Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- scheduler.pt +1 -1
- trainer_state.json +3503 -3
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1728658208
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:635a4a5eccec6e90fbd9e03e4a035575f6eb9aad65905ea7aa505ec343fa16cc
|
| 3 |
size 1728658208
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 878485380
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45a158cb4b1813a1b753cf0354b21f7f35de8aa54dd074770a6097fb67421644
|
| 3 |
size 878485380
|
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:e8cfa4bbfbaa016233c07ad7188a329dd106c70ae970f9e363d6f182d78c9d1c
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3507,6 +3507,3506 @@
|
|
| 3507 |
"learning_rate": 0.00047835781741867783,
|
| 3508 |
"loss": 1.046,
|
| 3509 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3510 |
}
|
| 3511 |
],
|
| 3512 |
"logging_steps": 1,
|
|
@@ -3526,7 +7026,7 @@
|
|
| 3526 |
"attributes": {}
|
| 3527 |
}
|
| 3528 |
},
|
| 3529 |
-
"total_flos":
|
| 3530 |
"train_batch_size": 2,
|
| 3531 |
"trial_name": null,
|
| 3532 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.08740303725554463,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3507 |
"learning_rate": 0.00047835781741867783,
|
| 3508 |
"loss": 1.046,
|
| 3509 |
"step": 500
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 0.04378892166502786,
|
| 3513 |
+
"grad_norm": 0.3747900128364563,
|
| 3514 |
+
"learning_rate": 0.00047831409583770553,
|
| 3515 |
+
"loss": 1.0201,
|
| 3516 |
+
"step": 501
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 0.043876324702283406,
|
| 3520 |
+
"grad_norm": 0.3672393262386322,
|
| 3521 |
+
"learning_rate": 0.0004782703742567331,
|
| 3522 |
+
"loss": 1.0021,
|
| 3523 |
+
"step": 502
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 0.04396372773953895,
|
| 3527 |
+
"grad_norm": 0.3505338132381439,
|
| 3528 |
+
"learning_rate": 0.00047822665267576077,
|
| 3529 |
+
"loss": 1.0002,
|
| 3530 |
+
"step": 503
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 0.04405113077679449,
|
| 3534 |
+
"grad_norm": 5.722542762756348,
|
| 3535 |
+
"learning_rate": 0.0004781829310947884,
|
| 3536 |
+
"loss": 2.5431,
|
| 3537 |
+
"step": 504
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 0.044138533814050036,
|
| 3541 |
+
"grad_norm": 0.5349693298339844,
|
| 3542 |
+
"learning_rate": 0.000478139209513816,
|
| 3543 |
+
"loss": 1.151,
|
| 3544 |
+
"step": 505
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 0.04422593685130558,
|
| 3548 |
+
"grad_norm": 0.4468895494937897,
|
| 3549 |
+
"learning_rate": 0.0004780954879328437,
|
| 3550 |
+
"loss": 0.9958,
|
| 3551 |
+
"step": 506
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 0.04431333988856113,
|
| 3555 |
+
"grad_norm": 0.47205036878585815,
|
| 3556 |
+
"learning_rate": 0.0004780517663518713,
|
| 3557 |
+
"loss": 0.9401,
|
| 3558 |
+
"step": 507
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 0.04440074292581667,
|
| 3562 |
+
"grad_norm": 0.35336941480636597,
|
| 3563 |
+
"learning_rate": 0.00047800804477089894,
|
| 3564 |
+
"loss": 1.0982,
|
| 3565 |
+
"step": 508
|
| 3566 |
+
},
|
| 3567 |
+
{
|
| 3568 |
+
"epoch": 0.04448814596307222,
|
| 3569 |
+
"grad_norm": 1.8884743452072144,
|
| 3570 |
+
"learning_rate": 0.00047796432318992653,
|
| 3571 |
+
"loss": 0.9199,
|
| 3572 |
+
"step": 509
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 0.044575549000327765,
|
| 3576 |
+
"grad_norm": 0.4091229736804962,
|
| 3577 |
+
"learning_rate": 0.0004779206016089542,
|
| 3578 |
+
"loss": 0.8953,
|
| 3579 |
+
"step": 510
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 0.0446629520375833,
|
| 3583 |
+
"grad_norm": 0.4730583131313324,
|
| 3584 |
+
"learning_rate": 0.0004778768800279818,
|
| 3585 |
+
"loss": 0.8085,
|
| 3586 |
+
"step": 511
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 0.04475035507483885,
|
| 3590 |
+
"grad_norm": 0.3801075220108032,
|
| 3591 |
+
"learning_rate": 0.00047783315844700946,
|
| 3592 |
+
"loss": 0.9914,
|
| 3593 |
+
"step": 512
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 0.044837758112094395,
|
| 3597 |
+
"grad_norm": 0.3660631477832794,
|
| 3598 |
+
"learning_rate": 0.0004777894368660371,
|
| 3599 |
+
"loss": 0.9804,
|
| 3600 |
+
"step": 513
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 0.04492516114934994,
|
| 3604 |
+
"grad_norm": 0.8466418981552124,
|
| 3605 |
+
"learning_rate": 0.0004777457152850647,
|
| 3606 |
+
"loss": 1.1207,
|
| 3607 |
+
"step": 514
|
| 3608 |
+
},
|
| 3609 |
+
{
|
| 3610 |
+
"epoch": 0.045012564186605486,
|
| 3611 |
+
"grad_norm": 0.3560774624347687,
|
| 3612 |
+
"learning_rate": 0.00047770199370409234,
|
| 3613 |
+
"loss": 0.8773,
|
| 3614 |
+
"step": 515
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 0.04509996722386103,
|
| 3618 |
+
"grad_norm": 0.49633318185806274,
|
| 3619 |
+
"learning_rate": 0.00047765827212312,
|
| 3620 |
+
"loss": 1.0111,
|
| 3621 |
+
"step": 516
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 0.04518737026111657,
|
| 3625 |
+
"grad_norm": 0.6001185178756714,
|
| 3626 |
+
"learning_rate": 0.00047761455054214764,
|
| 3627 |
+
"loss": 1.2566,
|
| 3628 |
+
"step": 517
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 0.045274773298372116,
|
| 3632 |
+
"grad_norm": 0.7423095703125,
|
| 3633 |
+
"learning_rate": 0.0004775708289611753,
|
| 3634 |
+
"loss": 1.1431,
|
| 3635 |
+
"step": 518
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 0.04536217633562766,
|
| 3639 |
+
"grad_norm": 0.34218892455101013,
|
| 3640 |
+
"learning_rate": 0.00047752710738020287,
|
| 3641 |
+
"loss": 0.9254,
|
| 3642 |
+
"step": 519
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 0.04544957937288321,
|
| 3646 |
+
"grad_norm": 0.336230605840683,
|
| 3647 |
+
"learning_rate": 0.0004774833857992305,
|
| 3648 |
+
"loss": 1.0015,
|
| 3649 |
+
"step": 520
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 0.04553698241013875,
|
| 3653 |
+
"grad_norm": 0.39158111810684204,
|
| 3654 |
+
"learning_rate": 0.0004774396642182581,
|
| 3655 |
+
"loss": 0.8319,
|
| 3656 |
+
"step": 521
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 0.0456243854473943,
|
| 3660 |
+
"grad_norm": 0.4045357406139374,
|
| 3661 |
+
"learning_rate": 0.00047739594263728575,
|
| 3662 |
+
"loss": 0.8531,
|
| 3663 |
+
"step": 522
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 0.045711788484649844,
|
| 3667 |
+
"grad_norm": 0.5861966013908386,
|
| 3668 |
+
"learning_rate": 0.00047735222105631345,
|
| 3669 |
+
"loss": 0.9975,
|
| 3670 |
+
"step": 523
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 0.04579919152190538,
|
| 3674 |
+
"grad_norm": 0.33865249156951904,
|
| 3675 |
+
"learning_rate": 0.00047730849947534104,
|
| 3676 |
+
"loss": 0.94,
|
| 3677 |
+
"step": 524
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 0.04588659455916093,
|
| 3681 |
+
"grad_norm": 0.4759502112865448,
|
| 3682 |
+
"learning_rate": 0.0004772647778943687,
|
| 3683 |
+
"loss": 0.9581,
|
| 3684 |
+
"step": 525
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 0.045973997596416474,
|
| 3688 |
+
"grad_norm": 0.492929607629776,
|
| 3689 |
+
"learning_rate": 0.0004772210563133963,
|
| 3690 |
+
"loss": 1.3563,
|
| 3691 |
+
"step": 526
|
| 3692 |
+
},
|
| 3693 |
+
{
|
| 3694 |
+
"epoch": 0.04606140063367202,
|
| 3695 |
+
"grad_norm": 0.31947705149650574,
|
| 3696 |
+
"learning_rate": 0.0004771773347324239,
|
| 3697 |
+
"loss": 0.8052,
|
| 3698 |
+
"step": 527
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 0.046148803670927566,
|
| 3702 |
+
"grad_norm": 0.3842394948005676,
|
| 3703 |
+
"learning_rate": 0.0004771336131514515,
|
| 3704 |
+
"loss": 0.9723,
|
| 3705 |
+
"step": 528
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 0.04623620670818311,
|
| 3709 |
+
"grad_norm": 0.338451623916626,
|
| 3710 |
+
"learning_rate": 0.0004770898915704792,
|
| 3711 |
+
"loss": 1.0315,
|
| 3712 |
+
"step": 529
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 0.04632360974543866,
|
| 3716 |
+
"grad_norm": 1.9640684127807617,
|
| 3717 |
+
"learning_rate": 0.00047704616998950686,
|
| 3718 |
+
"loss": 1.2013,
|
| 3719 |
+
"step": 530
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 0.046411012782694196,
|
| 3723 |
+
"grad_norm": 0.501758337020874,
|
| 3724 |
+
"learning_rate": 0.00047700244840853445,
|
| 3725 |
+
"loss": 1.0096,
|
| 3726 |
+
"step": 531
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 0.04649841581994974,
|
| 3730 |
+
"grad_norm": 0.5867491960525513,
|
| 3731 |
+
"learning_rate": 0.0004769587268275621,
|
| 3732 |
+
"loss": 0.9708,
|
| 3733 |
+
"step": 532
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 0.04658581885720529,
|
| 3737 |
+
"grad_norm": 2.1122539043426514,
|
| 3738 |
+
"learning_rate": 0.0004769150052465897,
|
| 3739 |
+
"loss": 0.8145,
|
| 3740 |
+
"step": 533
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 0.04667322189446083,
|
| 3744 |
+
"grad_norm": 0.7969621419906616,
|
| 3745 |
+
"learning_rate": 0.0004768712836656174,
|
| 3746 |
+
"loss": 0.829,
|
| 3747 |
+
"step": 534
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 0.04676062493171638,
|
| 3751 |
+
"grad_norm": 0.4205247461795807,
|
| 3752 |
+
"learning_rate": 0.00047682756208464503,
|
| 3753 |
+
"loss": 1.0063,
|
| 3754 |
+
"step": 535
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.046848027968971924,
|
| 3758 |
+
"grad_norm": 0.3231610059738159,
|
| 3759 |
+
"learning_rate": 0.0004767838405036726,
|
| 3760 |
+
"loss": 0.968,
|
| 3761 |
+
"step": 536
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.04693543100622747,
|
| 3765 |
+
"grad_norm": 1.369025707244873,
|
| 3766 |
+
"learning_rate": 0.00047674011892270027,
|
| 3767 |
+
"loss": 1.7445,
|
| 3768 |
+
"step": 537
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 0.04702283404348301,
|
| 3772 |
+
"grad_norm": 0.42706942558288574,
|
| 3773 |
+
"learning_rate": 0.00047669639734172786,
|
| 3774 |
+
"loss": 1.1781,
|
| 3775 |
+
"step": 538
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 0.047110237080738554,
|
| 3779 |
+
"grad_norm": 0.36257731914520264,
|
| 3780 |
+
"learning_rate": 0.0004766526757607555,
|
| 3781 |
+
"loss": 1.0557,
|
| 3782 |
+
"step": 539
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.0471976401179941,
|
| 3786 |
+
"grad_norm": 0.4783022105693817,
|
| 3787 |
+
"learning_rate": 0.00047660895417978315,
|
| 3788 |
+
"loss": 1.053,
|
| 3789 |
+
"step": 540
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 0.047285043155249645,
|
| 3793 |
+
"grad_norm": 0.3079909384250641,
|
| 3794 |
+
"learning_rate": 0.0004765652325988108,
|
| 3795 |
+
"loss": 1.1313,
|
| 3796 |
+
"step": 541
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 0.04737244619250519,
|
| 3800 |
+
"grad_norm": 0.4072510302066803,
|
| 3801 |
+
"learning_rate": 0.00047652151101783844,
|
| 3802 |
+
"loss": 0.8678,
|
| 3803 |
+
"step": 542
|
| 3804 |
+
},
|
| 3805 |
+
{
|
| 3806 |
+
"epoch": 0.04745984922976074,
|
| 3807 |
+
"grad_norm": 0.36985546350479126,
|
| 3808 |
+
"learning_rate": 0.00047647778943686603,
|
| 3809 |
+
"loss": 0.9387,
|
| 3810 |
+
"step": 543
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 0.04754725226701628,
|
| 3814 |
+
"grad_norm": 0.4222630262374878,
|
| 3815 |
+
"learning_rate": 0.0004764340678558937,
|
| 3816 |
+
"loss": 0.9083,
|
| 3817 |
+
"step": 544
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 0.04763465530427182,
|
| 3821 |
+
"grad_norm": 0.39896291494369507,
|
| 3822 |
+
"learning_rate": 0.0004763903462749213,
|
| 3823 |
+
"loss": 0.9773,
|
| 3824 |
+
"step": 545
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 0.04772205834152737,
|
| 3828 |
+
"grad_norm": 0.3235687017440796,
|
| 3829 |
+
"learning_rate": 0.00047634662469394896,
|
| 3830 |
+
"loss": 0.9484,
|
| 3831 |
+
"step": 546
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 0.04780946137878291,
|
| 3835 |
+
"grad_norm": 0.3377327620983124,
|
| 3836 |
+
"learning_rate": 0.0004763029031129766,
|
| 3837 |
+
"loss": 0.9319,
|
| 3838 |
+
"step": 547
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 0.04789686441603846,
|
| 3842 |
+
"grad_norm": 0.37998026609420776,
|
| 3843 |
+
"learning_rate": 0.0004762591815320042,
|
| 3844 |
+
"loss": 1.3499,
|
| 3845 |
+
"step": 548
|
| 3846 |
+
},
|
| 3847 |
+
{
|
| 3848 |
+
"epoch": 0.047984267453294004,
|
| 3849 |
+
"grad_norm": 0.37219107151031494,
|
| 3850 |
+
"learning_rate": 0.00047621545995103184,
|
| 3851 |
+
"loss": 1.1132,
|
| 3852 |
+
"step": 549
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 0.04807167049054955,
|
| 3856 |
+
"grad_norm": 0.3147220313549042,
|
| 3857 |
+
"learning_rate": 0.00047617173837005944,
|
| 3858 |
+
"loss": 0.9306,
|
| 3859 |
+
"step": 550
|
| 3860 |
+
},
|
| 3861 |
+
{
|
| 3862 |
+
"epoch": 0.04815907352780509,
|
| 3863 |
+
"grad_norm": 0.3832624852657318,
|
| 3864 |
+
"learning_rate": 0.00047612801678908713,
|
| 3865 |
+
"loss": 0.8518,
|
| 3866 |
+
"step": 551
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 0.048246476565060634,
|
| 3870 |
+
"grad_norm": 0.3098907172679901,
|
| 3871 |
+
"learning_rate": 0.0004760842952081147,
|
| 3872 |
+
"loss": 0.8183,
|
| 3873 |
+
"step": 552
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 0.04833387960231618,
|
| 3877 |
+
"grad_norm": 0.3062676191329956,
|
| 3878 |
+
"learning_rate": 0.00047604057362714237,
|
| 3879 |
+
"loss": 0.9226,
|
| 3880 |
+
"step": 553
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 0.048421282639571725,
|
| 3884 |
+
"grad_norm": 0.3292568624019623,
|
| 3885 |
+
"learning_rate": 0.00047599685204617,
|
| 3886 |
+
"loss": 0.9204,
|
| 3887 |
+
"step": 554
|
| 3888 |
+
},
|
| 3889 |
+
{
|
| 3890 |
+
"epoch": 0.04850868567682727,
|
| 3891 |
+
"grad_norm": 0.45942652225494385,
|
| 3892 |
+
"learning_rate": 0.0004759531304651976,
|
| 3893 |
+
"loss": 1.1571,
|
| 3894 |
+
"step": 555
|
| 3895 |
+
},
|
| 3896 |
+
{
|
| 3897 |
+
"epoch": 0.048596088714082816,
|
| 3898 |
+
"grad_norm": 0.3519571125507355,
|
| 3899 |
+
"learning_rate": 0.00047590940888422525,
|
| 3900 |
+
"loss": 0.9566,
|
| 3901 |
+
"step": 556
|
| 3902 |
+
},
|
| 3903 |
+
{
|
| 3904 |
+
"epoch": 0.04868349175133836,
|
| 3905 |
+
"grad_norm": 0.3418327569961548,
|
| 3906 |
+
"learning_rate": 0.0004758656873032529,
|
| 3907 |
+
"loss": 1.146,
|
| 3908 |
+
"step": 557
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 0.0487708947885939,
|
| 3912 |
+
"grad_norm": 0.3338674008846283,
|
| 3913 |
+
"learning_rate": 0.00047582196572228054,
|
| 3914 |
+
"loss": 1.0859,
|
| 3915 |
+
"step": 558
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 0.048858297825849446,
|
| 3919 |
+
"grad_norm": 1.2700949907302856,
|
| 3920 |
+
"learning_rate": 0.00047577824414130813,
|
| 3921 |
+
"loss": 1.3166,
|
| 3922 |
+
"step": 559
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 0.04894570086310499,
|
| 3926 |
+
"grad_norm": 0.706069827079773,
|
| 3927 |
+
"learning_rate": 0.0004757345225603358,
|
| 3928 |
+
"loss": 1.2259,
|
| 3929 |
+
"step": 560
|
| 3930 |
+
},
|
| 3931 |
+
{
|
| 3932 |
+
"epoch": 0.04903310390036054,
|
| 3933 |
+
"grad_norm": 0.5171198844909668,
|
| 3934 |
+
"learning_rate": 0.0004756908009793634,
|
| 3935 |
+
"loss": 0.7985,
|
| 3936 |
+
"step": 561
|
| 3937 |
+
},
|
| 3938 |
+
{
|
| 3939 |
+
"epoch": 0.04912050693761608,
|
| 3940 |
+
"grad_norm": 0.8621017932891846,
|
| 3941 |
+
"learning_rate": 0.00047564707939839107,
|
| 3942 |
+
"loss": 1.0042,
|
| 3943 |
+
"step": 562
|
| 3944 |
+
},
|
| 3945 |
+
{
|
| 3946 |
+
"epoch": 0.04920790997487163,
|
| 3947 |
+
"grad_norm": 0.926487922668457,
|
| 3948 |
+
"learning_rate": 0.0004756033578174187,
|
| 3949 |
+
"loss": 0.9945,
|
| 3950 |
+
"step": 563
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 0.049295313012127175,
|
| 3954 |
+
"grad_norm": 0.9586560726165771,
|
| 3955 |
+
"learning_rate": 0.0004755596362364463,
|
| 3956 |
+
"loss": 1.5266,
|
| 3957 |
+
"step": 564
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 0.04938271604938271,
|
| 3961 |
+
"grad_norm": 0.507824182510376,
|
| 3962 |
+
"learning_rate": 0.00047551591465547395,
|
| 3963 |
+
"loss": 0.8737,
|
| 3964 |
+
"step": 565
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 0.04947011908663826,
|
| 3968 |
+
"grad_norm": 0.38291049003601074,
|
| 3969 |
+
"learning_rate": 0.0004754721930745016,
|
| 3970 |
+
"loss": 0.7636,
|
| 3971 |
+
"step": 566
|
| 3972 |
+
},
|
| 3973 |
+
{
|
| 3974 |
+
"epoch": 0.049557522123893805,
|
| 3975 |
+
"grad_norm": 0.40479573607444763,
|
| 3976 |
+
"learning_rate": 0.0004754284714935292,
|
| 3977 |
+
"loss": 0.781,
|
| 3978 |
+
"step": 567
|
| 3979 |
+
},
|
| 3980 |
+
{
|
| 3981 |
+
"epoch": 0.04964492516114935,
|
| 3982 |
+
"grad_norm": 0.6375040411949158,
|
| 3983 |
+
"learning_rate": 0.0004753847499125569,
|
| 3984 |
+
"loss": 1.1493,
|
| 3985 |
+
"step": 568
|
| 3986 |
+
},
|
| 3987 |
+
{
|
| 3988 |
+
"epoch": 0.049732328198404896,
|
| 3989 |
+
"grad_norm": 0.3949948847293854,
|
| 3990 |
+
"learning_rate": 0.0004753410283315845,
|
| 3991 |
+
"loss": 0.9626,
|
| 3992 |
+
"step": 569
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 0.04981973123566044,
|
| 3996 |
+
"grad_norm": 0.3734526038169861,
|
| 3997 |
+
"learning_rate": 0.0004752973067506121,
|
| 3998 |
+
"loss": 0.9207,
|
| 3999 |
+
"step": 570
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 0.04990713427291599,
|
| 4003 |
+
"grad_norm": 0.5179705619812012,
|
| 4004 |
+
"learning_rate": 0.0004752535851696397,
|
| 4005 |
+
"loss": 1.3906,
|
| 4006 |
+
"step": 571
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.049994537310171526,
|
| 4010 |
+
"grad_norm": 0.4602389931678772,
|
| 4011 |
+
"learning_rate": 0.00047520986358866736,
|
| 4012 |
+
"loss": 1.0577,
|
| 4013 |
+
"step": 572
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.05008194034742707,
|
| 4017 |
+
"grad_norm": 0.30401960015296936,
|
| 4018 |
+
"learning_rate": 0.00047516614200769506,
|
| 4019 |
+
"loss": 1.13,
|
| 4020 |
+
"step": 573
|
| 4021 |
+
},
|
| 4022 |
+
{
|
| 4023 |
+
"epoch": 0.05016934338468262,
|
| 4024 |
+
"grad_norm": 0.3481753170490265,
|
| 4025 |
+
"learning_rate": 0.00047512242042672265,
|
| 4026 |
+
"loss": 0.857,
|
| 4027 |
+
"step": 574
|
| 4028 |
+
},
|
| 4029 |
+
{
|
| 4030 |
+
"epoch": 0.05025674642193816,
|
| 4031 |
+
"grad_norm": 0.4005964398384094,
|
| 4032 |
+
"learning_rate": 0.0004750786988457503,
|
| 4033 |
+
"loss": 0.9569,
|
| 4034 |
+
"step": 575
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.05034414945919371,
|
| 4038 |
+
"grad_norm": 0.43765851855278015,
|
| 4039 |
+
"learning_rate": 0.0004750349772647779,
|
| 4040 |
+
"loss": 1.2156,
|
| 4041 |
+
"step": 576
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 0.050431552496449254,
|
| 4045 |
+
"grad_norm": 0.3252186179161072,
|
| 4046 |
+
"learning_rate": 0.00047499125568380553,
|
| 4047 |
+
"loss": 1.0392,
|
| 4048 |
+
"step": 577
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 0.05051895553370479,
|
| 4052 |
+
"grad_norm": 0.3639061152935028,
|
| 4053 |
+
"learning_rate": 0.0004749475341028331,
|
| 4054 |
+
"loss": 0.914,
|
| 4055 |
+
"step": 578
|
| 4056 |
+
},
|
| 4057 |
+
{
|
| 4058 |
+
"epoch": 0.05060635857096034,
|
| 4059 |
+
"grad_norm": 0.3080824911594391,
|
| 4060 |
+
"learning_rate": 0.0004749038125218608,
|
| 4061 |
+
"loss": 0.9735,
|
| 4062 |
+
"step": 579
|
| 4063 |
+
},
|
| 4064 |
+
{
|
| 4065 |
+
"epoch": 0.050693761608215884,
|
| 4066 |
+
"grad_norm": 0.33566662669181824,
|
| 4067 |
+
"learning_rate": 0.00047486009094088846,
|
| 4068 |
+
"loss": 1.1619,
|
| 4069 |
+
"step": 580
|
| 4070 |
+
},
|
| 4071 |
+
{
|
| 4072 |
+
"epoch": 0.05078116464547143,
|
| 4073 |
+
"grad_norm": 0.2990110218524933,
|
| 4074 |
+
"learning_rate": 0.00047481636935991605,
|
| 4075 |
+
"loss": 0.97,
|
| 4076 |
+
"step": 581
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 0.050868567682726976,
|
| 4080 |
+
"grad_norm": 0.3264564871788025,
|
| 4081 |
+
"learning_rate": 0.0004747726477789437,
|
| 4082 |
+
"loss": 0.824,
|
| 4083 |
+
"step": 582
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 0.05095597071998252,
|
| 4087 |
+
"grad_norm": 0.37740233540534973,
|
| 4088 |
+
"learning_rate": 0.0004747289261979713,
|
| 4089 |
+
"loss": 1.1715,
|
| 4090 |
+
"step": 583
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 0.05104337375723807,
|
| 4094 |
+
"grad_norm": 0.39894765615463257,
|
| 4095 |
+
"learning_rate": 0.00047468520461699894,
|
| 4096 |
+
"loss": 1.3263,
|
| 4097 |
+
"step": 584
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 0.051130776794493606,
|
| 4101 |
+
"grad_norm": 0.3279603123664856,
|
| 4102 |
+
"learning_rate": 0.00047464148303602663,
|
| 4103 |
+
"loss": 0.8633,
|
| 4104 |
+
"step": 585
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 0.05121817983174915,
|
| 4108 |
+
"grad_norm": 0.30895987153053284,
|
| 4109 |
+
"learning_rate": 0.0004745977614550542,
|
| 4110 |
+
"loss": 0.9019,
|
| 4111 |
+
"step": 586
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 0.0513055828690047,
|
| 4115 |
+
"grad_norm": 0.8510332703590393,
|
| 4116 |
+
"learning_rate": 0.00047455403987408187,
|
| 4117 |
+
"loss": 0.9492,
|
| 4118 |
+
"step": 587
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.05139298590626024,
|
| 4122 |
+
"grad_norm": 0.5336425304412842,
|
| 4123 |
+
"learning_rate": 0.00047451031829310946,
|
| 4124 |
+
"loss": 0.8209,
|
| 4125 |
+
"step": 588
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 0.05148038894351579,
|
| 4129 |
+
"grad_norm": 0.3380926847457886,
|
| 4130 |
+
"learning_rate": 0.0004744665967121371,
|
| 4131 |
+
"loss": 0.8024,
|
| 4132 |
+
"step": 589
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.051567791980771334,
|
| 4136 |
+
"grad_norm": 0.3537689447402954,
|
| 4137 |
+
"learning_rate": 0.00047442287513116475,
|
| 4138 |
+
"loss": 1.1219,
|
| 4139 |
+
"step": 590
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.05165519501802688,
|
| 4143 |
+
"grad_norm": 0.5417413711547852,
|
| 4144 |
+
"learning_rate": 0.0004743791535501924,
|
| 4145 |
+
"loss": 1.0341,
|
| 4146 |
+
"step": 591
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 0.05174259805528242,
|
| 4150 |
+
"grad_norm": 0.4394038915634155,
|
| 4151 |
+
"learning_rate": 0.00047433543196922004,
|
| 4152 |
+
"loss": 0.934,
|
| 4153 |
+
"step": 592
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 0.051830001092537964,
|
| 4157 |
+
"grad_norm": 0.738370954990387,
|
| 4158 |
+
"learning_rate": 0.00047429171038824763,
|
| 4159 |
+
"loss": 1.1953,
|
| 4160 |
+
"step": 593
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.05191740412979351,
|
| 4164 |
+
"grad_norm": 0.33024734258651733,
|
| 4165 |
+
"learning_rate": 0.0004742479888072753,
|
| 4166 |
+
"loss": 0.687,
|
| 4167 |
+
"step": 594
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 0.052004807167049055,
|
| 4171 |
+
"grad_norm": 0.3696803152561188,
|
| 4172 |
+
"learning_rate": 0.00047420426722630287,
|
| 4173 |
+
"loss": 1.0533,
|
| 4174 |
+
"step": 595
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.0520922102043046,
|
| 4178 |
+
"grad_norm": 0.31398460268974304,
|
| 4179 |
+
"learning_rate": 0.00047416054564533057,
|
| 4180 |
+
"loss": 1.0434,
|
| 4181 |
+
"step": 596
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.05217961324156015,
|
| 4185 |
+
"grad_norm": 0.3482360541820526,
|
| 4186 |
+
"learning_rate": 0.0004741168240643582,
|
| 4187 |
+
"loss": 1.2415,
|
| 4188 |
+
"step": 597
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 0.05226701627881569,
|
| 4192 |
+
"grad_norm": 0.32207486033439636,
|
| 4193 |
+
"learning_rate": 0.0004740731024833858,
|
| 4194 |
+
"loss": 1.1465,
|
| 4195 |
+
"step": 598
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 0.05235441931607123,
|
| 4199 |
+
"grad_norm": 0.2964969277381897,
|
| 4200 |
+
"learning_rate": 0.00047402938090241345,
|
| 4201 |
+
"loss": 0.8746,
|
| 4202 |
+
"step": 599
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.05244182235332678,
|
| 4206 |
+
"grad_norm": 0.26993119716644287,
|
| 4207 |
+
"learning_rate": 0.00047398565932144104,
|
| 4208 |
+
"loss": 0.9161,
|
| 4209 |
+
"step": 600
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.05252922539058232,
|
| 4213 |
+
"grad_norm": 0.31088942289352417,
|
| 4214 |
+
"learning_rate": 0.00047394193774046874,
|
| 4215 |
+
"loss": 0.938,
|
| 4216 |
+
"step": 601
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.05261662842783787,
|
| 4220 |
+
"grad_norm": 0.2921091318130493,
|
| 4221 |
+
"learning_rate": 0.00047389821615949633,
|
| 4222 |
+
"loss": 0.914,
|
| 4223 |
+
"step": 602
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.052704031465093414,
|
| 4227 |
+
"grad_norm": 0.4693572223186493,
|
| 4228 |
+
"learning_rate": 0.000473854494578524,
|
| 4229 |
+
"loss": 0.9083,
|
| 4230 |
+
"step": 603
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.05279143450234896,
|
| 4234 |
+
"grad_norm": 0.6201152801513672,
|
| 4235 |
+
"learning_rate": 0.0004738107729975516,
|
| 4236 |
+
"loss": 1.1098,
|
| 4237 |
+
"step": 604
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.0528788375396045,
|
| 4241 |
+
"grad_norm": 0.48871442675590515,
|
| 4242 |
+
"learning_rate": 0.0004737670514165792,
|
| 4243 |
+
"loss": 1.1571,
|
| 4244 |
+
"step": 605
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.052966240576860044,
|
| 4248 |
+
"grad_norm": 0.26332658529281616,
|
| 4249 |
+
"learning_rate": 0.00047372332983560686,
|
| 4250 |
+
"loss": 0.995,
|
| 4251 |
+
"step": 606
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.05305364361411559,
|
| 4255 |
+
"grad_norm": 0.7663961052894592,
|
| 4256 |
+
"learning_rate": 0.0004736796082546345,
|
| 4257 |
+
"loss": 1.0206,
|
| 4258 |
+
"step": 607
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.053141046651371135,
|
| 4262 |
+
"grad_norm": 0.3350706100463867,
|
| 4263 |
+
"learning_rate": 0.00047363588667366215,
|
| 4264 |
+
"loss": 1.0328,
|
| 4265 |
+
"step": 608
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.05322844968862668,
|
| 4269 |
+
"grad_norm": 0.30147233605384827,
|
| 4270 |
+
"learning_rate": 0.00047359216509268974,
|
| 4271 |
+
"loss": 0.8874,
|
| 4272 |
+
"step": 609
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.053315852725882226,
|
| 4276 |
+
"grad_norm": 0.4487704038619995,
|
| 4277 |
+
"learning_rate": 0.0004735484435117174,
|
| 4278 |
+
"loss": 0.8327,
|
| 4279 |
+
"step": 610
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.05340325576313777,
|
| 4283 |
+
"grad_norm": 0.474685400724411,
|
| 4284 |
+
"learning_rate": 0.00047350472193074503,
|
| 4285 |
+
"loss": 0.8405,
|
| 4286 |
+
"step": 611
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.05349065880039331,
|
| 4290 |
+
"grad_norm": 0.6512682437896729,
|
| 4291 |
+
"learning_rate": 0.0004734610003497726,
|
| 4292 |
+
"loss": 1.418,
|
| 4293 |
+
"step": 612
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.053578061837648856,
|
| 4297 |
+
"grad_norm": 0.3829117715358734,
|
| 4298 |
+
"learning_rate": 0.0004734172787688003,
|
| 4299 |
+
"loss": 0.9036,
|
| 4300 |
+
"step": 613
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.0536654648749044,
|
| 4304 |
+
"grad_norm": 0.3626525402069092,
|
| 4305 |
+
"learning_rate": 0.0004733735571878279,
|
| 4306 |
+
"loss": 0.9919,
|
| 4307 |
+
"step": 614
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.05375286791215995,
|
| 4311 |
+
"grad_norm": 0.6899876594543457,
|
| 4312 |
+
"learning_rate": 0.00047332983560685555,
|
| 4313 |
+
"loss": 0.8781,
|
| 4314 |
+
"step": 615
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.05384027094941549,
|
| 4318 |
+
"grad_norm": 0.33936572074890137,
|
| 4319 |
+
"learning_rate": 0.0004732861140258832,
|
| 4320 |
+
"loss": 0.7375,
|
| 4321 |
+
"step": 616
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.05392767398667104,
|
| 4325 |
+
"grad_norm": 0.45376959443092346,
|
| 4326 |
+
"learning_rate": 0.0004732423924449108,
|
| 4327 |
+
"loss": 0.868,
|
| 4328 |
+
"step": 617
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.054015077023926585,
|
| 4332 |
+
"grad_norm": 0.5580937266349792,
|
| 4333 |
+
"learning_rate": 0.0004731986708639385,
|
| 4334 |
+
"loss": 1.182,
|
| 4335 |
+
"step": 618
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.05410248006118212,
|
| 4339 |
+
"grad_norm": 0.3207378685474396,
|
| 4340 |
+
"learning_rate": 0.0004731549492829661,
|
| 4341 |
+
"loss": 0.9069,
|
| 4342 |
+
"step": 619
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.05418988309843767,
|
| 4346 |
+
"grad_norm": 0.3553832769393921,
|
| 4347 |
+
"learning_rate": 0.0004731112277019937,
|
| 4348 |
+
"loss": 1.4,
|
| 4349 |
+
"step": 620
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.054277286135693215,
|
| 4353 |
+
"grad_norm": 0.3708738386631012,
|
| 4354 |
+
"learning_rate": 0.0004730675061210213,
|
| 4355 |
+
"loss": 1.1475,
|
| 4356 |
+
"step": 621
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.05436468917294876,
|
| 4360 |
+
"grad_norm": 0.35041436553001404,
|
| 4361 |
+
"learning_rate": 0.00047302378454004896,
|
| 4362 |
+
"loss": 0.9505,
|
| 4363 |
+
"step": 622
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.054452092210204306,
|
| 4367 |
+
"grad_norm": 0.37304723262786865,
|
| 4368 |
+
"learning_rate": 0.0004729800629590766,
|
| 4369 |
+
"loss": 0.8858,
|
| 4370 |
+
"step": 623
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.05453949524745985,
|
| 4374 |
+
"grad_norm": 0.34602999687194824,
|
| 4375 |
+
"learning_rate": 0.00047293634137810425,
|
| 4376 |
+
"loss": 1.0687,
|
| 4377 |
+
"step": 624
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.0546268982847154,
|
| 4381 |
+
"grad_norm": 0.3194156587123871,
|
| 4382 |
+
"learning_rate": 0.0004728926197971319,
|
| 4383 |
+
"loss": 0.9222,
|
| 4384 |
+
"step": 625
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.054714301321970936,
|
| 4388 |
+
"grad_norm": 0.34864407777786255,
|
| 4389 |
+
"learning_rate": 0.0004728488982161595,
|
| 4390 |
+
"loss": 1.1291,
|
| 4391 |
+
"step": 626
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.05480170435922648,
|
| 4395 |
+
"grad_norm": 0.27222639322280884,
|
| 4396 |
+
"learning_rate": 0.00047280517663518713,
|
| 4397 |
+
"loss": 0.9762,
|
| 4398 |
+
"step": 627
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.05488910739648203,
|
| 4402 |
+
"grad_norm": 0.289035826921463,
|
| 4403 |
+
"learning_rate": 0.0004727614550542148,
|
| 4404 |
+
"loss": 0.84,
|
| 4405 |
+
"step": 628
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.05497651043373757,
|
| 4409 |
+
"grad_norm": 1.1678911447525024,
|
| 4410 |
+
"learning_rate": 0.0004727177334732424,
|
| 4411 |
+
"loss": 0.8835,
|
| 4412 |
+
"step": 629
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.05506391347099312,
|
| 4416 |
+
"grad_norm": 0.32149800658226013,
|
| 4417 |
+
"learning_rate": 0.00047267401189227007,
|
| 4418 |
+
"loss": 0.8814,
|
| 4419 |
+
"step": 630
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.055151316508248664,
|
| 4423 |
+
"grad_norm": 0.3312610387802124,
|
| 4424 |
+
"learning_rate": 0.00047263029031129766,
|
| 4425 |
+
"loss": 0.9001,
|
| 4426 |
+
"step": 631
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.0552387195455042,
|
| 4430 |
+
"grad_norm": 0.32734236121177673,
|
| 4431 |
+
"learning_rate": 0.0004725865687303253,
|
| 4432 |
+
"loss": 0.6587,
|
| 4433 |
+
"step": 632
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.05532612258275975,
|
| 4437 |
+
"grad_norm": 0.780978798866272,
|
| 4438 |
+
"learning_rate": 0.0004725428471493529,
|
| 4439 |
+
"loss": 1.1513,
|
| 4440 |
+
"step": 633
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.055413525620015294,
|
| 4444 |
+
"grad_norm": 0.3088547885417938,
|
| 4445 |
+
"learning_rate": 0.00047249912556838054,
|
| 4446 |
+
"loss": 0.8629,
|
| 4447 |
+
"step": 634
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.05550092865727084,
|
| 4451 |
+
"grad_norm": 0.34646108746528625,
|
| 4452 |
+
"learning_rate": 0.00047245540398740824,
|
| 4453 |
+
"loss": 0.8972,
|
| 4454 |
+
"step": 635
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.055588331694526386,
|
| 4458 |
+
"grad_norm": 0.47034963965415955,
|
| 4459 |
+
"learning_rate": 0.00047241168240643583,
|
| 4460 |
+
"loss": 1.414,
|
| 4461 |
+
"step": 636
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.05567573473178193,
|
| 4465 |
+
"grad_norm": 0.3200039565563202,
|
| 4466 |
+
"learning_rate": 0.0004723679608254635,
|
| 4467 |
+
"loss": 1.0516,
|
| 4468 |
+
"step": 637
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.05576313776903748,
|
| 4472 |
+
"grad_norm": 0.3332134187221527,
|
| 4473 |
+
"learning_rate": 0.00047232423924449107,
|
| 4474 |
+
"loss": 0.9086,
|
| 4475 |
+
"step": 638
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.055850540806293016,
|
| 4479 |
+
"grad_norm": 0.4804655611515045,
|
| 4480 |
+
"learning_rate": 0.0004722805176635187,
|
| 4481 |
+
"loss": 0.9719,
|
| 4482 |
+
"step": 639
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.05593794384354856,
|
| 4486 |
+
"grad_norm": 0.3591998219490051,
|
| 4487 |
+
"learning_rate": 0.0004722367960825463,
|
| 4488 |
+
"loss": 0.7201,
|
| 4489 |
+
"step": 640
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.05602534688080411,
|
| 4493 |
+
"grad_norm": 0.3319551944732666,
|
| 4494 |
+
"learning_rate": 0.000472193074501574,
|
| 4495 |
+
"loss": 1.1264,
|
| 4496 |
+
"step": 641
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.05611274991805965,
|
| 4500 |
+
"grad_norm": 0.3312825858592987,
|
| 4501 |
+
"learning_rate": 0.00047214935292060165,
|
| 4502 |
+
"loss": 1.0482,
|
| 4503 |
+
"step": 642
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.0562001529553152,
|
| 4507 |
+
"grad_norm": 0.3713119328022003,
|
| 4508 |
+
"learning_rate": 0.00047210563133962924,
|
| 4509 |
+
"loss": 1.1576,
|
| 4510 |
+
"step": 643
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.056287555992570744,
|
| 4514 |
+
"grad_norm": 0.35899418592453003,
|
| 4515 |
+
"learning_rate": 0.0004720619097586569,
|
| 4516 |
+
"loss": 0.7906,
|
| 4517 |
+
"step": 644
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.05637495902982629,
|
| 4521 |
+
"grad_norm": 0.31557363271713257,
|
| 4522 |
+
"learning_rate": 0.0004720181881776845,
|
| 4523 |
+
"loss": 0.9632,
|
| 4524 |
+
"step": 645
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.05646236206708183,
|
| 4528 |
+
"grad_norm": 0.40129950642585754,
|
| 4529 |
+
"learning_rate": 0.00047197446659671217,
|
| 4530 |
+
"loss": 1.3243,
|
| 4531 |
+
"step": 646
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.056549765104337374,
|
| 4535 |
+
"grad_norm": 0.3548416495323181,
|
| 4536 |
+
"learning_rate": 0.0004719307450157398,
|
| 4537 |
+
"loss": 1.0228,
|
| 4538 |
+
"step": 647
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.05663716814159292,
|
| 4542 |
+
"grad_norm": 0.5984897017478943,
|
| 4543 |
+
"learning_rate": 0.0004718870234347674,
|
| 4544 |
+
"loss": 0.9532,
|
| 4545 |
+
"step": 648
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.056724571178848465,
|
| 4549 |
+
"grad_norm": 0.2719477117061615,
|
| 4550 |
+
"learning_rate": 0.00047184330185379505,
|
| 4551 |
+
"loss": 0.9909,
|
| 4552 |
+
"step": 649
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.05681197421610401,
|
| 4556 |
+
"grad_norm": 0.2690770626068115,
|
| 4557 |
+
"learning_rate": 0.00047179958027282264,
|
| 4558 |
+
"loss": 0.9754,
|
| 4559 |
+
"step": 650
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.05689937725335956,
|
| 4563 |
+
"grad_norm": 0.3287508189678192,
|
| 4564 |
+
"learning_rate": 0.0004717558586918503,
|
| 4565 |
+
"loss": 0.823,
|
| 4566 |
+
"step": 651
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.0569867802906151,
|
| 4570 |
+
"grad_norm": 0.6442591547966003,
|
| 4571 |
+
"learning_rate": 0.00047171213711087793,
|
| 4572 |
+
"loss": 1.1211,
|
| 4573 |
+
"step": 652
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.05707418332787064,
|
| 4577 |
+
"grad_norm": 0.3647923469543457,
|
| 4578 |
+
"learning_rate": 0.0004716684155299056,
|
| 4579 |
+
"loss": 0.8892,
|
| 4580 |
+
"step": 653
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.05716158636512619,
|
| 4584 |
+
"grad_norm": 0.3035934269428253,
|
| 4585 |
+
"learning_rate": 0.0004716246939489332,
|
| 4586 |
+
"loss": 0.9781,
|
| 4587 |
+
"step": 654
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.05724898940238173,
|
| 4591 |
+
"grad_norm": 0.2986050546169281,
|
| 4592 |
+
"learning_rate": 0.0004715809723679608,
|
| 4593 |
+
"loss": 0.873,
|
| 4594 |
+
"step": 655
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.05733639243963728,
|
| 4598 |
+
"grad_norm": 0.3101188540458679,
|
| 4599 |
+
"learning_rate": 0.00047153725078698846,
|
| 4600 |
+
"loss": 1.1788,
|
| 4601 |
+
"step": 656
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.057423795476892824,
|
| 4605 |
+
"grad_norm": 1.2602791786193848,
|
| 4606 |
+
"learning_rate": 0.0004714935292060161,
|
| 4607 |
+
"loss": 1.376,
|
| 4608 |
+
"step": 657
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.05751119851414837,
|
| 4612 |
+
"grad_norm": 0.374224454164505,
|
| 4613 |
+
"learning_rate": 0.00047144980762504375,
|
| 4614 |
+
"loss": 0.9379,
|
| 4615 |
+
"step": 658
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.05759860155140391,
|
| 4619 |
+
"grad_norm": 0.35825932025909424,
|
| 4620 |
+
"learning_rate": 0.0004714060860440714,
|
| 4621 |
+
"loss": 0.9601,
|
| 4622 |
+
"step": 659
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.057686004588659454,
|
| 4626 |
+
"grad_norm": 0.37547796964645386,
|
| 4627 |
+
"learning_rate": 0.000471362364463099,
|
| 4628 |
+
"loss": 1.5432,
|
| 4629 |
+
"step": 660
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.057773407625915,
|
| 4633 |
+
"grad_norm": 0.30925118923187256,
|
| 4634 |
+
"learning_rate": 0.00047131864288212663,
|
| 4635 |
+
"loss": 0.9129,
|
| 4636 |
+
"step": 661
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.057860810663170545,
|
| 4640 |
+
"grad_norm": 0.43315598368644714,
|
| 4641 |
+
"learning_rate": 0.0004712749213011542,
|
| 4642 |
+
"loss": 0.7993,
|
| 4643 |
+
"step": 662
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.05794821370042609,
|
| 4647 |
+
"grad_norm": 1.0459505319595337,
|
| 4648 |
+
"learning_rate": 0.0004712311997201819,
|
| 4649 |
+
"loss": 1.4232,
|
| 4650 |
+
"step": 663
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.058035616737681636,
|
| 4654 |
+
"grad_norm": 0.4363897740840912,
|
| 4655 |
+
"learning_rate": 0.0004711874781392095,
|
| 4656 |
+
"loss": 1.3812,
|
| 4657 |
+
"step": 664
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.05812301977493718,
|
| 4661 |
+
"grad_norm": 0.2475530058145523,
|
| 4662 |
+
"learning_rate": 0.00047114375655823716,
|
| 4663 |
+
"loss": 0.8574,
|
| 4664 |
+
"step": 665
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.05821042281219272,
|
| 4668 |
+
"grad_norm": 0.352760910987854,
|
| 4669 |
+
"learning_rate": 0.0004711000349772648,
|
| 4670 |
+
"loss": 1.1236,
|
| 4671 |
+
"step": 666
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.058297825849448266,
|
| 4675 |
+
"grad_norm": 0.5032192468643188,
|
| 4676 |
+
"learning_rate": 0.0004710563133962924,
|
| 4677 |
+
"loss": 1.1754,
|
| 4678 |
+
"step": 667
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.05838522888670381,
|
| 4682 |
+
"grad_norm": 0.35939404368400574,
|
| 4683 |
+
"learning_rate": 0.0004710125918153201,
|
| 4684 |
+
"loss": 0.963,
|
| 4685 |
+
"step": 668
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.05847263192395936,
|
| 4689 |
+
"grad_norm": 0.4467969834804535,
|
| 4690 |
+
"learning_rate": 0.0004709688702343477,
|
| 4691 |
+
"loss": 2.0293,
|
| 4692 |
+
"step": 669
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.0585600349612149,
|
| 4696 |
+
"grad_norm": 0.3420664966106415,
|
| 4697 |
+
"learning_rate": 0.00047092514865337533,
|
| 4698 |
+
"loss": 1.0342,
|
| 4699 |
+
"step": 670
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.05864743799847045,
|
| 4703 |
+
"grad_norm": 0.3728554844856262,
|
| 4704 |
+
"learning_rate": 0.0004708814270724029,
|
| 4705 |
+
"loss": 0.9747,
|
| 4706 |
+
"step": 671
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.058734841035725995,
|
| 4710 |
+
"grad_norm": 1.2405109405517578,
|
| 4711 |
+
"learning_rate": 0.00047083770549143057,
|
| 4712 |
+
"loss": 1.6034,
|
| 4713 |
+
"step": 672
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.05882224407298153,
|
| 4717 |
+
"grad_norm": 0.3643404543399811,
|
| 4718 |
+
"learning_rate": 0.0004707939839104582,
|
| 4719 |
+
"loss": 0.7948,
|
| 4720 |
+
"step": 673
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.05890964711023708,
|
| 4724 |
+
"grad_norm": 0.31262850761413574,
|
| 4725 |
+
"learning_rate": 0.00047075026232948586,
|
| 4726 |
+
"loss": 0.8154,
|
| 4727 |
+
"step": 674
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.058997050147492625,
|
| 4731 |
+
"grad_norm": 0.49073535203933716,
|
| 4732 |
+
"learning_rate": 0.0004707065407485135,
|
| 4733 |
+
"loss": 0.9082,
|
| 4734 |
+
"step": 675
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.05908445318474817,
|
| 4738 |
+
"grad_norm": 0.39412635564804077,
|
| 4739 |
+
"learning_rate": 0.0004706628191675411,
|
| 4740 |
+
"loss": 1.0025,
|
| 4741 |
+
"step": 676
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.059171856222003716,
|
| 4745 |
+
"grad_norm": 0.40831953287124634,
|
| 4746 |
+
"learning_rate": 0.00047061909758656874,
|
| 4747 |
+
"loss": 1.0005,
|
| 4748 |
+
"step": 677
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.05925925925925926,
|
| 4752 |
+
"grad_norm": 0.5391172766685486,
|
| 4753 |
+
"learning_rate": 0.0004705753760055964,
|
| 4754 |
+
"loss": 0.9031,
|
| 4755 |
+
"step": 678
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.05934666229651481,
|
| 4759 |
+
"grad_norm": 0.31176143884658813,
|
| 4760 |
+
"learning_rate": 0.000470531654424624,
|
| 4761 |
+
"loss": 0.9589,
|
| 4762 |
+
"step": 679
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.059434065333770346,
|
| 4766 |
+
"grad_norm": 0.4320748448371887,
|
| 4767 |
+
"learning_rate": 0.00047048793284365167,
|
| 4768 |
+
"loss": 1.0996,
|
| 4769 |
+
"step": 680
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.05952146837102589,
|
| 4773 |
+
"grad_norm": 0.4102902412414551,
|
| 4774 |
+
"learning_rate": 0.00047044421126267926,
|
| 4775 |
+
"loss": 2.0338,
|
| 4776 |
+
"step": 681
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.05960887140828144,
|
| 4780 |
+
"grad_norm": 0.36022135615348816,
|
| 4781 |
+
"learning_rate": 0.0004704004896817069,
|
| 4782 |
+
"loss": 0.9675,
|
| 4783 |
+
"step": 682
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.05969627444553698,
|
| 4787 |
+
"grad_norm": 0.34680843353271484,
|
| 4788 |
+
"learning_rate": 0.0004703567681007345,
|
| 4789 |
+
"loss": 0.8765,
|
| 4790 |
+
"step": 683
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.05978367748279253,
|
| 4794 |
+
"grad_norm": 0.29740166664123535,
|
| 4795 |
+
"learning_rate": 0.00047031304651976214,
|
| 4796 |
+
"loss": 1.0053,
|
| 4797 |
+
"step": 684
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.059871080520048074,
|
| 4801 |
+
"grad_norm": 0.31341496109962463,
|
| 4802 |
+
"learning_rate": 0.00047026932493878984,
|
| 4803 |
+
"loss": 1.0295,
|
| 4804 |
+
"step": 685
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.05995848355730361,
|
| 4808 |
+
"grad_norm": 2.076716184616089,
|
| 4809 |
+
"learning_rate": 0.00047022560335781743,
|
| 4810 |
+
"loss": 1.5646,
|
| 4811 |
+
"step": 686
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.06004588659455916,
|
| 4815 |
+
"grad_norm": 0.2896002531051636,
|
| 4816 |
+
"learning_rate": 0.0004701818817768451,
|
| 4817 |
+
"loss": 0.9136,
|
| 4818 |
+
"step": 687
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.060133289631814704,
|
| 4822 |
+
"grad_norm": 0.37143734097480774,
|
| 4823 |
+
"learning_rate": 0.00047013816019587267,
|
| 4824 |
+
"loss": 0.8871,
|
| 4825 |
+
"step": 688
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.06022069266907025,
|
| 4829 |
+
"grad_norm": 0.49429547786712646,
|
| 4830 |
+
"learning_rate": 0.0004700944386149003,
|
| 4831 |
+
"loss": 1.1602,
|
| 4832 |
+
"step": 689
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.060308095706325796,
|
| 4836 |
+
"grad_norm": 0.3905726671218872,
|
| 4837 |
+
"learning_rate": 0.0004700507170339279,
|
| 4838 |
+
"loss": 1.1543,
|
| 4839 |
+
"step": 690
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.06039549874358134,
|
| 4843 |
+
"grad_norm": 0.3924982249736786,
|
| 4844 |
+
"learning_rate": 0.0004700069954529556,
|
| 4845 |
+
"loss": 0.8275,
|
| 4846 |
+
"step": 691
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.06048290178083689,
|
| 4850 |
+
"grad_norm": 0.27903103828430176,
|
| 4851 |
+
"learning_rate": 0.00046996327387198325,
|
| 4852 |
+
"loss": 0.8494,
|
| 4853 |
+
"step": 692
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.060570304818092426,
|
| 4857 |
+
"grad_norm": 0.382907897233963,
|
| 4858 |
+
"learning_rate": 0.00046991955229101084,
|
| 4859 |
+
"loss": 0.9531,
|
| 4860 |
+
"step": 693
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.06065770785534797,
|
| 4864 |
+
"grad_norm": 0.37153640389442444,
|
| 4865 |
+
"learning_rate": 0.0004698758307100385,
|
| 4866 |
+
"loss": 0.9131,
|
| 4867 |
+
"step": 694
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.06074511089260352,
|
| 4871 |
+
"grad_norm": 0.3007877767086029,
|
| 4872 |
+
"learning_rate": 0.0004698321091290661,
|
| 4873 |
+
"loss": 0.9513,
|
| 4874 |
+
"step": 695
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.06083251392985906,
|
| 4878 |
+
"grad_norm": 0.2546001672744751,
|
| 4879 |
+
"learning_rate": 0.0004697883875480938,
|
| 4880 |
+
"loss": 0.944,
|
| 4881 |
+
"step": 696
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.06091991696711461,
|
| 4885 |
+
"grad_norm": 0.27665847539901733,
|
| 4886 |
+
"learning_rate": 0.0004697446659671214,
|
| 4887 |
+
"loss": 0.7422,
|
| 4888 |
+
"step": 697
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.061007320004370154,
|
| 4892 |
+
"grad_norm": 0.28401628136634827,
|
| 4893 |
+
"learning_rate": 0.000469700944386149,
|
| 4894 |
+
"loss": 0.8458,
|
| 4895 |
+
"step": 698
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.0610947230416257,
|
| 4899 |
+
"grad_norm": 0.5097898840904236,
|
| 4900 |
+
"learning_rate": 0.00046965722280517666,
|
| 4901 |
+
"loss": 1.0018,
|
| 4902 |
+
"step": 699
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.06118212607888124,
|
| 4906 |
+
"grad_norm": 0.44888317584991455,
|
| 4907 |
+
"learning_rate": 0.00046961350122420425,
|
| 4908 |
+
"loss": 1.1203,
|
| 4909 |
+
"step": 700
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.061269529116136784,
|
| 4913 |
+
"grad_norm": 0.25764307379722595,
|
| 4914 |
+
"learning_rate": 0.0004695697796432319,
|
| 4915 |
+
"loss": 1.0156,
|
| 4916 |
+
"step": 701
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.06135693215339233,
|
| 4920 |
+
"grad_norm": 0.31590837240219116,
|
| 4921 |
+
"learning_rate": 0.00046952605806225954,
|
| 4922 |
+
"loss": 0.8823,
|
| 4923 |
+
"step": 702
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.061444335190647875,
|
| 4927 |
+
"grad_norm": 0.6337835192680359,
|
| 4928 |
+
"learning_rate": 0.0004694823364812872,
|
| 4929 |
+
"loss": 1.1565,
|
| 4930 |
+
"step": 703
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.06153173822790342,
|
| 4934 |
+
"grad_norm": 0.34477898478507996,
|
| 4935 |
+
"learning_rate": 0.00046943861490031483,
|
| 4936 |
+
"loss": 0.7563,
|
| 4937 |
+
"step": 704
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.061619141265158967,
|
| 4941 |
+
"grad_norm": 0.39787057042121887,
|
| 4942 |
+
"learning_rate": 0.0004693948933193424,
|
| 4943 |
+
"loss": 0.9804,
|
| 4944 |
+
"step": 705
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.06170654430241451,
|
| 4948 |
+
"grad_norm": 0.28919321298599243,
|
| 4949 |
+
"learning_rate": 0.00046935117173837007,
|
| 4950 |
+
"loss": 1.0019,
|
| 4951 |
+
"step": 706
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.06179394733967005,
|
| 4955 |
+
"grad_norm": 0.25737130641937256,
|
| 4956 |
+
"learning_rate": 0.00046930745015739766,
|
| 4957 |
+
"loss": 0.8751,
|
| 4958 |
+
"step": 707
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.0618813503769256,
|
| 4962 |
+
"grad_norm": 0.2699412703514099,
|
| 4963 |
+
"learning_rate": 0.00046926372857642536,
|
| 4964 |
+
"loss": 0.8999,
|
| 4965 |
+
"step": 708
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.06196875341418114,
|
| 4969 |
+
"grad_norm": 0.2957920730113983,
|
| 4970 |
+
"learning_rate": 0.000469220006995453,
|
| 4971 |
+
"loss": 0.9083,
|
| 4972 |
+
"step": 709
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.06205615645143669,
|
| 4976 |
+
"grad_norm": 0.2826875150203705,
|
| 4977 |
+
"learning_rate": 0.0004691762854144806,
|
| 4978 |
+
"loss": 0.946,
|
| 4979 |
+
"step": 710
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.062143559488692234,
|
| 4983 |
+
"grad_norm": 0.29016223549842834,
|
| 4984 |
+
"learning_rate": 0.00046913256383350824,
|
| 4985 |
+
"loss": 0.8126,
|
| 4986 |
+
"step": 711
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.06223096252594778,
|
| 4990 |
+
"grad_norm": 0.3504863679409027,
|
| 4991 |
+
"learning_rate": 0.00046908884225253583,
|
| 4992 |
+
"loss": 0.9127,
|
| 4993 |
+
"step": 712
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.06231836556320332,
|
| 4997 |
+
"grad_norm": 0.2627776861190796,
|
| 4998 |
+
"learning_rate": 0.00046904512067156353,
|
| 4999 |
+
"loss": 0.9476,
|
| 5000 |
+
"step": 713
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.062405768600458864,
|
| 5004 |
+
"grad_norm": 0.3002050220966339,
|
| 5005 |
+
"learning_rate": 0.0004690013990905911,
|
| 5006 |
+
"loss": 0.9444,
|
| 5007 |
+
"step": 714
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.06249317163771441,
|
| 5011 |
+
"grad_norm": 0.8539018630981445,
|
| 5012 |
+
"learning_rate": 0.00046895767750961876,
|
| 5013 |
+
"loss": 0.8977,
|
| 5014 |
+
"step": 715
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.06258057467496995,
|
| 5018 |
+
"grad_norm": 0.25260186195373535,
|
| 5019 |
+
"learning_rate": 0.0004689139559286464,
|
| 5020 |
+
"loss": 0.9615,
|
| 5021 |
+
"step": 716
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.0626679777122255,
|
| 5025 |
+
"grad_norm": 0.25615084171295166,
|
| 5026 |
+
"learning_rate": 0.000468870234347674,
|
| 5027 |
+
"loss": 0.8912,
|
| 5028 |
+
"step": 717
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.06275538074948105,
|
| 5032 |
+
"grad_norm": 0.3263600170612335,
|
| 5033 |
+
"learning_rate": 0.00046882651276670164,
|
| 5034 |
+
"loss": 0.843,
|
| 5035 |
+
"step": 718
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.06284278378673659,
|
| 5039 |
+
"grad_norm": 0.5694889426231384,
|
| 5040 |
+
"learning_rate": 0.0004687827911857293,
|
| 5041 |
+
"loss": 1.1624,
|
| 5042 |
+
"step": 719
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.06293018682399214,
|
| 5046 |
+
"grad_norm": 0.3248819410800934,
|
| 5047 |
+
"learning_rate": 0.00046873906960475693,
|
| 5048 |
+
"loss": 0.9452,
|
| 5049 |
+
"step": 720
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.06301758986124768,
|
| 5053 |
+
"grad_norm": 0.40857037901878357,
|
| 5054 |
+
"learning_rate": 0.0004686953480237845,
|
| 5055 |
+
"loss": 0.9117,
|
| 5056 |
+
"step": 721
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.06310499289850323,
|
| 5060 |
+
"grad_norm": 0.3211118280887604,
|
| 5061 |
+
"learning_rate": 0.00046865162644281217,
|
| 5062 |
+
"loss": 0.794,
|
| 5063 |
+
"step": 722
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.06319239593575877,
|
| 5067 |
+
"grad_norm": 0.32386934757232666,
|
| 5068 |
+
"learning_rate": 0.0004686079048618398,
|
| 5069 |
+
"loss": 1.2288,
|
| 5070 |
+
"step": 723
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.0632797989730143,
|
| 5074 |
+
"grad_norm": 0.3044579029083252,
|
| 5075 |
+
"learning_rate": 0.00046856418328086746,
|
| 5076 |
+
"loss": 0.9187,
|
| 5077 |
+
"step": 724
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.06336720201026985,
|
| 5081 |
+
"grad_norm": 0.6175875067710876,
|
| 5082 |
+
"learning_rate": 0.0004685204616998951,
|
| 5083 |
+
"loss": 0.8695,
|
| 5084 |
+
"step": 725
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.0634546050475254,
|
| 5088 |
+
"grad_norm": 0.7931004166603088,
|
| 5089 |
+
"learning_rate": 0.0004684767401189227,
|
| 5090 |
+
"loss": 1.3616,
|
| 5091 |
+
"step": 726
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.06354200808478094,
|
| 5095 |
+
"grad_norm": 0.337348997592926,
|
| 5096 |
+
"learning_rate": 0.00046843301853795034,
|
| 5097 |
+
"loss": 0.8654,
|
| 5098 |
+
"step": 727
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.06362941112203649,
|
| 5102 |
+
"grad_norm": 0.4152870178222656,
|
| 5103 |
+
"learning_rate": 0.000468389296956978,
|
| 5104 |
+
"loss": 1.2349,
|
| 5105 |
+
"step": 728
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.06371681415929203,
|
| 5109 |
+
"grad_norm": 0.3474035859107971,
|
| 5110 |
+
"learning_rate": 0.0004683455753760056,
|
| 5111 |
+
"loss": 0.9225,
|
| 5112 |
+
"step": 729
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.06380421719654758,
|
| 5116 |
+
"grad_norm": 0.35225990414619446,
|
| 5117 |
+
"learning_rate": 0.0004683018537950333,
|
| 5118 |
+
"loss": 0.9248,
|
| 5119 |
+
"step": 730
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.06389162023380313,
|
| 5123 |
+
"grad_norm": 0.24920597672462463,
|
| 5124 |
+
"learning_rate": 0.00046825813221406087,
|
| 5125 |
+
"loss": 0.8138,
|
| 5126 |
+
"step": 731
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.06397902327105867,
|
| 5130 |
+
"grad_norm": 0.3522126376628876,
|
| 5131 |
+
"learning_rate": 0.0004682144106330885,
|
| 5132 |
+
"loss": 0.9314,
|
| 5133 |
+
"step": 732
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.06406642630831422,
|
| 5137 |
+
"grad_norm": 0.4510492980480194,
|
| 5138 |
+
"learning_rate": 0.0004681706890521161,
|
| 5139 |
+
"loss": 0.8733,
|
| 5140 |
+
"step": 733
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.06415382934556976,
|
| 5144 |
+
"grad_norm": 0.2538619935512543,
|
| 5145 |
+
"learning_rate": 0.00046812696747114375,
|
| 5146 |
+
"loss": 0.8893,
|
| 5147 |
+
"step": 734
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.06424123238282531,
|
| 5151 |
+
"grad_norm": 0.39753592014312744,
|
| 5152 |
+
"learning_rate": 0.0004680832458901714,
|
| 5153 |
+
"loss": 1.0493,
|
| 5154 |
+
"step": 735
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.06432863542008085,
|
| 5158 |
+
"grad_norm": 0.40073463320732117,
|
| 5159 |
+
"learning_rate": 0.00046803952430919904,
|
| 5160 |
+
"loss": 0.8895,
|
| 5161 |
+
"step": 736
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.06441603845733639,
|
| 5165 |
+
"grad_norm": 0.31110239028930664,
|
| 5166 |
+
"learning_rate": 0.0004679958027282267,
|
| 5167 |
+
"loss": 0.8689,
|
| 5168 |
+
"step": 737
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.06450344149459193,
|
| 5172 |
+
"grad_norm": 0.29956865310668945,
|
| 5173 |
+
"learning_rate": 0.0004679520811472543,
|
| 5174 |
+
"loss": 0.8385,
|
| 5175 |
+
"step": 738
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.06459084453184748,
|
| 5179 |
+
"grad_norm": 0.3735499382019043,
|
| 5180 |
+
"learning_rate": 0.0004679083595662819,
|
| 5181 |
+
"loss": 0.8552,
|
| 5182 |
+
"step": 739
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.06467824756910302,
|
| 5186 |
+
"grad_norm": 0.4668900966644287,
|
| 5187 |
+
"learning_rate": 0.0004678646379853095,
|
| 5188 |
+
"loss": 1.4957,
|
| 5189 |
+
"step": 740
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.06476565060635857,
|
| 5193 |
+
"grad_norm": 0.363799512386322,
|
| 5194 |
+
"learning_rate": 0.0004678209164043372,
|
| 5195 |
+
"loss": 1.0365,
|
| 5196 |
+
"step": 741
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.06485305364361411,
|
| 5200 |
+
"grad_norm": 0.3261052668094635,
|
| 5201 |
+
"learning_rate": 0.00046777719482336486,
|
| 5202 |
+
"loss": 0.8972,
|
| 5203 |
+
"step": 742
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.06494045668086966,
|
| 5207 |
+
"grad_norm": 0.27814945578575134,
|
| 5208 |
+
"learning_rate": 0.00046773347324239245,
|
| 5209 |
+
"loss": 0.8051,
|
| 5210 |
+
"step": 743
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.0650278597181252,
|
| 5214 |
+
"grad_norm": 0.37245509028434753,
|
| 5215 |
+
"learning_rate": 0.0004676897516614201,
|
| 5216 |
+
"loss": 0.9421,
|
| 5217 |
+
"step": 744
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.06511526275538075,
|
| 5221 |
+
"grad_norm": 0.2978193163871765,
|
| 5222 |
+
"learning_rate": 0.0004676460300804477,
|
| 5223 |
+
"loss": 0.8464,
|
| 5224 |
+
"step": 745
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.0652026657926363,
|
| 5228 |
+
"grad_norm": 0.41827908158302307,
|
| 5229 |
+
"learning_rate": 0.00046760230849947533,
|
| 5230 |
+
"loss": 1.3154,
|
| 5231 |
+
"step": 746
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.06529006882989184,
|
| 5235 |
+
"grad_norm": 0.28153055906295776,
|
| 5236 |
+
"learning_rate": 0.000467558586918503,
|
| 5237 |
+
"loss": 0.812,
|
| 5238 |
+
"step": 747
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.06537747186714739,
|
| 5242 |
+
"grad_norm": 0.3568740487098694,
|
| 5243 |
+
"learning_rate": 0.0004675148653375306,
|
| 5244 |
+
"loss": 0.9333,
|
| 5245 |
+
"step": 748
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.06546487490440293,
|
| 5249 |
+
"grad_norm": 0.5805249810218811,
|
| 5250 |
+
"learning_rate": 0.00046747114375655826,
|
| 5251 |
+
"loss": 1.3821,
|
| 5252 |
+
"step": 749
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.06555227794165848,
|
| 5256 |
+
"grad_norm": 0.30053797364234924,
|
| 5257 |
+
"learning_rate": 0.00046742742217558585,
|
| 5258 |
+
"loss": 0.9358,
|
| 5259 |
+
"step": 750
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.06563968097891401,
|
| 5263 |
+
"grad_norm": 0.3179711699485779,
|
| 5264 |
+
"learning_rate": 0.0004673837005946135,
|
| 5265 |
+
"loss": 0.9094,
|
| 5266 |
+
"step": 751
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.06572708401616956,
|
| 5270 |
+
"grad_norm": 0.2717473804950714,
|
| 5271 |
+
"learning_rate": 0.00046733997901364114,
|
| 5272 |
+
"loss": 0.7255,
|
| 5273 |
+
"step": 752
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.0658144870534251,
|
| 5277 |
+
"grad_norm": 0.24072229862213135,
|
| 5278 |
+
"learning_rate": 0.0004672962574326688,
|
| 5279 |
+
"loss": 1.1008,
|
| 5280 |
+
"step": 753
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.06590189009068065,
|
| 5284 |
+
"grad_norm": 0.3099074363708496,
|
| 5285 |
+
"learning_rate": 0.00046725253585169643,
|
| 5286 |
+
"loss": 0.8751,
|
| 5287 |
+
"step": 754
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.0659892931279362,
|
| 5291 |
+
"grad_norm": 0.31873032450675964,
|
| 5292 |
+
"learning_rate": 0.000467208814270724,
|
| 5293 |
+
"loss": 0.8932,
|
| 5294 |
+
"step": 755
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.06607669616519174,
|
| 5298 |
+
"grad_norm": 0.31468328833580017,
|
| 5299 |
+
"learning_rate": 0.00046716509268975167,
|
| 5300 |
+
"loss": 0.8792,
|
| 5301 |
+
"step": 756
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.06616409920244729,
|
| 5305 |
+
"grad_norm": 0.35658881068229675,
|
| 5306 |
+
"learning_rate": 0.00046712137110877926,
|
| 5307 |
+
"loss": 0.8955,
|
| 5308 |
+
"step": 757
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.06625150223970283,
|
| 5312 |
+
"grad_norm": 0.3107976019382477,
|
| 5313 |
+
"learning_rate": 0.00046707764952780696,
|
| 5314 |
+
"loss": 0.9174,
|
| 5315 |
+
"step": 758
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.06633890527695838,
|
| 5319 |
+
"grad_norm": 0.2277815192937851,
|
| 5320 |
+
"learning_rate": 0.0004670339279468346,
|
| 5321 |
+
"loss": 0.7611,
|
| 5322 |
+
"step": 759
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.06642630831421392,
|
| 5326 |
+
"grad_norm": 0.25561246275901794,
|
| 5327 |
+
"learning_rate": 0.0004669902063658622,
|
| 5328 |
+
"loss": 0.8041,
|
| 5329 |
+
"step": 760
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.06651371135146947,
|
| 5333 |
+
"grad_norm": 0.2826947271823883,
|
| 5334 |
+
"learning_rate": 0.00046694648478488984,
|
| 5335 |
+
"loss": 0.7732,
|
| 5336 |
+
"step": 761
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.06660111438872501,
|
| 5340 |
+
"grad_norm": 0.2515583038330078,
|
| 5341 |
+
"learning_rate": 0.00046690276320391743,
|
| 5342 |
+
"loss": 1.0321,
|
| 5343 |
+
"step": 762
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.06668851742598056,
|
| 5347 |
+
"grad_norm": 0.26518338918685913,
|
| 5348 |
+
"learning_rate": 0.0004668590416229451,
|
| 5349 |
+
"loss": 1.1347,
|
| 5350 |
+
"step": 763
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.06677592046323609,
|
| 5354 |
+
"grad_norm": 0.2963607609272003,
|
| 5355 |
+
"learning_rate": 0.0004668153200419727,
|
| 5356 |
+
"loss": 0.9982,
|
| 5357 |
+
"step": 764
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.06686332350049164,
|
| 5361 |
+
"grad_norm": 0.2876517176628113,
|
| 5362 |
+
"learning_rate": 0.00046677159846100037,
|
| 5363 |
+
"loss": 0.6918,
|
| 5364 |
+
"step": 765
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.06695072653774718,
|
| 5368 |
+
"grad_norm": 0.3714672923088074,
|
| 5369 |
+
"learning_rate": 0.000466727876880028,
|
| 5370 |
+
"loss": 0.9023,
|
| 5371 |
+
"step": 766
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.06703812957500273,
|
| 5375 |
+
"grad_norm": 0.3568623960018158,
|
| 5376 |
+
"learning_rate": 0.0004666841552990556,
|
| 5377 |
+
"loss": 0.8378,
|
| 5378 |
+
"step": 767
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.06712553261225827,
|
| 5382 |
+
"grad_norm": 0.4770544469356537,
|
| 5383 |
+
"learning_rate": 0.00046664043371808325,
|
| 5384 |
+
"loss": 1.0266,
|
| 5385 |
+
"step": 768
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.06721293564951382,
|
| 5389 |
+
"grad_norm": 0.2760886549949646,
|
| 5390 |
+
"learning_rate": 0.0004665967121371109,
|
| 5391 |
+
"loss": 0.8276,
|
| 5392 |
+
"step": 769
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.06730033868676936,
|
| 5396 |
+
"grad_norm": 0.31360816955566406,
|
| 5397 |
+
"learning_rate": 0.00046655299055613854,
|
| 5398 |
+
"loss": 0.8646,
|
| 5399 |
+
"step": 770
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.06738774172402491,
|
| 5403 |
+
"grad_norm": 0.3075156509876251,
|
| 5404 |
+
"learning_rate": 0.00046650926897516613,
|
| 5405 |
+
"loss": 1.1144,
|
| 5406 |
+
"step": 771
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.06747514476128046,
|
| 5410 |
+
"grad_norm": 0.3104390501976013,
|
| 5411 |
+
"learning_rate": 0.0004664655473941938,
|
| 5412 |
+
"loss": 0.8923,
|
| 5413 |
+
"step": 772
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.067562547798536,
|
| 5417 |
+
"grad_norm": 0.3964294493198395,
|
| 5418 |
+
"learning_rate": 0.0004664218258132214,
|
| 5419 |
+
"loss": 1.0969,
|
| 5420 |
+
"step": 773
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.06764995083579155,
|
| 5424 |
+
"grad_norm": 0.3698040843009949,
|
| 5425 |
+
"learning_rate": 0.000466378104232249,
|
| 5426 |
+
"loss": 0.9078,
|
| 5427 |
+
"step": 774
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.0677373538730471,
|
| 5431 |
+
"grad_norm": 0.28510838747024536,
|
| 5432 |
+
"learning_rate": 0.0004663343826512767,
|
| 5433 |
+
"loss": 1.0075,
|
| 5434 |
+
"step": 775
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.06782475691030264,
|
| 5438 |
+
"grad_norm": 0.25500908493995667,
|
| 5439 |
+
"learning_rate": 0.0004662906610703043,
|
| 5440 |
+
"loss": 0.8457,
|
| 5441 |
+
"step": 776
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.06791215994755818,
|
| 5445 |
+
"grad_norm": 0.27927708625793457,
|
| 5446 |
+
"learning_rate": 0.00046624693948933195,
|
| 5447 |
+
"loss": 1.01,
|
| 5448 |
+
"step": 777
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.06799956298481372,
|
| 5452 |
+
"grad_norm": 0.2683468461036682,
|
| 5453 |
+
"learning_rate": 0.0004662032179083596,
|
| 5454 |
+
"loss": 1.0491,
|
| 5455 |
+
"step": 778
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.06808696602206926,
|
| 5459 |
+
"grad_norm": 0.31843262910842896,
|
| 5460 |
+
"learning_rate": 0.0004661594963273872,
|
| 5461 |
+
"loss": 0.9467,
|
| 5462 |
+
"step": 779
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.06817436905932481,
|
| 5466 |
+
"grad_norm": 0.27564141154289246,
|
| 5467 |
+
"learning_rate": 0.0004661157747464149,
|
| 5468 |
+
"loss": 0.9487,
|
| 5469 |
+
"step": 780
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.06826177209658035,
|
| 5473 |
+
"grad_norm": 0.2407764047384262,
|
| 5474 |
+
"learning_rate": 0.00046607205316544247,
|
| 5475 |
+
"loss": 0.8939,
|
| 5476 |
+
"step": 781
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.0683491751338359,
|
| 5480 |
+
"grad_norm": 0.3025217652320862,
|
| 5481 |
+
"learning_rate": 0.0004660283315844701,
|
| 5482 |
+
"loss": 0.9859,
|
| 5483 |
+
"step": 782
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.06843657817109144,
|
| 5487 |
+
"grad_norm": 0.2979051470756531,
|
| 5488 |
+
"learning_rate": 0.0004659846100034977,
|
| 5489 |
+
"loss": 0.9136,
|
| 5490 |
+
"step": 783
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.06852398120834699,
|
| 5494 |
+
"grad_norm": 0.28788650035858154,
|
| 5495 |
+
"learning_rate": 0.00046594088842252535,
|
| 5496 |
+
"loss": 0.9734,
|
| 5497 |
+
"step": 784
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.06861138424560254,
|
| 5501 |
+
"grad_norm": 0.2947753667831421,
|
| 5502 |
+
"learning_rate": 0.000465897166841553,
|
| 5503 |
+
"loss": 0.735,
|
| 5504 |
+
"step": 785
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.06869878728285808,
|
| 5508 |
+
"grad_norm": 0.3203105032444,
|
| 5509 |
+
"learning_rate": 0.00046585344526058064,
|
| 5510 |
+
"loss": 0.8992,
|
| 5511 |
+
"step": 786
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.06878619032011363,
|
| 5515 |
+
"grad_norm": 0.2638401985168457,
|
| 5516 |
+
"learning_rate": 0.0004658097236796083,
|
| 5517 |
+
"loss": 0.8669,
|
| 5518 |
+
"step": 787
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.06887359335736917,
|
| 5522 |
+
"grad_norm": 0.26712629199028015,
|
| 5523 |
+
"learning_rate": 0.0004657660020986359,
|
| 5524 |
+
"loss": 0.9765,
|
| 5525 |
+
"step": 788
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.06896099639462472,
|
| 5529 |
+
"grad_norm": 0.4055823087692261,
|
| 5530 |
+
"learning_rate": 0.0004657222805176635,
|
| 5531 |
+
"loss": 0.8117,
|
| 5532 |
+
"step": 789
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.06904839943188026,
|
| 5536 |
+
"grad_norm": 0.2518852651119232,
|
| 5537 |
+
"learning_rate": 0.00046567855893669117,
|
| 5538 |
+
"loss": 0.9517,
|
| 5539 |
+
"step": 790
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.0691358024691358,
|
| 5543 |
+
"grad_norm": 0.27589836716651917,
|
| 5544 |
+
"learning_rate": 0.00046563483735571876,
|
| 5545 |
+
"loss": 0.7855,
|
| 5546 |
+
"step": 791
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.06922320550639134,
|
| 5550 |
+
"grad_norm": 0.2739314138889313,
|
| 5551 |
+
"learning_rate": 0.00046559111577474646,
|
| 5552 |
+
"loss": 0.8862,
|
| 5553 |
+
"step": 792
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.06931060854364689,
|
| 5557 |
+
"grad_norm": 0.3271756172180176,
|
| 5558 |
+
"learning_rate": 0.00046554739419377405,
|
| 5559 |
+
"loss": 1.2893,
|
| 5560 |
+
"step": 793
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.06939801158090243,
|
| 5564 |
+
"grad_norm": 0.27038949728012085,
|
| 5565 |
+
"learning_rate": 0.0004655036726128017,
|
| 5566 |
+
"loss": 0.8059,
|
| 5567 |
+
"step": 794
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.06948541461815798,
|
| 5571 |
+
"grad_norm": 0.2605447471141815,
|
| 5572 |
+
"learning_rate": 0.0004654599510318293,
|
| 5573 |
+
"loss": 0.8816,
|
| 5574 |
+
"step": 795
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.06957281765541352,
|
| 5578 |
+
"grad_norm": 0.2714409828186035,
|
| 5579 |
+
"learning_rate": 0.00046541622945085693,
|
| 5580 |
+
"loss": 0.9307,
|
| 5581 |
+
"step": 796
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.06966022069266907,
|
| 5585 |
+
"grad_norm": 0.2455201894044876,
|
| 5586 |
+
"learning_rate": 0.00046537250786988463,
|
| 5587 |
+
"loss": 0.8321,
|
| 5588 |
+
"step": 797
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.06974762372992462,
|
| 5592 |
+
"grad_norm": 0.29036253690719604,
|
| 5593 |
+
"learning_rate": 0.0004653287862889122,
|
| 5594 |
+
"loss": 0.8605,
|
| 5595 |
+
"step": 798
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.06983502676718016,
|
| 5599 |
+
"grad_norm": 0.24069538712501526,
|
| 5600 |
+
"learning_rate": 0.00046528506470793987,
|
| 5601 |
+
"loss": 1.0819,
|
| 5602 |
+
"step": 799
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.0699224298044357,
|
| 5606 |
+
"grad_norm": 0.254304975271225,
|
| 5607 |
+
"learning_rate": 0.00046524134312696746,
|
| 5608 |
+
"loss": 0.7388,
|
| 5609 |
+
"step": 800
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 0.07000983284169125,
|
| 5613 |
+
"grad_norm": 0.27309149503707886,
|
| 5614 |
+
"learning_rate": 0.0004651976215459951,
|
| 5615 |
+
"loss": 0.7796,
|
| 5616 |
+
"step": 801
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 0.0700972358789468,
|
| 5620 |
+
"grad_norm": 0.26903948187828064,
|
| 5621 |
+
"learning_rate": 0.0004651538999650227,
|
| 5622 |
+
"loss": 1.0103,
|
| 5623 |
+
"step": 802
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 0.07018463891620234,
|
| 5627 |
+
"grad_norm": 0.2526533901691437,
|
| 5628 |
+
"learning_rate": 0.0004651101783840504,
|
| 5629 |
+
"loss": 0.8566,
|
| 5630 |
+
"step": 803
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 0.07027204195345789,
|
| 5634 |
+
"grad_norm": 0.2822379469871521,
|
| 5635 |
+
"learning_rate": 0.00046506645680307804,
|
| 5636 |
+
"loss": 0.9441,
|
| 5637 |
+
"step": 804
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 0.07035944499071342,
|
| 5641 |
+
"grad_norm": 0.27883851528167725,
|
| 5642 |
+
"learning_rate": 0.00046502273522210563,
|
| 5643 |
+
"loss": 0.9006,
|
| 5644 |
+
"step": 805
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.07044684802796897,
|
| 5648 |
+
"grad_norm": 0.23839306831359863,
|
| 5649 |
+
"learning_rate": 0.0004649790136411333,
|
| 5650 |
+
"loss": 0.8387,
|
| 5651 |
+
"step": 806
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 0.07053425106522451,
|
| 5655 |
+
"grad_norm": 0.2352200597524643,
|
| 5656 |
+
"learning_rate": 0.00046493529206016087,
|
| 5657 |
+
"loss": 0.8228,
|
| 5658 |
+
"step": 807
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 0.07062165410248006,
|
| 5662 |
+
"grad_norm": 0.31958913803100586,
|
| 5663 |
+
"learning_rate": 0.00046489157047918857,
|
| 5664 |
+
"loss": 1.0312,
|
| 5665 |
+
"step": 808
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 0.0707090571397356,
|
| 5669 |
+
"grad_norm": 0.286045640707016,
|
| 5670 |
+
"learning_rate": 0.0004648478488982162,
|
| 5671 |
+
"loss": 0.8427,
|
| 5672 |
+
"step": 809
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 0.07079646017699115,
|
| 5676 |
+
"grad_norm": 0.24101607501506805,
|
| 5677 |
+
"learning_rate": 0.0004648041273172438,
|
| 5678 |
+
"loss": 0.9986,
|
| 5679 |
+
"step": 810
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 0.0708838632142467,
|
| 5683 |
+
"grad_norm": 0.28324073553085327,
|
| 5684 |
+
"learning_rate": 0.00046476040573627145,
|
| 5685 |
+
"loss": 0.778,
|
| 5686 |
+
"step": 811
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.07097126625150224,
|
| 5690 |
+
"grad_norm": 0.30368572473526,
|
| 5691 |
+
"learning_rate": 0.00046471668415529904,
|
| 5692 |
+
"loss": 0.9543,
|
| 5693 |
+
"step": 812
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 0.07105866928875779,
|
| 5697 |
+
"grad_norm": 0.3159104585647583,
|
| 5698 |
+
"learning_rate": 0.0004646729625743267,
|
| 5699 |
+
"loss": 0.9481,
|
| 5700 |
+
"step": 813
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 0.07114607232601333,
|
| 5704 |
+
"grad_norm": 0.2856074869632721,
|
| 5705 |
+
"learning_rate": 0.00046462924099335433,
|
| 5706 |
+
"loss": 1.0117,
|
| 5707 |
+
"step": 814
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 0.07123347536326888,
|
| 5711 |
+
"grad_norm": 0.32605329155921936,
|
| 5712 |
+
"learning_rate": 0.00046458551941238197,
|
| 5713 |
+
"loss": 0.8451,
|
| 5714 |
+
"step": 815
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 0.07132087840052442,
|
| 5718 |
+
"grad_norm": 0.22008907794952393,
|
| 5719 |
+
"learning_rate": 0.0004645417978314096,
|
| 5720 |
+
"loss": 0.8965,
|
| 5721 |
+
"step": 816
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 0.07140828143777997,
|
| 5725 |
+
"grad_norm": 0.26317551732063293,
|
| 5726 |
+
"learning_rate": 0.0004644980762504372,
|
| 5727 |
+
"loss": 0.8644,
|
| 5728 |
+
"step": 817
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.0714956844750355,
|
| 5732 |
+
"grad_norm": 0.22049389779567719,
|
| 5733 |
+
"learning_rate": 0.00046445435466946485,
|
| 5734 |
+
"loss": 0.8144,
|
| 5735 |
+
"step": 818
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 0.07158308751229105,
|
| 5739 |
+
"grad_norm": 0.2786102890968323,
|
| 5740 |
+
"learning_rate": 0.00046441063308849244,
|
| 5741 |
+
"loss": 0.8841,
|
| 5742 |
+
"step": 819
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 0.07167049054954659,
|
| 5746 |
+
"grad_norm": 0.31796136498451233,
|
| 5747 |
+
"learning_rate": 0.00046436691150752014,
|
| 5748 |
+
"loss": 1.0665,
|
| 5749 |
+
"step": 820
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 0.07175789358680214,
|
| 5753 |
+
"grad_norm": 0.29958993196487427,
|
| 5754 |
+
"learning_rate": 0.0004643231899265478,
|
| 5755 |
+
"loss": 0.8789,
|
| 5756 |
+
"step": 821
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 0.07184529662405768,
|
| 5760 |
+
"grad_norm": 0.2706652283668518,
|
| 5761 |
+
"learning_rate": 0.0004642794683455754,
|
| 5762 |
+
"loss": 0.8721,
|
| 5763 |
+
"step": 822
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 0.07193269966131323,
|
| 5767 |
+
"grad_norm": 0.22537319362163544,
|
| 5768 |
+
"learning_rate": 0.000464235746764603,
|
| 5769 |
+
"loss": 0.9403,
|
| 5770 |
+
"step": 823
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.07202010269856877,
|
| 5774 |
+
"grad_norm": 0.34331005811691284,
|
| 5775 |
+
"learning_rate": 0.0004641920251836306,
|
| 5776 |
+
"loss": 1.1497,
|
| 5777 |
+
"step": 824
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.07210750573582432,
|
| 5781 |
+
"grad_norm": 0.25914907455444336,
|
| 5782 |
+
"learning_rate": 0.0004641483036026583,
|
| 5783 |
+
"loss": 1.1589,
|
| 5784 |
+
"step": 825
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.07219490877307987,
|
| 5788 |
+
"grad_norm": 0.2956130802631378,
|
| 5789 |
+
"learning_rate": 0.0004641045820216859,
|
| 5790 |
+
"loss": 0.8587,
|
| 5791 |
+
"step": 826
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.07228231181033541,
|
| 5795 |
+
"grad_norm": 0.30292391777038574,
|
| 5796 |
+
"learning_rate": 0.00046406086044071355,
|
| 5797 |
+
"loss": 0.9224,
|
| 5798 |
+
"step": 827
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.07236971484759096,
|
| 5802 |
+
"grad_norm": 0.3101223409175873,
|
| 5803 |
+
"learning_rate": 0.0004640171388597412,
|
| 5804 |
+
"loss": 0.9115,
|
| 5805 |
+
"step": 828
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 0.0724571178848465,
|
| 5809 |
+
"grad_norm": 0.2720979154109955,
|
| 5810 |
+
"learning_rate": 0.0004639734172787688,
|
| 5811 |
+
"loss": 0.8112,
|
| 5812 |
+
"step": 829
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.07254452092210205,
|
| 5816 |
+
"grad_norm": 0.2774461507797241,
|
| 5817 |
+
"learning_rate": 0.00046392969569779643,
|
| 5818 |
+
"loss": 0.9776,
|
| 5819 |
+
"step": 830
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 0.0726319239593576,
|
| 5823 |
+
"grad_norm": 0.25150200724601746,
|
| 5824 |
+
"learning_rate": 0.0004638859741168241,
|
| 5825 |
+
"loss": 1.0255,
|
| 5826 |
+
"step": 831
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 0.07271932699661313,
|
| 5830 |
+
"grad_norm": 0.2526938319206238,
|
| 5831 |
+
"learning_rate": 0.0004638422525358517,
|
| 5832 |
+
"loss": 0.7242,
|
| 5833 |
+
"step": 832
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 0.07280673003386867,
|
| 5837 |
+
"grad_norm": 0.29642441868782043,
|
| 5838 |
+
"learning_rate": 0.0004637985309548793,
|
| 5839 |
+
"loss": 1.0944,
|
| 5840 |
+
"step": 833
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 0.07289413307112422,
|
| 5844 |
+
"grad_norm": 0.250478595495224,
|
| 5845 |
+
"learning_rate": 0.00046375480937390696,
|
| 5846 |
+
"loss": 0.8324,
|
| 5847 |
+
"step": 834
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 0.07298153610837976,
|
| 5851 |
+
"grad_norm": 0.28843697905540466,
|
| 5852 |
+
"learning_rate": 0.0004637110877929346,
|
| 5853 |
+
"loss": 0.8646,
|
| 5854 |
+
"step": 835
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.07306893914563531,
|
| 5858 |
+
"grad_norm": 0.22244645655155182,
|
| 5859 |
+
"learning_rate": 0.00046366736621196225,
|
| 5860 |
+
"loss": 0.7966,
|
| 5861 |
+
"step": 836
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 0.07315634218289085,
|
| 5865 |
+
"grad_norm": 0.2418157458305359,
|
| 5866 |
+
"learning_rate": 0.0004636236446309899,
|
| 5867 |
+
"loss": 0.8101,
|
| 5868 |
+
"step": 837
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 0.0732437452201464,
|
| 5872 |
+
"grad_norm": 0.2781657874584198,
|
| 5873 |
+
"learning_rate": 0.0004635799230500175,
|
| 5874 |
+
"loss": 0.9902,
|
| 5875 |
+
"step": 838
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 0.07333114825740195,
|
| 5879 |
+
"grad_norm": 0.24249030649662018,
|
| 5880 |
+
"learning_rate": 0.00046353620146904513,
|
| 5881 |
+
"loss": 0.7445,
|
| 5882 |
+
"step": 839
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 0.07341855129465749,
|
| 5886 |
+
"grad_norm": 0.23980437219142914,
|
| 5887 |
+
"learning_rate": 0.0004634924798880728,
|
| 5888 |
+
"loss": 0.8168,
|
| 5889 |
+
"step": 840
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 0.07350595433191304,
|
| 5893 |
+
"grad_norm": 0.3362947106361389,
|
| 5894 |
+
"learning_rate": 0.00046344875830710037,
|
| 5895 |
+
"loss": 1.1176,
|
| 5896 |
+
"step": 841
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.07359335736916858,
|
| 5900 |
+
"grad_norm": 0.23380422592163086,
|
| 5901 |
+
"learning_rate": 0.00046340503672612807,
|
| 5902 |
+
"loss": 0.8311,
|
| 5903 |
+
"step": 842
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 0.07368076040642413,
|
| 5907 |
+
"grad_norm": 0.2908138632774353,
|
| 5908 |
+
"learning_rate": 0.00046336131514515566,
|
| 5909 |
+
"loss": 0.8315,
|
| 5910 |
+
"step": 843
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 0.07376816344367967,
|
| 5914 |
+
"grad_norm": 0.2556897699832916,
|
| 5915 |
+
"learning_rate": 0.0004633175935641833,
|
| 5916 |
+
"loss": 0.939,
|
| 5917 |
+
"step": 844
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 0.0738555664809352,
|
| 5921 |
+
"grad_norm": 0.3416728079319,
|
| 5922 |
+
"learning_rate": 0.0004632738719832109,
|
| 5923 |
+
"loss": 0.746,
|
| 5924 |
+
"step": 845
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 0.07394296951819075,
|
| 5928 |
+
"grad_norm": 0.2219434678554535,
|
| 5929 |
+
"learning_rate": 0.00046323015040223854,
|
| 5930 |
+
"loss": 1.0259,
|
| 5931 |
+
"step": 846
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 0.0740303725554463,
|
| 5935 |
+
"grad_norm": 0.3327368497848511,
|
| 5936 |
+
"learning_rate": 0.0004631864288212662,
|
| 5937 |
+
"loss": 1.4831,
|
| 5938 |
+
"step": 847
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.07411777559270184,
|
| 5942 |
+
"grad_norm": 0.28128185868263245,
|
| 5943 |
+
"learning_rate": 0.00046314270724029383,
|
| 5944 |
+
"loss": 0.9478,
|
| 5945 |
+
"step": 848
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.07420517862995739,
|
| 5949 |
+
"grad_norm": 0.29582032561302185,
|
| 5950 |
+
"learning_rate": 0.00046309898565932147,
|
| 5951 |
+
"loss": 0.9397,
|
| 5952 |
+
"step": 849
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.07429258166721293,
|
| 5956 |
+
"grad_norm": 0.26146262884140015,
|
| 5957 |
+
"learning_rate": 0.00046305526407834906,
|
| 5958 |
+
"loss": 0.6904,
|
| 5959 |
+
"step": 850
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.07437998470446848,
|
| 5963 |
+
"grad_norm": 0.3188638389110565,
|
| 5964 |
+
"learning_rate": 0.0004630115424973767,
|
| 5965 |
+
"loss": 0.7268,
|
| 5966 |
+
"step": 851
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.07446738774172403,
|
| 5970 |
+
"grad_norm": 0.2691085934638977,
|
| 5971 |
+
"learning_rate": 0.0004629678209164043,
|
| 5972 |
+
"loss": 0.7836,
|
| 5973 |
+
"step": 852
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.07455479077897957,
|
| 5977 |
+
"grad_norm": 0.2730037569999695,
|
| 5978 |
+
"learning_rate": 0.000462924099335432,
|
| 5979 |
+
"loss": 0.8207,
|
| 5980 |
+
"step": 853
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.07464219381623512,
|
| 5984 |
+
"grad_norm": 0.23849952220916748,
|
| 5985 |
+
"learning_rate": 0.00046288037775445964,
|
| 5986 |
+
"loss": 0.9859,
|
| 5987 |
+
"step": 854
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.07472959685349066,
|
| 5991 |
+
"grad_norm": 0.24940194189548492,
|
| 5992 |
+
"learning_rate": 0.00046283665617348723,
|
| 5993 |
+
"loss": 0.7821,
|
| 5994 |
+
"step": 855
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.07481699989074621,
|
| 5998 |
+
"grad_norm": 0.23495396971702576,
|
| 5999 |
+
"learning_rate": 0.0004627929345925149,
|
| 6000 |
+
"loss": 0.8847,
|
| 6001 |
+
"step": 856
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.07490440292800175,
|
| 6005 |
+
"grad_norm": 0.25201091170310974,
|
| 6006 |
+
"learning_rate": 0.00046274921301154247,
|
| 6007 |
+
"loss": 0.8386,
|
| 6008 |
+
"step": 857
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.0749918059652573,
|
| 6012 |
+
"grad_norm": 0.25054988265037537,
|
| 6013 |
+
"learning_rate": 0.0004627054914305701,
|
| 6014 |
+
"loss": 0.9939,
|
| 6015 |
+
"step": 858
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.07507920900251283,
|
| 6019 |
+
"grad_norm": 0.39931726455688477,
|
| 6020 |
+
"learning_rate": 0.0004626617698495978,
|
| 6021 |
+
"loss": 1.1039,
|
| 6022 |
+
"step": 859
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.07516661203976838,
|
| 6026 |
+
"grad_norm": 0.2789982855319977,
|
| 6027 |
+
"learning_rate": 0.0004626180482686254,
|
| 6028 |
+
"loss": 1.1707,
|
| 6029 |
+
"step": 860
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.07525401507702392,
|
| 6033 |
+
"grad_norm": 0.282528817653656,
|
| 6034 |
+
"learning_rate": 0.00046257432668765305,
|
| 6035 |
+
"loss": 0.8738,
|
| 6036 |
+
"step": 861
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.07534141811427947,
|
| 6040 |
+
"grad_norm": 0.2707865536212921,
|
| 6041 |
+
"learning_rate": 0.00046253060510668064,
|
| 6042 |
+
"loss": 0.832,
|
| 6043 |
+
"step": 862
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.07542882115153501,
|
| 6047 |
+
"grad_norm": 0.19732601940631866,
|
| 6048 |
+
"learning_rate": 0.0004624868835257083,
|
| 6049 |
+
"loss": 0.8948,
|
| 6050 |
+
"step": 863
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.07551622418879056,
|
| 6054 |
+
"grad_norm": 0.2605394721031189,
|
| 6055 |
+
"learning_rate": 0.00046244316194473593,
|
| 6056 |
+
"loss": 0.7346,
|
| 6057 |
+
"step": 864
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.0756036272260461,
|
| 6061 |
+
"grad_norm": 0.26202288269996643,
|
| 6062 |
+
"learning_rate": 0.0004623994403637636,
|
| 6063 |
+
"loss": 0.8521,
|
| 6064 |
+
"step": 865
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.07569103026330165,
|
| 6068 |
+
"grad_norm": 0.3473947048187256,
|
| 6069 |
+
"learning_rate": 0.0004623557187827912,
|
| 6070 |
+
"loss": 1.043,
|
| 6071 |
+
"step": 866
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.0757784333005572,
|
| 6075 |
+
"grad_norm": 0.7824636697769165,
|
| 6076 |
+
"learning_rate": 0.0004623119972018188,
|
| 6077 |
+
"loss": 1.2121,
|
| 6078 |
+
"step": 867
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.07586583633781274,
|
| 6082 |
+
"grad_norm": 0.26076897978782654,
|
| 6083 |
+
"learning_rate": 0.00046226827562084646,
|
| 6084 |
+
"loss": 0.8669,
|
| 6085 |
+
"step": 868
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.07595323937506829,
|
| 6089 |
+
"grad_norm": 0.3360956013202667,
|
| 6090 |
+
"learning_rate": 0.00046222455403987405,
|
| 6091 |
+
"loss": 0.8806,
|
| 6092 |
+
"step": 869
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.07604064241232383,
|
| 6096 |
+
"grad_norm": 0.27572354674339294,
|
| 6097 |
+
"learning_rate": 0.00046218083245890175,
|
| 6098 |
+
"loss": 0.8105,
|
| 6099 |
+
"step": 870
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.07612804544957938,
|
| 6103 |
+
"grad_norm": 0.22802734375,
|
| 6104 |
+
"learning_rate": 0.0004621371108779294,
|
| 6105 |
+
"loss": 0.6879,
|
| 6106 |
+
"step": 871
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.07621544848683491,
|
| 6110 |
+
"grad_norm": 0.31544265151023865,
|
| 6111 |
+
"learning_rate": 0.000462093389296957,
|
| 6112 |
+
"loss": 0.835,
|
| 6113 |
+
"step": 872
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.07630285152409046,
|
| 6117 |
+
"grad_norm": 0.3530902564525604,
|
| 6118 |
+
"learning_rate": 0.00046204966771598463,
|
| 6119 |
+
"loss": 0.7543,
|
| 6120 |
+
"step": 873
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.076390254561346,
|
| 6124 |
+
"grad_norm": 0.28108978271484375,
|
| 6125 |
+
"learning_rate": 0.0004620059461350122,
|
| 6126 |
+
"loss": 0.9433,
|
| 6127 |
+
"step": 874
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.07647765759860155,
|
| 6131 |
+
"grad_norm": 0.2918491065502167,
|
| 6132 |
+
"learning_rate": 0.00046196222455403987,
|
| 6133 |
+
"loss": 0.9016,
|
| 6134 |
+
"step": 875
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.0765650606358571,
|
| 6138 |
+
"grad_norm": 0.3130475580692291,
|
| 6139 |
+
"learning_rate": 0.0004619185029730675,
|
| 6140 |
+
"loss": 0.8612,
|
| 6141 |
+
"step": 876
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.07665246367311264,
|
| 6145 |
+
"grad_norm": 0.2697352468967438,
|
| 6146 |
+
"learning_rate": 0.00046187478139209516,
|
| 6147 |
+
"loss": 1.0324,
|
| 6148 |
+
"step": 877
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.07673986671036818,
|
| 6152 |
+
"grad_norm": 0.3534733057022095,
|
| 6153 |
+
"learning_rate": 0.0004618310598111228,
|
| 6154 |
+
"loss": 0.7769,
|
| 6155 |
+
"step": 878
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.07682726974762373,
|
| 6159 |
+
"grad_norm": 0.46239951252937317,
|
| 6160 |
+
"learning_rate": 0.0004617873382301504,
|
| 6161 |
+
"loss": 0.8155,
|
| 6162 |
+
"step": 879
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.07691467278487928,
|
| 6166 |
+
"grad_norm": 0.2869885265827179,
|
| 6167 |
+
"learning_rate": 0.00046174361664917804,
|
| 6168 |
+
"loss": 0.8088,
|
| 6169 |
+
"step": 880
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.07700207582213482,
|
| 6173 |
+
"grad_norm": 0.544746458530426,
|
| 6174 |
+
"learning_rate": 0.0004616998950682057,
|
| 6175 |
+
"loss": 1.0332,
|
| 6176 |
+
"step": 881
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.07708947885939037,
|
| 6180 |
+
"grad_norm": 0.28001531958580017,
|
| 6181 |
+
"learning_rate": 0.0004616561734872333,
|
| 6182 |
+
"loss": 0.8363,
|
| 6183 |
+
"step": 882
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.07717688189664591,
|
| 6187 |
+
"grad_norm": 0.244185671210289,
|
| 6188 |
+
"learning_rate": 0.0004616124519062609,
|
| 6189 |
+
"loss": 0.8611,
|
| 6190 |
+
"step": 883
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.07726428493390146,
|
| 6194 |
+
"grad_norm": 0.3561322093009949,
|
| 6195 |
+
"learning_rate": 0.00046156873032528856,
|
| 6196 |
+
"loss": 0.9298,
|
| 6197 |
+
"step": 884
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.077351687971157,
|
| 6201 |
+
"grad_norm": 0.2852579355239868,
|
| 6202 |
+
"learning_rate": 0.0004615250087443162,
|
| 6203 |
+
"loss": 0.9415,
|
| 6204 |
+
"step": 885
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.07743909100841254,
|
| 6208 |
+
"grad_norm": 0.3458700180053711,
|
| 6209 |
+
"learning_rate": 0.0004614812871633438,
|
| 6210 |
+
"loss": 0.7855,
|
| 6211 |
+
"step": 886
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.07752649404566808,
|
| 6215 |
+
"grad_norm": 0.33211758732795715,
|
| 6216 |
+
"learning_rate": 0.0004614375655823715,
|
| 6217 |
+
"loss": 0.7652,
|
| 6218 |
+
"step": 887
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.07761389708292363,
|
| 6222 |
+
"grad_norm": 0.2643268406391144,
|
| 6223 |
+
"learning_rate": 0.0004613938440013991,
|
| 6224 |
+
"loss": 0.813,
|
| 6225 |
+
"step": 888
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.07770130012017917,
|
| 6229 |
+
"grad_norm": 0.26717138290405273,
|
| 6230 |
+
"learning_rate": 0.00046135012242042673,
|
| 6231 |
+
"loss": 0.673,
|
| 6232 |
+
"step": 889
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.07778870315743472,
|
| 6236 |
+
"grad_norm": 0.2716834843158722,
|
| 6237 |
+
"learning_rate": 0.0004613064008394544,
|
| 6238 |
+
"loss": 1.0343,
|
| 6239 |
+
"step": 890
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.07787610619469026,
|
| 6243 |
+
"grad_norm": 0.4963998794555664,
|
| 6244 |
+
"learning_rate": 0.00046126267925848197,
|
| 6245 |
+
"loss": 1.3856,
|
| 6246 |
+
"step": 891
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.07796350923194581,
|
| 6250 |
+
"grad_norm": 0.3124493360519409,
|
| 6251 |
+
"learning_rate": 0.00046121895767750967,
|
| 6252 |
+
"loss": 1.0451,
|
| 6253 |
+
"step": 892
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.07805091226920136,
|
| 6257 |
+
"grad_norm": 0.5837683081626892,
|
| 6258 |
+
"learning_rate": 0.00046117523609653726,
|
| 6259 |
+
"loss": 1.0501,
|
| 6260 |
+
"step": 893
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.0781383153064569,
|
| 6264 |
+
"grad_norm": 0.31839168071746826,
|
| 6265 |
+
"learning_rate": 0.0004611315145155649,
|
| 6266 |
+
"loss": 0.9903,
|
| 6267 |
+
"step": 894
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.07822571834371245,
|
| 6271 |
+
"grad_norm": 0.5437602996826172,
|
| 6272 |
+
"learning_rate": 0.0004610877929345925,
|
| 6273 |
+
"loss": 1.0399,
|
| 6274 |
+
"step": 895
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.07831312138096799,
|
| 6278 |
+
"grad_norm": 0.3862234354019165,
|
| 6279 |
+
"learning_rate": 0.00046104407135362014,
|
| 6280 |
+
"loss": 1.0355,
|
| 6281 |
+
"step": 896
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.07840052441822354,
|
| 6285 |
+
"grad_norm": 0.7273140549659729,
|
| 6286 |
+
"learning_rate": 0.0004610003497726478,
|
| 6287 |
+
"loss": 0.9339,
|
| 6288 |
+
"step": 897
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.07848792745547908,
|
| 6292 |
+
"grad_norm": 0.31776732206344604,
|
| 6293 |
+
"learning_rate": 0.00046095662819167543,
|
| 6294 |
+
"loss": 1.405,
|
| 6295 |
+
"step": 898
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.07857533049273462,
|
| 6299 |
+
"grad_norm": 0.33975592255592346,
|
| 6300 |
+
"learning_rate": 0.0004609129066107031,
|
| 6301 |
+
"loss": 0.9493,
|
| 6302 |
+
"step": 899
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.07866273352999016,
|
| 6306 |
+
"grad_norm": 0.3096635937690735,
|
| 6307 |
+
"learning_rate": 0.00046086918502973067,
|
| 6308 |
+
"loss": 0.8949,
|
| 6309 |
+
"step": 900
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.07875013656724571,
|
| 6313 |
+
"grad_norm": 0.22939470410346985,
|
| 6314 |
+
"learning_rate": 0.0004608254634487583,
|
| 6315 |
+
"loss": 1.0486,
|
| 6316 |
+
"step": 901
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.07883753960450125,
|
| 6320 |
+
"grad_norm": 0.27594518661499023,
|
| 6321 |
+
"learning_rate": 0.0004607817418677859,
|
| 6322 |
+
"loss": 0.7005,
|
| 6323 |
+
"step": 902
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.0789249426417568,
|
| 6327 |
+
"grad_norm": 0.38164445757865906,
|
| 6328 |
+
"learning_rate": 0.0004607380202868136,
|
| 6329 |
+
"loss": 1.2305,
|
| 6330 |
+
"step": 903
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.07901234567901234,
|
| 6334 |
+
"grad_norm": 0.26803824305534363,
|
| 6335 |
+
"learning_rate": 0.00046069429870584125,
|
| 6336 |
+
"loss": 0.824,
|
| 6337 |
+
"step": 904
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.07909974871626789,
|
| 6341 |
+
"grad_norm": 0.3049018085002899,
|
| 6342 |
+
"learning_rate": 0.00046065057712486884,
|
| 6343 |
+
"loss": 0.8824,
|
| 6344 |
+
"step": 905
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.07918715175352344,
|
| 6348 |
+
"grad_norm": 0.30478763580322266,
|
| 6349 |
+
"learning_rate": 0.0004606068555438965,
|
| 6350 |
+
"loss": 0.9809,
|
| 6351 |
+
"step": 906
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.07927455479077898,
|
| 6355 |
+
"grad_norm": 0.276212602853775,
|
| 6356 |
+
"learning_rate": 0.0004605631339629241,
|
| 6357 |
+
"loss": 0.8166,
|
| 6358 |
+
"step": 907
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.07936195782803453,
|
| 6362 |
+
"grad_norm": 0.8416312336921692,
|
| 6363 |
+
"learning_rate": 0.0004605194123819517,
|
| 6364 |
+
"loss": 1.5118,
|
| 6365 |
+
"step": 908
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.07944936086529007,
|
| 6369 |
+
"grad_norm": 0.3249102532863617,
|
| 6370 |
+
"learning_rate": 0.0004604756908009794,
|
| 6371 |
+
"loss": 0.905,
|
| 6372 |
+
"step": 909
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.07953676390254562,
|
| 6376 |
+
"grad_norm": 0.3695957064628601,
|
| 6377 |
+
"learning_rate": 0.000460431969220007,
|
| 6378 |
+
"loss": 0.809,
|
| 6379 |
+
"step": 910
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.07962416693980116,
|
| 6383 |
+
"grad_norm": 0.2533642649650574,
|
| 6384 |
+
"learning_rate": 0.00046038824763903466,
|
| 6385 |
+
"loss": 0.8706,
|
| 6386 |
+
"step": 911
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.07971156997705671,
|
| 6390 |
+
"grad_norm": 1.895600438117981,
|
| 6391 |
+
"learning_rate": 0.00046034452605806225,
|
| 6392 |
+
"loss": 0.906,
|
| 6393 |
+
"step": 912
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.07979897301431224,
|
| 6397 |
+
"grad_norm": 0.3041301369667053,
|
| 6398 |
+
"learning_rate": 0.0004603008044770899,
|
| 6399 |
+
"loss": 0.8028,
|
| 6400 |
+
"step": 913
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.07988637605156779,
|
| 6404 |
+
"grad_norm": 0.39580902457237244,
|
| 6405 |
+
"learning_rate": 0.0004602570828961175,
|
| 6406 |
+
"loss": 0.8785,
|
| 6407 |
+
"step": 914
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.07997377908882333,
|
| 6411 |
+
"grad_norm": 0.3260571360588074,
|
| 6412 |
+
"learning_rate": 0.0004602133613151452,
|
| 6413 |
+
"loss": 0.908,
|
| 6414 |
+
"step": 915
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.08006118212607888,
|
| 6418 |
+
"grad_norm": 0.3628925681114197,
|
| 6419 |
+
"learning_rate": 0.0004601696397341728,
|
| 6420 |
+
"loss": 0.8364,
|
| 6421 |
+
"step": 916
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.08014858516333442,
|
| 6425 |
+
"grad_norm": 0.4076823890209198,
|
| 6426 |
+
"learning_rate": 0.0004601259181532004,
|
| 6427 |
+
"loss": 1.93,
|
| 6428 |
+
"step": 917
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.08023598820058997,
|
| 6432 |
+
"grad_norm": 0.6916859149932861,
|
| 6433 |
+
"learning_rate": 0.00046008219657222806,
|
| 6434 |
+
"loss": 1.1446,
|
| 6435 |
+
"step": 918
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.08032339123784552,
|
| 6439 |
+
"grad_norm": 1.301007866859436,
|
| 6440 |
+
"learning_rate": 0.00046003847499125565,
|
| 6441 |
+
"loss": 1.117,
|
| 6442 |
+
"step": 919
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.08041079427510106,
|
| 6446 |
+
"grad_norm": 2.9351885318756104,
|
| 6447 |
+
"learning_rate": 0.00045999475341028335,
|
| 6448 |
+
"loss": 1.8147,
|
| 6449 |
+
"step": 920
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.0804981973123566,
|
| 6453 |
+
"grad_norm": 3.5363566875457764,
|
| 6454 |
+
"learning_rate": 0.000459951031829311,
|
| 6455 |
+
"loss": 1.4487,
|
| 6456 |
+
"step": 921
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.08058560034961215,
|
| 6460 |
+
"grad_norm": 1.0070669651031494,
|
| 6461 |
+
"learning_rate": 0.0004599073102483386,
|
| 6462 |
+
"loss": 0.9901,
|
| 6463 |
+
"step": 922
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.0806730033868677,
|
| 6467 |
+
"grad_norm": 0.42096540331840515,
|
| 6468 |
+
"learning_rate": 0.00045986358866736623,
|
| 6469 |
+
"loss": 0.8757,
|
| 6470 |
+
"step": 923
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.08076040642412324,
|
| 6474 |
+
"grad_norm": 0.7990926504135132,
|
| 6475 |
+
"learning_rate": 0.0004598198670863938,
|
| 6476 |
+
"loss": 1.1409,
|
| 6477 |
+
"step": 924
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.08084780946137879,
|
| 6481 |
+
"grad_norm": 0.6880809664726257,
|
| 6482 |
+
"learning_rate": 0.00045977614550542147,
|
| 6483 |
+
"loss": 0.9678,
|
| 6484 |
+
"step": 925
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.08093521249863432,
|
| 6488 |
+
"grad_norm": 0.7126320004463196,
|
| 6489 |
+
"learning_rate": 0.0004597324239244491,
|
| 6490 |
+
"loss": 0.8932,
|
| 6491 |
+
"step": 926
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.08102261553588987,
|
| 6495 |
+
"grad_norm": 1.2712117433547974,
|
| 6496 |
+
"learning_rate": 0.00045968870234347676,
|
| 6497 |
+
"loss": 1.7774,
|
| 6498 |
+
"step": 927
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.08111001857314541,
|
| 6502 |
+
"grad_norm": 1.9836965799331665,
|
| 6503 |
+
"learning_rate": 0.0004596449807625044,
|
| 6504 |
+
"loss": 1.1419,
|
| 6505 |
+
"step": 928
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.08119742161040096,
|
| 6509 |
+
"grad_norm": 0.6894294023513794,
|
| 6510 |
+
"learning_rate": 0.000459601259181532,
|
| 6511 |
+
"loss": 0.9666,
|
| 6512 |
+
"step": 929
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.0812848246476565,
|
| 6516 |
+
"grad_norm": 2.2530252933502197,
|
| 6517 |
+
"learning_rate": 0.00045955753760055964,
|
| 6518 |
+
"loss": 1.5093,
|
| 6519 |
+
"step": 930
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.08137222768491205,
|
| 6523 |
+
"grad_norm": 14.37427043914795,
|
| 6524 |
+
"learning_rate": 0.0004595138160195873,
|
| 6525 |
+
"loss": 1.3134,
|
| 6526 |
+
"step": 931
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.0814596307221676,
|
| 6530 |
+
"grad_norm": 3.392730236053467,
|
| 6531 |
+
"learning_rate": 0.00045947009443861493,
|
| 6532 |
+
"loss": 1.0883,
|
| 6533 |
+
"step": 932
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.08154703375942314,
|
| 6537 |
+
"grad_norm": 1.097122073173523,
|
| 6538 |
+
"learning_rate": 0.0004594263728576425,
|
| 6539 |
+
"loss": 1.0587,
|
| 6540 |
+
"step": 933
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.08163443679667869,
|
| 6544 |
+
"grad_norm": 0.7270208597183228,
|
| 6545 |
+
"learning_rate": 0.00045938265127667017,
|
| 6546 |
+
"loss": 1.1386,
|
| 6547 |
+
"step": 934
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.08172183983393423,
|
| 6551 |
+
"grad_norm": 3.5602266788482666,
|
| 6552 |
+
"learning_rate": 0.0004593389296956978,
|
| 6553 |
+
"loss": 1.1204,
|
| 6554 |
+
"step": 935
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.08180924287118978,
|
| 6558 |
+
"grad_norm": 1.953038215637207,
|
| 6559 |
+
"learning_rate": 0.0004592952081147254,
|
| 6560 |
+
"loss": 1.2367,
|
| 6561 |
+
"step": 936
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.08189664590844532,
|
| 6565 |
+
"grad_norm": 1.90444016456604,
|
| 6566 |
+
"learning_rate": 0.0004592514865337531,
|
| 6567 |
+
"loss": 1.1981,
|
| 6568 |
+
"step": 937
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.08198404894570087,
|
| 6572 |
+
"grad_norm": 9.526935577392578,
|
| 6573 |
+
"learning_rate": 0.0004592077649527807,
|
| 6574 |
+
"loss": 1.4363,
|
| 6575 |
+
"step": 938
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.08207145198295641,
|
| 6579 |
+
"grad_norm": 5.361575603485107,
|
| 6580 |
+
"learning_rate": 0.00045916404337180834,
|
| 6581 |
+
"loss": 1.4758,
|
| 6582 |
+
"step": 939
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.08215885502021195,
|
| 6586 |
+
"grad_norm": 49.836151123046875,
|
| 6587 |
+
"learning_rate": 0.000459120321790836,
|
| 6588 |
+
"loss": 3.2272,
|
| 6589 |
+
"step": 940
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.08224625805746749,
|
| 6593 |
+
"grad_norm": 6.1282877922058105,
|
| 6594 |
+
"learning_rate": 0.0004590766002098636,
|
| 6595 |
+
"loss": 2.0861,
|
| 6596 |
+
"step": 941
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.08233366109472304,
|
| 6600 |
+
"grad_norm": 9.320550918579102,
|
| 6601 |
+
"learning_rate": 0.0004590328786288912,
|
| 6602 |
+
"loss": 2.0217,
|
| 6603 |
+
"step": 942
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.08242106413197858,
|
| 6607 |
+
"grad_norm": 3.1131937503814697,
|
| 6608 |
+
"learning_rate": 0.00045898915704791887,
|
| 6609 |
+
"loss": 1.4848,
|
| 6610 |
+
"step": 943
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.08250846716923413,
|
| 6614 |
+
"grad_norm": 51.67763137817383,
|
| 6615 |
+
"learning_rate": 0.0004589454354669465,
|
| 6616 |
+
"loss": 3.2458,
|
| 6617 |
+
"step": 944
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.08259587020648967,
|
| 6621 |
+
"grad_norm": 7.247336387634277,
|
| 6622 |
+
"learning_rate": 0.0004589017138859741,
|
| 6623 |
+
"loss": 2.6957,
|
| 6624 |
+
"step": 945
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.08268327324374522,
|
| 6628 |
+
"grad_norm": 3.2208497524261475,
|
| 6629 |
+
"learning_rate": 0.00045885799230500175,
|
| 6630 |
+
"loss": 1.9059,
|
| 6631 |
+
"step": 946
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.08277067628100077,
|
| 6635 |
+
"grad_norm": 78.9037094116211,
|
| 6636 |
+
"learning_rate": 0.0004588142707240294,
|
| 6637 |
+
"loss": 5.5682,
|
| 6638 |
+
"step": 947
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.08285807931825631,
|
| 6642 |
+
"grad_norm": 4.832467079162598,
|
| 6643 |
+
"learning_rate": 0.00045877054914305704,
|
| 6644 |
+
"loss": 1.6731,
|
| 6645 |
+
"step": 948
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.08294548235551186,
|
| 6649 |
+
"grad_norm": 7.1308674812316895,
|
| 6650 |
+
"learning_rate": 0.0004587268275620847,
|
| 6651 |
+
"loss": 2.2772,
|
| 6652 |
+
"step": 949
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.0830328853927674,
|
| 6656 |
+
"grad_norm": 4.155465126037598,
|
| 6657 |
+
"learning_rate": 0.00045868310598111227,
|
| 6658 |
+
"loss": 2.2794,
|
| 6659 |
+
"step": 950
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.08312028843002295,
|
| 6663 |
+
"grad_norm": 51.88750457763672,
|
| 6664 |
+
"learning_rate": 0.0004586393844001399,
|
| 6665 |
+
"loss": 4.0774,
|
| 6666 |
+
"step": 951
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.0832076914672785,
|
| 6670 |
+
"grad_norm": 2.969212532043457,
|
| 6671 |
+
"learning_rate": 0.00045859566281916756,
|
| 6672 |
+
"loss": 1.9225,
|
| 6673 |
+
"step": 952
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.08329509450453403,
|
| 6677 |
+
"grad_norm": 3.454350233078003,
|
| 6678 |
+
"learning_rate": 0.00045855194123819515,
|
| 6679 |
+
"loss": 1.6258,
|
| 6680 |
+
"step": 953
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.08338249754178957,
|
| 6684 |
+
"grad_norm": 46.18666458129883,
|
| 6685 |
+
"learning_rate": 0.00045850821965722285,
|
| 6686 |
+
"loss": 1.7273,
|
| 6687 |
+
"step": 954
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.08346990057904512,
|
| 6691 |
+
"grad_norm": 13.307456016540527,
|
| 6692 |
+
"learning_rate": 0.00045846449807625044,
|
| 6693 |
+
"loss": 2.1933,
|
| 6694 |
+
"step": 955
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.08355730361630066,
|
| 6698 |
+
"grad_norm": 8.283126831054688,
|
| 6699 |
+
"learning_rate": 0.0004584207764952781,
|
| 6700 |
+
"loss": 2.499,
|
| 6701 |
+
"step": 956
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.08364470665355621,
|
| 6705 |
+
"grad_norm": 6.291905403137207,
|
| 6706 |
+
"learning_rate": 0.0004583770549143057,
|
| 6707 |
+
"loss": 1.8399,
|
| 6708 |
+
"step": 957
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.08373210969081175,
|
| 6712 |
+
"grad_norm": 19.28121566772461,
|
| 6713 |
+
"learning_rate": 0.0004583333333333333,
|
| 6714 |
+
"loss": 2.6815,
|
| 6715 |
+
"step": 958
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.0838195127280673,
|
| 6719 |
+
"grad_norm": 9.661205291748047,
|
| 6720 |
+
"learning_rate": 0.000458289611752361,
|
| 6721 |
+
"loss": 2.3274,
|
| 6722 |
+
"step": 959
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.08390691576532285,
|
| 6726 |
+
"grad_norm": 15.012873649597168,
|
| 6727 |
+
"learning_rate": 0.0004582458901713886,
|
| 6728 |
+
"loss": 2.1736,
|
| 6729 |
+
"step": 960
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.08399431880257839,
|
| 6733 |
+
"grad_norm": 10.02956485748291,
|
| 6734 |
+
"learning_rate": 0.00045820216859041626,
|
| 6735 |
+
"loss": 2.4168,
|
| 6736 |
+
"step": 961
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.08408172183983394,
|
| 6740 |
+
"grad_norm": 2.234221935272217,
|
| 6741 |
+
"learning_rate": 0.00045815844700944385,
|
| 6742 |
+
"loss": 1.7808,
|
| 6743 |
+
"step": 962
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.08416912487708948,
|
| 6747 |
+
"grad_norm": 7.04872989654541,
|
| 6748 |
+
"learning_rate": 0.0004581147254284715,
|
| 6749 |
+
"loss": 2.1456,
|
| 6750 |
+
"step": 963
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.08425652791434503,
|
| 6754 |
+
"grad_norm": 3.498042106628418,
|
| 6755 |
+
"learning_rate": 0.0004580710038474991,
|
| 6756 |
+
"loss": 1.6212,
|
| 6757 |
+
"step": 964
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.08434393095160057,
|
| 6761 |
+
"grad_norm": 2.731658935546875,
|
| 6762 |
+
"learning_rate": 0.0004580272822665268,
|
| 6763 |
+
"loss": 1.6905,
|
| 6764 |
+
"step": 965
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.08443133398885612,
|
| 6768 |
+
"grad_norm": 4.867488384246826,
|
| 6769 |
+
"learning_rate": 0.00045798356068555443,
|
| 6770 |
+
"loss": 1.4945,
|
| 6771 |
+
"step": 966
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.08451873702611165,
|
| 6775 |
+
"grad_norm": 10.225361824035645,
|
| 6776 |
+
"learning_rate": 0.000457939839104582,
|
| 6777 |
+
"loss": 2.4163,
|
| 6778 |
+
"step": 967
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.0846061400633672,
|
| 6782 |
+
"grad_norm": 2.749767780303955,
|
| 6783 |
+
"learning_rate": 0.00045789611752360967,
|
| 6784 |
+
"loss": 1.49,
|
| 6785 |
+
"step": 968
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.08469354310062274,
|
| 6789 |
+
"grad_norm": 14.945262908935547,
|
| 6790 |
+
"learning_rate": 0.00045785239594263726,
|
| 6791 |
+
"loss": 2.4579,
|
| 6792 |
+
"step": 969
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.08478094613787829,
|
| 6796 |
+
"grad_norm": 4.0551228523254395,
|
| 6797 |
+
"learning_rate": 0.0004578086743616649,
|
| 6798 |
+
"loss": 1.6358,
|
| 6799 |
+
"step": 970
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.08486834917513383,
|
| 6803 |
+
"grad_norm": 2.8462789058685303,
|
| 6804 |
+
"learning_rate": 0.0004577649527806926,
|
| 6805 |
+
"loss": 1.6568,
|
| 6806 |
+
"step": 971
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.08495575221238938,
|
| 6810 |
+
"grad_norm": 3.82456111907959,
|
| 6811 |
+
"learning_rate": 0.0004577212311997202,
|
| 6812 |
+
"loss": 1.696,
|
| 6813 |
+
"step": 972
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.08504315524964493,
|
| 6817 |
+
"grad_norm": 2.9463558197021484,
|
| 6818 |
+
"learning_rate": 0.00045767750961874784,
|
| 6819 |
+
"loss": 1.8359,
|
| 6820 |
+
"step": 973
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.08513055828690047,
|
| 6824 |
+
"grad_norm": 2.811894416809082,
|
| 6825 |
+
"learning_rate": 0.00045763378803777543,
|
| 6826 |
+
"loss": 1.369,
|
| 6827 |
+
"step": 974
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.08521796132415602,
|
| 6831 |
+
"grad_norm": 2.092231512069702,
|
| 6832 |
+
"learning_rate": 0.0004575900664568031,
|
| 6833 |
+
"loss": 1.5433,
|
| 6834 |
+
"step": 975
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.08530536436141156,
|
| 6838 |
+
"grad_norm": 4.028072357177734,
|
| 6839 |
+
"learning_rate": 0.0004575463448758307,
|
| 6840 |
+
"loss": 2.4999,
|
| 6841 |
+
"step": 976
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.08539276739866711,
|
| 6845 |
+
"grad_norm": 10.593165397644043,
|
| 6846 |
+
"learning_rate": 0.00045750262329485836,
|
| 6847 |
+
"loss": 1.5753,
|
| 6848 |
+
"step": 977
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.08548017043592265,
|
| 6852 |
+
"grad_norm": 6.811407089233398,
|
| 6853 |
+
"learning_rate": 0.000457458901713886,
|
| 6854 |
+
"loss": 1.7268,
|
| 6855 |
+
"step": 978
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.0855675734731782,
|
| 6859 |
+
"grad_norm": 2.3520467281341553,
|
| 6860 |
+
"learning_rate": 0.0004574151801329136,
|
| 6861 |
+
"loss": 1.4044,
|
| 6862 |
+
"step": 979
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.08565497651043373,
|
| 6866 |
+
"grad_norm": 3.668078660964966,
|
| 6867 |
+
"learning_rate": 0.00045737145855194125,
|
| 6868 |
+
"loss": 1.718,
|
| 6869 |
+
"step": 980
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.08574237954768928,
|
| 6873 |
+
"grad_norm": 10.229111671447754,
|
| 6874 |
+
"learning_rate": 0.00045732773697096884,
|
| 6875 |
+
"loss": 1.7006,
|
| 6876 |
+
"step": 981
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.08582978258494482,
|
| 6880 |
+
"grad_norm": 5.428765773773193,
|
| 6881 |
+
"learning_rate": 0.00045728401538999654,
|
| 6882 |
+
"loss": 2.2021,
|
| 6883 |
+
"step": 982
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.08591718562220037,
|
| 6887 |
+
"grad_norm": 2.0686569213867188,
|
| 6888 |
+
"learning_rate": 0.0004572402938090242,
|
| 6889 |
+
"loss": 1.687,
|
| 6890 |
+
"step": 983
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.08600458865945591,
|
| 6894 |
+
"grad_norm": 2.371243715286255,
|
| 6895 |
+
"learning_rate": 0.00045719657222805177,
|
| 6896 |
+
"loss": 1.6734,
|
| 6897 |
+
"step": 984
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.08609199169671146,
|
| 6901 |
+
"grad_norm": 1.6429576873779297,
|
| 6902 |
+
"learning_rate": 0.0004571528506470794,
|
| 6903 |
+
"loss": 1.8382,
|
| 6904 |
+
"step": 985
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.086179394733967,
|
| 6908 |
+
"grad_norm": 2.408743381500244,
|
| 6909 |
+
"learning_rate": 0.000457109129066107,
|
| 6910 |
+
"loss": 1.45,
|
| 6911 |
+
"step": 986
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.08626679777122255,
|
| 6915 |
+
"grad_norm": 4.068368434906006,
|
| 6916 |
+
"learning_rate": 0.0004570654074851347,
|
| 6917 |
+
"loss": 1.7464,
|
| 6918 |
+
"step": 987
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.0863542008084781,
|
| 6922 |
+
"grad_norm": 1.9330801963806152,
|
| 6923 |
+
"learning_rate": 0.0004570216859041623,
|
| 6924 |
+
"loss": 1.6335,
|
| 6925 |
+
"step": 988
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.08644160384573364,
|
| 6929 |
+
"grad_norm": 4.200726509094238,
|
| 6930 |
+
"learning_rate": 0.00045697796432318994,
|
| 6931 |
+
"loss": 1.6781,
|
| 6932 |
+
"step": 989
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.08652900688298919,
|
| 6936 |
+
"grad_norm": 4.335032939910889,
|
| 6937 |
+
"learning_rate": 0.0004569342427422176,
|
| 6938 |
+
"loss": 1.7382,
|
| 6939 |
+
"step": 990
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.08661640992024473,
|
| 6943 |
+
"grad_norm": 2.2428669929504395,
|
| 6944 |
+
"learning_rate": 0.0004568905211612452,
|
| 6945 |
+
"loss": 1.4791,
|
| 6946 |
+
"step": 991
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.08670381295750028,
|
| 6950 |
+
"grad_norm": 2.2247121334075928,
|
| 6951 |
+
"learning_rate": 0.0004568467995802728,
|
| 6952 |
+
"loss": 1.8668,
|
| 6953 |
+
"step": 992
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.08679121599475582,
|
| 6957 |
+
"grad_norm": 2.013319492340088,
|
| 6958 |
+
"learning_rate": 0.00045680307799930047,
|
| 6959 |
+
"loss": 1.4925,
|
| 6960 |
+
"step": 993
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.08687861903201136,
|
| 6964 |
+
"grad_norm": 1.5773614645004272,
|
| 6965 |
+
"learning_rate": 0.0004567593564183281,
|
| 6966 |
+
"loss": 1.3334,
|
| 6967 |
+
"step": 994
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.0869660220692669,
|
| 6971 |
+
"grad_norm": 1.1663486957550049,
|
| 6972 |
+
"learning_rate": 0.0004567156348373557,
|
| 6973 |
+
"loss": 1.5022,
|
| 6974 |
+
"step": 995
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.08705342510652245,
|
| 6978 |
+
"grad_norm": 1.763238549232483,
|
| 6979 |
+
"learning_rate": 0.00045667191325638335,
|
| 6980 |
+
"loss": 1.5118,
|
| 6981 |
+
"step": 996
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.08714082814377799,
|
| 6985 |
+
"grad_norm": 1.4888843297958374,
|
| 6986 |
+
"learning_rate": 0.000456628191675411,
|
| 6987 |
+
"loss": 1.6713,
|
| 6988 |
+
"step": 997
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.08722823118103354,
|
| 6992 |
+
"grad_norm": 2.5363516807556152,
|
| 6993 |
+
"learning_rate": 0.0004565844700944386,
|
| 6994 |
+
"loss": 1.4999,
|
| 6995 |
+
"step": 998
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.08731563421828908,
|
| 6999 |
+
"grad_norm": 2.134773015975952,
|
| 7000 |
+
"learning_rate": 0.0004565407485134663,
|
| 7001 |
+
"loss": 1.5086,
|
| 7002 |
+
"step": 999
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.08740303725554463,
|
| 7006 |
+
"grad_norm": 15.75776481628418,
|
| 7007 |
+
"learning_rate": 0.0004564970269324939,
|
| 7008 |
+
"loss": 2.11,
|
| 7009 |
+
"step": 1000
|
| 7010 |
}
|
| 7011 |
],
|
| 7012 |
"logging_steps": 1,
|
|
|
|
| 7026 |
"attributes": {}
|
| 7027 |
}
|
| 7028 |
},
|
| 7029 |
+
"total_flos": 8.60783873359872e+17,
|
| 7030 |
"train_batch_size": 2,
|
| 7031 |
"trial_name": null,
|
| 7032 |
"trial_params": null
|