Training in progress, step 4000, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1673342072
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bbfc8aad35faaf64abb513505043c047584775cadd7245019a5fe079204913c5
|
| 3 |
size 1673342072
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 194745274
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80b9fc16e20c7c3a4a8a32c70b58d54425e979f182d679aa58671e3ec7106544
|
| 3 |
size 194745274
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec8840281322a0a6eb80ad374fa65fffd4d0c70b4f8fda772212d4d71b733b12
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f809221ad225608f3be6038571a48fb2004d7ccafc2e2b8f01c991135b7010d6
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 900,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -24531,6 +24531,3514 @@
|
|
| 24531 |
"learning_rate": 4.4543429844098e-05,
|
| 24532 |
"loss": 0.6577,
|
| 24533 |
"step": 3500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24534 |
}
|
| 24535 |
],
|
| 24536 |
"logging_steps": 1,
|
|
@@ -24550,7 +28058,7 @@
|
|
| 24550 |
"attributes": {}
|
| 24551 |
}
|
| 24552 |
},
|
| 24553 |
-
"total_flos":
|
| 24554 |
"train_batch_size": 1,
|
| 24555 |
"trial_name": null,
|
| 24556 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8888888888888888,
|
| 5 |
"eval_steps": 900,
|
| 6 |
+
"global_step": 4000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 24531 |
"learning_rate": 4.4543429844098e-05,
|
| 24532 |
"loss": 0.6577,
|
| 24533 |
"step": 3500
|
| 24534 |
+
},
|
| 24535 |
+
{
|
| 24536 |
+
"epoch": 0.778,
|
| 24537 |
+
"grad_norm": 0.6440428495407104,
|
| 24538 |
+
"learning_rate": 4.44988864142539e-05,
|
| 24539 |
+
"loss": 0.9956,
|
| 24540 |
+
"step": 3501
|
| 24541 |
+
},
|
| 24542 |
+
{
|
| 24543 |
+
"epoch": 0.7782222222222223,
|
| 24544 |
+
"grad_norm": 0.5025835633277893,
|
| 24545 |
+
"learning_rate": 4.44543429844098e-05,
|
| 24546 |
+
"loss": 0.9893,
|
| 24547 |
+
"step": 3502
|
| 24548 |
+
},
|
| 24549 |
+
{
|
| 24550 |
+
"epoch": 0.7784444444444445,
|
| 24551 |
+
"grad_norm": 0.04510605335235596,
|
| 24552 |
+
"learning_rate": 4.4409799554565705e-05,
|
| 24553 |
+
"loss": 0.0106,
|
| 24554 |
+
"step": 3503
|
| 24555 |
+
},
|
| 24556 |
+
{
|
| 24557 |
+
"epoch": 0.7786666666666666,
|
| 24558 |
+
"grad_norm": 0.9423682689666748,
|
| 24559 |
+
"learning_rate": 4.43652561247216e-05,
|
| 24560 |
+
"loss": 2.3701,
|
| 24561 |
+
"step": 3504
|
| 24562 |
+
},
|
| 24563 |
+
{
|
| 24564 |
+
"epoch": 0.7788888888888889,
|
| 24565 |
+
"grad_norm": 0.9391410946846008,
|
| 24566 |
+
"learning_rate": 4.432071269487751e-05,
|
| 24567 |
+
"loss": 2.172,
|
| 24568 |
+
"step": 3505
|
| 24569 |
+
},
|
| 24570 |
+
{
|
| 24571 |
+
"epoch": 0.7791111111111111,
|
| 24572 |
+
"grad_norm": 0.9086732864379883,
|
| 24573 |
+
"learning_rate": 4.427616926503341e-05,
|
| 24574 |
+
"loss": 2.217,
|
| 24575 |
+
"step": 3506
|
| 24576 |
+
},
|
| 24577 |
+
{
|
| 24578 |
+
"epoch": 0.7793333333333333,
|
| 24579 |
+
"grad_norm": 0.9702697396278381,
|
| 24580 |
+
"learning_rate": 4.423162583518931e-05,
|
| 24581 |
+
"loss": 2.1024,
|
| 24582 |
+
"step": 3507
|
| 24583 |
+
},
|
| 24584 |
+
{
|
| 24585 |
+
"epoch": 0.7795555555555556,
|
| 24586 |
+
"grad_norm": 0.9364957809448242,
|
| 24587 |
+
"learning_rate": 4.418708240534522e-05,
|
| 24588 |
+
"loss": 1.9179,
|
| 24589 |
+
"step": 3508
|
| 24590 |
+
},
|
| 24591 |
+
{
|
| 24592 |
+
"epoch": 0.7797777777777778,
|
| 24593 |
+
"grad_norm": 0.9006823301315308,
|
| 24594 |
+
"learning_rate": 4.414253897550111e-05,
|
| 24595 |
+
"loss": 1.9431,
|
| 24596 |
+
"step": 3509
|
| 24597 |
+
},
|
| 24598 |
+
{
|
| 24599 |
+
"epoch": 0.78,
|
| 24600 |
+
"grad_norm": 0.8712829947471619,
|
| 24601 |
+
"learning_rate": 4.4097995545657015e-05,
|
| 24602 |
+
"loss": 1.7075,
|
| 24603 |
+
"step": 3510
|
| 24604 |
+
},
|
| 24605 |
+
{
|
| 24606 |
+
"epoch": 0.7802222222222223,
|
| 24607 |
+
"grad_norm": 0.8921668529510498,
|
| 24608 |
+
"learning_rate": 4.405345211581292e-05,
|
| 24609 |
+
"loss": 2.0187,
|
| 24610 |
+
"step": 3511
|
| 24611 |
+
},
|
| 24612 |
+
{
|
| 24613 |
+
"epoch": 0.7804444444444445,
|
| 24614 |
+
"grad_norm": 1.2319942712783813,
|
| 24615 |
+
"learning_rate": 4.400890868596882e-05,
|
| 24616 |
+
"loss": 2.2116,
|
| 24617 |
+
"step": 3512
|
| 24618 |
+
},
|
| 24619 |
+
{
|
| 24620 |
+
"epoch": 0.7806666666666666,
|
| 24621 |
+
"grad_norm": 1.0687848329544067,
|
| 24622 |
+
"learning_rate": 4.3964365256124724e-05,
|
| 24623 |
+
"loss": 1.827,
|
| 24624 |
+
"step": 3513
|
| 24625 |
+
},
|
| 24626 |
+
{
|
| 24627 |
+
"epoch": 0.7808888888888889,
|
| 24628 |
+
"grad_norm": 0.06872207671403885,
|
| 24629 |
+
"learning_rate": 4.391982182628063e-05,
|
| 24630 |
+
"loss": 0.016,
|
| 24631 |
+
"step": 3514
|
| 24632 |
+
},
|
| 24633 |
+
{
|
| 24634 |
+
"epoch": 0.7811111111111111,
|
| 24635 |
+
"grad_norm": 0.06946699321269989,
|
| 24636 |
+
"learning_rate": 4.387527839643653e-05,
|
| 24637 |
+
"loss": 0.016,
|
| 24638 |
+
"step": 3515
|
| 24639 |
+
},
|
| 24640 |
+
{
|
| 24641 |
+
"epoch": 0.7813333333333333,
|
| 24642 |
+
"grad_norm": 0.06873323768377304,
|
| 24643 |
+
"learning_rate": 4.383073496659243e-05,
|
| 24644 |
+
"loss": 0.0157,
|
| 24645 |
+
"step": 3516
|
| 24646 |
+
},
|
| 24647 |
+
{
|
| 24648 |
+
"epoch": 0.7815555555555556,
|
| 24649 |
+
"grad_norm": 0.7765884399414062,
|
| 24650 |
+
"learning_rate": 4.378619153674833e-05,
|
| 24651 |
+
"loss": 0.974,
|
| 24652 |
+
"step": 3517
|
| 24653 |
+
},
|
| 24654 |
+
{
|
| 24655 |
+
"epoch": 0.7817777777777778,
|
| 24656 |
+
"grad_norm": 0.7765089869499207,
|
| 24657 |
+
"learning_rate": 4.374164810690423e-05,
|
| 24658 |
+
"loss": 0.9251,
|
| 24659 |
+
"step": 3518
|
| 24660 |
+
},
|
| 24661 |
+
{
|
| 24662 |
+
"epoch": 0.782,
|
| 24663 |
+
"grad_norm": 0.6271977424621582,
|
| 24664 |
+
"learning_rate": 4.3697104677060136e-05,
|
| 24665 |
+
"loss": 0.8018,
|
| 24666 |
+
"step": 3519
|
| 24667 |
+
},
|
| 24668 |
+
{
|
| 24669 |
+
"epoch": 0.7822222222222223,
|
| 24670 |
+
"grad_norm": 0.726948618888855,
|
| 24671 |
+
"learning_rate": 4.365256124721604e-05,
|
| 24672 |
+
"loss": 0.8888,
|
| 24673 |
+
"step": 3520
|
| 24674 |
+
},
|
| 24675 |
+
{
|
| 24676 |
+
"epoch": 0.7824444444444445,
|
| 24677 |
+
"grad_norm": 0.9243329167366028,
|
| 24678 |
+
"learning_rate": 4.360801781737194e-05,
|
| 24679 |
+
"loss": 1.6066,
|
| 24680 |
+
"step": 3521
|
| 24681 |
+
},
|
| 24682 |
+
{
|
| 24683 |
+
"epoch": 0.7826666666666666,
|
| 24684 |
+
"grad_norm": 1.2513469457626343,
|
| 24685 |
+
"learning_rate": 4.356347438752784e-05,
|
| 24686 |
+
"loss": 1.9953,
|
| 24687 |
+
"step": 3522
|
| 24688 |
+
},
|
| 24689 |
+
{
|
| 24690 |
+
"epoch": 0.7828888888888889,
|
| 24691 |
+
"grad_norm": 0.9244915246963501,
|
| 24692 |
+
"learning_rate": 4.3518930957683744e-05,
|
| 24693 |
+
"loss": 1.8274,
|
| 24694 |
+
"step": 3523
|
| 24695 |
+
},
|
| 24696 |
+
{
|
| 24697 |
+
"epoch": 0.7831111111111111,
|
| 24698 |
+
"grad_norm": 1.0428435802459717,
|
| 24699 |
+
"learning_rate": 4.347438752783964e-05,
|
| 24700 |
+
"loss": 1.6283,
|
| 24701 |
+
"step": 3524
|
| 24702 |
+
},
|
| 24703 |
+
{
|
| 24704 |
+
"epoch": 0.7833333333333333,
|
| 24705 |
+
"grad_norm": 0.8906724452972412,
|
| 24706 |
+
"learning_rate": 4.342984409799555e-05,
|
| 24707 |
+
"loss": 1.6417,
|
| 24708 |
+
"step": 3525
|
| 24709 |
+
},
|
| 24710 |
+
{
|
| 24711 |
+
"epoch": 0.7835555555555556,
|
| 24712 |
+
"grad_norm": 1.084955096244812,
|
| 24713 |
+
"learning_rate": 4.3385300668151454e-05,
|
| 24714 |
+
"loss": 1.6931,
|
| 24715 |
+
"step": 3526
|
| 24716 |
+
},
|
| 24717 |
+
{
|
| 24718 |
+
"epoch": 0.7837777777777778,
|
| 24719 |
+
"grad_norm": 0.945055365562439,
|
| 24720 |
+
"learning_rate": 4.334075723830735e-05,
|
| 24721 |
+
"loss": 1.0096,
|
| 24722 |
+
"step": 3527
|
| 24723 |
+
},
|
| 24724 |
+
{
|
| 24725 |
+
"epoch": 0.784,
|
| 24726 |
+
"grad_norm": 0.0929998904466629,
|
| 24727 |
+
"learning_rate": 4.329621380846325e-05,
|
| 24728 |
+
"loss": 0.0173,
|
| 24729 |
+
"step": 3528
|
| 24730 |
+
},
|
| 24731 |
+
{
|
| 24732 |
+
"epoch": 0.7842222222222223,
|
| 24733 |
+
"grad_norm": 1.0231083631515503,
|
| 24734 |
+
"learning_rate": 4.3251670378619156e-05,
|
| 24735 |
+
"loss": 1.729,
|
| 24736 |
+
"step": 3529
|
| 24737 |
+
},
|
| 24738 |
+
{
|
| 24739 |
+
"epoch": 0.7844444444444445,
|
| 24740 |
+
"grad_norm": 1.0118030309677124,
|
| 24741 |
+
"learning_rate": 4.3207126948775055e-05,
|
| 24742 |
+
"loss": 1.6239,
|
| 24743 |
+
"step": 3530
|
| 24744 |
+
},
|
| 24745 |
+
{
|
| 24746 |
+
"epoch": 0.7846666666666666,
|
| 24747 |
+
"grad_norm": 0.712536096572876,
|
| 24748 |
+
"learning_rate": 4.316258351893096e-05,
|
| 24749 |
+
"loss": 0.7916,
|
| 24750 |
+
"step": 3531
|
| 24751 |
+
},
|
| 24752 |
+
{
|
| 24753 |
+
"epoch": 0.7848888888888889,
|
| 24754 |
+
"grad_norm": 0.0708150640130043,
|
| 24755 |
+
"learning_rate": 4.3118040089086865e-05,
|
| 24756 |
+
"loss": 0.0182,
|
| 24757 |
+
"step": 3532
|
| 24758 |
+
},
|
| 24759 |
+
{
|
| 24760 |
+
"epoch": 0.7851111111111111,
|
| 24761 |
+
"grad_norm": 0.08458317071199417,
|
| 24762 |
+
"learning_rate": 4.3073496659242764e-05,
|
| 24763 |
+
"loss": 0.0185,
|
| 24764 |
+
"step": 3533
|
| 24765 |
+
},
|
| 24766 |
+
{
|
| 24767 |
+
"epoch": 0.7853333333333333,
|
| 24768 |
+
"grad_norm": 1.0665457248687744,
|
| 24769 |
+
"learning_rate": 4.302895322939867e-05,
|
| 24770 |
+
"loss": 1.6411,
|
| 24771 |
+
"step": 3534
|
| 24772 |
+
},
|
| 24773 |
+
{
|
| 24774 |
+
"epoch": 0.7855555555555556,
|
| 24775 |
+
"grad_norm": 1.1385186910629272,
|
| 24776 |
+
"learning_rate": 4.298440979955457e-05,
|
| 24777 |
+
"loss": 1.3018,
|
| 24778 |
+
"step": 3535
|
| 24779 |
+
},
|
| 24780 |
+
{
|
| 24781 |
+
"epoch": 0.7857777777777778,
|
| 24782 |
+
"grad_norm": 1.0706120729446411,
|
| 24783 |
+
"learning_rate": 4.2939866369710466e-05,
|
| 24784 |
+
"loss": 1.5641,
|
| 24785 |
+
"step": 3536
|
| 24786 |
+
},
|
| 24787 |
+
{
|
| 24788 |
+
"epoch": 0.786,
|
| 24789 |
+
"grad_norm": 1.1300500631332397,
|
| 24790 |
+
"learning_rate": 4.289532293986637e-05,
|
| 24791 |
+
"loss": 1.8248,
|
| 24792 |
+
"step": 3537
|
| 24793 |
+
},
|
| 24794 |
+
{
|
| 24795 |
+
"epoch": 0.7862222222222223,
|
| 24796 |
+
"grad_norm": 1.3996295928955078,
|
| 24797 |
+
"learning_rate": 4.285077951002228e-05,
|
| 24798 |
+
"loss": 1.5244,
|
| 24799 |
+
"step": 3538
|
| 24800 |
+
},
|
| 24801 |
+
{
|
| 24802 |
+
"epoch": 0.7864444444444444,
|
| 24803 |
+
"grad_norm": 1.079788088798523,
|
| 24804 |
+
"learning_rate": 4.2806236080178176e-05,
|
| 24805 |
+
"loss": 1.6515,
|
| 24806 |
+
"step": 3539
|
| 24807 |
+
},
|
| 24808 |
+
{
|
| 24809 |
+
"epoch": 0.7866666666666666,
|
| 24810 |
+
"grad_norm": 1.3016208410263062,
|
| 24811 |
+
"learning_rate": 4.276169265033408e-05,
|
| 24812 |
+
"loss": 1.5737,
|
| 24813 |
+
"step": 3540
|
| 24814 |
+
},
|
| 24815 |
+
{
|
| 24816 |
+
"epoch": 0.7868888888888889,
|
| 24817 |
+
"grad_norm": 1.1378430128097534,
|
| 24818 |
+
"learning_rate": 4.271714922048998e-05,
|
| 24819 |
+
"loss": 1.4334,
|
| 24820 |
+
"step": 3541
|
| 24821 |
+
},
|
| 24822 |
+
{
|
| 24823 |
+
"epoch": 0.7871111111111111,
|
| 24824 |
+
"grad_norm": 1.0300368070602417,
|
| 24825 |
+
"learning_rate": 4.267260579064588e-05,
|
| 24826 |
+
"loss": 1.4333,
|
| 24827 |
+
"step": 3542
|
| 24828 |
+
},
|
| 24829 |
+
{
|
| 24830 |
+
"epoch": 0.7873333333333333,
|
| 24831 |
+
"grad_norm": 1.0914349555969238,
|
| 24832 |
+
"learning_rate": 4.2628062360801784e-05,
|
| 24833 |
+
"loss": 1.3632,
|
| 24834 |
+
"step": 3543
|
| 24835 |
+
},
|
| 24836 |
+
{
|
| 24837 |
+
"epoch": 0.7875555555555556,
|
| 24838 |
+
"grad_norm": 1.0840858221054077,
|
| 24839 |
+
"learning_rate": 4.258351893095768e-05,
|
| 24840 |
+
"loss": 1.471,
|
| 24841 |
+
"step": 3544
|
| 24842 |
+
},
|
| 24843 |
+
{
|
| 24844 |
+
"epoch": 0.7877777777777778,
|
| 24845 |
+
"grad_norm": 0.19576247036457062,
|
| 24846 |
+
"learning_rate": 4.253897550111359e-05,
|
| 24847 |
+
"loss": 0.0362,
|
| 24848 |
+
"step": 3545
|
| 24849 |
+
},
|
| 24850 |
+
{
|
| 24851 |
+
"epoch": 0.788,
|
| 24852 |
+
"grad_norm": 0.19378961622714996,
|
| 24853 |
+
"learning_rate": 4.249443207126949e-05,
|
| 24854 |
+
"loss": 0.035,
|
| 24855 |
+
"step": 3546
|
| 24856 |
+
},
|
| 24857 |
+
{
|
| 24858 |
+
"epoch": 0.7882222222222223,
|
| 24859 |
+
"grad_norm": 0.13323235511779785,
|
| 24860 |
+
"learning_rate": 4.244988864142539e-05,
|
| 24861 |
+
"loss": 0.0349,
|
| 24862 |
+
"step": 3547
|
| 24863 |
+
},
|
| 24864 |
+
{
|
| 24865 |
+
"epoch": 0.7884444444444444,
|
| 24866 |
+
"grad_norm": 0.1556854248046875,
|
| 24867 |
+
"learning_rate": 4.240534521158129e-05,
|
| 24868 |
+
"loss": 0.0353,
|
| 24869 |
+
"step": 3548
|
| 24870 |
+
},
|
| 24871 |
+
{
|
| 24872 |
+
"epoch": 0.7886666666666666,
|
| 24873 |
+
"grad_norm": 1.26119065284729,
|
| 24874 |
+
"learning_rate": 4.2360801781737195e-05,
|
| 24875 |
+
"loss": 1.1486,
|
| 24876 |
+
"step": 3549
|
| 24877 |
+
},
|
| 24878 |
+
{
|
| 24879 |
+
"epoch": 0.7888888888888889,
|
| 24880 |
+
"grad_norm": 0.8363838791847229,
|
| 24881 |
+
"learning_rate": 4.2316258351893094e-05,
|
| 24882 |
+
"loss": 0.5973,
|
| 24883 |
+
"step": 3550
|
| 24884 |
+
},
|
| 24885 |
+
{
|
| 24886 |
+
"epoch": 0.7891111111111111,
|
| 24887 |
+
"grad_norm": 0.6089571118354797,
|
| 24888 |
+
"learning_rate": 4.2271714922049e-05,
|
| 24889 |
+
"loss": 1.1064,
|
| 24890 |
+
"step": 3551
|
| 24891 |
+
},
|
| 24892 |
+
{
|
| 24893 |
+
"epoch": 0.7893333333333333,
|
| 24894 |
+
"grad_norm": 0.04548042267560959,
|
| 24895 |
+
"learning_rate": 4.2227171492204905e-05,
|
| 24896 |
+
"loss": 0.0104,
|
| 24897 |
+
"step": 3552
|
| 24898 |
+
},
|
| 24899 |
+
{
|
| 24900 |
+
"epoch": 0.7895555555555556,
|
| 24901 |
+
"grad_norm": 0.6471196413040161,
|
| 24902 |
+
"learning_rate": 4.21826280623608e-05,
|
| 24903 |
+
"loss": 1.1893,
|
| 24904 |
+
"step": 3553
|
| 24905 |
+
},
|
| 24906 |
+
{
|
| 24907 |
+
"epoch": 0.7897777777777778,
|
| 24908 |
+
"grad_norm": 0.046915166079998016,
|
| 24909 |
+
"learning_rate": 4.213808463251671e-05,
|
| 24910 |
+
"loss": 0.0107,
|
| 24911 |
+
"step": 3554
|
| 24912 |
+
},
|
| 24913 |
+
{
|
| 24914 |
+
"epoch": 0.79,
|
| 24915 |
+
"grad_norm": 0.5992786884307861,
|
| 24916 |
+
"learning_rate": 4.209354120267261e-05,
|
| 24917 |
+
"loss": 0.9432,
|
| 24918 |
+
"step": 3555
|
| 24919 |
+
},
|
| 24920 |
+
{
|
| 24921 |
+
"epoch": 0.7902222222222223,
|
| 24922 |
+
"grad_norm": 0.08186815679073334,
|
| 24923 |
+
"learning_rate": 4.2048997772828506e-05,
|
| 24924 |
+
"loss": 0.0118,
|
| 24925 |
+
"step": 3556
|
| 24926 |
+
},
|
| 24927 |
+
{
|
| 24928 |
+
"epoch": 0.7904444444444444,
|
| 24929 |
+
"grad_norm": 0.9210121035575867,
|
| 24930 |
+
"learning_rate": 4.200445434298441e-05,
|
| 24931 |
+
"loss": 2.1219,
|
| 24932 |
+
"step": 3557
|
| 24933 |
+
},
|
| 24934 |
+
{
|
| 24935 |
+
"epoch": 0.7906666666666666,
|
| 24936 |
+
"grad_norm": 1.006956696510315,
|
| 24937 |
+
"learning_rate": 4.1959910913140317e-05,
|
| 24938 |
+
"loss": 1.8555,
|
| 24939 |
+
"step": 3558
|
| 24940 |
+
},
|
| 24941 |
+
{
|
| 24942 |
+
"epoch": 0.7908888888888889,
|
| 24943 |
+
"grad_norm": 0.9316953420639038,
|
| 24944 |
+
"learning_rate": 4.1915367483296215e-05,
|
| 24945 |
+
"loss": 2.0922,
|
| 24946 |
+
"step": 3559
|
| 24947 |
+
},
|
| 24948 |
+
{
|
| 24949 |
+
"epoch": 0.7911111111111111,
|
| 24950 |
+
"grad_norm": 0.9348610639572144,
|
| 24951 |
+
"learning_rate": 4.187082405345212e-05,
|
| 24952 |
+
"loss": 2.2086,
|
| 24953 |
+
"step": 3560
|
| 24954 |
+
},
|
| 24955 |
+
{
|
| 24956 |
+
"epoch": 0.7913333333333333,
|
| 24957 |
+
"grad_norm": 0.884831964969635,
|
| 24958 |
+
"learning_rate": 4.182628062360802e-05,
|
| 24959 |
+
"loss": 1.712,
|
| 24960 |
+
"step": 3561
|
| 24961 |
+
},
|
| 24962 |
+
{
|
| 24963 |
+
"epoch": 0.7915555555555556,
|
| 24964 |
+
"grad_norm": 0.9388924837112427,
|
| 24965 |
+
"learning_rate": 4.178173719376392e-05,
|
| 24966 |
+
"loss": 1.9738,
|
| 24967 |
+
"step": 3562
|
| 24968 |
+
},
|
| 24969 |
+
{
|
| 24970 |
+
"epoch": 0.7917777777777778,
|
| 24971 |
+
"grad_norm": 0.8886390924453735,
|
| 24972 |
+
"learning_rate": 4.173719376391982e-05,
|
| 24973 |
+
"loss": 2.0557,
|
| 24974 |
+
"step": 3563
|
| 24975 |
+
},
|
| 24976 |
+
{
|
| 24977 |
+
"epoch": 0.792,
|
| 24978 |
+
"grad_norm": 0.9279087781906128,
|
| 24979 |
+
"learning_rate": 4.169265033407573e-05,
|
| 24980 |
+
"loss": 1.8811,
|
| 24981 |
+
"step": 3564
|
| 24982 |
+
},
|
| 24983 |
+
{
|
| 24984 |
+
"epoch": 0.7922222222222223,
|
| 24985 |
+
"grad_norm": 0.8624377250671387,
|
| 24986 |
+
"learning_rate": 4.164810690423163e-05,
|
| 24987 |
+
"loss": 1.7352,
|
| 24988 |
+
"step": 3565
|
| 24989 |
+
},
|
| 24990 |
+
{
|
| 24991 |
+
"epoch": 0.7924444444444444,
|
| 24992 |
+
"grad_norm": 0.9816845059394836,
|
| 24993 |
+
"learning_rate": 4.160356347438753e-05,
|
| 24994 |
+
"loss": 2.084,
|
| 24995 |
+
"step": 3566
|
| 24996 |
+
},
|
| 24997 |
+
{
|
| 24998 |
+
"epoch": 0.7926666666666666,
|
| 24999 |
+
"grad_norm": 1.113109827041626,
|
| 25000 |
+
"learning_rate": 4.155902004454343e-05,
|
| 25001 |
+
"loss": 1.8477,
|
| 25002 |
+
"step": 3567
|
| 25003 |
+
},
|
| 25004 |
+
{
|
| 25005 |
+
"epoch": 0.7928888888888889,
|
| 25006 |
+
"grad_norm": 0.9390388131141663,
|
| 25007 |
+
"learning_rate": 4.151447661469933e-05,
|
| 25008 |
+
"loss": 2.0081,
|
| 25009 |
+
"step": 3568
|
| 25010 |
+
},
|
| 25011 |
+
{
|
| 25012 |
+
"epoch": 0.7931111111111111,
|
| 25013 |
+
"grad_norm": 0.06967326253652573,
|
| 25014 |
+
"learning_rate": 4.1469933184855235e-05,
|
| 25015 |
+
"loss": 0.0158,
|
| 25016 |
+
"step": 3569
|
| 25017 |
+
},
|
| 25018 |
+
{
|
| 25019 |
+
"epoch": 0.7933333333333333,
|
| 25020 |
+
"grad_norm": 0.7333217859268188,
|
| 25021 |
+
"learning_rate": 4.142538975501114e-05,
|
| 25022 |
+
"loss": 0.959,
|
| 25023 |
+
"step": 3570
|
| 25024 |
+
},
|
| 25025 |
+
{
|
| 25026 |
+
"epoch": 0.7935555555555556,
|
| 25027 |
+
"grad_norm": 0.9319103360176086,
|
| 25028 |
+
"learning_rate": 4.138084632516704e-05,
|
| 25029 |
+
"loss": 1.7261,
|
| 25030 |
+
"step": 3571
|
| 25031 |
+
},
|
| 25032 |
+
{
|
| 25033 |
+
"epoch": 0.7937777777777778,
|
| 25034 |
+
"grad_norm": 1.187261700630188,
|
| 25035 |
+
"learning_rate": 4.1336302895322944e-05,
|
| 25036 |
+
"loss": 2.1856,
|
| 25037 |
+
"step": 3572
|
| 25038 |
+
},
|
| 25039 |
+
{
|
| 25040 |
+
"epoch": 0.794,
|
| 25041 |
+
"grad_norm": 0.9051704406738281,
|
| 25042 |
+
"learning_rate": 4.129175946547885e-05,
|
| 25043 |
+
"loss": 1.5193,
|
| 25044 |
+
"step": 3573
|
| 25045 |
+
},
|
| 25046 |
+
{
|
| 25047 |
+
"epoch": 0.7942222222222223,
|
| 25048 |
+
"grad_norm": 1.0516455173492432,
|
| 25049 |
+
"learning_rate": 4.124721603563475e-05,
|
| 25050 |
+
"loss": 1.8372,
|
| 25051 |
+
"step": 3574
|
| 25052 |
+
},
|
| 25053 |
+
{
|
| 25054 |
+
"epoch": 0.7944444444444444,
|
| 25055 |
+
"grad_norm": 0.9026862978935242,
|
| 25056 |
+
"learning_rate": 4.120267260579065e-05,
|
| 25057 |
+
"loss": 1.7081,
|
| 25058 |
+
"step": 3575
|
| 25059 |
+
},
|
| 25060 |
+
{
|
| 25061 |
+
"epoch": 0.7946666666666666,
|
| 25062 |
+
"grad_norm": 1.0022499561309814,
|
| 25063 |
+
"learning_rate": 4.115812917594655e-05,
|
| 25064 |
+
"loss": 1.8029,
|
| 25065 |
+
"step": 3576
|
| 25066 |
+
},
|
| 25067 |
+
{
|
| 25068 |
+
"epoch": 0.7948888888888889,
|
| 25069 |
+
"grad_norm": 1.0183892250061035,
|
| 25070 |
+
"learning_rate": 4.111358574610245e-05,
|
| 25071 |
+
"loss": 1.8359,
|
| 25072 |
+
"step": 3577
|
| 25073 |
+
},
|
| 25074 |
+
{
|
| 25075 |
+
"epoch": 0.7951111111111111,
|
| 25076 |
+
"grad_norm": 0.06487128883600235,
|
| 25077 |
+
"learning_rate": 4.1069042316258356e-05,
|
| 25078 |
+
"loss": 0.018,
|
| 25079 |
+
"step": 3578
|
| 25080 |
+
},
|
| 25081 |
+
{
|
| 25082 |
+
"epoch": 0.7953333333333333,
|
| 25083 |
+
"grad_norm": 0.07973368465900421,
|
| 25084 |
+
"learning_rate": 4.1024498886414255e-05,
|
| 25085 |
+
"loss": 0.0177,
|
| 25086 |
+
"step": 3579
|
| 25087 |
+
},
|
| 25088 |
+
{
|
| 25089 |
+
"epoch": 0.7955555555555556,
|
| 25090 |
+
"grad_norm": 0.6818245053291321,
|
| 25091 |
+
"learning_rate": 4.097995545657016e-05,
|
| 25092 |
+
"loss": 0.7042,
|
| 25093 |
+
"step": 3580
|
| 25094 |
+
},
|
| 25095 |
+
{
|
| 25096 |
+
"epoch": 0.7957777777777778,
|
| 25097 |
+
"grad_norm": 0.7157871127128601,
|
| 25098 |
+
"learning_rate": 4.093541202672606e-05,
|
| 25099 |
+
"loss": 0.9386,
|
| 25100 |
+
"step": 3581
|
| 25101 |
+
},
|
| 25102 |
+
{
|
| 25103 |
+
"epoch": 0.796,
|
| 25104 |
+
"grad_norm": 0.0813339352607727,
|
| 25105 |
+
"learning_rate": 4.089086859688196e-05,
|
| 25106 |
+
"loss": 0.0178,
|
| 25107 |
+
"step": 3582
|
| 25108 |
+
},
|
| 25109 |
+
{
|
| 25110 |
+
"epoch": 0.7962222222222223,
|
| 25111 |
+
"grad_norm": 0.07369329035282135,
|
| 25112 |
+
"learning_rate": 4.084632516703786e-05,
|
| 25113 |
+
"loss": 0.0175,
|
| 25114 |
+
"step": 3583
|
| 25115 |
+
},
|
| 25116 |
+
{
|
| 25117 |
+
"epoch": 0.7964444444444444,
|
| 25118 |
+
"grad_norm": 0.07479511946439743,
|
| 25119 |
+
"learning_rate": 4.080178173719377e-05,
|
| 25120 |
+
"loss": 0.0181,
|
| 25121 |
+
"step": 3584
|
| 25122 |
+
},
|
| 25123 |
+
{
|
| 25124 |
+
"epoch": 0.7966666666666666,
|
| 25125 |
+
"grad_norm": 0.07840964198112488,
|
| 25126 |
+
"learning_rate": 4.0757238307349666e-05,
|
| 25127 |
+
"loss": 0.0169,
|
| 25128 |
+
"step": 3585
|
| 25129 |
+
},
|
| 25130 |
+
{
|
| 25131 |
+
"epoch": 0.7968888888888889,
|
| 25132 |
+
"grad_norm": 1.3525023460388184,
|
| 25133 |
+
"learning_rate": 4.071269487750557e-05,
|
| 25134 |
+
"loss": 1.9031,
|
| 25135 |
+
"step": 3586
|
| 25136 |
+
},
|
| 25137 |
+
{
|
| 25138 |
+
"epoch": 0.7971111111111111,
|
| 25139 |
+
"grad_norm": 1.6226398944854736,
|
| 25140 |
+
"learning_rate": 4.066815144766147e-05,
|
| 25141 |
+
"loss": 1.763,
|
| 25142 |
+
"step": 3587
|
| 25143 |
+
},
|
| 25144 |
+
{
|
| 25145 |
+
"epoch": 0.7973333333333333,
|
| 25146 |
+
"grad_norm": 0.9485257267951965,
|
| 25147 |
+
"learning_rate": 4.062360801781737e-05,
|
| 25148 |
+
"loss": 1.4717,
|
| 25149 |
+
"step": 3588
|
| 25150 |
+
},
|
| 25151 |
+
{
|
| 25152 |
+
"epoch": 0.7975555555555556,
|
| 25153 |
+
"grad_norm": 0.9942083358764648,
|
| 25154 |
+
"learning_rate": 4.0579064587973274e-05,
|
| 25155 |
+
"loss": 1.61,
|
| 25156 |
+
"step": 3589
|
| 25157 |
+
},
|
| 25158 |
+
{
|
| 25159 |
+
"epoch": 0.7977777777777778,
|
| 25160 |
+
"grad_norm": 0.10811223834753036,
|
| 25161 |
+
"learning_rate": 4.053452115812918e-05,
|
| 25162 |
+
"loss": 0.0255,
|
| 25163 |
+
"step": 3590
|
| 25164 |
+
},
|
| 25165 |
+
{
|
| 25166 |
+
"epoch": 0.798,
|
| 25167 |
+
"grad_norm": 0.9328152537345886,
|
| 25168 |
+
"learning_rate": 4.048997772828508e-05,
|
| 25169 |
+
"loss": 1.6944,
|
| 25170 |
+
"step": 3591
|
| 25171 |
+
},
|
| 25172 |
+
{
|
| 25173 |
+
"epoch": 0.7982222222222223,
|
| 25174 |
+
"grad_norm": 1.1752344369888306,
|
| 25175 |
+
"learning_rate": 4.0445434298440984e-05,
|
| 25176 |
+
"loss": 1.5285,
|
| 25177 |
+
"step": 3592
|
| 25178 |
+
},
|
| 25179 |
+
{
|
| 25180 |
+
"epoch": 0.7984444444444444,
|
| 25181 |
+
"grad_norm": 1.2680740356445312,
|
| 25182 |
+
"learning_rate": 4.040089086859689e-05,
|
| 25183 |
+
"loss": 1.4009,
|
| 25184 |
+
"step": 3593
|
| 25185 |
+
},
|
| 25186 |
+
{
|
| 25187 |
+
"epoch": 0.7986666666666666,
|
| 25188 |
+
"grad_norm": 1.1118732690811157,
|
| 25189 |
+
"learning_rate": 4.035634743875278e-05,
|
| 25190 |
+
"loss": 1.3982,
|
| 25191 |
+
"step": 3594
|
| 25192 |
+
},
|
| 25193 |
+
{
|
| 25194 |
+
"epoch": 0.7988888888888889,
|
| 25195 |
+
"grad_norm": 1.389929175376892,
|
| 25196 |
+
"learning_rate": 4.0311804008908686e-05,
|
| 25197 |
+
"loss": 1.3998,
|
| 25198 |
+
"step": 3595
|
| 25199 |
+
},
|
| 25200 |
+
{
|
| 25201 |
+
"epoch": 0.7991111111111111,
|
| 25202 |
+
"grad_norm": 0.8431739807128906,
|
| 25203 |
+
"learning_rate": 4.026726057906459e-05,
|
| 25204 |
+
"loss": 0.8518,
|
| 25205 |
+
"step": 3596
|
| 25206 |
+
},
|
| 25207 |
+
{
|
| 25208 |
+
"epoch": 0.7993333333333333,
|
| 25209 |
+
"grad_norm": 1.3290144205093384,
|
| 25210 |
+
"learning_rate": 4.022271714922049e-05,
|
| 25211 |
+
"loss": 1.4607,
|
| 25212 |
+
"step": 3597
|
| 25213 |
+
},
|
| 25214 |
+
{
|
| 25215 |
+
"epoch": 0.7995555555555556,
|
| 25216 |
+
"grad_norm": 0.15491686761379242,
|
| 25217 |
+
"learning_rate": 4.0178173719376395e-05,
|
| 25218 |
+
"loss": 0.0335,
|
| 25219 |
+
"step": 3598
|
| 25220 |
+
},
|
| 25221 |
+
{
|
| 25222 |
+
"epoch": 0.7997777777777778,
|
| 25223 |
+
"grad_norm": 0.7583034038543701,
|
| 25224 |
+
"learning_rate": 4.01336302895323e-05,
|
| 25225 |
+
"loss": 0.4833,
|
| 25226 |
+
"step": 3599
|
| 25227 |
+
},
|
| 25228 |
+
{
|
| 25229 |
+
"epoch": 0.8,
|
| 25230 |
+
"grad_norm": 0.900272786617279,
|
| 25231 |
+
"learning_rate": 4.00890868596882e-05,
|
| 25232 |
+
"loss": 0.8159,
|
| 25233 |
+
"step": 3600
|
| 25234 |
+
},
|
| 25235 |
+
{
|
| 25236 |
+
"epoch": 0.8,
|
| 25237 |
+
"eval_loss": 1.1739096641540527,
|
| 25238 |
+
"eval_runtime": 239.8403,
|
| 25239 |
+
"eval_samples_per_second": 4.169,
|
| 25240 |
+
"eval_steps_per_second": 4.169,
|
| 25241 |
+
"step": 3600
|
| 25242 |
+
},
|
| 25243 |
+
{
|
| 25244 |
+
"epoch": 0.8002222222222222,
|
| 25245 |
+
"grad_norm": 0.5315160155296326,
|
| 25246 |
+
"learning_rate": 4.00445434298441e-05,
|
| 25247 |
+
"loss": 1.0474,
|
| 25248 |
+
"step": 3601
|
| 25249 |
+
},
|
| 25250 |
+
{
|
| 25251 |
+
"epoch": 0.8004444444444444,
|
| 25252 |
+
"grad_norm": 0.5535997152328491,
|
| 25253 |
+
"learning_rate": 4e-05,
|
| 25254 |
+
"loss": 0.9027,
|
| 25255 |
+
"step": 3602
|
| 25256 |
+
},
|
| 25257 |
+
{
|
| 25258 |
+
"epoch": 0.8006666666666666,
|
| 25259 |
+
"grad_norm": 0.7079357504844666,
|
| 25260 |
+
"learning_rate": 3.99554565701559e-05,
|
| 25261 |
+
"loss": 1.1432,
|
| 25262 |
+
"step": 3603
|
| 25263 |
+
},
|
| 25264 |
+
{
|
| 25265 |
+
"epoch": 0.8008888888888889,
|
| 25266 |
+
"grad_norm": 0.6734809875488281,
|
| 25267 |
+
"learning_rate": 3.991091314031181e-05,
|
| 25268 |
+
"loss": 1.0644,
|
| 25269 |
+
"step": 3604
|
| 25270 |
+
},
|
| 25271 |
+
{
|
| 25272 |
+
"epoch": 0.8011111111111111,
|
| 25273 |
+
"grad_norm": 0.6071887016296387,
|
| 25274 |
+
"learning_rate": 3.986636971046771e-05,
|
| 25275 |
+
"loss": 1.101,
|
| 25276 |
+
"step": 3605
|
| 25277 |
+
},
|
| 25278 |
+
{
|
| 25279 |
+
"epoch": 0.8013333333333333,
|
| 25280 |
+
"grad_norm": 0.5823980569839478,
|
| 25281 |
+
"learning_rate": 3.982182628062361e-05,
|
| 25282 |
+
"loss": 0.8761,
|
| 25283 |
+
"step": 3606
|
| 25284 |
+
},
|
| 25285 |
+
{
|
| 25286 |
+
"epoch": 0.8015555555555556,
|
| 25287 |
+
"grad_norm": 0.09482403099536896,
|
| 25288 |
+
"learning_rate": 3.977728285077951e-05,
|
| 25289 |
+
"loss": 0.0129,
|
| 25290 |
+
"step": 3607
|
| 25291 |
+
},
|
| 25292 |
+
{
|
| 25293 |
+
"epoch": 0.8017777777777778,
|
| 25294 |
+
"grad_norm": 0.588431179523468,
|
| 25295 |
+
"learning_rate": 3.9732739420935415e-05,
|
| 25296 |
+
"loss": 1.0989,
|
| 25297 |
+
"step": 3608
|
| 25298 |
+
},
|
| 25299 |
+
{
|
| 25300 |
+
"epoch": 0.802,
|
| 25301 |
+
"grad_norm": 0.9978165626525879,
|
| 25302 |
+
"learning_rate": 3.9688195991091314e-05,
|
| 25303 |
+
"loss": 1.6576,
|
| 25304 |
+
"step": 3609
|
| 25305 |
+
},
|
| 25306 |
+
{
|
| 25307 |
+
"epoch": 0.8022222222222222,
|
| 25308 |
+
"grad_norm": 0.9243869185447693,
|
| 25309 |
+
"learning_rate": 3.964365256124722e-05,
|
| 25310 |
+
"loss": 2.1462,
|
| 25311 |
+
"step": 3610
|
| 25312 |
+
},
|
| 25313 |
+
{
|
| 25314 |
+
"epoch": 0.8024444444444444,
|
| 25315 |
+
"grad_norm": 0.8666023015975952,
|
| 25316 |
+
"learning_rate": 3.9599109131403124e-05,
|
| 25317 |
+
"loss": 2.0262,
|
| 25318 |
+
"step": 3611
|
| 25319 |
+
},
|
| 25320 |
+
{
|
| 25321 |
+
"epoch": 0.8026666666666666,
|
| 25322 |
+
"grad_norm": 0.9008248448371887,
|
| 25323 |
+
"learning_rate": 3.955456570155902e-05,
|
| 25324 |
+
"loss": 2.0854,
|
| 25325 |
+
"step": 3612
|
| 25326 |
+
},
|
| 25327 |
+
{
|
| 25328 |
+
"epoch": 0.8028888888888889,
|
| 25329 |
+
"grad_norm": 0.6247386336326599,
|
| 25330 |
+
"learning_rate": 3.951002227171492e-05,
|
| 25331 |
+
"loss": 0.8506,
|
| 25332 |
+
"step": 3613
|
| 25333 |
+
},
|
| 25334 |
+
{
|
| 25335 |
+
"epoch": 0.8031111111111111,
|
| 25336 |
+
"grad_norm": 0.13101428747177124,
|
| 25337 |
+
"learning_rate": 3.946547884187082e-05,
|
| 25338 |
+
"loss": 0.0223,
|
| 25339 |
+
"step": 3614
|
| 25340 |
+
},
|
| 25341 |
+
{
|
| 25342 |
+
"epoch": 0.8033333333333333,
|
| 25343 |
+
"grad_norm": 0.12358218431472778,
|
| 25344 |
+
"learning_rate": 3.9420935412026726e-05,
|
| 25345 |
+
"loss": 0.0211,
|
| 25346 |
+
"step": 3615
|
| 25347 |
+
},
|
| 25348 |
+
{
|
| 25349 |
+
"epoch": 0.8035555555555556,
|
| 25350 |
+
"grad_norm": 0.9996263980865479,
|
| 25351 |
+
"learning_rate": 3.937639198218263e-05,
|
| 25352 |
+
"loss": 2.0538,
|
| 25353 |
+
"step": 3616
|
| 25354 |
+
},
|
| 25355 |
+
{
|
| 25356 |
+
"epoch": 0.8037777777777778,
|
| 25357 |
+
"grad_norm": 1.048120379447937,
|
| 25358 |
+
"learning_rate": 3.933184855233853e-05,
|
| 25359 |
+
"loss": 1.6907,
|
| 25360 |
+
"step": 3617
|
| 25361 |
+
},
|
| 25362 |
+
{
|
| 25363 |
+
"epoch": 0.804,
|
| 25364 |
+
"grad_norm": 0.9704152941703796,
|
| 25365 |
+
"learning_rate": 3.9287305122494435e-05,
|
| 25366 |
+
"loss": 2.025,
|
| 25367 |
+
"step": 3618
|
| 25368 |
+
},
|
| 25369 |
+
{
|
| 25370 |
+
"epoch": 0.8042222222222222,
|
| 25371 |
+
"grad_norm": 0.9544731378555298,
|
| 25372 |
+
"learning_rate": 3.924276169265034e-05,
|
| 25373 |
+
"loss": 1.8855,
|
| 25374 |
+
"step": 3619
|
| 25375 |
+
},
|
| 25376 |
+
{
|
| 25377 |
+
"epoch": 0.8044444444444444,
|
| 25378 |
+
"grad_norm": 0.869174599647522,
|
| 25379 |
+
"learning_rate": 3.919821826280624e-05,
|
| 25380 |
+
"loss": 2.1548,
|
| 25381 |
+
"step": 3620
|
| 25382 |
+
},
|
| 25383 |
+
{
|
| 25384 |
+
"epoch": 0.8046666666666666,
|
| 25385 |
+
"grad_norm": 0.7145273685455322,
|
| 25386 |
+
"learning_rate": 3.915367483296214e-05,
|
| 25387 |
+
"loss": 1.0897,
|
| 25388 |
+
"step": 3621
|
| 25389 |
+
},
|
| 25390 |
+
{
|
| 25391 |
+
"epoch": 0.8048888888888889,
|
| 25392 |
+
"grad_norm": 0.06968183070421219,
|
| 25393 |
+
"learning_rate": 3.910913140311804e-05,
|
| 25394 |
+
"loss": 0.0157,
|
| 25395 |
+
"step": 3622
|
| 25396 |
+
},
|
| 25397 |
+
{
|
| 25398 |
+
"epoch": 0.8051111111111111,
|
| 25399 |
+
"grad_norm": 0.636101484298706,
|
| 25400 |
+
"learning_rate": 3.906458797327394e-05,
|
| 25401 |
+
"loss": 0.8512,
|
| 25402 |
+
"step": 3623
|
| 25403 |
+
},
|
| 25404 |
+
{
|
| 25405 |
+
"epoch": 0.8053333333333333,
|
| 25406 |
+
"grad_norm": 1.1181496381759644,
|
| 25407 |
+
"learning_rate": 3.902004454342985e-05,
|
| 25408 |
+
"loss": 1.5164,
|
| 25409 |
+
"step": 3624
|
| 25410 |
+
},
|
| 25411 |
+
{
|
| 25412 |
+
"epoch": 0.8055555555555556,
|
| 25413 |
+
"grad_norm": 0.6359825730323792,
|
| 25414 |
+
"learning_rate": 3.897550111358575e-05,
|
| 25415 |
+
"loss": 1.0128,
|
| 25416 |
+
"step": 3625
|
| 25417 |
+
},
|
| 25418 |
+
{
|
| 25419 |
+
"epoch": 0.8057777777777778,
|
| 25420 |
+
"grad_norm": 0.943658173084259,
|
| 25421 |
+
"learning_rate": 3.893095768374165e-05,
|
| 25422 |
+
"loss": 1.582,
|
| 25423 |
+
"step": 3626
|
| 25424 |
+
},
|
| 25425 |
+
{
|
| 25426 |
+
"epoch": 0.806,
|
| 25427 |
+
"grad_norm": 1.047963261604309,
|
| 25428 |
+
"learning_rate": 3.888641425389755e-05,
|
| 25429 |
+
"loss": 1.704,
|
| 25430 |
+
"step": 3627
|
| 25431 |
+
},
|
| 25432 |
+
{
|
| 25433 |
+
"epoch": 0.8062222222222222,
|
| 25434 |
+
"grad_norm": 0.9852431416511536,
|
| 25435 |
+
"learning_rate": 3.8841870824053455e-05,
|
| 25436 |
+
"loss": 1.6806,
|
| 25437 |
+
"step": 3628
|
| 25438 |
+
},
|
| 25439 |
+
{
|
| 25440 |
+
"epoch": 0.8064444444444444,
|
| 25441 |
+
"grad_norm": 0.8574654459953308,
|
| 25442 |
+
"learning_rate": 3.879732739420935e-05,
|
| 25443 |
+
"loss": 0.916,
|
| 25444 |
+
"step": 3629
|
| 25445 |
+
},
|
| 25446 |
+
{
|
| 25447 |
+
"epoch": 0.8066666666666666,
|
| 25448 |
+
"grad_norm": 0.9380385279655457,
|
| 25449 |
+
"learning_rate": 3.875278396436526e-05,
|
| 25450 |
+
"loss": 0.9947,
|
| 25451 |
+
"step": 3630
|
| 25452 |
+
},
|
| 25453 |
+
{
|
| 25454 |
+
"epoch": 0.8068888888888889,
|
| 25455 |
+
"grad_norm": 0.06696880608797073,
|
| 25456 |
+
"learning_rate": 3.8708240534521164e-05,
|
| 25457 |
+
"loss": 0.017,
|
| 25458 |
+
"step": 3631
|
| 25459 |
+
},
|
| 25460 |
+
{
|
| 25461 |
+
"epoch": 0.8071111111111111,
|
| 25462 |
+
"grad_norm": 0.0664907768368721,
|
| 25463 |
+
"learning_rate": 3.866369710467706e-05,
|
| 25464 |
+
"loss": 0.0173,
|
| 25465 |
+
"step": 3632
|
| 25466 |
+
},
|
| 25467 |
+
{
|
| 25468 |
+
"epoch": 0.8073333333333333,
|
| 25469 |
+
"grad_norm": 0.06398806720972061,
|
| 25470 |
+
"learning_rate": 3.861915367483296e-05,
|
| 25471 |
+
"loss": 0.0171,
|
| 25472 |
+
"step": 3633
|
| 25473 |
+
},
|
| 25474 |
+
{
|
| 25475 |
+
"epoch": 0.8075555555555556,
|
| 25476 |
+
"grad_norm": 1.21707022190094,
|
| 25477 |
+
"learning_rate": 3.8574610244988866e-05,
|
| 25478 |
+
"loss": 2.0207,
|
| 25479 |
+
"step": 3634
|
| 25480 |
+
},
|
| 25481 |
+
{
|
| 25482 |
+
"epoch": 0.8077777777777778,
|
| 25483 |
+
"grad_norm": 1.04575777053833,
|
| 25484 |
+
"learning_rate": 3.8530066815144765e-05,
|
| 25485 |
+
"loss": 1.8143,
|
| 25486 |
+
"step": 3635
|
| 25487 |
+
},
|
| 25488 |
+
{
|
| 25489 |
+
"epoch": 0.808,
|
| 25490 |
+
"grad_norm": 0.0705951601266861,
|
| 25491 |
+
"learning_rate": 3.848552338530067e-05,
|
| 25492 |
+
"loss": 0.0166,
|
| 25493 |
+
"step": 3636
|
| 25494 |
+
},
|
| 25495 |
+
{
|
| 25496 |
+
"epoch": 0.8082222222222222,
|
| 25497 |
+
"grad_norm": 0.076121024787426,
|
| 25498 |
+
"learning_rate": 3.8440979955456576e-05,
|
| 25499 |
+
"loss": 0.0167,
|
| 25500 |
+
"step": 3637
|
| 25501 |
+
},
|
| 25502 |
+
{
|
| 25503 |
+
"epoch": 0.8084444444444444,
|
| 25504 |
+
"grad_norm": 1.0178598165512085,
|
| 25505 |
+
"learning_rate": 3.8396436525612474e-05,
|
| 25506 |
+
"loss": 1.6507,
|
| 25507 |
+
"step": 3638
|
| 25508 |
+
},
|
| 25509 |
+
{
|
| 25510 |
+
"epoch": 0.8086666666666666,
|
| 25511 |
+
"grad_norm": 1.102067232131958,
|
| 25512 |
+
"learning_rate": 3.835189309576838e-05,
|
| 25513 |
+
"loss": 1.7494,
|
| 25514 |
+
"step": 3639
|
| 25515 |
+
},
|
| 25516 |
+
{
|
| 25517 |
+
"epoch": 0.8088888888888889,
|
| 25518 |
+
"grad_norm": 0.10027281194925308,
|
| 25519 |
+
"learning_rate": 3.830734966592428e-05,
|
| 25520 |
+
"loss": 0.0251,
|
| 25521 |
+
"step": 3640
|
| 25522 |
+
},
|
| 25523 |
+
{
|
| 25524 |
+
"epoch": 0.8091111111111111,
|
| 25525 |
+
"grad_norm": 0.6546877026557922,
|
| 25526 |
+
"learning_rate": 3.826280623608018e-05,
|
| 25527 |
+
"loss": 0.6345,
|
| 25528 |
+
"step": 3641
|
| 25529 |
+
},
|
| 25530 |
+
{
|
| 25531 |
+
"epoch": 0.8093333333333333,
|
| 25532 |
+
"grad_norm": 1.25735342502594,
|
| 25533 |
+
"learning_rate": 3.821826280623608e-05,
|
| 25534 |
+
"loss": 1.7112,
|
| 25535 |
+
"step": 3642
|
| 25536 |
+
},
|
| 25537 |
+
{
|
| 25538 |
+
"epoch": 0.8095555555555556,
|
| 25539 |
+
"grad_norm": 1.1318788528442383,
|
| 25540 |
+
"learning_rate": 3.817371937639199e-05,
|
| 25541 |
+
"loss": 1.5739,
|
| 25542 |
+
"step": 3643
|
| 25543 |
+
},
|
| 25544 |
+
{
|
| 25545 |
+
"epoch": 0.8097777777777778,
|
| 25546 |
+
"grad_norm": 1.0459527969360352,
|
| 25547 |
+
"learning_rate": 3.8129175946547886e-05,
|
| 25548 |
+
"loss": 1.3376,
|
| 25549 |
+
"step": 3644
|
| 25550 |
+
},
|
| 25551 |
+
{
|
| 25552 |
+
"epoch": 0.81,
|
| 25553 |
+
"grad_norm": 1.0099126100540161,
|
| 25554 |
+
"learning_rate": 3.808463251670379e-05,
|
| 25555 |
+
"loss": 1.2814,
|
| 25556 |
+
"step": 3645
|
| 25557 |
+
},
|
| 25558 |
+
{
|
| 25559 |
+
"epoch": 0.8102222222222222,
|
| 25560 |
+
"grad_norm": 0.82170170545578,
|
| 25561 |
+
"learning_rate": 3.804008908685969e-05,
|
| 25562 |
+
"loss": 0.9743,
|
| 25563 |
+
"step": 3646
|
| 25564 |
+
},
|
| 25565 |
+
{
|
| 25566 |
+
"epoch": 0.8104444444444444,
|
| 25567 |
+
"grad_norm": 1.0343072414398193,
|
| 25568 |
+
"learning_rate": 3.799554565701559e-05,
|
| 25569 |
+
"loss": 1.4283,
|
| 25570 |
+
"step": 3647
|
| 25571 |
+
},
|
| 25572 |
+
{
|
| 25573 |
+
"epoch": 0.8106666666666666,
|
| 25574 |
+
"grad_norm": 0.944911777973175,
|
| 25575 |
+
"learning_rate": 3.7951002227171494e-05,
|
| 25576 |
+
"loss": 1.05,
|
| 25577 |
+
"step": 3648
|
| 25578 |
+
},
|
| 25579 |
+
{
|
| 25580 |
+
"epoch": 0.8108888888888889,
|
| 25581 |
+
"grad_norm": 0.7252945303916931,
|
| 25582 |
+
"learning_rate": 3.79064587973274e-05,
|
| 25583 |
+
"loss": 0.6662,
|
| 25584 |
+
"step": 3649
|
| 25585 |
+
},
|
| 25586 |
+
{
|
| 25587 |
+
"epoch": 0.8111111111111111,
|
| 25588 |
+
"grad_norm": 1.005825400352478,
|
| 25589 |
+
"learning_rate": 3.78619153674833e-05,
|
| 25590 |
+
"loss": 0.7958,
|
| 25591 |
+
"step": 3650
|
| 25592 |
+
},
|
| 25593 |
+
{
|
| 25594 |
+
"epoch": 0.8113333333333334,
|
| 25595 |
+
"grad_norm": 0.82235187292099,
|
| 25596 |
+
"learning_rate": 3.78173719376392e-05,
|
| 25597 |
+
"loss": 2.2468,
|
| 25598 |
+
"step": 3651
|
| 25599 |
+
},
|
| 25600 |
+
{
|
| 25601 |
+
"epoch": 0.8115555555555556,
|
| 25602 |
+
"grad_norm": 0.6531208157539368,
|
| 25603 |
+
"learning_rate": 3.77728285077951e-05,
|
| 25604 |
+
"loss": 0.9914,
|
| 25605 |
+
"step": 3652
|
| 25606 |
+
},
|
| 25607 |
+
{
|
| 25608 |
+
"epoch": 0.8117777777777778,
|
| 25609 |
+
"grad_norm": 0.8134424686431885,
|
| 25610 |
+
"learning_rate": 3.7728285077951e-05,
|
| 25611 |
+
"loss": 2.2763,
|
| 25612 |
+
"step": 3653
|
| 25613 |
+
},
|
| 25614 |
+
{
|
| 25615 |
+
"epoch": 0.812,
|
| 25616 |
+
"grad_norm": 0.043821610510349274,
|
| 25617 |
+
"learning_rate": 3.7683741648106906e-05,
|
| 25618 |
+
"loss": 0.0101,
|
| 25619 |
+
"step": 3654
|
| 25620 |
+
},
|
| 25621 |
+
{
|
| 25622 |
+
"epoch": 0.8122222222222222,
|
| 25623 |
+
"grad_norm": 0.6952782273292542,
|
| 25624 |
+
"learning_rate": 3.7639198218262804e-05,
|
| 25625 |
+
"loss": 1.234,
|
| 25626 |
+
"step": 3655
|
| 25627 |
+
},
|
| 25628 |
+
{
|
| 25629 |
+
"epoch": 0.8124444444444444,
|
| 25630 |
+
"grad_norm": 0.6162470579147339,
|
| 25631 |
+
"learning_rate": 3.759465478841871e-05,
|
| 25632 |
+
"loss": 1.0527,
|
| 25633 |
+
"step": 3656
|
| 25634 |
+
},
|
| 25635 |
+
{
|
| 25636 |
+
"epoch": 0.8126666666666666,
|
| 25637 |
+
"grad_norm": 1.0366730690002441,
|
| 25638 |
+
"learning_rate": 3.7550111358574615e-05,
|
| 25639 |
+
"loss": 2.3824,
|
| 25640 |
+
"step": 3657
|
| 25641 |
+
},
|
| 25642 |
+
{
|
| 25643 |
+
"epoch": 0.8128888888888889,
|
| 25644 |
+
"grad_norm": 0.08013699948787689,
|
| 25645 |
+
"learning_rate": 3.7505567928730514e-05,
|
| 25646 |
+
"loss": 0.0114,
|
| 25647 |
+
"step": 3658
|
| 25648 |
+
},
|
| 25649 |
+
{
|
| 25650 |
+
"epoch": 0.8131111111111111,
|
| 25651 |
+
"grad_norm": 0.07211296260356903,
|
| 25652 |
+
"learning_rate": 3.746102449888642e-05,
|
| 25653 |
+
"loss": 0.0114,
|
| 25654 |
+
"step": 3659
|
| 25655 |
+
},
|
| 25656 |
+
{
|
| 25657 |
+
"epoch": 0.8133333333333334,
|
| 25658 |
+
"grad_norm": 0.07523675262928009,
|
| 25659 |
+
"learning_rate": 3.741648106904232e-05,
|
| 25660 |
+
"loss": 0.0114,
|
| 25661 |
+
"step": 3660
|
| 25662 |
+
},
|
| 25663 |
+
{
|
| 25664 |
+
"epoch": 0.8135555555555556,
|
| 25665 |
+
"grad_norm": 0.4986688494682312,
|
| 25666 |
+
"learning_rate": 3.7371937639198216e-05,
|
| 25667 |
+
"loss": 0.946,
|
| 25668 |
+
"step": 3661
|
| 25669 |
+
},
|
| 25670 |
+
{
|
| 25671 |
+
"epoch": 0.8137777777777778,
|
| 25672 |
+
"grad_norm": 0.9343963265419006,
|
| 25673 |
+
"learning_rate": 3.732739420935412e-05,
|
| 25674 |
+
"loss": 2.0871,
|
| 25675 |
+
"step": 3662
|
| 25676 |
+
},
|
| 25677 |
+
{
|
| 25678 |
+
"epoch": 0.814,
|
| 25679 |
+
"grad_norm": 0.881712794303894,
|
| 25680 |
+
"learning_rate": 3.728285077951003e-05,
|
| 25681 |
+
"loss": 2.0004,
|
| 25682 |
+
"step": 3663
|
| 25683 |
+
},
|
| 25684 |
+
{
|
| 25685 |
+
"epoch": 0.8142222222222222,
|
| 25686 |
+
"grad_norm": 1.054946780204773,
|
| 25687 |
+
"learning_rate": 3.7238307349665925e-05,
|
| 25688 |
+
"loss": 2.1722,
|
| 25689 |
+
"step": 3664
|
| 25690 |
+
},
|
| 25691 |
+
{
|
| 25692 |
+
"epoch": 0.8144444444444444,
|
| 25693 |
+
"grad_norm": 0.9730517268180847,
|
| 25694 |
+
"learning_rate": 3.719376391982183e-05,
|
| 25695 |
+
"loss": 1.7299,
|
| 25696 |
+
"step": 3665
|
| 25697 |
+
},
|
| 25698 |
+
{
|
| 25699 |
+
"epoch": 0.8146666666666667,
|
| 25700 |
+
"grad_norm": 0.8659468293190002,
|
| 25701 |
+
"learning_rate": 3.714922048997773e-05,
|
| 25702 |
+
"loss": 2.1389,
|
| 25703 |
+
"step": 3666
|
| 25704 |
+
},
|
| 25705 |
+
{
|
| 25706 |
+
"epoch": 0.8148888888888889,
|
| 25707 |
+
"grad_norm": 0.6216636896133423,
|
| 25708 |
+
"learning_rate": 3.710467706013363e-05,
|
| 25709 |
+
"loss": 1.0111,
|
| 25710 |
+
"step": 3667
|
| 25711 |
+
},
|
| 25712 |
+
{
|
| 25713 |
+
"epoch": 0.8151111111111111,
|
| 25714 |
+
"grad_norm": 0.1096658706665039,
|
| 25715 |
+
"learning_rate": 3.706013363028953e-05,
|
| 25716 |
+
"loss": 0.0178,
|
| 25717 |
+
"step": 3668
|
| 25718 |
+
},
|
| 25719 |
+
{
|
| 25720 |
+
"epoch": 0.8153333333333334,
|
| 25721 |
+
"grad_norm": 0.6512637138366699,
|
| 25722 |
+
"learning_rate": 3.701559020044544e-05,
|
| 25723 |
+
"loss": 1.0441,
|
| 25724 |
+
"step": 3669
|
| 25725 |
+
},
|
| 25726 |
+
{
|
| 25727 |
+
"epoch": 0.8155555555555556,
|
| 25728 |
+
"grad_norm": 1.0497286319732666,
|
| 25729 |
+
"learning_rate": 3.697104677060134e-05,
|
| 25730 |
+
"loss": 1.7955,
|
| 25731 |
+
"step": 3670
|
| 25732 |
+
},
|
| 25733 |
+
{
|
| 25734 |
+
"epoch": 0.8157777777777778,
|
| 25735 |
+
"grad_norm": 0.9653757214546204,
|
| 25736 |
+
"learning_rate": 3.692650334075724e-05,
|
| 25737 |
+
"loss": 1.8275,
|
| 25738 |
+
"step": 3671
|
| 25739 |
+
},
|
| 25740 |
+
{
|
| 25741 |
+
"epoch": 0.816,
|
| 25742 |
+
"grad_norm": 0.6577117443084717,
|
| 25743 |
+
"learning_rate": 3.688195991091314e-05,
|
| 25744 |
+
"loss": 0.9564,
|
| 25745 |
+
"step": 3672
|
| 25746 |
+
},
|
| 25747 |
+
{
|
| 25748 |
+
"epoch": 0.8162222222222222,
|
| 25749 |
+
"grad_norm": 0.07206307351589203,
|
| 25750 |
+
"learning_rate": 3.683741648106904e-05,
|
| 25751 |
+
"loss": 0.0158,
|
| 25752 |
+
"step": 3673
|
| 25753 |
+
},
|
| 25754 |
+
{
|
| 25755 |
+
"epoch": 0.8164444444444444,
|
| 25756 |
+
"grad_norm": 0.06910723447799683,
|
| 25757 |
+
"learning_rate": 3.6792873051224945e-05,
|
| 25758 |
+
"loss": 0.0159,
|
| 25759 |
+
"step": 3674
|
| 25760 |
+
},
|
| 25761 |
+
{
|
| 25762 |
+
"epoch": 0.8166666666666667,
|
| 25763 |
+
"grad_norm": 0.9517031908035278,
|
| 25764 |
+
"learning_rate": 3.674832962138085e-05,
|
| 25765 |
+
"loss": 1.9874,
|
| 25766 |
+
"step": 3675
|
| 25767 |
+
},
|
| 25768 |
+
{
|
| 25769 |
+
"epoch": 0.8168888888888889,
|
| 25770 |
+
"grad_norm": 0.1267796903848648,
|
| 25771 |
+
"learning_rate": 3.670378619153675e-05,
|
| 25772 |
+
"loss": 0.0203,
|
| 25773 |
+
"step": 3676
|
| 25774 |
+
},
|
| 25775 |
+
{
|
| 25776 |
+
"epoch": 0.8171111111111111,
|
| 25777 |
+
"grad_norm": 0.6326009631156921,
|
| 25778 |
+
"learning_rate": 3.6659242761692654e-05,
|
| 25779 |
+
"loss": 0.8404,
|
| 25780 |
+
"step": 3677
|
| 25781 |
+
},
|
| 25782 |
+
{
|
| 25783 |
+
"epoch": 0.8173333333333334,
|
| 25784 |
+
"grad_norm": 0.950645387172699,
|
| 25785 |
+
"learning_rate": 3.661469933184856e-05,
|
| 25786 |
+
"loss": 1.7112,
|
| 25787 |
+
"step": 3678
|
| 25788 |
+
},
|
| 25789 |
+
{
|
| 25790 |
+
"epoch": 0.8175555555555556,
|
| 25791 |
+
"grad_norm": 1.134836196899414,
|
| 25792 |
+
"learning_rate": 3.657015590200445e-05,
|
| 25793 |
+
"loss": 1.8671,
|
| 25794 |
+
"step": 3679
|
| 25795 |
+
},
|
| 25796 |
+
{
|
| 25797 |
+
"epoch": 0.8177777777777778,
|
| 25798 |
+
"grad_norm": 0.9950535893440247,
|
| 25799 |
+
"learning_rate": 3.652561247216036e-05,
|
| 25800 |
+
"loss": 1.9121,
|
| 25801 |
+
"step": 3680
|
| 25802 |
+
},
|
| 25803 |
+
{
|
| 25804 |
+
"epoch": 0.818,
|
| 25805 |
+
"grad_norm": 0.06296125799417496,
|
| 25806 |
+
"learning_rate": 3.648106904231626e-05,
|
| 25807 |
+
"loss": 0.0175,
|
| 25808 |
+
"step": 3681
|
| 25809 |
+
},
|
| 25810 |
+
{
|
| 25811 |
+
"epoch": 0.8182222222222222,
|
| 25812 |
+
"grad_norm": 0.9528807401657104,
|
| 25813 |
+
"learning_rate": 3.643652561247216e-05,
|
| 25814 |
+
"loss": 1.7288,
|
| 25815 |
+
"step": 3682
|
| 25816 |
+
},
|
| 25817 |
+
{
|
| 25818 |
+
"epoch": 0.8184444444444444,
|
| 25819 |
+
"grad_norm": 1.0748895406723022,
|
| 25820 |
+
"learning_rate": 3.6391982182628066e-05,
|
| 25821 |
+
"loss": 1.3809,
|
| 25822 |
+
"step": 3683
|
| 25823 |
+
},
|
| 25824 |
+
{
|
| 25825 |
+
"epoch": 0.8186666666666667,
|
| 25826 |
+
"grad_norm": 0.07369447499513626,
|
| 25827 |
+
"learning_rate": 3.634743875278397e-05,
|
| 25828 |
+
"loss": 0.0173,
|
| 25829 |
+
"step": 3684
|
| 25830 |
+
},
|
| 25831 |
+
{
|
| 25832 |
+
"epoch": 0.8188888888888889,
|
| 25833 |
+
"grad_norm": 0.07020772248506546,
|
| 25834 |
+
"learning_rate": 3.630289532293987e-05,
|
| 25835 |
+
"loss": 0.0172,
|
| 25836 |
+
"step": 3685
|
| 25837 |
+
},
|
| 25838 |
+
{
|
| 25839 |
+
"epoch": 0.8191111111111111,
|
| 25840 |
+
"grad_norm": 0.7225638628005981,
|
| 25841 |
+
"learning_rate": 3.625835189309577e-05,
|
| 25842 |
+
"loss": 0.9078,
|
| 25843 |
+
"step": 3686
|
| 25844 |
+
},
|
| 25845 |
+
{
|
| 25846 |
+
"epoch": 0.8193333333333334,
|
| 25847 |
+
"grad_norm": 1.205776333808899,
|
| 25848 |
+
"learning_rate": 3.621380846325167e-05,
|
| 25849 |
+
"loss": 2.0067,
|
| 25850 |
+
"step": 3687
|
| 25851 |
+
},
|
| 25852 |
+
{
|
| 25853 |
+
"epoch": 0.8195555555555556,
|
| 25854 |
+
"grad_norm": 1.0239263772964478,
|
| 25855 |
+
"learning_rate": 3.616926503340757e-05,
|
| 25856 |
+
"loss": 1.3173,
|
| 25857 |
+
"step": 3688
|
| 25858 |
+
},
|
| 25859 |
+
{
|
| 25860 |
+
"epoch": 0.8197777777777778,
|
| 25861 |
+
"grad_norm": 0.09923997521400452,
|
| 25862 |
+
"learning_rate": 3.612472160356348e-05,
|
| 25863 |
+
"loss": 0.0243,
|
| 25864 |
+
"step": 3689
|
| 25865 |
+
},
|
| 25866 |
+
{
|
| 25867 |
+
"epoch": 0.82,
|
| 25868 |
+
"grad_norm": 1.150796890258789,
|
| 25869 |
+
"learning_rate": 3.608017817371938e-05,
|
| 25870 |
+
"loss": 1.5561,
|
| 25871 |
+
"step": 3690
|
| 25872 |
+
},
|
| 25873 |
+
{
|
| 25874 |
+
"epoch": 0.8202222222222222,
|
| 25875 |
+
"grad_norm": 1.0021523237228394,
|
| 25876 |
+
"learning_rate": 3.603563474387528e-05,
|
| 25877 |
+
"loss": 1.6326,
|
| 25878 |
+
"step": 3691
|
| 25879 |
+
},
|
| 25880 |
+
{
|
| 25881 |
+
"epoch": 0.8204444444444444,
|
| 25882 |
+
"grad_norm": 0.9547725319862366,
|
| 25883 |
+
"learning_rate": 3.599109131403118e-05,
|
| 25884 |
+
"loss": 1.2198,
|
| 25885 |
+
"step": 3692
|
| 25886 |
+
},
|
| 25887 |
+
{
|
| 25888 |
+
"epoch": 0.8206666666666667,
|
| 25889 |
+
"grad_norm": 0.9803183674812317,
|
| 25890 |
+
"learning_rate": 3.594654788418708e-05,
|
| 25891 |
+
"loss": 1.4813,
|
| 25892 |
+
"step": 3693
|
| 25893 |
+
},
|
| 25894 |
+
{
|
| 25895 |
+
"epoch": 0.8208888888888889,
|
| 25896 |
+
"grad_norm": 1.140236258506775,
|
| 25897 |
+
"learning_rate": 3.5902004454342985e-05,
|
| 25898 |
+
"loss": 1.5703,
|
| 25899 |
+
"step": 3694
|
| 25900 |
+
},
|
| 25901 |
+
{
|
| 25902 |
+
"epoch": 0.8211111111111111,
|
| 25903 |
+
"grad_norm": 0.1800074428319931,
|
| 25904 |
+
"learning_rate": 3.585746102449889e-05,
|
| 25905 |
+
"loss": 0.0296,
|
| 25906 |
+
"step": 3695
|
| 25907 |
+
},
|
| 25908 |
+
{
|
| 25909 |
+
"epoch": 0.8213333333333334,
|
| 25910 |
+
"grad_norm": 0.744806170463562,
|
| 25911 |
+
"learning_rate": 3.581291759465479e-05,
|
| 25912 |
+
"loss": 0.7234,
|
| 25913 |
+
"step": 3696
|
| 25914 |
+
},
|
| 25915 |
+
{
|
| 25916 |
+
"epoch": 0.8215555555555556,
|
| 25917 |
+
"grad_norm": 1.0382987260818481,
|
| 25918 |
+
"learning_rate": 3.5768374164810694e-05,
|
| 25919 |
+
"loss": 1.0165,
|
| 25920 |
+
"step": 3697
|
| 25921 |
+
},
|
| 25922 |
+
{
|
| 25923 |
+
"epoch": 0.8217777777777778,
|
| 25924 |
+
"grad_norm": 1.01142156124115,
|
| 25925 |
+
"learning_rate": 3.572383073496659e-05,
|
| 25926 |
+
"loss": 1.2093,
|
| 25927 |
+
"step": 3698
|
| 25928 |
+
},
|
| 25929 |
+
{
|
| 25930 |
+
"epoch": 0.822,
|
| 25931 |
+
"grad_norm": 0.575733482837677,
|
| 25932 |
+
"learning_rate": 3.567928730512249e-05,
|
| 25933 |
+
"loss": 0.4409,
|
| 25934 |
+
"step": 3699
|
| 25935 |
+
},
|
| 25936 |
+
{
|
| 25937 |
+
"epoch": 0.8222222222222222,
|
| 25938 |
+
"grad_norm": 0.9521045684814453,
|
| 25939 |
+
"learning_rate": 3.5634743875278396e-05,
|
| 25940 |
+
"loss": 0.6805,
|
| 25941 |
+
"step": 3700
|
| 25942 |
+
},
|
| 25943 |
+
{
|
| 25944 |
+
"epoch": 0.8224444444444444,
|
| 25945 |
+
"grad_norm": 0.9057009220123291,
|
| 25946 |
+
"learning_rate": 3.55902004454343e-05,
|
| 25947 |
+
"loss": 2.3256,
|
| 25948 |
+
"step": 3701
|
| 25949 |
+
},
|
| 25950 |
+
{
|
| 25951 |
+
"epoch": 0.8226666666666667,
|
| 25952 |
+
"grad_norm": 0.8539500832557678,
|
| 25953 |
+
"learning_rate": 3.55456570155902e-05,
|
| 25954 |
+
"loss": 2.1458,
|
| 25955 |
+
"step": 3702
|
| 25956 |
+
},
|
| 25957 |
+
{
|
| 25958 |
+
"epoch": 0.8228888888888889,
|
| 25959 |
+
"grad_norm": 0.598690390586853,
|
| 25960 |
+
"learning_rate": 3.5501113585746106e-05,
|
| 25961 |
+
"loss": 1.164,
|
| 25962 |
+
"step": 3703
|
| 25963 |
+
},
|
| 25964 |
+
{
|
| 25965 |
+
"epoch": 0.8231111111111111,
|
| 25966 |
+
"grad_norm": 1.0174680948257446,
|
| 25967 |
+
"learning_rate": 3.545657015590201e-05,
|
| 25968 |
+
"loss": 2.2128,
|
| 25969 |
+
"step": 3704
|
| 25970 |
+
},
|
| 25971 |
+
{
|
| 25972 |
+
"epoch": 0.8233333333333334,
|
| 25973 |
+
"grad_norm": 0.5973488092422485,
|
| 25974 |
+
"learning_rate": 3.541202672605791e-05,
|
| 25975 |
+
"loss": 1.1449,
|
| 25976 |
+
"step": 3705
|
| 25977 |
+
},
|
| 25978 |
+
{
|
| 25979 |
+
"epoch": 0.8235555555555556,
|
| 25980 |
+
"grad_norm": 0.07344137132167816,
|
| 25981 |
+
"learning_rate": 3.536748329621381e-05,
|
| 25982 |
+
"loss": 0.0108,
|
| 25983 |
+
"step": 3706
|
| 25984 |
+
},
|
| 25985 |
+
{
|
| 25986 |
+
"epoch": 0.8237777777777778,
|
| 25987 |
+
"grad_norm": 0.08576471358537674,
|
| 25988 |
+
"learning_rate": 3.5322939866369714e-05,
|
| 25989 |
+
"loss": 0.0111,
|
| 25990 |
+
"step": 3707
|
| 25991 |
+
},
|
| 25992 |
+
{
|
| 25993 |
+
"epoch": 0.824,
|
| 25994 |
+
"grad_norm": 0.0801275447010994,
|
| 25995 |
+
"learning_rate": 3.527839643652561e-05,
|
| 25996 |
+
"loss": 0.0111,
|
| 25997 |
+
"step": 3708
|
| 25998 |
+
},
|
| 25999 |
+
{
|
| 26000 |
+
"epoch": 0.8242222222222222,
|
| 26001 |
+
"grad_norm": 0.08194973319768906,
|
| 26002 |
+
"learning_rate": 3.523385300668152e-05,
|
| 26003 |
+
"loss": 0.0109,
|
| 26004 |
+
"step": 3709
|
| 26005 |
+
},
|
| 26006 |
+
{
|
| 26007 |
+
"epoch": 0.8244444444444444,
|
| 26008 |
+
"grad_norm": 1.0708433389663696,
|
| 26009 |
+
"learning_rate": 3.518930957683742e-05,
|
| 26010 |
+
"loss": 2.033,
|
| 26011 |
+
"step": 3710
|
| 26012 |
+
},
|
| 26013 |
+
{
|
| 26014 |
+
"epoch": 0.8246666666666667,
|
| 26015 |
+
"grad_norm": 1.263612985610962,
|
| 26016 |
+
"learning_rate": 3.514476614699332e-05,
|
| 26017 |
+
"loss": 2.6208,
|
| 26018 |
+
"step": 3711
|
| 26019 |
+
},
|
| 26020 |
+
{
|
| 26021 |
+
"epoch": 0.8248888888888889,
|
| 26022 |
+
"grad_norm": 0.9804373383522034,
|
| 26023 |
+
"learning_rate": 3.510022271714922e-05,
|
| 26024 |
+
"loss": 1.8539,
|
| 26025 |
+
"step": 3712
|
| 26026 |
+
},
|
| 26027 |
+
{
|
| 26028 |
+
"epoch": 0.8251111111111111,
|
| 26029 |
+
"grad_norm": 0.8642978668212891,
|
| 26030 |
+
"learning_rate": 3.5055679287305125e-05,
|
| 26031 |
+
"loss": 1.7259,
|
| 26032 |
+
"step": 3713
|
| 26033 |
+
},
|
| 26034 |
+
{
|
| 26035 |
+
"epoch": 0.8253333333333334,
|
| 26036 |
+
"grad_norm": 0.9090456962585449,
|
| 26037 |
+
"learning_rate": 3.5011135857461024e-05,
|
| 26038 |
+
"loss": 1.8745,
|
| 26039 |
+
"step": 3714
|
| 26040 |
+
},
|
| 26041 |
+
{
|
| 26042 |
+
"epoch": 0.8255555555555556,
|
| 26043 |
+
"grad_norm": 0.9151667356491089,
|
| 26044 |
+
"learning_rate": 3.496659242761693e-05,
|
| 26045 |
+
"loss": 1.9903,
|
| 26046 |
+
"step": 3715
|
| 26047 |
+
},
|
| 26048 |
+
{
|
| 26049 |
+
"epoch": 0.8257777777777778,
|
| 26050 |
+
"grad_norm": 0.6366732120513916,
|
| 26051 |
+
"learning_rate": 3.4922048997772835e-05,
|
| 26052 |
+
"loss": 0.9141,
|
| 26053 |
+
"step": 3716
|
| 26054 |
+
},
|
| 26055 |
+
{
|
| 26056 |
+
"epoch": 0.826,
|
| 26057 |
+
"grad_norm": 0.7700564861297607,
|
| 26058 |
+
"learning_rate": 3.487750556792873e-05,
|
| 26059 |
+
"loss": 0.9737,
|
| 26060 |
+
"step": 3717
|
| 26061 |
+
},
|
| 26062 |
+
{
|
| 26063 |
+
"epoch": 0.8262222222222222,
|
| 26064 |
+
"grad_norm": 0.9940738081932068,
|
| 26065 |
+
"learning_rate": 3.483296213808463e-05,
|
| 26066 |
+
"loss": 2.0155,
|
| 26067 |
+
"step": 3718
|
| 26068 |
+
},
|
| 26069 |
+
{
|
| 26070 |
+
"epoch": 0.8264444444444444,
|
| 26071 |
+
"grad_norm": 0.8816027641296387,
|
| 26072 |
+
"learning_rate": 3.478841870824054e-05,
|
| 26073 |
+
"loss": 1.7297,
|
| 26074 |
+
"step": 3719
|
| 26075 |
+
},
|
| 26076 |
+
{
|
| 26077 |
+
"epoch": 0.8266666666666667,
|
| 26078 |
+
"grad_norm": 1.3799381256103516,
|
| 26079 |
+
"learning_rate": 3.4743875278396436e-05,
|
| 26080 |
+
"loss": 2.159,
|
| 26081 |
+
"step": 3720
|
| 26082 |
+
},
|
| 26083 |
+
{
|
| 26084 |
+
"epoch": 0.8268888888888889,
|
| 26085 |
+
"grad_norm": 1.0440785884857178,
|
| 26086 |
+
"learning_rate": 3.469933184855234e-05,
|
| 26087 |
+
"loss": 1.7563,
|
| 26088 |
+
"step": 3721
|
| 26089 |
+
},
|
| 26090 |
+
{
|
| 26091 |
+
"epoch": 0.8271111111111111,
|
| 26092 |
+
"grad_norm": 0.6796101331710815,
|
| 26093 |
+
"learning_rate": 3.465478841870824e-05,
|
| 26094 |
+
"loss": 0.9447,
|
| 26095 |
+
"step": 3722
|
| 26096 |
+
},
|
| 26097 |
+
{
|
| 26098 |
+
"epoch": 0.8273333333333334,
|
| 26099 |
+
"grad_norm": 0.06959293782711029,
|
| 26100 |
+
"learning_rate": 3.4610244988864145e-05,
|
| 26101 |
+
"loss": 0.0163,
|
| 26102 |
+
"step": 3723
|
| 26103 |
+
},
|
| 26104 |
+
{
|
| 26105 |
+
"epoch": 0.8275555555555556,
|
| 26106 |
+
"grad_norm": 0.06835558265447617,
|
| 26107 |
+
"learning_rate": 3.456570155902005e-05,
|
| 26108 |
+
"loss": 0.0164,
|
| 26109 |
+
"step": 3724
|
| 26110 |
+
},
|
| 26111 |
+
{
|
| 26112 |
+
"epoch": 0.8277777777777777,
|
| 26113 |
+
"grad_norm": 1.080851435661316,
|
| 26114 |
+
"learning_rate": 3.452115812917594e-05,
|
| 26115 |
+
"loss": 1.9956,
|
| 26116 |
+
"step": 3725
|
| 26117 |
+
},
|
| 26118 |
+
{
|
| 26119 |
+
"epoch": 0.828,
|
| 26120 |
+
"grad_norm": 0.6416263580322266,
|
| 26121 |
+
"learning_rate": 3.447661469933185e-05,
|
| 26122 |
+
"loss": 0.6763,
|
| 26123 |
+
"step": 3726
|
| 26124 |
+
},
|
| 26125 |
+
{
|
| 26126 |
+
"epoch": 0.8282222222222222,
|
| 26127 |
+
"grad_norm": 1.0977882146835327,
|
| 26128 |
+
"learning_rate": 3.443207126948775e-05,
|
| 26129 |
+
"loss": 1.7526,
|
| 26130 |
+
"step": 3727
|
| 26131 |
+
},
|
| 26132 |
+
{
|
| 26133 |
+
"epoch": 0.8284444444444444,
|
| 26134 |
+
"grad_norm": 0.8835853338241577,
|
| 26135 |
+
"learning_rate": 3.438752783964365e-05,
|
| 26136 |
+
"loss": 1.7257,
|
| 26137 |
+
"step": 3728
|
| 26138 |
+
},
|
| 26139 |
+
{
|
| 26140 |
+
"epoch": 0.8286666666666667,
|
| 26141 |
+
"grad_norm": 1.0617055892944336,
|
| 26142 |
+
"learning_rate": 3.434298440979956e-05,
|
| 26143 |
+
"loss": 1.9876,
|
| 26144 |
+
"step": 3729
|
| 26145 |
+
},
|
| 26146 |
+
{
|
| 26147 |
+
"epoch": 0.8288888888888889,
|
| 26148 |
+
"grad_norm": 1.1054614782333374,
|
| 26149 |
+
"learning_rate": 3.429844097995546e-05,
|
| 26150 |
+
"loss": 1.4692,
|
| 26151 |
+
"step": 3730
|
| 26152 |
+
},
|
| 26153 |
+
{
|
| 26154 |
+
"epoch": 0.8291111111111111,
|
| 26155 |
+
"grad_norm": 0.06595687568187714,
|
| 26156 |
+
"learning_rate": 3.425389755011136e-05,
|
| 26157 |
+
"loss": 0.0171,
|
| 26158 |
+
"step": 3731
|
| 26159 |
+
},
|
| 26160 |
+
{
|
| 26161 |
+
"epoch": 0.8293333333333334,
|
| 26162 |
+
"grad_norm": 0.06783387809991837,
|
| 26163 |
+
"learning_rate": 3.420935412026726e-05,
|
| 26164 |
+
"loss": 0.0172,
|
| 26165 |
+
"step": 3732
|
| 26166 |
+
},
|
| 26167 |
+
{
|
| 26168 |
+
"epoch": 0.8295555555555556,
|
| 26169 |
+
"grad_norm": 0.7828741073608398,
|
| 26170 |
+
"learning_rate": 3.4164810690423165e-05,
|
| 26171 |
+
"loss": 0.8107,
|
| 26172 |
+
"step": 3733
|
| 26173 |
+
},
|
| 26174 |
+
{
|
| 26175 |
+
"epoch": 0.8297777777777777,
|
| 26176 |
+
"grad_norm": 0.07702479511499405,
|
| 26177 |
+
"learning_rate": 3.4120267260579063e-05,
|
| 26178 |
+
"loss": 0.0177,
|
| 26179 |
+
"step": 3734
|
| 26180 |
+
},
|
| 26181 |
+
{
|
| 26182 |
+
"epoch": 0.83,
|
| 26183 |
+
"grad_norm": 0.6449767351150513,
|
| 26184 |
+
"learning_rate": 3.407572383073497e-05,
|
| 26185 |
+
"loss": 0.7656,
|
| 26186 |
+
"step": 3735
|
| 26187 |
+
},
|
| 26188 |
+
{
|
| 26189 |
+
"epoch": 0.8302222222222222,
|
| 26190 |
+
"grad_norm": 1.3564252853393555,
|
| 26191 |
+
"learning_rate": 3.4031180400890874e-05,
|
| 26192 |
+
"loss": 1.9644,
|
| 26193 |
+
"step": 3736
|
| 26194 |
+
},
|
| 26195 |
+
{
|
| 26196 |
+
"epoch": 0.8304444444444444,
|
| 26197 |
+
"grad_norm": 0.7747462391853333,
|
| 26198 |
+
"learning_rate": 3.398663697104677e-05,
|
| 26199 |
+
"loss": 1.0032,
|
| 26200 |
+
"step": 3737
|
| 26201 |
+
},
|
| 26202 |
+
{
|
| 26203 |
+
"epoch": 0.8306666666666667,
|
| 26204 |
+
"grad_norm": 1.13922119140625,
|
| 26205 |
+
"learning_rate": 3.394209354120267e-05,
|
| 26206 |
+
"loss": 1.545,
|
| 26207 |
+
"step": 3738
|
| 26208 |
+
},
|
| 26209 |
+
{
|
| 26210 |
+
"epoch": 0.8308888888888889,
|
| 26211 |
+
"grad_norm": 0.9685484766960144,
|
| 26212 |
+
"learning_rate": 3.389755011135858e-05,
|
| 26213 |
+
"loss": 1.3771,
|
| 26214 |
+
"step": 3739
|
| 26215 |
+
},
|
| 26216 |
+
{
|
| 26217 |
+
"epoch": 0.8311111111111111,
|
| 26218 |
+
"grad_norm": 0.7709338665008545,
|
| 26219 |
+
"learning_rate": 3.3853006681514475e-05,
|
| 26220 |
+
"loss": 0.696,
|
| 26221 |
+
"step": 3740
|
| 26222 |
+
},
|
| 26223 |
+
{
|
| 26224 |
+
"epoch": 0.8313333333333334,
|
| 26225 |
+
"grad_norm": 1.0451520681381226,
|
| 26226 |
+
"learning_rate": 3.380846325167038e-05,
|
| 26227 |
+
"loss": 1.564,
|
| 26228 |
+
"step": 3741
|
| 26229 |
+
},
|
| 26230 |
+
{
|
| 26231 |
+
"epoch": 0.8315555555555556,
|
| 26232 |
+
"grad_norm": 1.1514785289764404,
|
| 26233 |
+
"learning_rate": 3.3763919821826286e-05,
|
| 26234 |
+
"loss": 1.4297,
|
| 26235 |
+
"step": 3742
|
| 26236 |
+
},
|
| 26237 |
+
{
|
| 26238 |
+
"epoch": 0.8317777777777777,
|
| 26239 |
+
"grad_norm": 1.3433165550231934,
|
| 26240 |
+
"learning_rate": 3.3719376391982185e-05,
|
| 26241 |
+
"loss": 1.7462,
|
| 26242 |
+
"step": 3743
|
| 26243 |
+
},
|
| 26244 |
+
{
|
| 26245 |
+
"epoch": 0.832,
|
| 26246 |
+
"grad_norm": 0.9320109486579895,
|
| 26247 |
+
"learning_rate": 3.367483296213808e-05,
|
| 26248 |
+
"loss": 0.95,
|
| 26249 |
+
"step": 3744
|
| 26250 |
+
},
|
| 26251 |
+
{
|
| 26252 |
+
"epoch": 0.8322222222222222,
|
| 26253 |
+
"grad_norm": 0.9835542440414429,
|
| 26254 |
+
"learning_rate": 3.363028953229399e-05,
|
| 26255 |
+
"loss": 1.1297,
|
| 26256 |
+
"step": 3745
|
| 26257 |
+
},
|
| 26258 |
+
{
|
| 26259 |
+
"epoch": 0.8324444444444444,
|
| 26260 |
+
"grad_norm": 0.18634167313575745,
|
| 26261 |
+
"learning_rate": 3.358574610244989e-05,
|
| 26262 |
+
"loss": 0.0346,
|
| 26263 |
+
"step": 3746
|
| 26264 |
+
},
|
| 26265 |
+
{
|
| 26266 |
+
"epoch": 0.8326666666666667,
|
| 26267 |
+
"grad_norm": 0.1696268618106842,
|
| 26268 |
+
"learning_rate": 3.354120267260579e-05,
|
| 26269 |
+
"loss": 0.034,
|
| 26270 |
+
"step": 3747
|
| 26271 |
+
},
|
| 26272 |
+
{
|
| 26273 |
+
"epoch": 0.8328888888888889,
|
| 26274 |
+
"grad_norm": 1.043257236480713,
|
| 26275 |
+
"learning_rate": 3.34966592427617e-05,
|
| 26276 |
+
"loss": 0.9229,
|
| 26277 |
+
"step": 3748
|
| 26278 |
+
},
|
| 26279 |
+
{
|
| 26280 |
+
"epoch": 0.8331111111111111,
|
| 26281 |
+
"grad_norm": 1.21977698802948,
|
| 26282 |
+
"learning_rate": 3.3452115812917596e-05,
|
| 26283 |
+
"loss": 1.0685,
|
| 26284 |
+
"step": 3749
|
| 26285 |
+
},
|
| 26286 |
+
{
|
| 26287 |
+
"epoch": 0.8333333333333334,
|
| 26288 |
+
"grad_norm": 1.1700232028961182,
|
| 26289 |
+
"learning_rate": 3.34075723830735e-05,
|
| 26290 |
+
"loss": 0.7485,
|
| 26291 |
+
"step": 3750
|
| 26292 |
+
},
|
| 26293 |
+
{
|
| 26294 |
+
"epoch": 0.8335555555555556,
|
| 26295 |
+
"grad_norm": 0.4857214093208313,
|
| 26296 |
+
"learning_rate": 3.33630289532294e-05,
|
| 26297 |
+
"loss": 0.9854,
|
| 26298 |
+
"step": 3751
|
| 26299 |
+
},
|
| 26300 |
+
{
|
| 26301 |
+
"epoch": 0.8337777777777777,
|
| 26302 |
+
"grad_norm": 0.8265995383262634,
|
| 26303 |
+
"learning_rate": 3.33184855233853e-05,
|
| 26304 |
+
"loss": 1.9879,
|
| 26305 |
+
"step": 3752
|
| 26306 |
+
},
|
| 26307 |
+
{
|
| 26308 |
+
"epoch": 0.834,
|
| 26309 |
+
"grad_norm": 0.48844748735427856,
|
| 26310 |
+
"learning_rate": 3.3273942093541204e-05,
|
| 26311 |
+
"loss": 1.0029,
|
| 26312 |
+
"step": 3753
|
| 26313 |
+
},
|
| 26314 |
+
{
|
| 26315 |
+
"epoch": 0.8342222222222222,
|
| 26316 |
+
"grad_norm": 0.6757328510284424,
|
| 26317 |
+
"learning_rate": 3.322939866369711e-05,
|
| 26318 |
+
"loss": 1.0597,
|
| 26319 |
+
"step": 3754
|
| 26320 |
+
},
|
| 26321 |
+
{
|
| 26322 |
+
"epoch": 0.8344444444444444,
|
| 26323 |
+
"grad_norm": 1.0443055629730225,
|
| 26324 |
+
"learning_rate": 3.318485523385301e-05,
|
| 26325 |
+
"loss": 2.3737,
|
| 26326 |
+
"step": 3755
|
| 26327 |
+
},
|
| 26328 |
+
{
|
| 26329 |
+
"epoch": 0.8346666666666667,
|
| 26330 |
+
"grad_norm": 0.8254187703132629,
|
| 26331 |
+
"learning_rate": 3.3140311804008914e-05,
|
| 26332 |
+
"loss": 1.9924,
|
| 26333 |
+
"step": 3756
|
| 26334 |
+
},
|
| 26335 |
+
{
|
| 26336 |
+
"epoch": 0.8348888888888889,
|
| 26337 |
+
"grad_norm": 0.09497911483049393,
|
| 26338 |
+
"learning_rate": 3.309576837416481e-05,
|
| 26339 |
+
"loss": 0.0113,
|
| 26340 |
+
"step": 3757
|
| 26341 |
+
},
|
| 26342 |
+
{
|
| 26343 |
+
"epoch": 0.8351111111111111,
|
| 26344 |
+
"grad_norm": 0.09401866048574448,
|
| 26345 |
+
"learning_rate": 3.305122494432071e-05,
|
| 26346 |
+
"loss": 0.0116,
|
| 26347 |
+
"step": 3758
|
| 26348 |
+
},
|
| 26349 |
+
{
|
| 26350 |
+
"epoch": 0.8353333333333334,
|
| 26351 |
+
"grad_norm": 0.09458266943693161,
|
| 26352 |
+
"learning_rate": 3.3006681514476616e-05,
|
| 26353 |
+
"loss": 0.0116,
|
| 26354 |
+
"step": 3759
|
| 26355 |
+
},
|
| 26356 |
+
{
|
| 26357 |
+
"epoch": 0.8355555555555556,
|
| 26358 |
+
"grad_norm": 0.09231861680746078,
|
| 26359 |
+
"learning_rate": 3.2962138084632515e-05,
|
| 26360 |
+
"loss": 0.0114,
|
| 26361 |
+
"step": 3760
|
| 26362 |
+
},
|
| 26363 |
+
{
|
| 26364 |
+
"epoch": 0.8357777777777777,
|
| 26365 |
+
"grad_norm": 1.012135624885559,
|
| 26366 |
+
"learning_rate": 3.291759465478842e-05,
|
| 26367 |
+
"loss": 2.2142,
|
| 26368 |
+
"step": 3761
|
| 26369 |
+
},
|
| 26370 |
+
{
|
| 26371 |
+
"epoch": 0.836,
|
| 26372 |
+
"grad_norm": 0.8345160484313965,
|
| 26373 |
+
"learning_rate": 3.2873051224944325e-05,
|
| 26374 |
+
"loss": 2.0887,
|
| 26375 |
+
"step": 3762
|
| 26376 |
+
},
|
| 26377 |
+
{
|
| 26378 |
+
"epoch": 0.8362222222222222,
|
| 26379 |
+
"grad_norm": 0.777621865272522,
|
| 26380 |
+
"learning_rate": 3.2828507795100224e-05,
|
| 26381 |
+
"loss": 1.7456,
|
| 26382 |
+
"step": 3763
|
| 26383 |
+
},
|
| 26384 |
+
{
|
| 26385 |
+
"epoch": 0.8364444444444444,
|
| 26386 |
+
"grad_norm": 0.9471651911735535,
|
| 26387 |
+
"learning_rate": 3.278396436525612e-05,
|
| 26388 |
+
"loss": 1.9976,
|
| 26389 |
+
"step": 3764
|
| 26390 |
+
},
|
| 26391 |
+
{
|
| 26392 |
+
"epoch": 0.8366666666666667,
|
| 26393 |
+
"grad_norm": 0.9481960535049438,
|
| 26394 |
+
"learning_rate": 3.273942093541203e-05,
|
| 26395 |
+
"loss": 2.1314,
|
| 26396 |
+
"step": 3765
|
| 26397 |
+
},
|
| 26398 |
+
{
|
| 26399 |
+
"epoch": 0.8368888888888889,
|
| 26400 |
+
"grad_norm": 1.0837010145187378,
|
| 26401 |
+
"learning_rate": 3.2694877505567926e-05,
|
| 26402 |
+
"loss": 2.0858,
|
| 26403 |
+
"step": 3766
|
| 26404 |
+
},
|
| 26405 |
+
{
|
| 26406 |
+
"epoch": 0.8371111111111111,
|
| 26407 |
+
"grad_norm": 1.0105607509613037,
|
| 26408 |
+
"learning_rate": 3.265033407572383e-05,
|
| 26409 |
+
"loss": 2.1732,
|
| 26410 |
+
"step": 3767
|
| 26411 |
+
},
|
| 26412 |
+
{
|
| 26413 |
+
"epoch": 0.8373333333333334,
|
| 26414 |
+
"grad_norm": 0.5977281928062439,
|
| 26415 |
+
"learning_rate": 3.260579064587974e-05,
|
| 26416 |
+
"loss": 0.8871,
|
| 26417 |
+
"step": 3768
|
| 26418 |
+
},
|
| 26419 |
+
{
|
| 26420 |
+
"epoch": 0.8375555555555556,
|
| 26421 |
+
"grad_norm": 1.2047114372253418,
|
| 26422 |
+
"learning_rate": 3.2561247216035636e-05,
|
| 26423 |
+
"loss": 2.0168,
|
| 26424 |
+
"step": 3769
|
| 26425 |
+
},
|
| 26426 |
+
{
|
| 26427 |
+
"epoch": 0.8377777777777777,
|
| 26428 |
+
"grad_norm": 0.8766410946846008,
|
| 26429 |
+
"learning_rate": 3.251670378619154e-05,
|
| 26430 |
+
"loss": 1.6475,
|
| 26431 |
+
"step": 3770
|
| 26432 |
+
},
|
| 26433 |
+
{
|
| 26434 |
+
"epoch": 0.838,
|
| 26435 |
+
"grad_norm": 0.9371228218078613,
|
| 26436 |
+
"learning_rate": 3.247216035634744e-05,
|
| 26437 |
+
"loss": 1.8688,
|
| 26438 |
+
"step": 3771
|
| 26439 |
+
},
|
| 26440 |
+
{
|
| 26441 |
+
"epoch": 0.8382222222222222,
|
| 26442 |
+
"grad_norm": 1.0611170530319214,
|
| 26443 |
+
"learning_rate": 3.242761692650334e-05,
|
| 26444 |
+
"loss": 1.8825,
|
| 26445 |
+
"step": 3772
|
| 26446 |
+
},
|
| 26447 |
+
{
|
| 26448 |
+
"epoch": 0.8384444444444444,
|
| 26449 |
+
"grad_norm": 1.0075304508209229,
|
| 26450 |
+
"learning_rate": 3.2383073496659244e-05,
|
| 26451 |
+
"loss": 2.1026,
|
| 26452 |
+
"step": 3773
|
| 26453 |
+
},
|
| 26454 |
+
{
|
| 26455 |
+
"epoch": 0.8386666666666667,
|
| 26456 |
+
"grad_norm": 0.06949839740991592,
|
| 26457 |
+
"learning_rate": 3.233853006681515e-05,
|
| 26458 |
+
"loss": 0.0162,
|
| 26459 |
+
"step": 3774
|
| 26460 |
+
},
|
| 26461 |
+
{
|
| 26462 |
+
"epoch": 0.8388888888888889,
|
| 26463 |
+
"grad_norm": 0.06908978521823883,
|
| 26464 |
+
"learning_rate": 3.229398663697105e-05,
|
| 26465 |
+
"loss": 0.0159,
|
| 26466 |
+
"step": 3775
|
| 26467 |
+
},
|
| 26468 |
+
{
|
| 26469 |
+
"epoch": 0.8391111111111111,
|
| 26470 |
+
"grad_norm": 0.905208945274353,
|
| 26471 |
+
"learning_rate": 3.224944320712695e-05,
|
| 26472 |
+
"loss": 1.7153,
|
| 26473 |
+
"step": 3776
|
| 26474 |
+
},
|
| 26475 |
+
{
|
| 26476 |
+
"epoch": 0.8393333333333334,
|
| 26477 |
+
"grad_norm": 0.8136224150657654,
|
| 26478 |
+
"learning_rate": 3.220489977728285e-05,
|
| 26479 |
+
"loss": 0.903,
|
| 26480 |
+
"step": 3777
|
| 26481 |
+
},
|
| 26482 |
+
{
|
| 26483 |
+
"epoch": 0.8395555555555556,
|
| 26484 |
+
"grad_norm": 0.9069592356681824,
|
| 26485 |
+
"learning_rate": 3.216035634743875e-05,
|
| 26486 |
+
"loss": 1.6758,
|
| 26487 |
+
"step": 3778
|
| 26488 |
+
},
|
| 26489 |
+
{
|
| 26490 |
+
"epoch": 0.8397777777777777,
|
| 26491 |
+
"grad_norm": 1.0851026773452759,
|
| 26492 |
+
"learning_rate": 3.2115812917594655e-05,
|
| 26493 |
+
"loss": 1.8553,
|
| 26494 |
+
"step": 3779
|
| 26495 |
+
},
|
| 26496 |
+
{
|
| 26497 |
+
"epoch": 0.84,
|
| 26498 |
+
"grad_norm": 0.998877763748169,
|
| 26499 |
+
"learning_rate": 3.207126948775056e-05,
|
| 26500 |
+
"loss": 1.8056,
|
| 26501 |
+
"step": 3780
|
| 26502 |
+
},
|
| 26503 |
+
{
|
| 26504 |
+
"epoch": 0.8402222222222222,
|
| 26505 |
+
"grad_norm": 0.0660950317978859,
|
| 26506 |
+
"learning_rate": 3.202672605790646e-05,
|
| 26507 |
+
"loss": 0.0175,
|
| 26508 |
+
"step": 3781
|
| 26509 |
+
},
|
| 26510 |
+
{
|
| 26511 |
+
"epoch": 0.8404444444444444,
|
| 26512 |
+
"grad_norm": 0.0653744786977768,
|
| 26513 |
+
"learning_rate": 3.1982182628062365e-05,
|
| 26514 |
+
"loss": 0.0173,
|
| 26515 |
+
"step": 3782
|
| 26516 |
+
},
|
| 26517 |
+
{
|
| 26518 |
+
"epoch": 0.8406666666666667,
|
| 26519 |
+
"grad_norm": 0.06411214917898178,
|
| 26520 |
+
"learning_rate": 3.193763919821826e-05,
|
| 26521 |
+
"loss": 0.0175,
|
| 26522 |
+
"step": 3783
|
| 26523 |
+
},
|
| 26524 |
+
{
|
| 26525 |
+
"epoch": 0.8408888888888889,
|
| 26526 |
+
"grad_norm": 0.06954985857009888,
|
| 26527 |
+
"learning_rate": 3.189309576837416e-05,
|
| 26528 |
+
"loss": 0.0174,
|
| 26529 |
+
"step": 3784
|
| 26530 |
+
},
|
| 26531 |
+
{
|
| 26532 |
+
"epoch": 0.8411111111111111,
|
| 26533 |
+
"grad_norm": 0.7279578447341919,
|
| 26534 |
+
"learning_rate": 3.184855233853007e-05,
|
| 26535 |
+
"loss": 1.0075,
|
| 26536 |
+
"step": 3785
|
| 26537 |
+
},
|
| 26538 |
+
{
|
| 26539 |
+
"epoch": 0.8413333333333334,
|
| 26540 |
+
"grad_norm": 1.0854923725128174,
|
| 26541 |
+
"learning_rate": 3.180400890868597e-05,
|
| 26542 |
+
"loss": 1.8502,
|
| 26543 |
+
"step": 3786
|
| 26544 |
+
},
|
| 26545 |
+
{
|
| 26546 |
+
"epoch": 0.8415555555555555,
|
| 26547 |
+
"grad_norm": 1.1248599290847778,
|
| 26548 |
+
"learning_rate": 3.175946547884187e-05,
|
| 26549 |
+
"loss": 1.769,
|
| 26550 |
+
"step": 3787
|
| 26551 |
+
},
|
| 26552 |
+
{
|
| 26553 |
+
"epoch": 0.8417777777777777,
|
| 26554 |
+
"grad_norm": 0.6535754203796387,
|
| 26555 |
+
"learning_rate": 3.1714922048997777e-05,
|
| 26556 |
+
"loss": 0.922,
|
| 26557 |
+
"step": 3788
|
| 26558 |
+
},
|
| 26559 |
+
{
|
| 26560 |
+
"epoch": 0.842,
|
| 26561 |
+
"grad_norm": 1.0051473379135132,
|
| 26562 |
+
"learning_rate": 3.167037861915368e-05,
|
| 26563 |
+
"loss": 1.6551,
|
| 26564 |
+
"step": 3789
|
| 26565 |
+
},
|
| 26566 |
+
{
|
| 26567 |
+
"epoch": 0.8422222222222222,
|
| 26568 |
+
"grad_norm": 1.0957441329956055,
|
| 26569 |
+
"learning_rate": 3.162583518930958e-05,
|
| 26570 |
+
"loss": 1.7058,
|
| 26571 |
+
"step": 3790
|
| 26572 |
+
},
|
| 26573 |
+
{
|
| 26574 |
+
"epoch": 0.8424444444444444,
|
| 26575 |
+
"grad_norm": 1.0055428743362427,
|
| 26576 |
+
"learning_rate": 3.158129175946548e-05,
|
| 26577 |
+
"loss": 1.5342,
|
| 26578 |
+
"step": 3791
|
| 26579 |
+
},
|
| 26580 |
+
{
|
| 26581 |
+
"epoch": 0.8426666666666667,
|
| 26582 |
+
"grad_norm": 0.8064576387405396,
|
| 26583 |
+
"learning_rate": 3.1536748329621384e-05,
|
| 26584 |
+
"loss": 0.8161,
|
| 26585 |
+
"step": 3792
|
| 26586 |
+
},
|
| 26587 |
+
{
|
| 26588 |
+
"epoch": 0.8428888888888889,
|
| 26589 |
+
"grad_norm": 1.1807235479354858,
|
| 26590 |
+
"learning_rate": 3.149220489977728e-05,
|
| 26591 |
+
"loss": 1.8853,
|
| 26592 |
+
"step": 3793
|
| 26593 |
+
},
|
| 26594 |
+
{
|
| 26595 |
+
"epoch": 0.8431111111111111,
|
| 26596 |
+
"grad_norm": 1.0103986263275146,
|
| 26597 |
+
"learning_rate": 3.144766146993319e-05,
|
| 26598 |
+
"loss": 1.3899,
|
| 26599 |
+
"step": 3794
|
| 26600 |
+
},
|
| 26601 |
+
{
|
| 26602 |
+
"epoch": 0.8433333333333334,
|
| 26603 |
+
"grad_norm": 0.2762221693992615,
|
| 26604 |
+
"learning_rate": 3.140311804008909e-05,
|
| 26605 |
+
"loss": 0.0367,
|
| 26606 |
+
"step": 3795
|
| 26607 |
+
},
|
| 26608 |
+
{
|
| 26609 |
+
"epoch": 0.8435555555555555,
|
| 26610 |
+
"grad_norm": 0.7262986302375793,
|
| 26611 |
+
"learning_rate": 3.135857461024499e-05,
|
| 26612 |
+
"loss": 0.7566,
|
| 26613 |
+
"step": 3796
|
| 26614 |
+
},
|
| 26615 |
+
{
|
| 26616 |
+
"epoch": 0.8437777777777777,
|
| 26617 |
+
"grad_norm": 1.0480473041534424,
|
| 26618 |
+
"learning_rate": 3.131403118040089e-05,
|
| 26619 |
+
"loss": 1.6141,
|
| 26620 |
+
"step": 3797
|
| 26621 |
+
},
|
| 26622 |
+
{
|
| 26623 |
+
"epoch": 0.844,
|
| 26624 |
+
"grad_norm": 1.0954132080078125,
|
| 26625 |
+
"learning_rate": 3.126948775055679e-05,
|
| 26626 |
+
"loss": 1.269,
|
| 26627 |
+
"step": 3798
|
| 26628 |
+
},
|
| 26629 |
+
{
|
| 26630 |
+
"epoch": 0.8442222222222222,
|
| 26631 |
+
"grad_norm": 0.7563920021057129,
|
| 26632 |
+
"learning_rate": 3.1224944320712695e-05,
|
| 26633 |
+
"loss": 0.6253,
|
| 26634 |
+
"step": 3799
|
| 26635 |
+
},
|
| 26636 |
+
{
|
| 26637 |
+
"epoch": 0.8444444444444444,
|
| 26638 |
+
"grad_norm": 0.7246300578117371,
|
| 26639 |
+
"learning_rate": 3.11804008908686e-05,
|
| 26640 |
+
"loss": 0.3778,
|
| 26641 |
+
"step": 3800
|
| 26642 |
+
},
|
| 26643 |
+
{
|
| 26644 |
+
"epoch": 0.8446666666666667,
|
| 26645 |
+
"grad_norm": 0.9608231782913208,
|
| 26646 |
+
"learning_rate": 3.11358574610245e-05,
|
| 26647 |
+
"loss": 2.1959,
|
| 26648 |
+
"step": 3801
|
| 26649 |
+
},
|
| 26650 |
+
{
|
| 26651 |
+
"epoch": 0.8448888888888889,
|
| 26652 |
+
"grad_norm": 0.04354199394583702,
|
| 26653 |
+
"learning_rate": 3.1091314031180404e-05,
|
| 26654 |
+
"loss": 0.0106,
|
| 26655 |
+
"step": 3802
|
| 26656 |
+
},
|
| 26657 |
+
{
|
| 26658 |
+
"epoch": 0.8451111111111111,
|
| 26659 |
+
"grad_norm": 0.8725544810295105,
|
| 26660 |
+
"learning_rate": 3.10467706013363e-05,
|
| 26661 |
+
"loss": 2.4019,
|
| 26662 |
+
"step": 3803
|
| 26663 |
+
},
|
| 26664 |
+
{
|
| 26665 |
+
"epoch": 0.8453333333333334,
|
| 26666 |
+
"grad_norm": 0.04339034482836723,
|
| 26667 |
+
"learning_rate": 3.10022271714922e-05,
|
| 26668 |
+
"loss": 0.0107,
|
| 26669 |
+
"step": 3804
|
| 26670 |
+
},
|
| 26671 |
+
{
|
| 26672 |
+
"epoch": 0.8455555555555555,
|
| 26673 |
+
"grad_norm": 0.8996299505233765,
|
| 26674 |
+
"learning_rate": 3.095768374164811e-05,
|
| 26675 |
+
"loss": 1.7978,
|
| 26676 |
+
"step": 3805
|
| 26677 |
+
},
|
| 26678 |
+
{
|
| 26679 |
+
"epoch": 0.8457777777777777,
|
| 26680 |
+
"grad_norm": 0.08517049998044968,
|
| 26681 |
+
"learning_rate": 3.091314031180401e-05,
|
| 26682 |
+
"loss": 0.0108,
|
| 26683 |
+
"step": 3806
|
| 26684 |
+
},
|
| 26685 |
+
{
|
| 26686 |
+
"epoch": 0.846,
|
| 26687 |
+
"grad_norm": 0.08175533264875412,
|
| 26688 |
+
"learning_rate": 3.086859688195991e-05,
|
| 26689 |
+
"loss": 0.0109,
|
| 26690 |
+
"step": 3807
|
| 26691 |
+
},
|
| 26692 |
+
{
|
| 26693 |
+
"epoch": 0.8462222222222222,
|
| 26694 |
+
"grad_norm": 0.06705193221569061,
|
| 26695 |
+
"learning_rate": 3.0824053452115816e-05,
|
| 26696 |
+
"loss": 0.0103,
|
| 26697 |
+
"step": 3808
|
| 26698 |
+
},
|
| 26699 |
+
{
|
| 26700 |
+
"epoch": 0.8464444444444444,
|
| 26701 |
+
"grad_norm": 0.08038879185914993,
|
| 26702 |
+
"learning_rate": 3.077951002227172e-05,
|
| 26703 |
+
"loss": 0.0106,
|
| 26704 |
+
"step": 3809
|
| 26705 |
+
},
|
| 26706 |
+
{
|
| 26707 |
+
"epoch": 0.8466666666666667,
|
| 26708 |
+
"grad_norm": 0.617675244808197,
|
| 26709 |
+
"learning_rate": 3.073496659242761e-05,
|
| 26710 |
+
"loss": 1.0167,
|
| 26711 |
+
"step": 3810
|
| 26712 |
+
},
|
| 26713 |
+
{
|
| 26714 |
+
"epoch": 0.8468888888888889,
|
| 26715 |
+
"grad_norm": 0.8487913012504578,
|
| 26716 |
+
"learning_rate": 3.069042316258352e-05,
|
| 26717 |
+
"loss": 1.8088,
|
| 26718 |
+
"step": 3811
|
| 26719 |
+
},
|
| 26720 |
+
{
|
| 26721 |
+
"epoch": 0.8471111111111111,
|
| 26722 |
+
"grad_norm": 0.8923436403274536,
|
| 26723 |
+
"learning_rate": 3.0645879732739424e-05,
|
| 26724 |
+
"loss": 1.8555,
|
| 26725 |
+
"step": 3812
|
| 26726 |
+
},
|
| 26727 |
+
{
|
| 26728 |
+
"epoch": 0.8473333333333334,
|
| 26729 |
+
"grad_norm": 0.9946725368499756,
|
| 26730 |
+
"learning_rate": 3.060133630289532e-05,
|
| 26731 |
+
"loss": 2.1355,
|
| 26732 |
+
"step": 3813
|
| 26733 |
+
},
|
| 26734 |
+
{
|
| 26735 |
+
"epoch": 0.8475555555555555,
|
| 26736 |
+
"grad_norm": 0.8608193397521973,
|
| 26737 |
+
"learning_rate": 3.055679287305123e-05,
|
| 26738 |
+
"loss": 1.8501,
|
| 26739 |
+
"step": 3814
|
| 26740 |
+
},
|
| 26741 |
+
{
|
| 26742 |
+
"epoch": 0.8477777777777777,
|
| 26743 |
+
"grad_norm": 1.0533936023712158,
|
| 26744 |
+
"learning_rate": 3.051224944320713e-05,
|
| 26745 |
+
"loss": 2.3759,
|
| 26746 |
+
"step": 3815
|
| 26747 |
+
},
|
| 26748 |
+
{
|
| 26749 |
+
"epoch": 0.848,
|
| 26750 |
+
"grad_norm": 0.9395473003387451,
|
| 26751 |
+
"learning_rate": 3.046770601336303e-05,
|
| 26752 |
+
"loss": 1.8055,
|
| 26753 |
+
"step": 3816
|
| 26754 |
+
},
|
| 26755 |
+
{
|
| 26756 |
+
"epoch": 0.8482222222222222,
|
| 26757 |
+
"grad_norm": 0.868739902973175,
|
| 26758 |
+
"learning_rate": 3.0423162583518934e-05,
|
| 26759 |
+
"loss": 1.9014,
|
| 26760 |
+
"step": 3817
|
| 26761 |
+
},
|
| 26762 |
+
{
|
| 26763 |
+
"epoch": 0.8484444444444444,
|
| 26764 |
+
"grad_norm": 0.9286447167396545,
|
| 26765 |
+
"learning_rate": 3.0378619153674836e-05,
|
| 26766 |
+
"loss": 2.1256,
|
| 26767 |
+
"step": 3818
|
| 26768 |
+
},
|
| 26769 |
+
{
|
| 26770 |
+
"epoch": 0.8486666666666667,
|
| 26771 |
+
"grad_norm": 0.9453836679458618,
|
| 26772 |
+
"learning_rate": 3.0334075723830734e-05,
|
| 26773 |
+
"loss": 1.9572,
|
| 26774 |
+
"step": 3819
|
| 26775 |
+
},
|
| 26776 |
+
{
|
| 26777 |
+
"epoch": 0.8488888888888889,
|
| 26778 |
+
"grad_norm": 1.1995909214019775,
|
| 26779 |
+
"learning_rate": 3.028953229398664e-05,
|
| 26780 |
+
"loss": 1.7376,
|
| 26781 |
+
"step": 3820
|
| 26782 |
+
},
|
| 26783 |
+
{
|
| 26784 |
+
"epoch": 0.8491111111111111,
|
| 26785 |
+
"grad_norm": 1.2044036388397217,
|
| 26786 |
+
"learning_rate": 3.024498886414254e-05,
|
| 26787 |
+
"loss": 2.1343,
|
| 26788 |
+
"step": 3821
|
| 26789 |
+
},
|
| 26790 |
+
{
|
| 26791 |
+
"epoch": 0.8493333333333334,
|
| 26792 |
+
"grad_norm": 0.8160643577575684,
|
| 26793 |
+
"learning_rate": 3.020044543429844e-05,
|
| 26794 |
+
"loss": 0.9794,
|
| 26795 |
+
"step": 3822
|
| 26796 |
+
},
|
| 26797 |
+
{
|
| 26798 |
+
"epoch": 0.8495555555555555,
|
| 26799 |
+
"grad_norm": 0.06719803065061569,
|
| 26800 |
+
"learning_rate": 3.0155902004454346e-05,
|
| 26801 |
+
"loss": 0.0154,
|
| 26802 |
+
"step": 3823
|
| 26803 |
+
},
|
| 26804 |
+
{
|
| 26805 |
+
"epoch": 0.8497777777777777,
|
| 26806 |
+
"grad_norm": 0.06740820407867432,
|
| 26807 |
+
"learning_rate": 3.0111358574610248e-05,
|
| 26808 |
+
"loss": 0.0154,
|
| 26809 |
+
"step": 3824
|
| 26810 |
+
},
|
| 26811 |
+
{
|
| 26812 |
+
"epoch": 0.85,
|
| 26813 |
+
"grad_norm": 1.0504337549209595,
|
| 26814 |
+
"learning_rate": 3.0066815144766146e-05,
|
| 26815 |
+
"loss": 1.8658,
|
| 26816 |
+
"step": 3825
|
| 26817 |
+
},
|
| 26818 |
+
{
|
| 26819 |
+
"epoch": 0.8502222222222222,
|
| 26820 |
+
"grad_norm": 0.7439045906066895,
|
| 26821 |
+
"learning_rate": 3.002227171492205e-05,
|
| 26822 |
+
"loss": 1.1039,
|
| 26823 |
+
"step": 3826
|
| 26824 |
+
},
|
| 26825 |
+
{
|
| 26826 |
+
"epoch": 0.8504444444444444,
|
| 26827 |
+
"grad_norm": 0.1253952980041504,
|
| 26828 |
+
"learning_rate": 2.9977728285077953e-05,
|
| 26829 |
+
"loss": 0.0206,
|
| 26830 |
+
"step": 3827
|
| 26831 |
+
},
|
| 26832 |
+
{
|
| 26833 |
+
"epoch": 0.8506666666666667,
|
| 26834 |
+
"grad_norm": 1.0089833736419678,
|
| 26835 |
+
"learning_rate": 2.9933184855233852e-05,
|
| 26836 |
+
"loss": 2.0178,
|
| 26837 |
+
"step": 3828
|
| 26838 |
+
},
|
| 26839 |
+
{
|
| 26840 |
+
"epoch": 0.8508888888888889,
|
| 26841 |
+
"grad_norm": 1.0395070314407349,
|
| 26842 |
+
"learning_rate": 2.9888641425389757e-05,
|
| 26843 |
+
"loss": 1.9692,
|
| 26844 |
+
"step": 3829
|
| 26845 |
+
},
|
| 26846 |
+
{
|
| 26847 |
+
"epoch": 0.8511111111111112,
|
| 26848 |
+
"grad_norm": 1.0526185035705566,
|
| 26849 |
+
"learning_rate": 2.9844097995545663e-05,
|
| 26850 |
+
"loss": 1.8065,
|
| 26851 |
+
"step": 3830
|
| 26852 |
+
},
|
| 26853 |
+
{
|
| 26854 |
+
"epoch": 0.8513333333333334,
|
| 26855 |
+
"grad_norm": 1.0034129619598389,
|
| 26856 |
+
"learning_rate": 2.9799554565701558e-05,
|
| 26857 |
+
"loss": 1.679,
|
| 26858 |
+
"step": 3831
|
| 26859 |
+
},
|
| 26860 |
+
{
|
| 26861 |
+
"epoch": 0.8515555555555555,
|
| 26862 |
+
"grad_norm": 0.06555074453353882,
|
| 26863 |
+
"learning_rate": 2.9755011135857463e-05,
|
| 26864 |
+
"loss": 0.017,
|
| 26865 |
+
"step": 3832
|
| 26866 |
+
},
|
| 26867 |
+
{
|
| 26868 |
+
"epoch": 0.8517777777777777,
|
| 26869 |
+
"grad_norm": 0.8089559078216553,
|
| 26870 |
+
"learning_rate": 2.9710467706013362e-05,
|
| 26871 |
+
"loss": 0.9377,
|
| 26872 |
+
"step": 3833
|
| 26873 |
+
},
|
| 26874 |
+
{
|
| 26875 |
+
"epoch": 0.852,
|
| 26876 |
+
"grad_norm": 0.7607543468475342,
|
| 26877 |
+
"learning_rate": 2.9665924276169267e-05,
|
| 26878 |
+
"loss": 0.873,
|
| 26879 |
+
"step": 3834
|
| 26880 |
+
},
|
| 26881 |
+
{
|
| 26882 |
+
"epoch": 0.8522222222222222,
|
| 26883 |
+
"grad_norm": 0.08201993256807327,
|
| 26884 |
+
"learning_rate": 2.962138084632517e-05,
|
| 26885 |
+
"loss": 0.0183,
|
| 26886 |
+
"step": 3835
|
| 26887 |
+
},
|
| 26888 |
+
{
|
| 26889 |
+
"epoch": 0.8524444444444444,
|
| 26890 |
+
"grad_norm": 0.6691009402275085,
|
| 26891 |
+
"learning_rate": 2.9576837416481068e-05,
|
| 26892 |
+
"loss": 0.6488,
|
| 26893 |
+
"step": 3836
|
| 26894 |
+
},
|
| 26895 |
+
{
|
| 26896 |
+
"epoch": 0.8526666666666667,
|
| 26897 |
+
"grad_norm": 1.0818275213241577,
|
| 26898 |
+
"learning_rate": 2.9532293986636973e-05,
|
| 26899 |
+
"loss": 1.6231,
|
| 26900 |
+
"step": 3837
|
| 26901 |
+
},
|
| 26902 |
+
{
|
| 26903 |
+
"epoch": 0.8528888888888889,
|
| 26904 |
+
"grad_norm": 0.10234081745147705,
|
| 26905 |
+
"learning_rate": 2.9487750556792875e-05,
|
| 26906 |
+
"loss": 0.0249,
|
| 26907 |
+
"step": 3838
|
| 26908 |
+
},
|
| 26909 |
+
{
|
| 26910 |
+
"epoch": 0.8531111111111112,
|
| 26911 |
+
"grad_norm": 0.7301368117332458,
|
| 26912 |
+
"learning_rate": 2.9443207126948774e-05,
|
| 26913 |
+
"loss": 0.7285,
|
| 26914 |
+
"step": 3839
|
| 26915 |
+
},
|
| 26916 |
+
{
|
| 26917 |
+
"epoch": 0.8533333333333334,
|
| 26918 |
+
"grad_norm": 1.020973563194275,
|
| 26919 |
+
"learning_rate": 2.939866369710468e-05,
|
| 26920 |
+
"loss": 1.5453,
|
| 26921 |
+
"step": 3840
|
| 26922 |
+
},
|
| 26923 |
+
{
|
| 26924 |
+
"epoch": 0.8535555555555555,
|
| 26925 |
+
"grad_norm": 1.161118745803833,
|
| 26926 |
+
"learning_rate": 2.935412026726058e-05,
|
| 26927 |
+
"loss": 1.5789,
|
| 26928 |
+
"step": 3841
|
| 26929 |
+
},
|
| 26930 |
+
{
|
| 26931 |
+
"epoch": 0.8537777777777777,
|
| 26932 |
+
"grad_norm": 1.1855006217956543,
|
| 26933 |
+
"learning_rate": 2.930957683741648e-05,
|
| 26934 |
+
"loss": 1.443,
|
| 26935 |
+
"step": 3842
|
| 26936 |
+
},
|
| 26937 |
+
{
|
| 26938 |
+
"epoch": 0.854,
|
| 26939 |
+
"grad_norm": 1.0485907793045044,
|
| 26940 |
+
"learning_rate": 2.9265033407572385e-05,
|
| 26941 |
+
"loss": 1.2626,
|
| 26942 |
+
"step": 3843
|
| 26943 |
+
},
|
| 26944 |
+
{
|
| 26945 |
+
"epoch": 0.8542222222222222,
|
| 26946 |
+
"grad_norm": 1.0825096368789673,
|
| 26947 |
+
"learning_rate": 2.9220489977728287e-05,
|
| 26948 |
+
"loss": 1.4062,
|
| 26949 |
+
"step": 3844
|
| 26950 |
+
},
|
| 26951 |
+
{
|
| 26952 |
+
"epoch": 0.8544444444444445,
|
| 26953 |
+
"grad_norm": 0.16687047481536865,
|
| 26954 |
+
"learning_rate": 2.9175946547884186e-05,
|
| 26955 |
+
"loss": 0.0262,
|
| 26956 |
+
"step": 3845
|
| 26957 |
+
},
|
| 26958 |
+
{
|
| 26959 |
+
"epoch": 0.8546666666666667,
|
| 26960 |
+
"grad_norm": 1.0403611660003662,
|
| 26961 |
+
"learning_rate": 2.913140311804009e-05,
|
| 26962 |
+
"loss": 1.4555,
|
| 26963 |
+
"step": 3846
|
| 26964 |
+
},
|
| 26965 |
+
{
|
| 26966 |
+
"epoch": 0.8548888888888889,
|
| 26967 |
+
"grad_norm": 1.069176197052002,
|
| 26968 |
+
"learning_rate": 2.9086859688195993e-05,
|
| 26969 |
+
"loss": 1.2518,
|
| 26970 |
+
"step": 3847
|
| 26971 |
+
},
|
| 26972 |
+
{
|
| 26973 |
+
"epoch": 0.8551111111111112,
|
| 26974 |
+
"grad_norm": 1.2168667316436768,
|
| 26975 |
+
"learning_rate": 2.904231625835189e-05,
|
| 26976 |
+
"loss": 1.289,
|
| 26977 |
+
"step": 3848
|
| 26978 |
+
},
|
| 26979 |
+
{
|
| 26980 |
+
"epoch": 0.8553333333333333,
|
| 26981 |
+
"grad_norm": 0.5977094769477844,
|
| 26982 |
+
"learning_rate": 2.8997772828507797e-05,
|
| 26983 |
+
"loss": 0.4731,
|
| 26984 |
+
"step": 3849
|
| 26985 |
+
},
|
| 26986 |
+
{
|
| 26987 |
+
"epoch": 0.8555555555555555,
|
| 26988 |
+
"grad_norm": 0.543451189994812,
|
| 26989 |
+
"learning_rate": 2.89532293986637e-05,
|
| 26990 |
+
"loss": 0.3367,
|
| 26991 |
+
"step": 3850
|
| 26992 |
+
},
|
| 26993 |
+
{
|
| 26994 |
+
"epoch": 0.8557777777777777,
|
| 26995 |
+
"grad_norm": 0.044504791498184204,
|
| 26996 |
+
"learning_rate": 2.8908685968819597e-05,
|
| 26997 |
+
"loss": 0.0103,
|
| 26998 |
+
"step": 3851
|
| 26999 |
+
},
|
| 27000 |
+
{
|
| 27001 |
+
"epoch": 0.856,
|
| 27002 |
+
"grad_norm": 0.8173375725746155,
|
| 27003 |
+
"learning_rate": 2.8864142538975503e-05,
|
| 27004 |
+
"loss": 2.0658,
|
| 27005 |
+
"step": 3852
|
| 27006 |
+
},
|
| 27007 |
+
{
|
| 27008 |
+
"epoch": 0.8562222222222222,
|
| 27009 |
+
"grad_norm": 0.6008175015449524,
|
| 27010 |
+
"learning_rate": 2.8819599109131408e-05,
|
| 27011 |
+
"loss": 1.0048,
|
| 27012 |
+
"step": 3853
|
| 27013 |
+
},
|
| 27014 |
+
{
|
| 27015 |
+
"epoch": 0.8564444444444445,
|
| 27016 |
+
"grad_norm": 0.6246810555458069,
|
| 27017 |
+
"learning_rate": 2.8775055679287303e-05,
|
| 27018 |
+
"loss": 1.0661,
|
| 27019 |
+
"step": 3854
|
| 27020 |
+
},
|
| 27021 |
+
{
|
| 27022 |
+
"epoch": 0.8566666666666667,
|
| 27023 |
+
"grad_norm": 0.9632955193519592,
|
| 27024 |
+
"learning_rate": 2.873051224944321e-05,
|
| 27025 |
+
"loss": 2.273,
|
| 27026 |
+
"step": 3855
|
| 27027 |
+
},
|
| 27028 |
+
{
|
| 27029 |
+
"epoch": 0.8568888888888889,
|
| 27030 |
+
"grad_norm": 0.8222072720527649,
|
| 27031 |
+
"learning_rate": 2.8685968819599114e-05,
|
| 27032 |
+
"loss": 2.0065,
|
| 27033 |
+
"step": 3856
|
| 27034 |
+
},
|
| 27035 |
+
{
|
| 27036 |
+
"epoch": 0.8571111111111112,
|
| 27037 |
+
"grad_norm": 0.6057097911834717,
|
| 27038 |
+
"learning_rate": 2.8641425389755013e-05,
|
| 27039 |
+
"loss": 1.1629,
|
| 27040 |
+
"step": 3857
|
| 27041 |
+
},
|
| 27042 |
+
{
|
| 27043 |
+
"epoch": 0.8573333333333333,
|
| 27044 |
+
"grad_norm": 0.06852009892463684,
|
| 27045 |
+
"learning_rate": 2.8596881959910915e-05,
|
| 27046 |
+
"loss": 0.0104,
|
| 27047 |
+
"step": 3858
|
| 27048 |
+
},
|
| 27049 |
+
{
|
| 27050 |
+
"epoch": 0.8575555555555555,
|
| 27051 |
+
"grad_norm": 0.06707873195409775,
|
| 27052 |
+
"learning_rate": 2.855233853006682e-05,
|
| 27053 |
+
"loss": 0.0103,
|
| 27054 |
+
"step": 3859
|
| 27055 |
+
},
|
| 27056 |
+
{
|
| 27057 |
+
"epoch": 0.8577777777777778,
|
| 27058 |
+
"grad_norm": 0.0674692690372467,
|
| 27059 |
+
"learning_rate": 2.850779510022272e-05,
|
| 27060 |
+
"loss": 0.0099,
|
| 27061 |
+
"step": 3860
|
| 27062 |
+
},
|
| 27063 |
+
{
|
| 27064 |
+
"epoch": 0.858,
|
| 27065 |
+
"grad_norm": 0.5297547578811646,
|
| 27066 |
+
"learning_rate": 2.846325167037862e-05,
|
| 27067 |
+
"loss": 1.1045,
|
| 27068 |
+
"step": 3861
|
| 27069 |
+
},
|
| 27070 |
+
{
|
| 27071 |
+
"epoch": 0.8582222222222222,
|
| 27072 |
+
"grad_norm": 0.9173485040664673,
|
| 27073 |
+
"learning_rate": 2.8418708240534526e-05,
|
| 27074 |
+
"loss": 1.9225,
|
| 27075 |
+
"step": 3862
|
| 27076 |
+
},
|
| 27077 |
+
{
|
| 27078 |
+
"epoch": 0.8584444444444445,
|
| 27079 |
+
"grad_norm": 0.9960424900054932,
|
| 27080 |
+
"learning_rate": 2.8374164810690424e-05,
|
| 27081 |
+
"loss": 2.0529,
|
| 27082 |
+
"step": 3863
|
| 27083 |
+
},
|
| 27084 |
+
{
|
| 27085 |
+
"epoch": 0.8586666666666667,
|
| 27086 |
+
"grad_norm": 0.6144242286682129,
|
| 27087 |
+
"learning_rate": 2.8329621380846326e-05,
|
| 27088 |
+
"loss": 1.0756,
|
| 27089 |
+
"step": 3864
|
| 27090 |
+
},
|
| 27091 |
+
{
|
| 27092 |
+
"epoch": 0.8588888888888889,
|
| 27093 |
+
"grad_norm": 0.9492395520210266,
|
| 27094 |
+
"learning_rate": 2.8285077951002232e-05,
|
| 27095 |
+
"loss": 1.5811,
|
| 27096 |
+
"step": 3865
|
| 27097 |
+
},
|
| 27098 |
+
{
|
| 27099 |
+
"epoch": 0.8591111111111112,
|
| 27100 |
+
"grad_norm": 1.0924068689346313,
|
| 27101 |
+
"learning_rate": 2.824053452115813e-05,
|
| 27102 |
+
"loss": 2.1395,
|
| 27103 |
+
"step": 3866
|
| 27104 |
+
},
|
| 27105 |
+
{
|
| 27106 |
+
"epoch": 0.8593333333333333,
|
| 27107 |
+
"grad_norm": 1.214991569519043,
|
| 27108 |
+
"learning_rate": 2.8195991091314032e-05,
|
| 27109 |
+
"loss": 2.3004,
|
| 27110 |
+
"step": 3867
|
| 27111 |
+
},
|
| 27112 |
+
{
|
| 27113 |
+
"epoch": 0.8595555555555555,
|
| 27114 |
+
"grad_norm": 0.9233739972114563,
|
| 27115 |
+
"learning_rate": 2.815144766146993e-05,
|
| 27116 |
+
"loss": 1.9741,
|
| 27117 |
+
"step": 3868
|
| 27118 |
+
},
|
| 27119 |
+
{
|
| 27120 |
+
"epoch": 0.8597777777777778,
|
| 27121 |
+
"grad_norm": 0.9544225931167603,
|
| 27122 |
+
"learning_rate": 2.8106904231625836e-05,
|
| 27123 |
+
"loss": 2.0073,
|
| 27124 |
+
"step": 3869
|
| 27125 |
+
},
|
| 27126 |
+
{
|
| 27127 |
+
"epoch": 0.86,
|
| 27128 |
+
"grad_norm": 1.0409972667694092,
|
| 27129 |
+
"learning_rate": 2.8062360801781738e-05,
|
| 27130 |
+
"loss": 2.0343,
|
| 27131 |
+
"step": 3870
|
| 27132 |
+
},
|
| 27133 |
+
{
|
| 27134 |
+
"epoch": 0.8602222222222222,
|
| 27135 |
+
"grad_norm": 1.0500308275222778,
|
| 27136 |
+
"learning_rate": 2.8017817371937637e-05,
|
| 27137 |
+
"loss": 1.8736,
|
| 27138 |
+
"step": 3871
|
| 27139 |
+
},
|
| 27140 |
+
{
|
| 27141 |
+
"epoch": 0.8604444444444445,
|
| 27142 |
+
"grad_norm": 1.0792829990386963,
|
| 27143 |
+
"learning_rate": 2.7973273942093542e-05,
|
| 27144 |
+
"loss": 1.9845,
|
| 27145 |
+
"step": 3872
|
| 27146 |
+
},
|
| 27147 |
+
{
|
| 27148 |
+
"epoch": 0.8606666666666667,
|
| 27149 |
+
"grad_norm": 1.1363780498504639,
|
| 27150 |
+
"learning_rate": 2.7928730512249447e-05,
|
| 27151 |
+
"loss": 1.8806,
|
| 27152 |
+
"step": 3873
|
| 27153 |
+
},
|
| 27154 |
+
{
|
| 27155 |
+
"epoch": 0.8608888888888889,
|
| 27156 |
+
"grad_norm": 0.6764510869979858,
|
| 27157 |
+
"learning_rate": 2.7884187082405343e-05,
|
| 27158 |
+
"loss": 0.9852,
|
| 27159 |
+
"step": 3874
|
| 27160 |
+
},
|
| 27161 |
+
{
|
| 27162 |
+
"epoch": 0.8611111111111112,
|
| 27163 |
+
"grad_norm": 0.06674336642026901,
|
| 27164 |
+
"learning_rate": 2.7839643652561248e-05,
|
| 27165 |
+
"loss": 0.0153,
|
| 27166 |
+
"step": 3875
|
| 27167 |
+
},
|
| 27168 |
+
{
|
| 27169 |
+
"epoch": 0.8613333333333333,
|
| 27170 |
+
"grad_norm": 0.06754778325557709,
|
| 27171 |
+
"learning_rate": 2.7795100222717153e-05,
|
| 27172 |
+
"loss": 0.0156,
|
| 27173 |
+
"step": 3876
|
| 27174 |
+
},
|
| 27175 |
+
{
|
| 27176 |
+
"epoch": 0.8615555555555555,
|
| 27177 |
+
"grad_norm": 0.6871387362480164,
|
| 27178 |
+
"learning_rate": 2.7750556792873052e-05,
|
| 27179 |
+
"loss": 0.8689,
|
| 27180 |
+
"step": 3877
|
| 27181 |
+
},
|
| 27182 |
+
{
|
| 27183 |
+
"epoch": 0.8617777777777778,
|
| 27184 |
+
"grad_norm": 1.1654753684997559,
|
| 27185 |
+
"learning_rate": 2.7706013363028954e-05,
|
| 27186 |
+
"loss": 1.9632,
|
| 27187 |
+
"step": 3878
|
| 27188 |
+
},
|
| 27189 |
+
{
|
| 27190 |
+
"epoch": 0.862,
|
| 27191 |
+
"grad_norm": 0.13107286393642426,
|
| 27192 |
+
"learning_rate": 2.766146993318486e-05,
|
| 27193 |
+
"loss": 0.0207,
|
| 27194 |
+
"step": 3879
|
| 27195 |
+
},
|
| 27196 |
+
{
|
| 27197 |
+
"epoch": 0.8622222222222222,
|
| 27198 |
+
"grad_norm": 1.0098730325698853,
|
| 27199 |
+
"learning_rate": 2.7616926503340758e-05,
|
| 27200 |
+
"loss": 1.7388,
|
| 27201 |
+
"step": 3880
|
| 27202 |
+
},
|
| 27203 |
+
{
|
| 27204 |
+
"epoch": 0.8624444444444445,
|
| 27205 |
+
"grad_norm": 0.9629087448120117,
|
| 27206 |
+
"learning_rate": 2.757238307349666e-05,
|
| 27207 |
+
"loss": 1.7969,
|
| 27208 |
+
"step": 3881
|
| 27209 |
+
},
|
| 27210 |
+
{
|
| 27211 |
+
"epoch": 0.8626666666666667,
|
| 27212 |
+
"grad_norm": 0.9217532873153687,
|
| 27213 |
+
"learning_rate": 2.7527839643652565e-05,
|
| 27214 |
+
"loss": 1.9322,
|
| 27215 |
+
"step": 3882
|
| 27216 |
+
},
|
| 27217 |
+
{
|
| 27218 |
+
"epoch": 0.8628888888888889,
|
| 27219 |
+
"grad_norm": 1.0283830165863037,
|
| 27220 |
+
"learning_rate": 2.7483296213808464e-05,
|
| 27221 |
+
"loss": 1.7681,
|
| 27222 |
+
"step": 3883
|
| 27223 |
+
},
|
| 27224 |
+
{
|
| 27225 |
+
"epoch": 0.8631111111111112,
|
| 27226 |
+
"grad_norm": 0.701818585395813,
|
| 27227 |
+
"learning_rate": 2.7438752783964366e-05,
|
| 27228 |
+
"loss": 0.8642,
|
| 27229 |
+
"step": 3884
|
| 27230 |
+
},
|
| 27231 |
+
{
|
| 27232 |
+
"epoch": 0.8633333333333333,
|
| 27233 |
+
"grad_norm": 0.7634962201118469,
|
| 27234 |
+
"learning_rate": 2.739420935412027e-05,
|
| 27235 |
+
"loss": 0.7195,
|
| 27236 |
+
"step": 3885
|
| 27237 |
+
},
|
| 27238 |
+
{
|
| 27239 |
+
"epoch": 0.8635555555555555,
|
| 27240 |
+
"grad_norm": 0.9613010287284851,
|
| 27241 |
+
"learning_rate": 2.734966592427617e-05,
|
| 27242 |
+
"loss": 1.7299,
|
| 27243 |
+
"step": 3886
|
| 27244 |
+
},
|
| 27245 |
+
{
|
| 27246 |
+
"epoch": 0.8637777777777778,
|
| 27247 |
+
"grad_norm": 0.8127443790435791,
|
| 27248 |
+
"learning_rate": 2.730512249443207e-05,
|
| 27249 |
+
"loss": 0.9909,
|
| 27250 |
+
"step": 3887
|
| 27251 |
+
},
|
| 27252 |
+
{
|
| 27253 |
+
"epoch": 0.864,
|
| 27254 |
+
"grad_norm": 0.7633342146873474,
|
| 27255 |
+
"learning_rate": 2.7260579064587977e-05,
|
| 27256 |
+
"loss": 0.7784,
|
| 27257 |
+
"step": 3888
|
| 27258 |
+
},
|
| 27259 |
+
{
|
| 27260 |
+
"epoch": 0.8642222222222222,
|
| 27261 |
+
"grad_norm": 0.8209825754165649,
|
| 27262 |
+
"learning_rate": 2.7216035634743876e-05,
|
| 27263 |
+
"loss": 0.8605,
|
| 27264 |
+
"step": 3889
|
| 27265 |
+
},
|
| 27266 |
+
{
|
| 27267 |
+
"epoch": 0.8644444444444445,
|
| 27268 |
+
"grad_norm": 1.1006879806518555,
|
| 27269 |
+
"learning_rate": 2.7171492204899778e-05,
|
| 27270 |
+
"loss": 1.4274,
|
| 27271 |
+
"step": 3890
|
| 27272 |
+
},
|
| 27273 |
+
{
|
| 27274 |
+
"epoch": 0.8646666666666667,
|
| 27275 |
+
"grad_norm": 0.9458972811698914,
|
| 27276 |
+
"learning_rate": 2.7126948775055683e-05,
|
| 27277 |
+
"loss": 1.4813,
|
| 27278 |
+
"step": 3891
|
| 27279 |
+
},
|
| 27280 |
+
{
|
| 27281 |
+
"epoch": 0.8648888888888889,
|
| 27282 |
+
"grad_norm": 1.3641767501831055,
|
| 27283 |
+
"learning_rate": 2.708240534521158e-05,
|
| 27284 |
+
"loss": 1.5427,
|
| 27285 |
+
"step": 3892
|
| 27286 |
+
},
|
| 27287 |
+
{
|
| 27288 |
+
"epoch": 0.8651111111111112,
|
| 27289 |
+
"grad_norm": 1.0138379335403442,
|
| 27290 |
+
"learning_rate": 2.7037861915367484e-05,
|
| 27291 |
+
"loss": 1.4602,
|
| 27292 |
+
"step": 3893
|
| 27293 |
+
},
|
| 27294 |
+
{
|
| 27295 |
+
"epoch": 0.8653333333333333,
|
| 27296 |
+
"grad_norm": 1.0641552209854126,
|
| 27297 |
+
"learning_rate": 2.699331848552339e-05,
|
| 27298 |
+
"loss": 1.2653,
|
| 27299 |
+
"step": 3894
|
| 27300 |
+
},
|
| 27301 |
+
{
|
| 27302 |
+
"epoch": 0.8655555555555555,
|
| 27303 |
+
"grad_norm": 1.2748581171035767,
|
| 27304 |
+
"learning_rate": 2.6948775055679287e-05,
|
| 27305 |
+
"loss": 1.3019,
|
| 27306 |
+
"step": 3895
|
| 27307 |
+
},
|
| 27308 |
+
{
|
| 27309 |
+
"epoch": 0.8657777777777778,
|
| 27310 |
+
"grad_norm": 1.1393803358078003,
|
| 27311 |
+
"learning_rate": 2.6904231625835193e-05,
|
| 27312 |
+
"loss": 1.331,
|
| 27313 |
+
"step": 3896
|
| 27314 |
+
},
|
| 27315 |
+
{
|
| 27316 |
+
"epoch": 0.866,
|
| 27317 |
+
"grad_norm": 1.0819334983825684,
|
| 27318 |
+
"learning_rate": 2.6859688195991095e-05,
|
| 27319 |
+
"loss": 1.1292,
|
| 27320 |
+
"step": 3897
|
| 27321 |
+
},
|
| 27322 |
+
{
|
| 27323 |
+
"epoch": 0.8662222222222222,
|
| 27324 |
+
"grad_norm": 0.14373019337654114,
|
| 27325 |
+
"learning_rate": 2.6815144766146993e-05,
|
| 27326 |
+
"loss": 0.0321,
|
| 27327 |
+
"step": 3898
|
| 27328 |
+
},
|
| 27329 |
+
{
|
| 27330 |
+
"epoch": 0.8664444444444445,
|
| 27331 |
+
"grad_norm": 0.14658400416374207,
|
| 27332 |
+
"learning_rate": 2.67706013363029e-05,
|
| 27333 |
+
"loss": 0.032,
|
| 27334 |
+
"step": 3899
|
| 27335 |
+
},
|
| 27336 |
+
{
|
| 27337 |
+
"epoch": 0.8666666666666667,
|
| 27338 |
+
"grad_norm": 1.1421598196029663,
|
| 27339 |
+
"learning_rate": 2.67260579064588e-05,
|
| 27340 |
+
"loss": 1.0447,
|
| 27341 |
+
"step": 3900
|
| 27342 |
+
},
|
| 27343 |
+
{
|
| 27344 |
+
"epoch": 0.8668888888888889,
|
| 27345 |
+
"grad_norm": 0.6876357793807983,
|
| 27346 |
+
"learning_rate": 2.66815144766147e-05,
|
| 27347 |
+
"loss": 1.2227,
|
| 27348 |
+
"step": 3901
|
| 27349 |
+
},
|
| 27350 |
+
{
|
| 27351 |
+
"epoch": 0.8671111111111112,
|
| 27352 |
+
"grad_norm": 0.0448576956987381,
|
| 27353 |
+
"learning_rate": 2.6636971046770605e-05,
|
| 27354 |
+
"loss": 0.0104,
|
| 27355 |
+
"step": 3902
|
| 27356 |
+
},
|
| 27357 |
+
{
|
| 27358 |
+
"epoch": 0.8673333333333333,
|
| 27359 |
+
"grad_norm": 0.6660778522491455,
|
| 27360 |
+
"learning_rate": 2.6592427616926503e-05,
|
| 27361 |
+
"loss": 1.121,
|
| 27362 |
+
"step": 3903
|
| 27363 |
+
},
|
| 27364 |
+
{
|
| 27365 |
+
"epoch": 0.8675555555555555,
|
| 27366 |
+
"grad_norm": 0.043087027966976166,
|
| 27367 |
+
"learning_rate": 2.6547884187082405e-05,
|
| 27368 |
+
"loss": 0.0103,
|
| 27369 |
+
"step": 3904
|
| 27370 |
+
},
|
| 27371 |
+
{
|
| 27372 |
+
"epoch": 0.8677777777777778,
|
| 27373 |
+
"grad_norm": 0.5372818112373352,
|
| 27374 |
+
"learning_rate": 2.650334075723831e-05,
|
| 27375 |
+
"loss": 1.0993,
|
| 27376 |
+
"step": 3905
|
| 27377 |
+
},
|
| 27378 |
+
{
|
| 27379 |
+
"epoch": 0.868,
|
| 27380 |
+
"grad_norm": 0.9083240628242493,
|
| 27381 |
+
"learning_rate": 2.645879732739421e-05,
|
| 27382 |
+
"loss": 2.459,
|
| 27383 |
+
"step": 3906
|
| 27384 |
+
},
|
| 27385 |
+
{
|
| 27386 |
+
"epoch": 0.8682222222222222,
|
| 27387 |
+
"grad_norm": 0.8523256182670593,
|
| 27388 |
+
"learning_rate": 2.641425389755011e-05,
|
| 27389 |
+
"loss": 2.1683,
|
| 27390 |
+
"step": 3907
|
| 27391 |
+
},
|
| 27392 |
+
{
|
| 27393 |
+
"epoch": 0.8684444444444445,
|
| 27394 |
+
"grad_norm": 0.6197808384895325,
|
| 27395 |
+
"learning_rate": 2.6369710467706016e-05,
|
| 27396 |
+
"loss": 1.0535,
|
| 27397 |
+
"step": 3908
|
| 27398 |
+
},
|
| 27399 |
+
{
|
| 27400 |
+
"epoch": 0.8686666666666667,
|
| 27401 |
+
"grad_norm": 0.8953803181648254,
|
| 27402 |
+
"learning_rate": 2.6325167037861915e-05,
|
| 27403 |
+
"loss": 1.9434,
|
| 27404 |
+
"step": 3909
|
| 27405 |
+
},
|
| 27406 |
+
{
|
| 27407 |
+
"epoch": 0.8688888888888889,
|
| 27408 |
+
"grad_norm": 0.9139788150787354,
|
| 27409 |
+
"learning_rate": 2.6280623608017817e-05,
|
| 27410 |
+
"loss": 1.8545,
|
| 27411 |
+
"step": 3910
|
| 27412 |
+
},
|
| 27413 |
+
{
|
| 27414 |
+
"epoch": 0.8691111111111111,
|
| 27415 |
+
"grad_norm": 0.8638214468955994,
|
| 27416 |
+
"learning_rate": 2.6236080178173722e-05,
|
| 27417 |
+
"loss": 1.8329,
|
| 27418 |
+
"step": 3911
|
| 27419 |
+
},
|
| 27420 |
+
{
|
| 27421 |
+
"epoch": 0.8693333333333333,
|
| 27422 |
+
"grad_norm": 0.8344167470932007,
|
| 27423 |
+
"learning_rate": 2.619153674832962e-05,
|
| 27424 |
+
"loss": 1.7991,
|
| 27425 |
+
"step": 3912
|
| 27426 |
+
},
|
| 27427 |
+
{
|
| 27428 |
+
"epoch": 0.8695555555555555,
|
| 27429 |
+
"grad_norm": 0.96803879737854,
|
| 27430 |
+
"learning_rate": 2.6146993318485523e-05,
|
| 27431 |
+
"loss": 1.9849,
|
| 27432 |
+
"step": 3913
|
| 27433 |
+
},
|
| 27434 |
+
{
|
| 27435 |
+
"epoch": 0.8697777777777778,
|
| 27436 |
+
"grad_norm": 1.0239784717559814,
|
| 27437 |
+
"learning_rate": 2.6102449888641428e-05,
|
| 27438 |
+
"loss": 1.9256,
|
| 27439 |
+
"step": 3914
|
| 27440 |
+
},
|
| 27441 |
+
{
|
| 27442 |
+
"epoch": 0.87,
|
| 27443 |
+
"grad_norm": 0.8905801177024841,
|
| 27444 |
+
"learning_rate": 2.6057906458797327e-05,
|
| 27445 |
+
"loss": 1.6746,
|
| 27446 |
+
"step": 3915
|
| 27447 |
+
},
|
| 27448 |
+
{
|
| 27449 |
+
"epoch": 0.8702222222222222,
|
| 27450 |
+
"grad_norm": 1.0133596658706665,
|
| 27451 |
+
"learning_rate": 2.601336302895323e-05,
|
| 27452 |
+
"loss": 2.1594,
|
| 27453 |
+
"step": 3916
|
| 27454 |
+
},
|
| 27455 |
+
{
|
| 27456 |
+
"epoch": 0.8704444444444445,
|
| 27457 |
+
"grad_norm": 0.06942284107208252,
|
| 27458 |
+
"learning_rate": 2.5968819599109134e-05,
|
| 27459 |
+
"loss": 0.015,
|
| 27460 |
+
"step": 3917
|
| 27461 |
+
},
|
| 27462 |
+
{
|
| 27463 |
+
"epoch": 0.8706666666666667,
|
| 27464 |
+
"grad_norm": 0.07573316246271133,
|
| 27465 |
+
"learning_rate": 2.5924276169265033e-05,
|
| 27466 |
+
"loss": 0.0148,
|
| 27467 |
+
"step": 3918
|
| 27468 |
+
},
|
| 27469 |
+
{
|
| 27470 |
+
"epoch": 0.8708888888888889,
|
| 27471 |
+
"grad_norm": 0.07226064801216125,
|
| 27472 |
+
"learning_rate": 2.5879732739420938e-05,
|
| 27473 |
+
"loss": 0.015,
|
| 27474 |
+
"step": 3919
|
| 27475 |
+
},
|
| 27476 |
+
{
|
| 27477 |
+
"epoch": 0.8711111111111111,
|
| 27478 |
+
"grad_norm": 1.0551682710647583,
|
| 27479 |
+
"learning_rate": 2.583518930957684e-05,
|
| 27480 |
+
"loss": 1.9451,
|
| 27481 |
+
"step": 3920
|
| 27482 |
+
},
|
| 27483 |
+
{
|
| 27484 |
+
"epoch": 0.8713333333333333,
|
| 27485 |
+
"grad_norm": 1.0881084203720093,
|
| 27486 |
+
"learning_rate": 2.579064587973274e-05,
|
| 27487 |
+
"loss": 1.9361,
|
| 27488 |
+
"step": 3921
|
| 27489 |
+
},
|
| 27490 |
+
{
|
| 27491 |
+
"epoch": 0.8715555555555555,
|
| 27492 |
+
"grad_norm": 1.029228925704956,
|
| 27493 |
+
"learning_rate": 2.5746102449888644e-05,
|
| 27494 |
+
"loss": 1.97,
|
| 27495 |
+
"step": 3922
|
| 27496 |
+
},
|
| 27497 |
+
{
|
| 27498 |
+
"epoch": 0.8717777777777778,
|
| 27499 |
+
"grad_norm": 0.9416628479957581,
|
| 27500 |
+
"learning_rate": 2.5701559020044546e-05,
|
| 27501 |
+
"loss": 1.4815,
|
| 27502 |
+
"step": 3923
|
| 27503 |
+
},
|
| 27504 |
+
{
|
| 27505 |
+
"epoch": 0.872,
|
| 27506 |
+
"grad_norm": 1.8949933052062988,
|
| 27507 |
+
"learning_rate": 2.5657015590200445e-05,
|
| 27508 |
+
"loss": 2.0094,
|
| 27509 |
+
"step": 3924
|
| 27510 |
+
},
|
| 27511 |
+
{
|
| 27512 |
+
"epoch": 0.8722222222222222,
|
| 27513 |
+
"grad_norm": 0.9487776756286621,
|
| 27514 |
+
"learning_rate": 2.561247216035635e-05,
|
| 27515 |
+
"loss": 1.8348,
|
| 27516 |
+
"step": 3925
|
| 27517 |
+
},
|
| 27518 |
+
{
|
| 27519 |
+
"epoch": 0.8724444444444445,
|
| 27520 |
+
"grad_norm": 0.865877091884613,
|
| 27521 |
+
"learning_rate": 2.5567928730512252e-05,
|
| 27522 |
+
"loss": 1.5909,
|
| 27523 |
+
"step": 3926
|
| 27524 |
+
},
|
| 27525 |
+
{
|
| 27526 |
+
"epoch": 0.8726666666666667,
|
| 27527 |
+
"grad_norm": 0.9927725195884705,
|
| 27528 |
+
"learning_rate": 2.552338530066815e-05,
|
| 27529 |
+
"loss": 1.9465,
|
| 27530 |
+
"step": 3927
|
| 27531 |
+
},
|
| 27532 |
+
{
|
| 27533 |
+
"epoch": 0.8728888888888889,
|
| 27534 |
+
"grad_norm": 0.9912342429161072,
|
| 27535 |
+
"learning_rate": 2.5478841870824056e-05,
|
| 27536 |
+
"loss": 1.3772,
|
| 27537 |
+
"step": 3928
|
| 27538 |
+
},
|
| 27539 |
+
{
|
| 27540 |
+
"epoch": 0.8731111111111111,
|
| 27541 |
+
"grad_norm": 0.9611807465553284,
|
| 27542 |
+
"learning_rate": 2.5434298440979958e-05,
|
| 27543 |
+
"loss": 1.5165,
|
| 27544 |
+
"step": 3929
|
| 27545 |
+
},
|
| 27546 |
+
{
|
| 27547 |
+
"epoch": 0.8733333333333333,
|
| 27548 |
+
"grad_norm": 0.9328694343566895,
|
| 27549 |
+
"learning_rate": 2.5389755011135856e-05,
|
| 27550 |
+
"loss": 1.3826,
|
| 27551 |
+
"step": 3930
|
| 27552 |
+
},
|
| 27553 |
+
{
|
| 27554 |
+
"epoch": 0.8735555555555555,
|
| 27555 |
+
"grad_norm": 0.9587991237640381,
|
| 27556 |
+
"learning_rate": 2.5345211581291762e-05,
|
| 27557 |
+
"loss": 1.6781,
|
| 27558 |
+
"step": 3931
|
| 27559 |
+
},
|
| 27560 |
+
{
|
| 27561 |
+
"epoch": 0.8737777777777778,
|
| 27562 |
+
"grad_norm": 0.06626418977975845,
|
| 27563 |
+
"learning_rate": 2.5300668151447664e-05,
|
| 27564 |
+
"loss": 0.0176,
|
| 27565 |
+
"step": 3932
|
| 27566 |
+
},
|
| 27567 |
+
{
|
| 27568 |
+
"epoch": 0.874,
|
| 27569 |
+
"grad_norm": 0.06570940464735031,
|
| 27570 |
+
"learning_rate": 2.5256124721603562e-05,
|
| 27571 |
+
"loss": 0.0179,
|
| 27572 |
+
"step": 3933
|
| 27573 |
+
},
|
| 27574 |
+
{
|
| 27575 |
+
"epoch": 0.8742222222222222,
|
| 27576 |
+
"grad_norm": 0.06534791737794876,
|
| 27577 |
+
"learning_rate": 2.5211581291759468e-05,
|
| 27578 |
+
"loss": 0.0175,
|
| 27579 |
+
"step": 3934
|
| 27580 |
+
},
|
| 27581 |
+
{
|
| 27582 |
+
"epoch": 0.8744444444444445,
|
| 27583 |
+
"grad_norm": 0.07688681036233902,
|
| 27584 |
+
"learning_rate": 2.516703786191537e-05,
|
| 27585 |
+
"loss": 0.0177,
|
| 27586 |
+
"step": 3935
|
| 27587 |
+
},
|
| 27588 |
+
{
|
| 27589 |
+
"epoch": 0.8746666666666667,
|
| 27590 |
+
"grad_norm": 0.060970455408096313,
|
| 27591 |
+
"learning_rate": 2.5122494432071268e-05,
|
| 27592 |
+
"loss": 0.0177,
|
| 27593 |
+
"step": 3936
|
| 27594 |
+
},
|
| 27595 |
+
{
|
| 27596 |
+
"epoch": 0.8748888888888889,
|
| 27597 |
+
"grad_norm": 1.0032833814620972,
|
| 27598 |
+
"learning_rate": 2.5077951002227174e-05,
|
| 27599 |
+
"loss": 1.7855,
|
| 27600 |
+
"step": 3937
|
| 27601 |
+
},
|
| 27602 |
+
{
|
| 27603 |
+
"epoch": 0.8751111111111111,
|
| 27604 |
+
"grad_norm": 0.9916431903839111,
|
| 27605 |
+
"learning_rate": 2.503340757238308e-05,
|
| 27606 |
+
"loss": 1.7102,
|
| 27607 |
+
"step": 3938
|
| 27608 |
+
},
|
| 27609 |
+
{
|
| 27610 |
+
"epoch": 0.8753333333333333,
|
| 27611 |
+
"grad_norm": 0.6338675618171692,
|
| 27612 |
+
"learning_rate": 2.4988864142538974e-05,
|
| 27613 |
+
"loss": 0.744,
|
| 27614 |
+
"step": 3939
|
| 27615 |
+
},
|
| 27616 |
+
{
|
| 27617 |
+
"epoch": 0.8755555555555555,
|
| 27618 |
+
"grad_norm": 0.732306957244873,
|
| 27619 |
+
"learning_rate": 2.494432071269488e-05,
|
| 27620 |
+
"loss": 0.7834,
|
| 27621 |
+
"step": 3940
|
| 27622 |
+
},
|
| 27623 |
+
{
|
| 27624 |
+
"epoch": 0.8757777777777778,
|
| 27625 |
+
"grad_norm": 0.9343276619911194,
|
| 27626 |
+
"learning_rate": 2.489977728285078e-05,
|
| 27627 |
+
"loss": 1.6322,
|
| 27628 |
+
"step": 3941
|
| 27629 |
+
},
|
| 27630 |
+
{
|
| 27631 |
+
"epoch": 0.876,
|
| 27632 |
+
"grad_norm": 1.0164755582809448,
|
| 27633 |
+
"learning_rate": 2.4855233853006683e-05,
|
| 27634 |
+
"loss": 1.561,
|
| 27635 |
+
"step": 3942
|
| 27636 |
+
},
|
| 27637 |
+
{
|
| 27638 |
+
"epoch": 0.8762222222222222,
|
| 27639 |
+
"grad_norm": 0.968427300453186,
|
| 27640 |
+
"learning_rate": 2.4810690423162585e-05,
|
| 27641 |
+
"loss": 1.2585,
|
| 27642 |
+
"step": 3943
|
| 27643 |
+
},
|
| 27644 |
+
{
|
| 27645 |
+
"epoch": 0.8764444444444445,
|
| 27646 |
+
"grad_norm": 0.10157324373722076,
|
| 27647 |
+
"learning_rate": 2.4766146993318487e-05,
|
| 27648 |
+
"loss": 0.025,
|
| 27649 |
+
"step": 3944
|
| 27650 |
+
},
|
| 27651 |
+
{
|
| 27652 |
+
"epoch": 0.8766666666666667,
|
| 27653 |
+
"grad_norm": 1.4210426807403564,
|
| 27654 |
+
"learning_rate": 2.472160356347439e-05,
|
| 27655 |
+
"loss": 1.3668,
|
| 27656 |
+
"step": 3945
|
| 27657 |
+
},
|
| 27658 |
+
{
|
| 27659 |
+
"epoch": 0.8768888888888889,
|
| 27660 |
+
"grad_norm": 1.2264833450317383,
|
| 27661 |
+
"learning_rate": 2.467706013363029e-05,
|
| 27662 |
+
"loss": 1.2792,
|
| 27663 |
+
"step": 3946
|
| 27664 |
+
},
|
| 27665 |
+
{
|
| 27666 |
+
"epoch": 0.8771111111111111,
|
| 27667 |
+
"grad_norm": 1.10524582862854,
|
| 27668 |
+
"learning_rate": 2.4632516703786193e-05,
|
| 27669 |
+
"loss": 1.3331,
|
| 27670 |
+
"step": 3947
|
| 27671 |
+
},
|
| 27672 |
+
{
|
| 27673 |
+
"epoch": 0.8773333333333333,
|
| 27674 |
+
"grad_norm": 0.8390571475028992,
|
| 27675 |
+
"learning_rate": 2.4587973273942095e-05,
|
| 27676 |
+
"loss": 0.6529,
|
| 27677 |
+
"step": 3948
|
| 27678 |
+
},
|
| 27679 |
+
{
|
| 27680 |
+
"epoch": 0.8775555555555555,
|
| 27681 |
+
"grad_norm": 1.1716080904006958,
|
| 27682 |
+
"learning_rate": 2.4543429844097994e-05,
|
| 27683 |
+
"loss": 1.2058,
|
| 27684 |
+
"step": 3949
|
| 27685 |
+
},
|
| 27686 |
+
{
|
| 27687 |
+
"epoch": 0.8777777777777778,
|
| 27688 |
+
"grad_norm": 1.0907095670700073,
|
| 27689 |
+
"learning_rate": 2.44988864142539e-05,
|
| 27690 |
+
"loss": 0.9754,
|
| 27691 |
+
"step": 3950
|
| 27692 |
+
},
|
| 27693 |
+
{
|
| 27694 |
+
"epoch": 0.878,
|
| 27695 |
+
"grad_norm": 0.5868102312088013,
|
| 27696 |
+
"learning_rate": 2.44543429844098e-05,
|
| 27697 |
+
"loss": 1.1827,
|
| 27698 |
+
"step": 3951
|
| 27699 |
+
},
|
| 27700 |
+
{
|
| 27701 |
+
"epoch": 0.8782222222222222,
|
| 27702 |
+
"grad_norm": 0.8409274220466614,
|
| 27703 |
+
"learning_rate": 2.4409799554565703e-05,
|
| 27704 |
+
"loss": 2.1622,
|
| 27705 |
+
"step": 3952
|
| 27706 |
+
},
|
| 27707 |
+
{
|
| 27708 |
+
"epoch": 0.8784444444444445,
|
| 27709 |
+
"grad_norm": 0.6952332258224487,
|
| 27710 |
+
"learning_rate": 2.4365256124721605e-05,
|
| 27711 |
+
"loss": 1.1901,
|
| 27712 |
+
"step": 3953
|
| 27713 |
+
},
|
| 27714 |
+
{
|
| 27715 |
+
"epoch": 0.8786666666666667,
|
| 27716 |
+
"grad_norm": 0.519538164138794,
|
| 27717 |
+
"learning_rate": 2.4320712694877507e-05,
|
| 27718 |
+
"loss": 1.2036,
|
| 27719 |
+
"step": 3954
|
| 27720 |
+
},
|
| 27721 |
+
{
|
| 27722 |
+
"epoch": 0.8788888888888889,
|
| 27723 |
+
"grad_norm": 0.6376737356185913,
|
| 27724 |
+
"learning_rate": 2.427616926503341e-05,
|
| 27725 |
+
"loss": 1.1078,
|
| 27726 |
+
"step": 3955
|
| 27727 |
+
},
|
| 27728 |
+
{
|
| 27729 |
+
"epoch": 0.8791111111111111,
|
| 27730 |
+
"grad_norm": 0.044897519052028656,
|
| 27731 |
+
"learning_rate": 2.423162583518931e-05,
|
| 27732 |
+
"loss": 0.0104,
|
| 27733 |
+
"step": 3956
|
| 27734 |
+
},
|
| 27735 |
+
{
|
| 27736 |
+
"epoch": 0.8793333333333333,
|
| 27737 |
+
"grad_norm": 0.9802849292755127,
|
| 27738 |
+
"learning_rate": 2.4187082405345213e-05,
|
| 27739 |
+
"loss": 1.8767,
|
| 27740 |
+
"step": 3957
|
| 27741 |
+
},
|
| 27742 |
+
{
|
| 27743 |
+
"epoch": 0.8795555555555555,
|
| 27744 |
+
"grad_norm": 0.08882291615009308,
|
| 27745 |
+
"learning_rate": 2.4142538975501115e-05,
|
| 27746 |
+
"loss": 0.0114,
|
| 27747 |
+
"step": 3958
|
| 27748 |
+
},
|
| 27749 |
+
{
|
| 27750 |
+
"epoch": 0.8797777777777778,
|
| 27751 |
+
"grad_norm": 0.9282602667808533,
|
| 27752 |
+
"learning_rate": 2.4097995545657017e-05,
|
| 27753 |
+
"loss": 1.8114,
|
| 27754 |
+
"step": 3959
|
| 27755 |
+
},
|
| 27756 |
+
{
|
| 27757 |
+
"epoch": 0.88,
|
| 27758 |
+
"grad_norm": 0.9374412894248962,
|
| 27759 |
+
"learning_rate": 2.405345211581292e-05,
|
| 27760 |
+
"loss": 1.9425,
|
| 27761 |
+
"step": 3960
|
| 27762 |
+
},
|
| 27763 |
+
{
|
| 27764 |
+
"epoch": 0.8802222222222222,
|
| 27765 |
+
"grad_norm": 1.0642507076263428,
|
| 27766 |
+
"learning_rate": 2.400890868596882e-05,
|
| 27767 |
+
"loss": 2.585,
|
| 27768 |
+
"step": 3961
|
| 27769 |
+
},
|
| 27770 |
+
{
|
| 27771 |
+
"epoch": 0.8804444444444445,
|
| 27772 |
+
"grad_norm": 0.8070052862167358,
|
| 27773 |
+
"learning_rate": 2.3964365256124723e-05,
|
| 27774 |
+
"loss": 1.908,
|
| 27775 |
+
"step": 3962
|
| 27776 |
+
},
|
| 27777 |
+
{
|
| 27778 |
+
"epoch": 0.8806666666666667,
|
| 27779 |
+
"grad_norm": 0.8729952573776245,
|
| 27780 |
+
"learning_rate": 2.3919821826280625e-05,
|
| 27781 |
+
"loss": 1.9254,
|
| 27782 |
+
"step": 3963
|
| 27783 |
+
},
|
| 27784 |
+
{
|
| 27785 |
+
"epoch": 0.8808888888888889,
|
| 27786 |
+
"grad_norm": 0.12168601900339127,
|
| 27787 |
+
"learning_rate": 2.3875278396436527e-05,
|
| 27788 |
+
"loss": 0.0192,
|
| 27789 |
+
"step": 3964
|
| 27790 |
+
},
|
| 27791 |
+
{
|
| 27792 |
+
"epoch": 0.8811111111111111,
|
| 27793 |
+
"grad_norm": 0.6381791830062866,
|
| 27794 |
+
"learning_rate": 2.383073496659243e-05,
|
| 27795 |
+
"loss": 0.9943,
|
| 27796 |
+
"step": 3965
|
| 27797 |
+
},
|
| 27798 |
+
{
|
| 27799 |
+
"epoch": 0.8813333333333333,
|
| 27800 |
+
"grad_norm": 1.2023353576660156,
|
| 27801 |
+
"learning_rate": 2.378619153674833e-05,
|
| 27802 |
+
"loss": 1.9608,
|
| 27803 |
+
"step": 3966
|
| 27804 |
+
},
|
| 27805 |
+
{
|
| 27806 |
+
"epoch": 0.8815555555555555,
|
| 27807 |
+
"grad_norm": 0.9587229490280151,
|
| 27808 |
+
"learning_rate": 2.3741648106904233e-05,
|
| 27809 |
+
"loss": 1.8919,
|
| 27810 |
+
"step": 3967
|
| 27811 |
+
},
|
| 27812 |
+
{
|
| 27813 |
+
"epoch": 0.8817777777777778,
|
| 27814 |
+
"grad_norm": 1.0025968551635742,
|
| 27815 |
+
"learning_rate": 2.3697104677060135e-05,
|
| 27816 |
+
"loss": 1.953,
|
| 27817 |
+
"step": 3968
|
| 27818 |
+
},
|
| 27819 |
+
{
|
| 27820 |
+
"epoch": 0.882,
|
| 27821 |
+
"grad_norm": 0.9075009226799011,
|
| 27822 |
+
"learning_rate": 2.3652561247216037e-05,
|
| 27823 |
+
"loss": 1.9855,
|
| 27824 |
+
"step": 3969
|
| 27825 |
+
},
|
| 27826 |
+
{
|
| 27827 |
+
"epoch": 0.8822222222222222,
|
| 27828 |
+
"grad_norm": 0.06732242554426193,
|
| 27829 |
+
"learning_rate": 2.360801781737194e-05,
|
| 27830 |
+
"loss": 0.0149,
|
| 27831 |
+
"step": 3970
|
| 27832 |
+
},
|
| 27833 |
+
{
|
| 27834 |
+
"epoch": 0.8824444444444445,
|
| 27835 |
+
"grad_norm": 0.06586241722106934,
|
| 27836 |
+
"learning_rate": 2.356347438752784e-05,
|
| 27837 |
+
"loss": 0.015,
|
| 27838 |
+
"step": 3971
|
| 27839 |
+
},
|
| 27840 |
+
{
|
| 27841 |
+
"epoch": 0.8826666666666667,
|
| 27842 |
+
"grad_norm": 0.06589429080486298,
|
| 27843 |
+
"learning_rate": 2.3518930957683743e-05,
|
| 27844 |
+
"loss": 0.015,
|
| 27845 |
+
"step": 3972
|
| 27846 |
+
},
|
| 27847 |
+
{
|
| 27848 |
+
"epoch": 0.8828888888888888,
|
| 27849 |
+
"grad_norm": 0.7274507284164429,
|
| 27850 |
+
"learning_rate": 2.3474387527839645e-05,
|
| 27851 |
+
"loss": 0.9097,
|
| 27852 |
+
"step": 3973
|
| 27853 |
+
},
|
| 27854 |
+
{
|
| 27855 |
+
"epoch": 0.8831111111111111,
|
| 27856 |
+
"grad_norm": 0.9447082877159119,
|
| 27857 |
+
"learning_rate": 2.3429844097995547e-05,
|
| 27858 |
+
"loss": 1.8058,
|
| 27859 |
+
"step": 3974
|
| 27860 |
+
},
|
| 27861 |
+
{
|
| 27862 |
+
"epoch": 0.8833333333333333,
|
| 27863 |
+
"grad_norm": 0.13066767156124115,
|
| 27864 |
+
"learning_rate": 2.338530066815145e-05,
|
| 27865 |
+
"loss": 0.0211,
|
| 27866 |
+
"step": 3975
|
| 27867 |
+
},
|
| 27868 |
+
{
|
| 27869 |
+
"epoch": 0.8835555555555555,
|
| 27870 |
+
"grad_norm": 0.7804778218269348,
|
| 27871 |
+
"learning_rate": 2.334075723830735e-05,
|
| 27872 |
+
"loss": 1.0144,
|
| 27873 |
+
"step": 3976
|
| 27874 |
+
},
|
| 27875 |
+
{
|
| 27876 |
+
"epoch": 0.8837777777777778,
|
| 27877 |
+
"grad_norm": 1.0640380382537842,
|
| 27878 |
+
"learning_rate": 2.3296213808463252e-05,
|
| 27879 |
+
"loss": 1.7064,
|
| 27880 |
+
"step": 3977
|
| 27881 |
+
},
|
| 27882 |
+
{
|
| 27883 |
+
"epoch": 0.884,
|
| 27884 |
+
"grad_norm": 1.0175601243972778,
|
| 27885 |
+
"learning_rate": 2.3251670378619154e-05,
|
| 27886 |
+
"loss": 1.9517,
|
| 27887 |
+
"step": 3978
|
| 27888 |
+
},
|
| 27889 |
+
{
|
| 27890 |
+
"epoch": 0.8842222222222222,
|
| 27891 |
+
"grad_norm": 1.1040079593658447,
|
| 27892 |
+
"learning_rate": 2.3207126948775056e-05,
|
| 27893 |
+
"loss": 1.8058,
|
| 27894 |
+
"step": 3979
|
| 27895 |
+
},
|
| 27896 |
+
{
|
| 27897 |
+
"epoch": 0.8844444444444445,
|
| 27898 |
+
"grad_norm": 0.728284478187561,
|
| 27899 |
+
"learning_rate": 2.316258351893096e-05,
|
| 27900 |
+
"loss": 0.7271,
|
| 27901 |
+
"step": 3980
|
| 27902 |
+
},
|
| 27903 |
+
{
|
| 27904 |
+
"epoch": 0.8846666666666667,
|
| 27905 |
+
"grad_norm": 0.9347479939460754,
|
| 27906 |
+
"learning_rate": 2.3118040089086864e-05,
|
| 27907 |
+
"loss": 1.4855,
|
| 27908 |
+
"step": 3981
|
| 27909 |
+
},
|
| 27910 |
+
{
|
| 27911 |
+
"epoch": 0.8848888888888888,
|
| 27912 |
+
"grad_norm": 0.07804767787456512,
|
| 27913 |
+
"learning_rate": 2.3073496659242762e-05,
|
| 27914 |
+
"loss": 0.0192,
|
| 27915 |
+
"step": 3982
|
| 27916 |
+
},
|
| 27917 |
+
{
|
| 27918 |
+
"epoch": 0.8851111111111111,
|
| 27919 |
+
"grad_norm": 0.08066914230585098,
|
| 27920 |
+
"learning_rate": 2.3028953229398664e-05,
|
| 27921 |
+
"loss": 0.0188,
|
| 27922 |
+
"step": 3983
|
| 27923 |
+
},
|
| 27924 |
+
{
|
| 27925 |
+
"epoch": 0.8853333333333333,
|
| 27926 |
+
"grad_norm": 0.7905464768409729,
|
| 27927 |
+
"learning_rate": 2.298440979955457e-05,
|
| 27928 |
+
"loss": 0.8492,
|
| 27929 |
+
"step": 3984
|
| 27930 |
+
},
|
| 27931 |
+
{
|
| 27932 |
+
"epoch": 0.8855555555555555,
|
| 27933 |
+
"grad_norm": 1.2028931379318237,
|
| 27934 |
+
"learning_rate": 2.2939866369710468e-05,
|
| 27935 |
+
"loss": 0.9801,
|
| 27936 |
+
"step": 3985
|
| 27937 |
+
},
|
| 27938 |
+
{
|
| 27939 |
+
"epoch": 0.8857777777777778,
|
| 27940 |
+
"grad_norm": 0.10054640471935272,
|
| 27941 |
+
"learning_rate": 2.289532293986637e-05,
|
| 27942 |
+
"loss": 0.0247,
|
| 27943 |
+
"step": 3986
|
| 27944 |
+
},
|
| 27945 |
+
{
|
| 27946 |
+
"epoch": 0.886,
|
| 27947 |
+
"grad_norm": 1.1364169120788574,
|
| 27948 |
+
"learning_rate": 2.2850779510022272e-05,
|
| 27949 |
+
"loss": 1.6362,
|
| 27950 |
+
"step": 3987
|
| 27951 |
+
},
|
| 27952 |
+
{
|
| 27953 |
+
"epoch": 0.8862222222222222,
|
| 27954 |
+
"grad_norm": 1.2560831308364868,
|
| 27955 |
+
"learning_rate": 2.2806236080178174e-05,
|
| 27956 |
+
"loss": 1.8384,
|
| 27957 |
+
"step": 3988
|
| 27958 |
+
},
|
| 27959 |
+
{
|
| 27960 |
+
"epoch": 0.8864444444444445,
|
| 27961 |
+
"grad_norm": 1.1757941246032715,
|
| 27962 |
+
"learning_rate": 2.2761692650334076e-05,
|
| 27963 |
+
"loss": 1.4828,
|
| 27964 |
+
"step": 3989
|
| 27965 |
+
},
|
| 27966 |
+
{
|
| 27967 |
+
"epoch": 0.8866666666666667,
|
| 27968 |
+
"grad_norm": 1.120353102684021,
|
| 27969 |
+
"learning_rate": 2.2717149220489978e-05,
|
| 27970 |
+
"loss": 1.3649,
|
| 27971 |
+
"step": 3990
|
| 27972 |
+
},
|
| 27973 |
+
{
|
| 27974 |
+
"epoch": 0.8868888888888888,
|
| 27975 |
+
"grad_norm": 1.0847200155258179,
|
| 27976 |
+
"learning_rate": 2.267260579064588e-05,
|
| 27977 |
+
"loss": 1.7217,
|
| 27978 |
+
"step": 3991
|
| 27979 |
+
},
|
| 27980 |
+
{
|
| 27981 |
+
"epoch": 0.8871111111111111,
|
| 27982 |
+
"grad_norm": 1.1451468467712402,
|
| 27983 |
+
"learning_rate": 2.2628062360801782e-05,
|
| 27984 |
+
"loss": 1.6961,
|
| 27985 |
+
"step": 3992
|
| 27986 |
+
},
|
| 27987 |
+
{
|
| 27988 |
+
"epoch": 0.8873333333333333,
|
| 27989 |
+
"grad_norm": 1.0738978385925293,
|
| 27990 |
+
"learning_rate": 2.2583518930957684e-05,
|
| 27991 |
+
"loss": 1.4236,
|
| 27992 |
+
"step": 3993
|
| 27993 |
+
},
|
| 27994 |
+
{
|
| 27995 |
+
"epoch": 0.8875555555555555,
|
| 27996 |
+
"grad_norm": 1.3635321855545044,
|
| 27997 |
+
"learning_rate": 2.253897550111359e-05,
|
| 27998 |
+
"loss": 1.4417,
|
| 27999 |
+
"step": 3994
|
| 28000 |
+
},
|
| 28001 |
+
{
|
| 28002 |
+
"epoch": 0.8877777777777778,
|
| 28003 |
+
"grad_norm": 0.19308915734291077,
|
| 28004 |
+
"learning_rate": 2.2494432071269488e-05,
|
| 28005 |
+
"loss": 0.0304,
|
| 28006 |
+
"step": 3995
|
| 28007 |
+
},
|
| 28008 |
+
{
|
| 28009 |
+
"epoch": 0.888,
|
| 28010 |
+
"grad_norm": 1.2087732553482056,
|
| 28011 |
+
"learning_rate": 2.244988864142539e-05,
|
| 28012 |
+
"loss": 1.4158,
|
| 28013 |
+
"step": 3996
|
| 28014 |
+
},
|
| 28015 |
+
{
|
| 28016 |
+
"epoch": 0.8882222222222222,
|
| 28017 |
+
"grad_norm": 1.0328425168991089,
|
| 28018 |
+
"learning_rate": 2.2405345211581295e-05,
|
| 28019 |
+
"loss": 1.014,
|
| 28020 |
+
"step": 3997
|
| 28021 |
+
},
|
| 28022 |
+
{
|
| 28023 |
+
"epoch": 0.8884444444444445,
|
| 28024 |
+
"grad_norm": 0.8065721988677979,
|
| 28025 |
+
"learning_rate": 2.2360801781737194e-05,
|
| 28026 |
+
"loss": 0.6411,
|
| 28027 |
+
"step": 3998
|
| 28028 |
+
},
|
| 28029 |
+
{
|
| 28030 |
+
"epoch": 0.8886666666666667,
|
| 28031 |
+
"grad_norm": 0.6682571172714233,
|
| 28032 |
+
"learning_rate": 2.2316258351893096e-05,
|
| 28033 |
+
"loss": 0.5168,
|
| 28034 |
+
"step": 3999
|
| 28035 |
+
},
|
| 28036 |
+
{
|
| 28037 |
+
"epoch": 0.8888888888888888,
|
| 28038 |
+
"grad_norm": 0.9756750464439392,
|
| 28039 |
+
"learning_rate": 2.2271714922049e-05,
|
| 28040 |
+
"loss": 0.4815,
|
| 28041 |
+
"step": 4000
|
| 28042 |
}
|
| 28043 |
],
|
| 28044 |
"logging_steps": 1,
|
|
|
|
| 28058 |
"attributes": {}
|
| 28059 |
}
|
| 28060 |
},
|
| 28061 |
+
"total_flos": 4.305961824831898e+16,
|
| 28062 |
"train_batch_size": 1,
|
| 28063 |
"trial_name": null,
|
| 28064 |
"trial_params": null
|