 3.3226592540740967,
|
| 5176 |
+
"learning_rate": 9.746850343199995e-05,
|
| 5177 |
+
"loss": 1.6785,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.0621031536757726,
|
| 5182 |
+
"grad_norm": 3.230769634246826,
|
| 5183 |
+
"learning_rate": 9.724842664355347e-05,
|
| 5184 |
+
"loss": 1.6788,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.06218753296065816,
|
| 5189 |
+
"grad_norm": 2.9634227752685547,
|
| 5190 |
+
"learning_rate": 9.702836319132597e-05,
|
| 5191 |
+
"loss": 1.3981,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.062271912245543716,
|
| 5196 |
+
"grad_norm": 3.766829490661621,
|
| 5197 |
+
"learning_rate": 9.680831414191253e-05,
|
| 5198 |
+
"loss": 1.4034,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.06235629153042928,
|
| 5203 |
+
"grad_norm": 3.2265536785125732,
|
| 5204 |
+
"learning_rate": 9.658828056183828e-05,
|
| 5205 |
+
"loss": 1.3725,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.06244067081531484,
|
| 5210 |
+
"grad_norm": 3.632070779800415,
|
| 5211 |
+
"learning_rate": 9.636826351755346e-05,
|
| 5212 |
+
"loss": 1.9203,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.0625250501002004,
|
| 5217 |
+
"grad_norm": 2.820812463760376,
|
| 5218 |
+
"learning_rate": 9.61482640754282e-05,
|
| 5219 |
+
"loss": 1.1133,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.06260942938508596,
|
| 5224 |
+
"grad_norm": 3.2152464389801025,
|
| 5225 |
+
"learning_rate": 9.592828330174726e-05,
|
| 5226 |
+
"loss": 1.4804,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.06269380866997153,
|
| 5231 |
+
"grad_norm": 3.831986904144287,
|
| 5232 |
+
"learning_rate": 9.570832226270482e-05,
|
| 5233 |
+
"loss": 1.6262,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.06277818795485708,
|
| 5238 |
+
"grad_norm": 3.158581256866455,
|
| 5239 |
+
"learning_rate": 9.54883820243996e-05,
|
| 5240 |
+
"loss": 1.5282,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.06286256723974264,
|
| 5245 |
+
"grad_norm": 3.6407358646392822,
|
| 5246 |
+
"learning_rate": 9.526846365282937e-05,
|
| 5247 |
+
"loss": 1.5901,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.0629469465246282,
|
| 5252 |
+
"grad_norm": 3.8346829414367676,
|
| 5253 |
+
"learning_rate": 9.504856821388602e-05,
|
| 5254 |
+
"loss": 1.6533,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.06303132580951376,
|
| 5259 |
+
"grad_norm": 3.2862420082092285,
|
| 5260 |
+
"learning_rate": 9.482869677335017e-05,
|
| 5261 |
+
"loss": 1.5965,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.06311570509439933,
|
| 5266 |
+
"grad_norm": 2.801699638366699,
|
| 5267 |
+
"learning_rate": 9.460885039688625e-05,
|
| 5268 |
+
"loss": 1.1895,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.06320008437928488,
|
| 5273 |
+
"grad_norm": 3.4097049236297607,
|
| 5274 |
+
"learning_rate": 9.438903015003714e-05,
|
| 5275 |
+
"loss": 1.8107,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.06328446366417044,
|
| 5280 |
+
"grad_norm": 4.033748626708984,
|
| 5281 |
+
"learning_rate": 9.416923709821903e-05,
|
| 5282 |
+
"loss": 1.5959,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.06336884294905601,
|
| 5287 |
+
"grad_norm": 3.375924825668335,
|
| 5288 |
+
"learning_rate": 9.394947230671638e-05,
|
| 5289 |
+
"loss": 1.2444,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.06345322223394156,
|
| 5294 |
+
"grad_norm": 3.985337734222412,
|
| 5295 |
+
"learning_rate": 9.372973684067671e-05,
|
| 5296 |
+
"loss": 1.3358,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.06353760151882713,
|
| 5301 |
+
"grad_norm": 3.5320260524749756,
|
| 5302 |
+
"learning_rate": 9.35100317651053e-05,
|
| 5303 |
+
"loss": 1.6331,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.06362198080371269,
|
| 5308 |
+
"grad_norm": 4.086630344390869,
|
| 5309 |
+
"learning_rate": 9.329035814486021e-05,
|
| 5310 |
+
"loss": 1.5098,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.06370636008859824,
|
| 5315 |
+
"grad_norm": 3.4603257179260254,
|
| 5316 |
+
"learning_rate": 9.307071704464701e-05,
|
| 5317 |
+
"loss": 1.317,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.06379073937348381,
|
| 5322 |
+
"grad_norm": 4.04933500289917,
|
| 5323 |
+
"learning_rate": 9.285110952901369e-05,
|
| 5324 |
+
"loss": 1.8193,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.06387511865836937,
|
| 5329 |
+
"grad_norm": 3.3762123584747314,
|
| 5330 |
+
"learning_rate": 9.263153666234545e-05,
|
| 5331 |
+
"loss": 1.7771,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.06395949794325494,
|
| 5336 |
+
"grad_norm": 2.8218278884887695,
|
| 5337 |
+
"learning_rate": 9.241199950885949e-05,
|
| 5338 |
+
"loss": 1.3342,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.06404387722814049,
|
| 5343 |
+
"grad_norm": 3.081434726715088,
|
| 5344 |
+
"learning_rate": 9.219249913259998e-05,
|
| 5345 |
+
"loss": 1.4174,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.06412825651302605,
|
| 5350 |
+
"grad_norm": 4.025990962982178,
|
| 5351 |
+
"learning_rate": 9.197303659743286e-05,
|
| 5352 |
+
"loss": 1.7749,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.06421263579791162,
|
| 5357 |
+
"grad_norm": 3.2301442623138428,
|
| 5358 |
+
"learning_rate": 9.17536129670406e-05,
|
| 5359 |
+
"loss": 1.5955,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.06429701508279717,
|
| 5364 |
+
"grad_norm": 3.809837818145752,
|
| 5365 |
+
"learning_rate": 9.15342293049172e-05,
|
| 5366 |
+
"loss": 2.047,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.06438139436768274,
|
| 5371 |
+
"grad_norm": 3.9897491931915283,
|
| 5372 |
+
"learning_rate": 9.131488667436281e-05,
|
| 5373 |
+
"loss": 1.6496,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.0644657736525683,
|
| 5378 |
+
"grad_norm": 4.438673496246338,
|
| 5379 |
+
"learning_rate": 9.109558613847887e-05,
|
| 5380 |
+
"loss": 1.5308,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.06455015293745385,
|
| 5385 |
+
"grad_norm": 3.345057249069214,
|
| 5386 |
+
"learning_rate": 9.087632876016264e-05,
|
| 5387 |
+
"loss": 2.0642,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.06463453222233942,
|
| 5392 |
+
"grad_norm": 3.56162166595459,
|
| 5393 |
+
"learning_rate": 9.065711560210227e-05,
|
| 5394 |
+
"loss": 1.619,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.06471891150722497,
|
| 5399 |
+
"grad_norm": 3.0354530811309814,
|
| 5400 |
+
"learning_rate": 9.043794772677164e-05,
|
| 5401 |
+
"loss": 1.582,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.06480329079211054,
|
| 5406 |
+
"grad_norm": 3.2199714183807373,
|
| 5407 |
+
"learning_rate": 9.021882619642507e-05,
|
| 5408 |
+
"loss": 1.4944,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.0648876700769961,
|
| 5413 |
+
"grad_norm": 3.431900978088379,
|
| 5414 |
+
"learning_rate": 8.999975207309235e-05,
|
| 5415 |
+
"loss": 1.5814,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.06497204936188165,
|
| 5420 |
+
"grad_norm": 3.4392762184143066,
|
| 5421 |
+
"learning_rate": 8.978072641857336e-05,
|
| 5422 |
+
"loss": 1.549,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.06505642864676722,
|
| 5427 |
+
"grad_norm": 3.4558968544006348,
|
| 5428 |
+
"learning_rate": 8.956175029443325e-05,
|
| 5429 |
+
"loss": 1.6068,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.06514080793165278,
|
| 5434 |
+
"grad_norm": 3.6479742527008057,
|
| 5435 |
+
"learning_rate": 8.93428247619969e-05,
|
| 5436 |
+
"loss": 1.5165,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.06522518721653835,
|
| 5441 |
+
"grad_norm": 3.4587035179138184,
|
| 5442 |
+
"learning_rate": 8.912395088234417e-05,
|
| 5443 |
+
"loss": 1.7611,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.0653095665014239,
|
| 5448 |
+
"grad_norm": 3.351179838180542,
|
| 5449 |
+
"learning_rate": 8.890512971630442e-05,
|
| 5450 |
+
"loss": 1.8494,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.06539394578630946,
|
| 5455 |
+
"grad_norm": 4.118631839752197,
|
| 5456 |
+
"learning_rate": 8.868636232445165e-05,
|
| 5457 |
+
"loss": 1.6619,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.06547832507119503,
|
| 5462 |
+
"grad_norm": 3.9293482303619385,
|
| 5463 |
+
"learning_rate": 8.846764976709914e-05,
|
| 5464 |
+
"loss": 1.511,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.06556270435608058,
|
| 5469 |
+
"grad_norm": 3.27178692817688,
|
| 5470 |
+
"learning_rate": 8.824899310429445e-05,
|
| 5471 |
+
"loss": 1.933,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.06564708364096614,
|
| 5476 |
+
"grad_norm": 3.6199560165405273,
|
| 5477 |
+
"learning_rate": 8.803039339581418e-05,
|
| 5478 |
+
"loss": 1.4583,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.0657314629258517,
|
| 5483 |
+
"grad_norm": 3.488508701324463,
|
| 5484 |
+
"learning_rate": 8.781185170115898e-05,
|
| 5485 |
+
"loss": 1.3298,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.06581584221073726,
|
| 5490 |
+
"grad_norm": 3.3315086364746094,
|
| 5491 |
+
"learning_rate": 8.75933690795482e-05,
|
| 5492 |
+
"loss": 1.4671,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.06590022149562283,
|
| 5497 |
+
"grad_norm": 3.379645586013794,
|
| 5498 |
+
"learning_rate": 8.737494658991493e-05,
|
| 5499 |
+
"loss": 1.5694,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.06598460078050838,
|
| 5504 |
+
"grad_norm": 3.764233112335205,
|
| 5505 |
+
"learning_rate": 8.715658529090083e-05,
|
| 5506 |
+
"loss": 1.722,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.06606898006539394,
|
| 5511 |
+
"grad_norm": 3.8047001361846924,
|
| 5512 |
+
"learning_rate": 8.693828624085104e-05,
|
| 5513 |
+
"loss": 1.7402,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.06615335935027951,
|
| 5518 |
+
"grad_norm": 3.627662420272827,
|
| 5519 |
+
"learning_rate": 8.672005049780884e-05,
|
| 5520 |
+
"loss": 1.6401,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.06623773863516506,
|
| 5525 |
+
"grad_norm": 4.321152687072754,
|
| 5526 |
+
"learning_rate": 8.650187911951082e-05,
|
| 5527 |
+
"loss": 2.2155,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.06632211792005063,
|
| 5532 |
+
"grad_norm": 3.7158429622650146,
|
| 5533 |
+
"learning_rate": 8.628377316338156e-05,
|
| 5534 |
+
"loss": 1.7856,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.06640649720493619,
|
| 5539 |
+
"grad_norm": 2.9161453247070312,
|
| 5540 |
+
"learning_rate": 8.606573368652848e-05,
|
| 5541 |
+
"loss": 1.2219,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.06649087648982174,
|
| 5546 |
+
"grad_norm": 3.9051358699798584,
|
| 5547 |
+
"learning_rate": 8.584776174573692e-05,
|
| 5548 |
+
"loss": 1.808,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.06657525577470731,
|
| 5553 |
+
"grad_norm": 4.269354343414307,
|
| 5554 |
+
"learning_rate": 8.56298583974648e-05,
|
| 5555 |
+
"loss": 1.9708,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.06665963505959287,
|
| 5560 |
+
"grad_norm": 4.096782684326172,
|
| 5561 |
+
"learning_rate": 8.541202469783766e-05,
|
| 5562 |
+
"loss": 1.894,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.06674401434447844,
|
| 5567 |
+
"grad_norm": 3.182770252227783,
|
| 5568 |
+
"learning_rate": 8.519426170264336e-05,
|
| 5569 |
+
"loss": 2.0521,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.06682839362936399,
|
| 5574 |
+
"grad_norm": 3.3189237117767334,
|
| 5575 |
+
"learning_rate": 8.497657046732718e-05,
|
| 5576 |
+
"loss": 1.6191,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.06691277291424955,
|
| 5581 |
+
"grad_norm": 2.698390483856201,
|
| 5582 |
+
"learning_rate": 8.475895204698659e-05,
|
| 5583 |
+
"loss": 1.2702,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.06699715219913512,
|
| 5588 |
+
"grad_norm": 3.1540493965148926,
|
| 5589 |
+
"learning_rate": 8.454140749636605e-05,
|
| 5590 |
+
"loss": 1.4738,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.06708153148402067,
|
| 5595 |
+
"grad_norm": 3.229556083679199,
|
| 5596 |
+
"learning_rate": 8.432393786985204e-05,
|
| 5597 |
+
"loss": 1.6078,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.06716591076890624,
|
| 5602 |
+
"grad_norm": 3.086798667907715,
|
| 5603 |
+
"learning_rate": 8.410654422146798e-05,
|
| 5604 |
+
"loss": 1.4351,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.0672502900537918,
|
| 5609 |
+
"grad_norm": 3.499110698699951,
|
| 5610 |
+
"learning_rate": 8.388922760486891e-05,
|
| 5611 |
+
"loss": 2.0398,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.06733466933867735,
|
| 5616 |
+
"grad_norm": 4.102422714233398,
|
| 5617 |
+
"learning_rate": 8.36719890733366e-05,
|
| 5618 |
+
"loss": 1.486,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.06741904862356292,
|
| 5623 |
+
"grad_norm": 4.713635444641113,
|
| 5624 |
+
"learning_rate": 8.345482967977434e-05,
|
| 5625 |
+
"loss": 1.9632,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.06750342790844847,
|
| 5630 |
+
"grad_norm": 3.019648551940918,
|
| 5631 |
+
"learning_rate": 8.323775047670189e-05,
|
| 5632 |
+
"loss": 1.3464,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.06758780719333404,
|
| 5637 |
+
"grad_norm": 3.2463009357452393,
|
| 5638 |
+
"learning_rate": 8.302075251625026e-05,
|
| 5639 |
+
"loss": 1.1357,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.0676721864782196,
|
| 5644 |
+
"grad_norm": 3.1642446517944336,
|
| 5645 |
+
"learning_rate": 8.280383685015673e-05,
|
| 5646 |
+
"loss": 1.746,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.06775656576310515,
|
| 5651 |
+
"grad_norm": 3.4454307556152344,
|
| 5652 |
+
"learning_rate": 8.258700452975979e-05,
|
| 5653 |
+
"loss": 1.6777,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.06784094504799072,
|
| 5658 |
+
"grad_norm": 3.307554244995117,
|
| 5659 |
+
"learning_rate": 8.237025660599393e-05,
|
| 5660 |
+
"loss": 1.9145,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.06792532433287628,
|
| 5665 |
+
"grad_norm": 2.7016055583953857,
|
| 5666 |
+
"learning_rate": 8.215359412938451e-05,
|
| 5667 |
+
"loss": 1.3254,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.06800970361776185,
|
| 5672 |
+
"grad_norm": 2.5887274742126465,
|
| 5673 |
+
"learning_rate": 8.19370181500429e-05,
|
| 5674 |
+
"loss": 1.1497,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.0680940829026474,
|
| 5679 |
+
"grad_norm": 3.540260076522827,
|
| 5680 |
+
"learning_rate": 8.172052971766109e-05,
|
| 5681 |
+
"loss": 1.6842,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.06817846218753296,
|
| 5686 |
+
"grad_norm": 3.2746472358703613,
|
| 5687 |
+
"learning_rate": 8.150412988150689e-05,
|
| 5688 |
+
"loss": 1.7623,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.06826284147241853,
|
| 5693 |
+
"grad_norm": 3.496894598007202,
|
| 5694 |
+
"learning_rate": 8.128781969041856e-05,
|
| 5695 |
+
"loss": 1.4555,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.06834722075730408,
|
| 5700 |
+
"grad_norm": 4.125760078430176,
|
| 5701 |
+
"learning_rate": 8.107160019279996e-05,
|
| 5702 |
+
"loss": 1.77,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.06843160004218964,
|
| 5707 |
+
"grad_norm": 3.6408426761627197,
|
| 5708 |
+
"learning_rate": 8.085547243661537e-05,
|
| 5709 |
+
"loss": 1.511,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.0685159793270752,
|
| 5714 |
+
"grad_norm": 3.1120951175689697,
|
| 5715 |
+
"learning_rate": 8.063943746938443e-05,
|
| 5716 |
+
"loss": 1.4816,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.06860035861196076,
|
| 5721 |
+
"grad_norm": 4.043081283569336,
|
| 5722 |
+
"learning_rate": 8.042349633817703e-05,
|
| 5723 |
+
"loss": 1.6875,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.06868473789684633,
|
| 5728 |
+
"grad_norm": 2.890449047088623,
|
| 5729 |
+
"learning_rate": 8.020765008960828e-05,
|
| 5730 |
+
"loss": 1.3573,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.06876911718173188,
|
| 5735 |
+
"grad_norm": 3.8338587284088135,
|
| 5736 |
+
"learning_rate": 7.999189976983343e-05,
|
| 5737 |
+
"loss": 1.8795,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.06885349646661744,
|
| 5742 |
+
"grad_norm": 3.022118330001831,
|
| 5743 |
+
"learning_rate": 7.977624642454268e-05,
|
| 5744 |
+
"loss": 1.4375,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.06893787575150301,
|
| 5749 |
+
"grad_norm": 3.2548141479492188,
|
| 5750 |
+
"learning_rate": 7.956069109895636e-05,
|
| 5751 |
+
"loss": 1.3097,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.06902225503638856,
|
| 5756 |
+
"grad_norm": 3.541154623031616,
|
| 5757 |
+
"learning_rate": 7.934523483781963e-05,
|
| 5758 |
+
"loss": 1.5465,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.06910663432127413,
|
| 5763 |
+
"grad_norm": 3.054385185241699,
|
| 5764 |
+
"learning_rate": 7.912987868539759e-05,
|
| 5765 |
+
"loss": 1.4008,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.06919101360615969,
|
| 5770 |
+
"grad_norm": 3.5634217262268066,
|
| 5771 |
+
"learning_rate": 7.891462368547003e-05,
|
| 5772 |
+
"loss": 1.8095,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.06927539289104524,
|
| 5777 |
+
"grad_norm": 3.556551218032837,
|
| 5778 |
+
"learning_rate": 7.869947088132657e-05,
|
| 5779 |
+
"loss": 1.8197,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.06935977217593081,
|
| 5784 |
+
"grad_norm": 4.121778964996338,
|
| 5785 |
+
"learning_rate": 7.84844213157615e-05,
|
| 5786 |
+
"loss": 1.2268,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.06944415146081637,
|
| 5791 |
+
"grad_norm": 3.446765661239624,
|
| 5792 |
+
"learning_rate": 7.826947603106864e-05,
|
| 5793 |
+
"loss": 1.1298,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.06952853074570194,
|
| 5798 |
+
"grad_norm": 3.5260965824127197,
|
| 5799 |
+
"learning_rate": 7.80546360690365e-05,
|
| 5800 |
+
"loss": 1.6826,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.06961291003058749,
|
| 5805 |
+
"grad_norm": 3.636963129043579,
|
| 5806 |
+
"learning_rate": 7.783990247094308e-05,
|
| 5807 |
+
"loss": 1.8052,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.06969728931547305,
|
| 5812 |
+
"grad_norm": 3.8106868267059326,
|
| 5813 |
+
"learning_rate": 7.762527627755083e-05,
|
| 5814 |
+
"loss": 1.603,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.06978166860035862,
|
| 5819 |
+
"grad_norm": 3.557321786880493,
|
| 5820 |
+
"learning_rate": 7.741075852910172e-05,
|
| 5821 |
+
"loss": 1.5966,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.06986604788524417,
|
| 5826 |
+
"grad_norm": 3.3957102298736572,
|
| 5827 |
+
"learning_rate": 7.719635026531195e-05,
|
| 5828 |
+
"loss": 1.987,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.06995042717012974,
|
| 5833 |
+
"grad_norm": 3.3128857612609863,
|
| 5834 |
+
"learning_rate": 7.698205252536728e-05,
|
| 5835 |
+
"loss": 1.5101,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.0700348064550153,
|
| 5840 |
+
"grad_norm": 4.3435797691345215,
|
| 5841 |
+
"learning_rate": 7.676786634791766e-05,
|
| 5842 |
+
"loss": 1.7899,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.07011918573990085,
|
| 5847 |
+
"grad_norm": 3.399648666381836,
|
| 5848 |
+
"learning_rate": 7.655379277107228e-05,
|
| 5849 |
+
"loss": 1.3155,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.07020356502478642,
|
| 5854 |
+
"grad_norm": 3.2626960277557373,
|
| 5855 |
+
"learning_rate": 7.633983283239471e-05,
|
| 5856 |
+
"loss": 1.6369,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.07028794430967197,
|
| 5861 |
+
"grad_norm": 4.113405704498291,
|
| 5862 |
+
"learning_rate": 7.612598756889768e-05,
|
| 5863 |
+
"loss": 1.4886,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.07037232359455754,
|
| 5868 |
+
"grad_norm": 4.464077949523926,
|
| 5869 |
+
"learning_rate": 7.591225801703809e-05,
|
| 5870 |
+
"loss": 1.475,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.0704567028794431,
|
| 5875 |
+
"grad_norm": 4.5221686363220215,
|
| 5876 |
+
"learning_rate": 7.569864521271207e-05,
|
| 5877 |
+
"loss": 2.0369,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.07054108216432865,
|
| 5882 |
+
"grad_norm": 3.127971887588501,
|
| 5883 |
+
"learning_rate": 7.548515019124986e-05,
|
| 5884 |
+
"loss": 1.1989,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.07062546144921422,
|
| 5889 |
+
"grad_norm": 3.166808605194092,
|
| 5890 |
+
"learning_rate": 7.527177398741088e-05,
|
| 5891 |
+
"loss": 1.573,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.07070984073409978,
|
| 5896 |
+
"grad_norm": 3.741363286972046,
|
| 5897 |
+
"learning_rate": 7.505851763537861e-05,
|
| 5898 |
+
"loss": 1.544,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.07079422001898535,
|
| 5903 |
+
"grad_norm": 3.4282264709472656,
|
| 5904 |
+
"learning_rate": 7.484538216875563e-05,
|
| 5905 |
+
"loss": 1.7555,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.0708785993038709,
|
| 5910 |
+
"grad_norm": 4.068146228790283,
|
| 5911 |
+
"learning_rate": 7.463236862055872e-05,
|
| 5912 |
+
"loss": 1.7943,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.07096297858875646,
|
| 5917 |
+
"grad_norm": 3.7871479988098145,
|
| 5918 |
+
"learning_rate": 7.441947802321363e-05,
|
| 5919 |
+
"loss": 1.8306,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.07104735787364203,
|
| 5924 |
+
"grad_norm": 3.1276440620422363,
|
| 5925 |
+
"learning_rate": 7.420671140855026e-05,
|
| 5926 |
+
"loss": 1.4611,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.07113173715852758,
|
| 5931 |
+
"grad_norm": 3.529446601867676,
|
| 5932 |
+
"learning_rate": 7.399406980779765e-05,
|
| 5933 |
+
"loss": 1.2342,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.07121611644341314,
|
| 5938 |
+
"grad_norm": 3.3045966625213623,
|
| 5939 |
+
"learning_rate": 7.378155425157883e-05,
|
| 5940 |
+
"loss": 1.4644,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.0713004957282987,
|
| 5945 |
+
"grad_norm": 3.1722381114959717,
|
| 5946 |
+
"learning_rate": 7.356916576990587e-05,
|
| 5947 |
+
"loss": 1.2117,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.07138487501318426,
|
| 5952 |
+
"grad_norm": 3.494398355484009,
|
| 5953 |
+
"learning_rate": 7.335690539217513e-05,
|
| 5954 |
+
"loss": 1.7005,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.07146925429806983,
|
| 5959 |
+
"grad_norm": 4.649952411651611,
|
| 5960 |
+
"learning_rate": 7.314477414716192e-05,
|
| 5961 |
+
"loss": 2.0836,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.07155363358295538,
|
| 5966 |
+
"grad_norm": 3.2355849742889404,
|
| 5967 |
+
"learning_rate": 7.293277306301575e-05,
|
| 5968 |
+
"loss": 1.8021,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.07163801286784094,
|
| 5973 |
+
"grad_norm": 3.792217254638672,
|
| 5974 |
+
"learning_rate": 7.272090316725527e-05,
|
| 5975 |
+
"loss": 1.9283,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.07172239215272651,
|
| 5980 |
+
"grad_norm": 3.2298521995544434,
|
| 5981 |
+
"learning_rate": 7.250916548676323e-05,
|
| 5982 |
+
"loss": 1.3729,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.07180677143761206,
|
| 5987 |
+
"grad_norm": 3.544177770614624,
|
| 5988 |
+
"learning_rate": 7.229756104778167e-05,
|
| 5989 |
+
"loss": 1.6331,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.07189115072249763,
|
| 5994 |
+
"grad_norm": 3.6409194469451904,
|
| 5995 |
+
"learning_rate": 7.20860908759067e-05,
|
| 5996 |
+
"loss": 1.8094,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.07197553000738319,
|
| 6001 |
+
"grad_norm": 2.7580833435058594,
|
| 6002 |
+
"learning_rate": 7.187475599608378e-05,
|
| 6003 |
+
"loss": 1.3586,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.07205990929226874,
|
| 6008 |
+
"grad_norm": 3.81415057182312,
|
| 6009 |
+
"learning_rate": 7.166355743260265e-05,
|
| 6010 |
+
"loss": 2.0804,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.07214428857715431,
|
| 6015 |
+
"grad_norm": 2.594660758972168,
|
| 6016 |
+
"learning_rate": 7.145249620909224e-05,
|
| 6017 |
+
"loss": 1.2372,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.07222866786203987,
|
| 6022 |
+
"grad_norm": 3.071242332458496,
|
| 6023 |
+
"learning_rate": 7.124157334851597e-05,
|
| 6024 |
+
"loss": 1.1069,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.07231304714692544,
|
| 6029 |
+
"grad_norm": 3.43115234375,
|
| 6030 |
+
"learning_rate": 7.103078987316652e-05,
|
| 6031 |
+
"loss": 1.6771,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.07239742643181099,
|
| 6036 |
+
"grad_norm": 3.099498987197876,
|
| 6037 |
+
"learning_rate": 7.082014680466112e-05,
|
| 6038 |
+
"loss": 1.4458,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.07248180571669655,
|
| 6043 |
+
"grad_norm": 3.128213882446289,
|
| 6044 |
+
"learning_rate": 7.060964516393642e-05,
|
| 6045 |
+
"loss": 1.2736,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.07256618500158211,
|
| 6050 |
+
"grad_norm": 3.516040325164795,
|
| 6051 |
+
"learning_rate": 7.039928597124354e-05,
|
| 6052 |
+
"loss": 1.6611,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.07265056428646767,
|
| 6057 |
+
"grad_norm": 3.657384157180786,
|
| 6058 |
+
"learning_rate": 7.018907024614334e-05,
|
| 6059 |
+
"loss": 1.5757,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.07273494357135324,
|
| 6064 |
+
"grad_norm": 3.818807363510132,
|
| 6065 |
+
"learning_rate": 6.997899900750118e-05,
|
| 6066 |
+
"loss": 1.4448,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.0728193228562388,
|
| 6071 |
+
"grad_norm": 3.4620652198791504,
|
| 6072 |
+
"learning_rate": 6.97690732734822e-05,
|
| 6073 |
+
"loss": 1.7199,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.07290370214112435,
|
| 6078 |
+
"grad_norm": 4.390957355499268,
|
| 6079 |
+
"learning_rate": 6.955929406154636e-05,
|
| 6080 |
+
"loss": 1.5205,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.07298808142600992,
|
| 6085 |
+
"grad_norm": 3.043879985809326,
|
| 6086 |
+
"learning_rate": 6.934966238844332e-05,
|
| 6087 |
+
"loss": 1.3107,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.07307246071089547,
|
| 6092 |
+
"grad_norm": 4.120273113250732,
|
| 6093 |
+
"learning_rate": 6.914017927020786e-05,
|
| 6094 |
+
"loss": 2.0678,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.07315683999578104,
|
| 6099 |
+
"grad_norm": 2.8248279094696045,
|
| 6100 |
+
"learning_rate": 6.893084572215453e-05,
|
| 6101 |
+
"loss": 1.4521,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.0732412192806666,
|
| 6106 |
+
"grad_norm": 3.080636739730835,
|
| 6107 |
+
"learning_rate": 6.872166275887306e-05,
|
| 6108 |
+
"loss": 1.769,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.07332559856555215,
|
| 6113 |
+
"grad_norm": 3.1468093395233154,
|
| 6114 |
+
"learning_rate": 6.851263139422339e-05,
|
| 6115 |
+
"loss": 1.223,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.07340997785043772,
|
| 6120 |
+
"grad_norm": 3.397599458694458,
|
| 6121 |
+
"learning_rate": 6.830375264133055e-05,
|
| 6122 |
+
"loss": 1.7281,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.07349435713532328,
|
| 6127 |
+
"grad_norm": 3.566293716430664,
|
| 6128 |
+
"learning_rate": 6.809502751258003e-05,
|
| 6129 |
+
"loss": 1.3899,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.07357873642020885,
|
| 6134 |
+
"grad_norm": 3.603502035140991,
|
| 6135 |
+
"learning_rate": 6.788645701961267e-05,
|
| 6136 |
+
"loss": 1.7868,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.0736631157050944,
|
| 6141 |
+
"grad_norm": 2.914247751235962,
|
| 6142 |
+
"learning_rate": 6.767804217331989e-05,
|
| 6143 |
+
"loss": 1.4781,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.07374749498997996,
|
| 6148 |
+
"grad_norm": 3.851073980331421,
|
| 6149 |
+
"learning_rate": 6.74697839838386e-05,
|
| 6150 |
+
"loss": 1.9343,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.07383187427486552,
|
| 6155 |
+
"grad_norm": 2.637230396270752,
|
| 6156 |
+
"learning_rate": 6.726168346054657e-05,
|
| 6157 |
+
"loss": 1.2372,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.07391625355975108,
|
| 6162 |
+
"grad_norm": 3.256101131439209,
|
| 6163 |
+
"learning_rate": 6.705374161205732e-05,
|
| 6164 |
+
"loss": 1.4526,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.07400063284463664,
|
| 6169 |
+
"grad_norm": 2.6738524436950684,
|
| 6170 |
+
"learning_rate": 6.684595944621536e-05,
|
| 6171 |
+
"loss": 1.2832,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.0740850121295222,
|
| 6176 |
+
"grad_norm": 2.684112310409546,
|
| 6177 |
+
"learning_rate": 6.663833797009123e-05,
|
| 6178 |
+
"loss": 1.0865,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.07416939141440776,
|
| 6183 |
+
"grad_norm": 3.5784056186676025,
|
| 6184 |
+
"learning_rate": 6.643087818997663e-05,
|
| 6185 |
+
"loss": 1.3953,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.07425377069929333,
|
| 6190 |
+
"grad_norm": 2.9961676597595215,
|
| 6191 |
+
"learning_rate": 6.622358111137958e-05,
|
| 6192 |
+
"loss": 1.3251,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.07433814998417888,
|
| 6197 |
+
"grad_norm": 2.8923776149749756,
|
| 6198 |
+
"learning_rate": 6.601644773901957e-05,
|
| 6199 |
+
"loss": 1.0592,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.07442252926906444,
|
| 6204 |
+
"grad_norm": 2.8989806175231934,
|
| 6205 |
+
"learning_rate": 6.58094790768225e-05,
|
| 6206 |
+
"loss": 1.7374,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.07450690855395001,
|
| 6211 |
+
"grad_norm": 3.5653345584869385,
|
| 6212 |
+
"learning_rate": 6.56026761279161e-05,
|
| 6213 |
+
"loss": 1.5435,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.07459128783883556,
|
| 6218 |
+
"grad_norm": 3.112036943435669,
|
| 6219 |
+
"learning_rate": 6.53960398946249e-05,
|
| 6220 |
+
"loss": 1.2511,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.07467566712372113,
|
| 6225 |
+
"grad_norm": 2.459752082824707,
|
| 6226 |
+
"learning_rate": 6.518957137846538e-05,
|
| 6227 |
+
"loss": 1.1427,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.07476004640860669,
|
| 6232 |
+
"grad_norm": 3.9591658115386963,
|
| 6233 |
+
"learning_rate": 6.498327158014108e-05,
|
| 6234 |
+
"loss": 1.984,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.07484442569349224,
|
| 6239 |
+
"grad_norm": 3.097809076309204,
|
| 6240 |
+
"learning_rate": 6.477714149953795e-05,
|
| 6241 |
+
"loss": 1.2613,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.07492880497837781,
|
| 6246 |
+
"grad_norm": 2.4523212909698486,
|
| 6247 |
+
"learning_rate": 6.457118213571922e-05,
|
| 6248 |
+
"loss": 1.2974,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.07501318426326337,
|
| 6253 |
+
"grad_norm": 3.101086378097534,
|
| 6254 |
+
"learning_rate": 6.436539448692073e-05,
|
| 6255 |
+
"loss": 1.6382,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.07509756354814894,
|
| 6260 |
+
"grad_norm": 3.1861653327941895,
|
| 6261 |
+
"learning_rate": 6.415977955054611e-05,
|
| 6262 |
+
"loss": 1.5635,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.07518194283303449,
|
| 6267 |
+
"grad_norm": 3.8832900524139404,
|
| 6268 |
+
"learning_rate": 6.395433832316181e-05,
|
| 6269 |
+
"loss": 1.5918,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.07526632211792005,
|
| 6274 |
+
"grad_norm": 3.0145103931427,
|
| 6275 |
+
"learning_rate": 6.374907180049246e-05,
|
| 6276 |
+
"loss": 1.1065,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.07535070140280561,
|
| 6281 |
+
"grad_norm": 3.072340726852417,
|
| 6282 |
+
"learning_rate": 6.354398097741585e-05,
|
| 6283 |
+
"loss": 1.5317,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.07543508068769117,
|
| 6288 |
+
"grad_norm": 4.860788822174072,
|
| 6289 |
+
"learning_rate": 6.333906684795821e-05,
|
| 6290 |
+
"loss": 1.5721,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.07551945997257674,
|
| 6295 |
+
"grad_norm": 3.3973536491394043,
|
| 6296 |
+
"learning_rate": 6.313433040528947e-05,
|
| 6297 |
+
"loss": 1.7848,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.0756038392574623,
|
| 6302 |
+
"grad_norm": 3.1489346027374268,
|
| 6303 |
+
"learning_rate": 6.292977264171819e-05,
|
| 6304 |
+
"loss": 1.6759,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.07568821854234785,
|
| 6309 |
+
"grad_norm": 3.1691362857818604,
|
| 6310 |
+
"learning_rate": 6.272539454868702e-05,
|
| 6311 |
+
"loss": 1.5443,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.07577259782723342,
|
| 6316 |
+
"grad_norm": 3.704650640487671,
|
| 6317 |
+
"learning_rate": 6.252119711676782e-05,
|
| 6318 |
+
"loss": 1.9884,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.07585697711211897,
|
| 6323 |
+
"grad_norm": 2.927281141281128,
|
| 6324 |
+
"learning_rate": 6.231718133565674e-05,
|
| 6325 |
+
"loss": 1.3909,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.07594135639700454,
|
| 6330 |
+
"grad_norm": 3.6570024490356445,
|
| 6331 |
+
"learning_rate": 6.21133481941696e-05,
|
| 6332 |
+
"loss": 1.5397,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.0760257356818901,
|
| 6337 |
+
"grad_norm": 3.40569806098938,
|
| 6338 |
+
"learning_rate": 6.190969868023691e-05,
|
| 6339 |
+
"loss": 1.5339,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.07611011496677565,
|
| 6344 |
+
"grad_norm": 2.9963929653167725,
|
| 6345 |
+
"learning_rate": 6.17062337808993e-05,
|
| 6346 |
+
"loss": 1.3681,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.07619449425166122,
|
| 6351 |
+
"grad_norm": 3.280752420425415,
|
| 6352 |
+
"learning_rate": 6.150295448230248e-05,
|
| 6353 |
+
"loss": 1.6603,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.07627887353654678,
|
| 6358 |
+
"grad_norm": 3.2442002296447754,
|
| 6359 |
+
"learning_rate": 6.129986176969269e-05,
|
| 6360 |
+
"loss": 1.7117,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.07636325282143235,
|
| 6365 |
+
"grad_norm": 3.50140380859375,
|
| 6366 |
+
"learning_rate": 6.109695662741183e-05,
|
| 6367 |
+
"loss": 1.8245,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.0764476321063179,
|
| 6372 |
+
"grad_norm": 3.7245430946350098,
|
| 6373 |
+
"learning_rate": 6.089424003889268e-05,
|
| 6374 |
+
"loss": 1.474,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.07653201139120346,
|
| 6379 |
+
"grad_norm": 4.147988796234131,
|
| 6380 |
+
"learning_rate": 6.0691712986654124e-05,
|
| 6381 |
+
"loss": 1.7769,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.07661639067608902,
|
| 6386 |
+
"grad_norm": 3.388179302215576,
|
| 6387 |
+
"learning_rate": 6.048937645229644e-05,
|
| 6388 |
+
"loss": 1.5102,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.07670076996097458,
|
| 6393 |
+
"grad_norm": 3.4448678493499756,
|
| 6394 |
+
"learning_rate": 6.028723141649646e-05,
|
| 6395 |
+
"loss": 1.8252,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.07678514924586013,
|
| 6400 |
+
"grad_norm": 3.0707340240478516,
|
| 6401 |
+
"learning_rate": 6.008527885900296e-05,
|
| 6402 |
+
"loss": 1.5494,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.0768695285307457,
|
| 6407 |
+
"grad_norm": 3.1965088844299316,
|
| 6408 |
+
"learning_rate": 5.988351975863171e-05,
|
| 6409 |
+
"loss": 1.3532,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.07695390781563126,
|
| 6414 |
+
"grad_norm": 2.9536924362182617,
|
| 6415 |
+
"learning_rate": 5.968195509326089e-05,
|
| 6416 |
+
"loss": 1.5506,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.07703828710051683,
|
| 6421 |
+
"grad_norm": 2.2717971801757812,
|
| 6422 |
+
"learning_rate": 5.9480585839826343e-05,
|
| 6423 |
+
"loss": 1.0923,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.07712266638540238,
|
| 6428 |
+
"grad_norm": 3.312793254852295,
|
| 6429 |
+
"learning_rate": 5.927941297431672e-05,
|
| 6430 |
+
"loss": 1.8926,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.07720704567028794,
|
| 6435 |
+
"grad_norm": 3.4543073177337646,
|
| 6436 |
+
"learning_rate": 5.9078437471768865e-05,
|
| 6437 |
+
"loss": 1.3719,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.07729142495517351,
|
| 6442 |
+
"grad_norm": 3.4697766304016113,
|
| 6443 |
+
"learning_rate": 5.887766030626308e-05,
|
| 6444 |
+
"loss": 1.97,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.07737580424005906,
|
| 6449 |
+
"grad_norm": 3.1399221420288086,
|
| 6450 |
+
"learning_rate": 5.8677082450918354e-05,
|
| 6451 |
+
"loss": 1.3203,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.07746018352494463,
|
| 6456 |
+
"grad_norm": 4.258705139160156,
|
| 6457 |
+
"learning_rate": 5.847670487788758e-05,
|
| 6458 |
+
"loss": 1.798,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.07754456280983019,
|
| 6463 |
+
"grad_norm": 4.258029460906982,
|
| 6464 |
+
"learning_rate": 5.827652855835308e-05,
|
| 6465 |
+
"loss": 1.7186,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.07762894209471574,
|
| 6470 |
+
"grad_norm": 3.9962995052337646,
|
| 6471 |
+
"learning_rate": 5.807655446252163e-05,
|
| 6472 |
+
"loss": 1.5903,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.07771332137960131,
|
| 6477 |
+
"grad_norm": 4.5598626136779785,
|
| 6478 |
+
"learning_rate": 5.787678355961993e-05,
|
| 6479 |
+
"loss": 2.0344,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.07779770066448687,
|
| 6484 |
+
"grad_norm": 3.61279559135437,
|
| 6485 |
+
"learning_rate": 5.76772168178899e-05,
|
| 6486 |
+
"loss": 2.022,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.07788207994937243,
|
| 6491 |
+
"grad_norm": 3.357609987258911,
|
| 6492 |
+
"learning_rate": 5.74778552045838e-05,
|
| 6493 |
+
"loss": 1.089,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.07796645923425799,
|
| 6498 |
+
"grad_norm": 3.9268362522125244,
|
| 6499 |
+
"learning_rate": 5.7278699685959835e-05,
|
| 6500 |
+
"loss": 1.9794,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.07805083851914354,
|
| 6505 |
+
"grad_norm": 3.3527932167053223,
|
| 6506 |
+
"learning_rate": 5.7079751227277176e-05,
|
| 6507 |
+
"loss": 1.5096,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.07813521780402911,
|
| 6512 |
+
"grad_norm": 3.296321392059326,
|
| 6513 |
+
"learning_rate": 5.6881010792791546e-05,
|
| 6514 |
+
"loss": 1.3736,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.07821959708891467,
|
| 6519 |
+
"grad_norm": 3.662139892578125,
|
| 6520 |
+
"learning_rate": 5.668247934575035e-05,
|
| 6521 |
+
"loss": 1.6151,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.07830397637380024,
|
| 6526 |
+
"grad_norm": 3.816279888153076,
|
| 6527 |
+
"learning_rate": 5.6484157848388165e-05,
|
| 6528 |
+
"loss": 1.7165,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.0783883556586858,
|
| 6533 |
+
"grad_norm": 3.0304577350616455,
|
| 6534 |
+
"learning_rate": 5.628604726192186e-05,
|
| 6535 |
+
"loss": 1.4309,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.07847273494357135,
|
| 6540 |
+
"grad_norm": 3.2069170475006104,
|
| 6541 |
+
"learning_rate": 5.6088148546546146e-05,
|
| 6542 |
+
"loss": 1.283,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.07855711422845692,
|
| 6547 |
+
"grad_norm": 2.918621063232422,
|
| 6548 |
+
"learning_rate": 5.589046266142891e-05,
|
| 6549 |
+
"loss": 1.4255,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.07864149351334247,
|
| 6554 |
+
"grad_norm": 3.786069869995117,
|
| 6555 |
+
"learning_rate": 5.569299056470635e-05,
|
| 6556 |
+
"loss": 1.5281,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.07872587279822804,
|
| 6561 |
+
"grad_norm": 3.4659860134124756,
|
| 6562 |
+
"learning_rate": 5.549573321347863e-05,
|
| 6563 |
+
"loss": 1.7832,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.0788102520831136,
|
| 6568 |
+
"grad_norm": 3.6599302291870117,
|
| 6569 |
+
"learning_rate": 5.529869156380496e-05,
|
| 6570 |
+
"loss": 1.7159,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.07889463136799915,
|
| 6575 |
+
"grad_norm": 3.3843908309936523,
|
| 6576 |
+
"learning_rate": 5.510186657069921e-05,
|
| 6577 |
+
"loss": 1.226,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.07897901065288472,
|
| 6582 |
+
"grad_norm": 3.5379812717437744,
|
| 6583 |
+
"learning_rate": 5.490525918812512e-05,
|
| 6584 |
+
"loss": 1.7822,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.07906338993777028,
|
| 6589 |
+
"grad_norm": 3.178551197052002,
|
| 6590 |
+
"learning_rate": 5.4708870368991715e-05,
|
| 6591 |
+
"loss": 1.6024,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.07914776922265584,
|
| 6596 |
+
"grad_norm": 3.1955838203430176,
|
| 6597 |
+
"learning_rate": 5.451270106514875e-05,
|
| 6598 |
+
"loss": 2.0264,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.0792321485075414,
|
| 6603 |
+
"grad_norm": 3.555262565612793,
|
| 6604 |
+
"learning_rate": 5.431675222738197e-05,
|
| 6605 |
+
"loss": 1.5953,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.07931652779242696,
|
| 6610 |
+
"grad_norm": 3.2316176891326904,
|
| 6611 |
+
"learning_rate": 5.412102480540855e-05,
|
| 6612 |
+
"loss": 1.715,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.07940090707731252,
|
| 6617 |
+
"grad_norm": 3.2542645931243896,
|
| 6618 |
+
"learning_rate": 5.392551974787261e-05,
|
| 6619 |
+
"loss": 1.701,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.07948528636219808,
|
| 6624 |
+
"grad_norm": 2.499079942703247,
|
| 6625 |
+
"learning_rate": 5.373023800234048e-05,
|
| 6626 |
+
"loss": 0.9561,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.07956966564708363,
|
| 6631 |
+
"grad_norm": 3.4604427814483643,
|
| 6632 |
+
"learning_rate": 5.353518051529619e-05,
|
| 6633 |
+
"loss": 1.6331,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.0796540449319692,
|
| 6638 |
+
"grad_norm": 2.747849464416504,
|
| 6639 |
+
"learning_rate": 5.334034823213673e-05,
|
| 6640 |
+
"loss": 1.4567,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.07973842421685476,
|
| 6645 |
+
"grad_norm": 3.287957191467285,
|
| 6646 |
+
"learning_rate": 5.314574209716767e-05,
|
| 6647 |
+
"loss": 1.5251,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.07982280350174033,
|
| 6652 |
+
"grad_norm": 2.9816641807556152,
|
| 6653 |
+
"learning_rate": 5.295136305359856e-05,
|
| 6654 |
+
"loss": 1.5381,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.07990718278662588,
|
| 6659 |
+
"grad_norm": 3.083892583847046,
|
| 6660 |
+
"learning_rate": 5.275721204353807e-05,
|
| 6661 |
+
"loss": 1.9563,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.07999156207151144,
|
| 6666 |
+
"grad_norm": 3.9584100246429443,
|
| 6667 |
+
"learning_rate": 5.256329000798988e-05,
|
| 6668 |
+
"loss": 1.6158,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.080075941356397,
|
| 6673 |
+
"grad_norm": 3.2833170890808105,
|
| 6674 |
+
"learning_rate": 5.23695978868478e-05,
|
| 6675 |
+
"loss": 1.5464,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.08016032064128256,
|
| 6680 |
+
"grad_norm": 3.2715957164764404,
|
| 6681 |
+
"learning_rate": 5.2176136618891225e-05,
|
| 6682 |
+
"loss": 1.6228,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.08024469992616813,
|
| 6687 |
+
"grad_norm": 4.0370049476623535,
|
| 6688 |
+
"learning_rate": 5.1982907141780744e-05,
|
| 6689 |
+
"loss": 1.7749,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.08032907921105369,
|
| 6694 |
+
"grad_norm": 3.634587049484253,
|
| 6695 |
+
"learning_rate": 5.178991039205352e-05,
|
| 6696 |
+
"loss": 1.9981,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.08041345849593924,
|
| 6701 |
+
"grad_norm": 2.6632728576660156,
|
| 6702 |
+
"learning_rate": 5.159714730511873e-05,
|
| 6703 |
+
"loss": 1.3034,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.08049783778082481,
|
| 6708 |
+
"grad_norm": 3.3665788173675537,
|
| 6709 |
+
"learning_rate": 5.140461881525299e-05,
|
| 6710 |
+
"loss": 1.716,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.08058221706571037,
|
| 6715 |
+
"grad_norm": 3.839794874191284,
|
| 6716 |
+
"learning_rate": 5.121232585559589e-05,
|
| 6717 |
+
"loss": 1.8759,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.08066659635059593,
|
| 6722 |
+
"grad_norm": 3.8704581260681152,
|
| 6723 |
+
"learning_rate": 5.1020269358145525e-05,
|
| 6724 |
+
"loss": 1.9303,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.08075097563548149,
|
| 6729 |
+
"grad_norm": 3.7815945148468018,
|
| 6730 |
+
"learning_rate": 5.0828450253753866e-05,
|
| 6731 |
+
"loss": 1.5441,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.08083535492036704,
|
| 6736 |
+
"grad_norm": 3.394934892654419,
|
| 6737 |
+
"learning_rate": 5.0636869472122326e-05,
|
| 6738 |
+
"loss": 1.9486,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.08091973420525261,
|
| 6743 |
+
"grad_norm": 4.189752578735352,
|
| 6744 |
+
"learning_rate": 5.0445527941797236e-05,
|
| 6745 |
+
"loss": 1.4785,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.08100411349013817,
|
| 6750 |
+
"grad_norm": 3.6348304748535156,
|
| 6751 |
+
"learning_rate": 5.0254426590165214e-05,
|
| 6752 |
+
"loss": 1.811,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.08108849277502374,
|
| 6757 |
+
"grad_norm": 3.7563483715057373,
|
| 6758 |
+
"learning_rate": 5.006356634344898e-05,
|
| 6759 |
+
"loss": 1.6477,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.08117287205990929,
|
| 6764 |
+
"grad_norm": 2.8033125400543213,
|
| 6765 |
+
"learning_rate": 4.987294812670249e-05,
|
| 6766 |
+
"loss": 1.1668,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.08125725134479485,
|
| 6771 |
+
"grad_norm": 3.1571547985076904,
|
| 6772 |
+
"learning_rate": 4.9682572863806755e-05,
|
| 6773 |
+
"loss": 1.6883,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.08134163062968042,
|
| 6778 |
+
"grad_norm": 3.429401159286499,
|
| 6779 |
+
"learning_rate": 4.949244147746525e-05,
|
| 6780 |
+
"loss": 1.5639,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.08142600991456597,
|
| 6785 |
+
"grad_norm": 3.838519811630249,
|
| 6786 |
+
"learning_rate": 4.930255488919934e-05,
|
| 6787 |
+
"loss": 1.6605,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.08151038919945154,
|
| 6792 |
+
"grad_norm": 3.1025607585906982,
|
| 6793 |
+
"learning_rate": 4.9112914019344003e-05,
|
| 6794 |
+
"loss": 1.8484,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.0815947684843371,
|
| 6799 |
+
"grad_norm": 2.9810519218444824,
|
| 6800 |
+
"learning_rate": 4.892351978704326e-05,
|
| 6801 |
+
"loss": 1.3605,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.08167914776922265,
|
| 6806 |
+
"grad_norm": 3.149088144302368,
|
| 6807 |
+
"learning_rate": 4.873437311024577e-05,
|
| 6808 |
+
"loss": 1.5338,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.08176352705410822,
|
| 6813 |
+
"grad_norm": 3.399979829788208,
|
| 6814 |
+
"learning_rate": 4.854547490570022e-05,
|
| 6815 |
+
"loss": 1.3578,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.08184790633899378,
|
| 6820 |
+
"grad_norm": 2.430124282836914,
|
| 6821 |
+
"learning_rate": 4.835682608895118e-05,
|
| 6822 |
+
"loss": 1.1392,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.08193228562387934,
|
| 6827 |
+
"grad_norm": 3.357212781906128,
|
| 6828 |
+
"learning_rate": 4.816842757433436e-05,
|
| 6829 |
+
"loss": 1.248,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.0820166649087649,
|
| 6834 |
+
"grad_norm": 2.704453945159912,
|
| 6835 |
+
"learning_rate": 4.79802802749724e-05,
|
| 6836 |
+
"loss": 1.3,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.08210104419365045,
|
| 6841 |
+
"grad_norm": 3.4953999519348145,
|
| 6842 |
+
"learning_rate": 4.7792385102770324e-05,
|
| 6843 |
+
"loss": 1.3193,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.08218542347853602,
|
| 6848 |
+
"grad_norm": 3.4437851905822754,
|
| 6849 |
+
"learning_rate": 4.760474296841121e-05,
|
| 6850 |
+
"loss": 1.607,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.08226980276342158,
|
| 6855 |
+
"grad_norm": 3.2445197105407715,
|
| 6856 |
+
"learning_rate": 4.741735478135164e-05,
|
| 6857 |
+
"loss": 2.194,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.08235418204830713,
|
| 6862 |
+
"grad_norm": 2.820953369140625,
|
| 6863 |
+
"learning_rate": 4.723022144981738e-05,
|
| 6864 |
+
"loss": 1.2643,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.0824385613331927,
|
| 6869 |
+
"grad_norm": 3.4670326709747314,
|
| 6870 |
+
"learning_rate": 4.704334388079903e-05,
|
| 6871 |
+
"loss": 1.4809,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.08252294061807826,
|
| 6876 |
+
"grad_norm": 3.0149405002593994,
|
| 6877 |
+
"learning_rate": 4.685672298004754e-05,
|
| 6878 |
+
"loss": 1.3301,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.08260731990296383,
|
| 6883 |
+
"grad_norm": 2.8329551219940186,
|
| 6884 |
+
"learning_rate": 4.6670359652069863e-05,
|
| 6885 |
+
"loss": 1.6225,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.08269169918784938,
|
| 6890 |
+
"grad_norm": 2.9316213130950928,
|
| 6891 |
+
"learning_rate": 4.648425480012458e-05,
|
| 6892 |
+
"loss": 1.4515,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.08277607847273494,
|
| 6897 |
+
"grad_norm": 4.188765525817871,
|
| 6898 |
+
"learning_rate": 4.629840932621741e-05,
|
| 6899 |
+
"loss": 1.7886,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.0828604577576205,
|
| 6904 |
+
"grad_norm": 3.880164861679077,
|
| 6905 |
+
"learning_rate": 4.611282413109704e-05,
|
| 6906 |
+
"loss": 1.7346,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.08294483704250606,
|
| 6911 |
+
"grad_norm": 3.7946267127990723,
|
| 6912 |
+
"learning_rate": 4.592750011425053e-05,
|
| 6913 |
+
"loss": 2.0565,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.08302921632739163,
|
| 6918 |
+
"grad_norm": 3.2721166610717773,
|
| 6919 |
+
"learning_rate": 4.574243817389917e-05,
|
| 6920 |
+
"loss": 1.419,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.08311359561227719,
|
| 6925 |
+
"grad_norm": 3.628910779953003,
|
| 6926 |
+
"learning_rate": 4.555763920699404e-05,
|
| 6927 |
+
"loss": 1.6117,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.08319797489716274,
|
| 6932 |
+
"grad_norm": 4.114333152770996,
|
| 6933 |
+
"learning_rate": 4.537310410921153e-05,
|
| 6934 |
+
"loss": 1.8862,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.08328235418204831,
|
| 6939 |
+
"grad_norm": 3.2277815341949463,
|
| 6940 |
+
"learning_rate": 4.5188833774949224e-05,
|
| 6941 |
+
"loss": 2.1243,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.08336673346693386,
|
| 6946 |
+
"grad_norm": 3.812563419342041,
|
| 6947 |
+
"learning_rate": 4.500482909732144e-05,
|
| 6948 |
+
"loss": 1.8358,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.08345111275181943,
|
| 6953 |
+
"grad_norm": 3.6660330295562744,
|
| 6954 |
+
"learning_rate": 4.482109096815491e-05,
|
| 6955 |
+
"loss": 1.5744,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.08353549203670499,
|
| 6960 |
+
"grad_norm": 3.3940811157226562,
|
| 6961 |
+
"learning_rate": 4.4637620277984506e-05,
|
| 6962 |
+
"loss": 1.8974,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.08361987132159054,
|
| 6967 |
+
"grad_norm": 3.1708710193634033,
|
| 6968 |
+
"learning_rate": 4.4454417916048804e-05,
|
| 6969 |
+
"loss": 1.565,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.08370425060647611,
|
| 6974 |
+
"grad_norm": 3.2875232696533203,
|
| 6975 |
+
"learning_rate": 4.427148477028589e-05,
|
| 6976 |
+
"loss": 1.4847,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.08378862989136167,
|
| 6981 |
+
"grad_norm": 2.535433292388916,
|
| 6982 |
+
"learning_rate": 4.4088821727329067e-05,
|
| 6983 |
+
"loss": 1.3308,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.08387300917624724,
|
| 6988 |
+
"grad_norm": 3.839012622833252,
|
| 6989 |
+
"learning_rate": 4.390642967250249e-05,
|
| 6990 |
+
"loss": 1.6541,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.08395738846113279,
|
| 6995 |
+
"grad_norm": 3.5515658855438232,
|
| 6996 |
+
"learning_rate": 4.372430948981693e-05,
|
| 6997 |
+
"loss": 1.8802,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.08404176774601835,
|
| 7002 |
+
"grad_norm": 4.180967807769775,
|
| 7003 |
+
"learning_rate": 4.354246206196535e-05,
|
| 7004 |
+
"loss": 1.7852,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.08412614703090392,
|
| 7009 |
+
"grad_norm": 3.247296094894409,
|
| 7010 |
+
"learning_rate": 4.336088827031889e-05,
|
| 7011 |
+
"loss": 1.7732,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.08421052631578947,
|
| 7016 |
+
"grad_norm": 3.3442742824554443,
|
| 7017 |
+
"learning_rate": 4.3179588994922274e-05,
|
| 7018 |
+
"loss": 1.3819,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.08429490560067504,
|
| 7023 |
+
"grad_norm": 3.765320062637329,
|
| 7024 |
+
"learning_rate": 4.2998565114489855e-05,
|
| 7025 |
+
"loss": 1.5747,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.0843792848855606,
|
| 7030 |
+
"grad_norm": 3.264878273010254,
|
| 7031 |
+
"learning_rate": 4.281781750640112e-05,
|
| 7032 |
+
"loss": 1.5075,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.08446366417044615,
|
| 7037 |
+
"grad_norm": 3.3386402130126953,
|
| 7038 |
+
"learning_rate": 4.263734704669665e-05,
|
| 7039 |
+
"loss": 1.5605,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.08454804345533172,
|
| 7044 |
+
"grad_norm": 2.855348587036133,
|
| 7045 |
+
"learning_rate": 4.245715461007357e-05,
|
| 7046 |
+
"loss": 1.2566,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.08463242274021728,
|
| 7051 |
+
"grad_norm": 3.1625888347625732,
|
| 7052 |
+
"learning_rate": 4.2277241069881615e-05,
|
| 7053 |
+
"loss": 1.3671,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.08471680202510284,
|
| 7058 |
+
"grad_norm": 3.769261360168457,
|
| 7059 |
+
"learning_rate": 4.209760729811881e-05,
|
| 7060 |
+
"loss": 1.9523,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.0848011813099884,
|
| 7065 |
+
"grad_norm": 2.738025426864624,
|
| 7066 |
+
"learning_rate": 4.191825416542708e-05,
|
| 7067 |
+
"loss": 1.2554,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.08488556059487395,
|
| 7072 |
+
"grad_norm": 4.013035297393799,
|
| 7073 |
+
"learning_rate": 4.1739182541088305e-05,
|
| 7074 |
+
"loss": 1.7184,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.08496993987975952,
|
| 7079 |
+
"grad_norm": 3.0900278091430664,
|
| 7080 |
+
"learning_rate": 4.15603932930198e-05,
|
| 7081 |
+
"loss": 1.7968,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.08505431916464508,
|
| 7086 |
+
"grad_norm": 3.4534189701080322,
|
| 7087 |
+
"learning_rate": 4.138188728777041e-05,
|
| 7088 |
+
"loss": 1.9061,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.08513869844953063,
|
| 7093 |
+
"grad_norm": 2.792163848876953,
|
| 7094 |
+
"learning_rate": 4.120366539051612e-05,
|
| 7095 |
+
"loss": 1.0116,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.0852230777344162,
|
| 7100 |
+
"grad_norm": 2.787792205810547,
|
| 7101 |
+
"learning_rate": 4.102572846505589e-05,
|
| 7102 |
+
"loss": 1.7397,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.08530745701930176,
|
| 7107 |
+
"grad_norm": 3.7990641593933105,
|
| 7108 |
+
"learning_rate": 4.0848077373807556e-05,
|
| 7109 |
+
"loss": 1.6142,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.08539183630418733,
|
| 7114 |
+
"grad_norm": 3.2461421489715576,
|
| 7115 |
+
"learning_rate": 4.067071297780351e-05,
|
| 7116 |
+
"loss": 1.5828,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.08547621558907288,
|
| 7121 |
+
"grad_norm": 3.261294364929199,
|
| 7122 |
+
"learning_rate": 4.0493636136686574e-05,
|
| 7123 |
+
"loss": 1.4872,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.08556059487395844,
|
| 7128 |
+
"grad_norm": 3.000673294067383,
|
| 7129 |
+
"learning_rate": 4.0316847708705964e-05,
|
| 7130 |
+
"loss": 1.482,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.085644974158844,
|
| 7135 |
+
"grad_norm": 3.1343703269958496,
|
| 7136 |
+
"learning_rate": 4.0140348550712955e-05,
|
| 7137 |
+
"loss": 1.2417,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.08572935344372956,
|
| 7142 |
+
"grad_norm": 4.0371880531311035,
|
| 7143 |
+
"learning_rate": 3.996413951815685e-05,
|
| 7144 |
+
"loss": 1.6358,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.08581373272861513,
|
| 7149 |
+
"grad_norm": 3.639571189880371,
|
| 7150 |
+
"learning_rate": 3.9788221465080764e-05,
|
| 7151 |
+
"loss": 1.6611,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.08589811201350069,
|
| 7156 |
+
"grad_norm": 3.0644125938415527,
|
| 7157 |
+
"learning_rate": 3.9612595244117444e-05,
|
| 7158 |
+
"loss": 1.1822,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.08598249129838624,
|
| 7163 |
+
"grad_norm": 3.3109662532806396,
|
| 7164 |
+
"learning_rate": 3.9437261706485295e-05,
|
| 7165 |
+
"loss": 1.6849,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.08606687058327181,
|
| 7170 |
+
"grad_norm": 2.721682548522949,
|
| 7171 |
+
"learning_rate": 3.926222170198404e-05,
|
| 7172 |
+
"loss": 1.2948,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.08615124986815736,
|
| 7177 |
+
"grad_norm": 3.4836831092834473,
|
| 7178 |
+
"learning_rate": 3.9087476078990826e-05,
|
| 7179 |
+
"loss": 1.7757,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.08623562915304293,
|
| 7184 |
+
"grad_norm": 3.04488205909729,
|
| 7185 |
+
"learning_rate": 3.8913025684456005e-05,
|
| 7186 |
+
"loss": 1.7447,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.08632000843792849,
|
| 7191 |
+
"grad_norm": 2.8262956142425537,
|
| 7192 |
+
"learning_rate": 3.873887136389889e-05,
|
| 7193 |
+
"loss": 1.2663,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.08640438772281404,
|
| 7198 |
+
"grad_norm": 3.794739246368408,
|
| 7199 |
+
"learning_rate": 3.856501396140394e-05,
|
| 7200 |
+
"loss": 1.9884,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.08648876700769961,
|
| 7205 |
+
"grad_norm": 3.342715263366699,
|
| 7206 |
+
"learning_rate": 3.839145431961646e-05,
|
| 7207 |
+
"loss": 1.9778,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.08657314629258517,
|
| 7212 |
+
"grad_norm": 2.6708970069885254,
|
| 7213 |
+
"learning_rate": 3.821819327973862e-05,
|
| 7214 |
+
"loss": 1.4457,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.08665752557747074,
|
| 7219 |
+
"grad_norm": 4.427373886108398,
|
| 7220 |
+
"learning_rate": 3.804523168152526e-05,
|
| 7221 |
+
"loss": 1.85,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.08674190486235629,
|
| 7226 |
+
"grad_norm": 3.5284879207611084,
|
| 7227 |
+
"learning_rate": 3.787257036327999e-05,
|
| 7228 |
+
"loss": 1.4974,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.08682628414724185,
|
| 7233 |
+
"grad_norm": 3.2620925903320312,
|
| 7234 |
+
"learning_rate": 3.770021016185094e-05,
|
| 7235 |
+
"loss": 1.5186,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.08691066343212742,
|
| 7240 |
+
"grad_norm": 3.066788911819458,
|
| 7241 |
+
"learning_rate": 3.7528151912626884e-05,
|
| 7242 |
+
"loss": 1.2812,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.08699504271701297,
|
| 7247 |
+
"grad_norm": 3.0070440769195557,
|
| 7248 |
+
"learning_rate": 3.7356396449533046e-05,
|
| 7249 |
+
"loss": 1.5595,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.08707942200189854,
|
| 7254 |
+
"grad_norm": 3.7550156116485596,
|
| 7255 |
+
"learning_rate": 3.718494460502721e-05,
|
| 7256 |
+
"loss": 1.6685,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.0871638012867841,
|
| 7261 |
+
"grad_norm": 2.7150115966796875,
|
| 7262 |
+
"learning_rate": 3.701379721009547e-05,
|
| 7263 |
+
"loss": 1.4513,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.08724818057166965,
|
| 7268 |
+
"grad_norm": 2.8110036849975586,
|
| 7269 |
+
"learning_rate": 3.6842955094248356e-05,
|
| 7270 |
+
"loss": 1.4277,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.08733255985655522,
|
| 7275 |
+
"grad_norm": 2.648404359817505,
|
| 7276 |
+
"learning_rate": 3.667241908551683e-05,
|
| 7277 |
+
"loss": 1.2026,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.08741693914144077,
|
| 7282 |
+
"grad_norm": 3.5289621353149414,
|
| 7283 |
+
"learning_rate": 3.650219001044822e-05,
|
| 7284 |
+
"loss": 1.9194,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.08750131842632633,
|
| 7289 |
+
"grad_norm": 3.684596300125122,
|
| 7290 |
+
"learning_rate": 3.63322686941022e-05,
|
| 7291 |
+
"loss": 1.5032,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.0875856977112119,
|
| 7296 |
+
"grad_norm": 3.558779716491699,
|
| 7297 |
+
"learning_rate": 3.616265596004686e-05,
|
| 7298 |
+
"loss": 1.6558,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.08767007699609745,
|
| 7303 |
+
"grad_norm": 2.632128953933716,
|
| 7304 |
+
"learning_rate": 3.599335263035453e-05,
|
| 7305 |
+
"loss": 1.3601,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.08775445628098302,
|
| 7310 |
+
"grad_norm": 3.0218188762664795,
|
| 7311 |
+
"learning_rate": 3.582435952559806e-05,
|
| 7312 |
+
"loss": 1.6693,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.08783883556586858,
|
| 7317 |
+
"grad_norm": 3.2471461296081543,
|
| 7318 |
+
"learning_rate": 3.5655677464846716e-05,
|
| 7319 |
+
"loss": 1.5243,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.08792321485075413,
|
| 7324 |
+
"grad_norm": 3.517889976501465,
|
| 7325 |
+
"learning_rate": 3.5487307265662106e-05,
|
| 7326 |
+
"loss": 1.7958,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.0880075941356397,
|
| 7331 |
+
"grad_norm": 4.443117618560791,
|
| 7332 |
+
"learning_rate": 3.531924974409445e-05,
|
| 7333 |
+
"loss": 1.5469,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.08809197342052526,
|
| 7338 |
+
"grad_norm": 2.9980645179748535,
|
| 7339 |
+
"learning_rate": 3.5151505714678334e-05,
|
| 7340 |
+
"loss": 1.7784,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.08817635270541083,
|
| 7345 |
+
"grad_norm": 3.082932472229004,
|
| 7346 |
+
"learning_rate": 3.498407599042908e-05,
|
| 7347 |
+
"loss": 1.2205,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.08826073199029638,
|
| 7352 |
+
"grad_norm": 3.174506425857544,
|
| 7353 |
+
"learning_rate": 3.481696138283855e-05,
|
| 7354 |
+
"loss": 1.3015,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.08834511127518194,
|
| 7359 |
+
"grad_norm": 2.673461437225342,
|
| 7360 |
+
"learning_rate": 3.4650162701871367e-05,
|
| 7361 |
+
"loss": 1.4825,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.0884294905600675,
|
| 7366 |
+
"grad_norm": 3.3988704681396484,
|
| 7367 |
+
"learning_rate": 3.448368075596092e-05,
|
| 7368 |
+
"loss": 1.8758,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.08851386984495306,
|
| 7373 |
+
"grad_norm": 3.7007596492767334,
|
| 7374 |
+
"learning_rate": 3.431751635200544e-05,
|
| 7375 |
+
"loss": 1.5928,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.08859824912983863,
|
| 7380 |
+
"grad_norm": 3.5111403465270996,
|
| 7381 |
+
"learning_rate": 3.415167029536406e-05,
|
| 7382 |
+
"loss": 1.918,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.08868262841472419,
|
| 7387 |
+
"grad_norm": 3.1500866413116455,
|
| 7388 |
+
"learning_rate": 3.3986143389853054e-05,
|
| 7389 |
+
"loss": 1.4275,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.08876700769960974,
|
| 7394 |
+
"grad_norm": 2.674935817718506,
|
| 7395 |
+
"learning_rate": 3.3820936437741814e-05,
|
| 7396 |
+
"loss": 1.332,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.08885138698449531,
|
| 7401 |
+
"grad_norm": 3.5701823234558105,
|
| 7402 |
+
"learning_rate": 3.3656050239749004e-05,
|
| 7403 |
+
"loss": 1.9382,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.08893576626938086,
|
| 7408 |
+
"grad_norm": 3.080343008041382,
|
| 7409 |
+
"learning_rate": 3.349148559503861e-05,
|
| 7410 |
+
"loss": 1.2917,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.08902014555426643,
|
| 7415 |
+
"grad_norm": 3.38442063331604,
|
| 7416 |
+
"learning_rate": 3.332724330121622e-05,
|
| 7417 |
+
"loss": 1.7187,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.08910452483915199,
|
| 7422 |
+
"grad_norm": 3.537364959716797,
|
| 7423 |
+
"learning_rate": 3.316332415432495e-05,
|
| 7424 |
+
"loss": 1.6425,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.08918890412403754,
|
| 7429 |
+
"grad_norm": 3.537048816680908,
|
| 7430 |
+
"learning_rate": 3.29997289488418e-05,
|
| 7431 |
+
"loss": 1.7704,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.08927328340892311,
|
| 7436 |
+
"grad_norm": 3.23457407951355,
|
| 7437 |
+
"learning_rate": 3.283645847767366e-05,
|
| 7438 |
+
"loss": 1.4781,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.08935766269380867,
|
| 7443 |
+
"grad_norm": 3.4182138442993164,
|
| 7444 |
+
"learning_rate": 3.267351353215355e-05,
|
| 7445 |
+
"loss": 1.5074,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.08944204197869424,
|
| 7450 |
+
"grad_norm": 3.326955556869507,
|
| 7451 |
+
"learning_rate": 3.251089490203666e-05,
|
| 7452 |
+
"loss": 1.6414,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.08952642126357979,
|
| 7457 |
+
"grad_norm": 3.422151803970337,
|
| 7458 |
+
"learning_rate": 3.234860337549668e-05,
|
| 7459 |
+
"loss": 1.4056,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.08961080054846535,
|
| 7464 |
+
"grad_norm": 3.5641722679138184,
|
| 7465 |
+
"learning_rate": 3.218663973912187e-05,
|
| 7466 |
+
"loss": 1.7189,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.08969517983335092,
|
| 7471 |
+
"grad_norm": 3.1099095344543457,
|
| 7472 |
+
"learning_rate": 3.202500477791137e-05,
|
| 7473 |
+
"loss": 1.7291,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.08977955911823647,
|
| 7478 |
+
"grad_norm": 2.243972063064575,
|
| 7479 |
+
"learning_rate": 3.186369927527119e-05,
|
| 7480 |
+
"loss": 1.101,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.08986393840312204,
|
| 7485 |
+
"grad_norm": 3.1082541942596436,
|
| 7486 |
+
"learning_rate": 3.170272401301053e-05,
|
| 7487 |
+
"loss": 1.4632,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.0899483176880076,
|
| 7492 |
+
"grad_norm": 3.0708813667297363,
|
| 7493 |
+
"learning_rate": 3.154207977133812e-05,
|
| 7494 |
+
"loss": 1.183,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.09003269697289315,
|
| 7499 |
+
"grad_norm": 3.0313644409179688,
|
| 7500 |
+
"learning_rate": 3.138176732885821e-05,
|
| 7501 |
+
"loss": 1.0117,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.09011707625777872,
|
| 7506 |
+
"grad_norm": 3.0188441276550293,
|
| 7507 |
+
"learning_rate": 3.122178746256694e-05,
|
| 7508 |
+
"loss": 1.4957,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.09020145554266427,
|
| 7513 |
+
"grad_norm": 2.6221067905426025,
|
| 7514 |
+
"learning_rate": 3.106214094784855e-05,
|
| 7515 |
+
"loss": 1.3605,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.09028583482754983,
|
| 7520 |
+
"grad_norm": 3.322061061859131,
|
| 7521 |
+
"learning_rate": 3.090282855847155e-05,
|
| 7522 |
+
"loss": 1.5937,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.0903702141124354,
|
| 7527 |
+
"grad_norm": 2.7745418548583984,
|
| 7528 |
+
"learning_rate": 3.0743851066585004e-05,
|
| 7529 |
+
"loss": 1.3412,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.09045459339732095,
|
| 7534 |
+
"grad_norm": 3.403449296951294,
|
| 7535 |
+
"learning_rate": 3.05852092427149e-05,
|
| 7536 |
+
"loss": 1.3846,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.09053897268220652,
|
| 7541 |
+
"grad_norm": 5.336084842681885,
|
| 7542 |
+
"learning_rate": 3.042690385576027e-05,
|
| 7543 |
+
"loss": 2.3124,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.09062335196709208,
|
| 7548 |
+
"grad_norm": 3.023393154144287,
|
| 7549 |
+
"learning_rate": 3.026893567298955e-05,
|
| 7550 |
+
"loss": 1.3777,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.09070773125197763,
|
| 7555 |
+
"grad_norm": 3.2764523029327393,
|
| 7556 |
+
"learning_rate": 3.0111305460036744e-05,
|
| 7557 |
+
"loss": 1.6167,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.0907921105368632,
|
| 7562 |
+
"grad_norm": 3.874765396118164,
|
| 7563 |
+
"learning_rate": 2.995401398089789e-05,
|
| 7564 |
+
"loss": 1.6595,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.09087648982174876,
|
| 7569 |
+
"grad_norm": 2.6643738746643066,
|
| 7570 |
+
"learning_rate": 2.9797061997927245e-05,
|
| 7571 |
+
"loss": 0.9284,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.09096086910663433,
|
| 7576 |
+
"grad_norm": 3.545632839202881,
|
| 7577 |
+
"learning_rate": 2.9640450271833552e-05,
|
| 7578 |
+
"loss": 1.6928,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.09104524839151988,
|
| 7583 |
+
"grad_norm": 3.3833420276641846,
|
| 7584 |
+
"learning_rate": 2.9484179561676462e-05,
|
| 7585 |
+
"loss": 1.6922,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.09112962767640544,
|
| 7590 |
+
"grad_norm": 3.4521713256835938,
|
| 7591 |
+
"learning_rate": 2.932825062486283e-05,
|
| 7592 |
+
"loss": 1.8018,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.09112962767640544,
|
| 7597 |
+
"eval_loss": 1.5595093965530396,
|
| 7598 |
+
"eval_runtime": 177.3501,
|
| 7599 |
+
"eval_samples_per_second": 28.136,
|
| 7600 |
+
"eval_steps_per_second": 14.068,
|
| 7601 |
+
"step": 1080
|
| 7602 |
}
|
| 7603 |
],
|
| 7604 |
"logging_steps": 1,
|
|
|
|
| 7618 |
"attributes": {}
|
| 7619 |
}
|
| 7620 |
},
|
| 7621 |
+
"total_flos": 2.0873895994195968e+17,
|
| 7622 |
"train_batch_size": 2,
|
| 7623 |
"trial_name": null,
|
| 7624 |
"trial_params": null
|