Training in progress, step 1312, 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 5327496
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8154f422c806e03cfbaa960ceae5cb57305b7a91cd33712440268468179c6e7d
|
| 3 |
size 5327496
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2857850
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0da9173aa2b1588c0646b562c824b5b7c358046890288caa386b54efdc0b3688
|
| 3 |
size 2857850
|
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:db8ae9be8cf8744709ff35d36b7b7c1157f26ed4a20388d517ead00ad8f57cbe
|
| 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 sha256:7692a22ce4a153f8e1646f4dea36295220df47a3d86a6c0e7c97a15691a7dfa6
|
| 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": 328,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6927,6 +6927,2310 @@
|
|
| 6927 |
"eval_samples_per_second": 106.804,
|
| 6928 |
"eval_steps_per_second": 53.408,
|
| 6929 |
"step": 984
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6930 |
}
|
| 6931 |
],
|
| 6932 |
"logging_steps": 1,
|
|
@@ -6941,12 +9245,12 @@
|
|
| 6941 |
"should_evaluate": false,
|
| 6942 |
"should_log": false,
|
| 6943 |
"should_save": true,
|
| 6944 |
-
"should_training_stop":
|
| 6945 |
},
|
| 6946 |
"attributes": {}
|
| 6947 |
}
|
| 6948 |
},
|
| 6949 |
-
"total_flos":
|
| 6950 |
"train_batch_size": 2,
|
| 6951 |
"trial_name": null,
|
| 6952 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.06272859840787932,
|
| 5 |
"eval_steps": 328,
|
| 6 |
+
"global_step": 1312,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6927 |
"eval_samples_per_second": 106.804,
|
| 6928 |
"eval_steps_per_second": 53.408,
|
| 6929 |
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.047094260237622816,
|
| 6933 |
+
"grad_norm": 7.109315395355225,
|
| 6934 |
+
"learning_rate": 2.954571090182149e-05,
|
| 6935 |
+
"loss": 11.0976,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.047142071669336136,
|
| 6940 |
+
"grad_norm": 5.694741249084473,
|
| 6941 |
+
"learning_rate": 2.9374682133177546e-05,
|
| 6942 |
+
"loss": 12.443,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.04718988310104946,
|
| 6947 |
+
"grad_norm": 7.9805097579956055,
|
| 6948 |
+
"learning_rate": 2.920406455021385e-05,
|
| 6949 |
+
"loss": 11.8897,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.04723769453276278,
|
| 6954 |
+
"grad_norm": 7.192514419555664,
|
| 6955 |
+
"learning_rate": 2.9033859146278197e-05,
|
| 6956 |
+
"loss": 13.6973,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.04728550596447611,
|
| 6961 |
+
"grad_norm": 6.8400959968566895,
|
| 6962 |
+
"learning_rate": 2.88640669123187e-05,
|
| 6963 |
+
"loss": 11.4047,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.04733331739618943,
|
| 6968 |
+
"grad_norm": 7.250961780548096,
|
| 6969 |
+
"learning_rate": 2.869468883687798e-05,
|
| 6970 |
+
"loss": 14.0175,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.047381128827902755,
|
| 6975 |
+
"grad_norm": 6.001102924346924,
|
| 6976 |
+
"learning_rate": 2.852572590608735e-05,
|
| 6977 |
+
"loss": 9.9436,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.047428940259616074,
|
| 6982 |
+
"grad_norm": 7.87460994720459,
|
| 6983 |
+
"learning_rate": 2.835717910366117e-05,
|
| 6984 |
+
"loss": 10.5558,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.047476751691329394,
|
| 6989 |
+
"grad_norm": 6.603124141693115,
|
| 6990 |
+
"learning_rate": 2.8189049410891032e-05,
|
| 6991 |
+
"loss": 10.5087,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.04752456312304272,
|
| 6996 |
+
"grad_norm": 7.176062107086182,
|
| 6997 |
+
"learning_rate": 2.8021337806640135e-05,
|
| 6998 |
+
"loss": 10.9031,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.04757237455475604,
|
| 7003 |
+
"grad_norm": 8.618760108947754,
|
| 7004 |
+
"learning_rate": 2.7854045267337436e-05,
|
| 7005 |
+
"loss": 12.6463,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.04762018598646937,
|
| 7010 |
+
"grad_norm": 5.61967134475708,
|
| 7011 |
+
"learning_rate": 2.768717276697217e-05,
|
| 7012 |
+
"loss": 10.0929,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.04766799741818269,
|
| 7017 |
+
"grad_norm": 5.709712028503418,
|
| 7018 |
+
"learning_rate": 2.7520721277088024e-05,
|
| 7019 |
+
"loss": 11.1073,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.04771580884989601,
|
| 7024 |
+
"grad_norm": 7.481429576873779,
|
| 7025 |
+
"learning_rate": 2.735469176677755e-05,
|
| 7026 |
+
"loss": 12.3471,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.04776362028160933,
|
| 7031 |
+
"grad_norm": 6.881556987762451,
|
| 7032 |
+
"learning_rate": 2.718908520267649e-05,
|
| 7033 |
+
"loss": 10.5751,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.04781143171332265,
|
| 7038 |
+
"grad_norm": 7.028878688812256,
|
| 7039 |
+
"learning_rate": 2.702390254895819e-05,
|
| 7040 |
+
"loss": 10.0113,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.04785924314503598,
|
| 7045 |
+
"grad_norm": 6.605225086212158,
|
| 7046 |
+
"learning_rate": 2.685914476732794e-05,
|
| 7047 |
+
"loss": 12.4407,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.0479070545767493,
|
| 7052 |
+
"grad_norm": 7.520063877105713,
|
| 7053 |
+
"learning_rate": 2.669481281701739e-05,
|
| 7054 |
+
"loss": 10.8321,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.047954866008462625,
|
| 7059 |
+
"grad_norm": 6.562614440917969,
|
| 7060 |
+
"learning_rate": 2.6530907654778958e-05,
|
| 7061 |
+
"loss": 11.7903,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.048002677440175945,
|
| 7066 |
+
"grad_norm": 8.194293022155762,
|
| 7067 |
+
"learning_rate": 2.6367430234880284e-05,
|
| 7068 |
+
"loss": 11.2306,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.04805048887188927,
|
| 7073 |
+
"grad_norm": 6.811942100524902,
|
| 7074 |
+
"learning_rate": 2.6204381509098684e-05,
|
| 7075 |
+
"loss": 10.7551,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.04809830030360259,
|
| 7080 |
+
"grad_norm": 6.931473255157471,
|
| 7081 |
+
"learning_rate": 2.6041762426715566e-05,
|
| 7082 |
+
"loss": 12.1181,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.04814611173531591,
|
| 7087 |
+
"grad_norm": 6.276373386383057,
|
| 7088 |
+
"learning_rate": 2.5879573934510913e-05,
|
| 7089 |
+
"loss": 12.2452,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.04819392316702924,
|
| 7094 |
+
"grad_norm": 7.315033912658691,
|
| 7095 |
+
"learning_rate": 2.57178169767578e-05,
|
| 7096 |
+
"loss": 11.5262,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.04824173459874256,
|
| 7101 |
+
"grad_norm": 7.419189453125,
|
| 7102 |
+
"learning_rate": 2.5556492495216867e-05,
|
| 7103 |
+
"loss": 13.3119,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.048289546030455884,
|
| 7108 |
+
"grad_norm": 6.809630393981934,
|
| 7109 |
+
"learning_rate": 2.5395601429130876e-05,
|
| 7110 |
+
"loss": 13.1221,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.0483373574621692,
|
| 7115 |
+
"grad_norm": 6.1475419998168945,
|
| 7116 |
+
"learning_rate": 2.523514471521913e-05,
|
| 7117 |
+
"loss": 11.0054,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.04838516889388253,
|
| 7122 |
+
"grad_norm": 8.833901405334473,
|
| 7123 |
+
"learning_rate": 2.5075123287672175e-05,
|
| 7124 |
+
"loss": 13.8518,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.04843298032559585,
|
| 7129 |
+
"grad_norm": 6.803923606872559,
|
| 7130 |
+
"learning_rate": 2.4915538078146294e-05,
|
| 7131 |
+
"loss": 9.5813,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.048480791757309176,
|
| 7136 |
+
"grad_norm": 6.928596019744873,
|
| 7137 |
+
"learning_rate": 2.4756390015758037e-05,
|
| 7138 |
+
"loss": 13.0504,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.048528603189022496,
|
| 7143 |
+
"grad_norm": 6.18769645690918,
|
| 7144 |
+
"learning_rate": 2.4597680027078873e-05,
|
| 7145 |
+
"loss": 10.7243,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.048576414620735815,
|
| 7150 |
+
"grad_norm": 7.228739261627197,
|
| 7151 |
+
"learning_rate": 2.443940903612978e-05,
|
| 7152 |
+
"loss": 12.2092,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.04862422605244914,
|
| 7157 |
+
"grad_norm": 6.727472305297852,
|
| 7158 |
+
"learning_rate": 2.4281577964375845e-05,
|
| 7159 |
+
"loss": 11.2015,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.04867203748416246,
|
| 7164 |
+
"grad_norm": 6.819626331329346,
|
| 7165 |
+
"learning_rate": 2.4124187730720917e-05,
|
| 7166 |
+
"loss": 11.4536,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.04871984891587579,
|
| 7171 |
+
"grad_norm": 8.001144409179688,
|
| 7172 |
+
"learning_rate": 2.396723925150225e-05,
|
| 7173 |
+
"loss": 13.981,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.04876766034758911,
|
| 7178 |
+
"grad_norm": 6.684787750244141,
|
| 7179 |
+
"learning_rate": 2.381073344048519e-05,
|
| 7180 |
+
"loss": 10.4641,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.048815471779302434,
|
| 7185 |
+
"grad_norm": 6.8459086418151855,
|
| 7186 |
+
"learning_rate": 2.3654671208857826e-05,
|
| 7187 |
+
"loss": 10.3555,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.048863283211015754,
|
| 7192 |
+
"grad_norm": 6.782165050506592,
|
| 7193 |
+
"learning_rate": 2.3499053465225685e-05,
|
| 7194 |
+
"loss": 10.8759,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.048911094642729074,
|
| 7199 |
+
"grad_norm": 8.837566375732422,
|
| 7200 |
+
"learning_rate": 2.334388111560647e-05,
|
| 7201 |
+
"loss": 12.8346,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.0489589060744424,
|
| 7206 |
+
"grad_norm": 5.899895191192627,
|
| 7207 |
+
"learning_rate": 2.3189155063424782e-05,
|
| 7208 |
+
"loss": 10.5408,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.04900671750615572,
|
| 7213 |
+
"grad_norm": 6.8309760093688965,
|
| 7214 |
+
"learning_rate": 2.3034876209506772e-05,
|
| 7215 |
+
"loss": 11.2973,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.049054528937869046,
|
| 7220 |
+
"grad_norm": 6.9437055587768555,
|
| 7221 |
+
"learning_rate": 2.2881045452075067e-05,
|
| 7222 |
+
"loss": 11.7396,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.049102340369582366,
|
| 7227 |
+
"grad_norm": 7.168252468109131,
|
| 7228 |
+
"learning_rate": 2.2727663686743385e-05,
|
| 7229 |
+
"loss": 10.8321,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.04915015180129569,
|
| 7234 |
+
"grad_norm": 7.503252983093262,
|
| 7235 |
+
"learning_rate": 2.2574731806511397e-05,
|
| 7236 |
+
"loss": 11.1285,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.04919796323300901,
|
| 7241 |
+
"grad_norm": 6.952978134155273,
|
| 7242 |
+
"learning_rate": 2.2422250701759506e-05,
|
| 7243 |
+
"loss": 10.7459,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.04924577466472233,
|
| 7248 |
+
"grad_norm": 6.832451820373535,
|
| 7249 |
+
"learning_rate": 2.2270221260243673e-05,
|
| 7250 |
+
"loss": 10.5416,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.04929358609643566,
|
| 7255 |
+
"grad_norm": 6.886053562164307,
|
| 7256 |
+
"learning_rate": 2.211864436709028e-05,
|
| 7257 |
+
"loss": 12.82,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.04934139752814898,
|
| 7262 |
+
"grad_norm": 6.735972881317139,
|
| 7263 |
+
"learning_rate": 2.1967520904790827e-05,
|
| 7264 |
+
"loss": 12.3402,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.049389208959862305,
|
| 7269 |
+
"grad_norm": 7.045289993286133,
|
| 7270 |
+
"learning_rate": 2.181685175319702e-05,
|
| 7271 |
+
"loss": 9.2852,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.049437020391575624,
|
| 7276 |
+
"grad_norm": 7.346778392791748,
|
| 7277 |
+
"learning_rate": 2.1666637789515498e-05,
|
| 7278 |
+
"loss": 12.6252,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.04948483182328895,
|
| 7283 |
+
"grad_norm": 7.470715522766113,
|
| 7284 |
+
"learning_rate": 2.1516879888302764e-05,
|
| 7285 |
+
"loss": 9.5202,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.04953264325500227,
|
| 7290 |
+
"grad_norm": 7.316771507263184,
|
| 7291 |
+
"learning_rate": 2.1367578921460074e-05,
|
| 7292 |
+
"loss": 10.0187,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.0495804546867156,
|
| 7297 |
+
"grad_norm": 6.675178527832031,
|
| 7298 |
+
"learning_rate": 2.1218735758228403e-05,
|
| 7299 |
+
"loss": 12.6647,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.04962826611842892,
|
| 7304 |
+
"grad_norm": 10.01680850982666,
|
| 7305 |
+
"learning_rate": 2.1070351265183386e-05,
|
| 7306 |
+
"loss": 12.8561,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.049676077550142236,
|
| 7311 |
+
"grad_norm": 6.511804103851318,
|
| 7312 |
+
"learning_rate": 2.092242630623016e-05,
|
| 7313 |
+
"loss": 10.6495,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.04972388898185556,
|
| 7318 |
+
"grad_norm": 6.2225871086120605,
|
| 7319 |
+
"learning_rate": 2.0774961742598488e-05,
|
| 7320 |
+
"loss": 10.7088,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.04977170041356888,
|
| 7325 |
+
"grad_norm": 7.0774970054626465,
|
| 7326 |
+
"learning_rate": 2.062795843283768e-05,
|
| 7327 |
+
"loss": 11.7864,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.04981951184528221,
|
| 7332 |
+
"grad_norm": 6.972522258758545,
|
| 7333 |
+
"learning_rate": 2.0481417232811573e-05,
|
| 7334 |
+
"loss": 10.9509,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.04986732327699553,
|
| 7339 |
+
"grad_norm": 8.037018775939941,
|
| 7340 |
+
"learning_rate": 2.0335338995693564e-05,
|
| 7341 |
+
"loss": 12.8436,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.049915134708708855,
|
| 7346 |
+
"grad_norm": 7.327417373657227,
|
| 7347 |
+
"learning_rate": 2.0189724571961677e-05,
|
| 7348 |
+
"loss": 11.537,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.049962946140422175,
|
| 7353 |
+
"grad_norm": 7.174251079559326,
|
| 7354 |
+
"learning_rate": 2.0044574809393547e-05,
|
| 7355 |
+
"loss": 9.8398,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.050010757572135495,
|
| 7360 |
+
"grad_norm": 6.9801716804504395,
|
| 7361 |
+
"learning_rate": 1.9899890553061562e-05,
|
| 7362 |
+
"loss": 11.7782,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.05005856900384882,
|
| 7367 |
+
"grad_norm": 7.83244514465332,
|
| 7368 |
+
"learning_rate": 1.9755672645327804e-05,
|
| 7369 |
+
"loss": 11.4249,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.05010638043556214,
|
| 7374 |
+
"grad_norm": 6.077456474304199,
|
| 7375 |
+
"learning_rate": 1.961192192583934e-05,
|
| 7376 |
+
"loss": 10.2965,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.05015419186727547,
|
| 7381 |
+
"grad_norm": 6.794162273406982,
|
| 7382 |
+
"learning_rate": 1.94686392315232e-05,
|
| 7383 |
+
"loss": 10.2771,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.05020200329898879,
|
| 7388 |
+
"grad_norm": 9.769984245300293,
|
| 7389 |
+
"learning_rate": 1.9325825396581543e-05,
|
| 7390 |
+
"loss": 14.9416,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.050249814730702114,
|
| 7395 |
+
"grad_norm": 6.598973751068115,
|
| 7396 |
+
"learning_rate": 1.9183481252486767e-05,
|
| 7397 |
+
"loss": 10.7436,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.05029762616241543,
|
| 7402 |
+
"grad_norm": 8.009735107421875,
|
| 7403 |
+
"learning_rate": 1.9041607627976732e-05,
|
| 7404 |
+
"loss": 14.2835,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.05034543759412875,
|
| 7409 |
+
"grad_norm": 6.3866167068481445,
|
| 7410 |
+
"learning_rate": 1.8900205349049904e-05,
|
| 7411 |
+
"loss": 9.5854,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.05039324902584208,
|
| 7416 |
+
"grad_norm": 6.82936954498291,
|
| 7417 |
+
"learning_rate": 1.8759275238960473e-05,
|
| 7418 |
+
"loss": 12.8392,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.0504410604575554,
|
| 7423 |
+
"grad_norm": 8.580596923828125,
|
| 7424 |
+
"learning_rate": 1.8618818118213698e-05,
|
| 7425 |
+
"loss": 10.9765,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.050488871889268726,
|
| 7430 |
+
"grad_norm": 7.819889068603516,
|
| 7431 |
+
"learning_rate": 1.847883480456104e-05,
|
| 7432 |
+
"loss": 13.7809,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.050536683320982045,
|
| 7437 |
+
"grad_norm": 7.687258720397949,
|
| 7438 |
+
"learning_rate": 1.8339326112995425e-05,
|
| 7439 |
+
"loss": 11.0958,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.05058449475269537,
|
| 7444 |
+
"grad_norm": 8.695372581481934,
|
| 7445 |
+
"learning_rate": 1.820029285574647e-05,
|
| 7446 |
+
"loss": 16.2532,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.05063230618440869,
|
| 7451 |
+
"grad_norm": 6.339588165283203,
|
| 7452 |
+
"learning_rate": 1.806173584227583e-05,
|
| 7453 |
+
"loss": 11.2791,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.05068011761612202,
|
| 7458 |
+
"grad_norm": 6.770627975463867,
|
| 7459 |
+
"learning_rate": 1.7923655879272393e-05,
|
| 7460 |
+
"loss": 11.3009,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.05072792904783534,
|
| 7465 |
+
"grad_norm": 6.357767105102539,
|
| 7466 |
+
"learning_rate": 1.7786053770647615e-05,
|
| 7467 |
+
"loss": 11.1885,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.05077574047954866,
|
| 7472 |
+
"grad_norm": 7.122130870819092,
|
| 7473 |
+
"learning_rate": 1.7648930317530865e-05,
|
| 7474 |
+
"loss": 9.7104,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.050823551911261984,
|
| 7479 |
+
"grad_norm": 6.65751314163208,
|
| 7480 |
+
"learning_rate": 1.751228631826478e-05,
|
| 7481 |
+
"loss": 11.9357,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.050871363342975304,
|
| 7486 |
+
"grad_norm": 7.130817413330078,
|
| 7487 |
+
"learning_rate": 1.7376122568400532e-05,
|
| 7488 |
+
"loss": 14.0172,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.05091917477468863,
|
| 7493 |
+
"grad_norm": 7.422727108001709,
|
| 7494 |
+
"learning_rate": 1.724043986069327e-05,
|
| 7495 |
+
"loss": 11.0116,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.05096698620640195,
|
| 7500 |
+
"grad_norm": 7.368555545806885,
|
| 7501 |
+
"learning_rate": 1.7105238985097472e-05,
|
| 7502 |
+
"loss": 12.0141,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.051014797638115277,
|
| 7507 |
+
"grad_norm": 6.818940162658691,
|
| 7508 |
+
"learning_rate": 1.6970520728762375e-05,
|
| 7509 |
+
"loss": 13.2337,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.051062609069828596,
|
| 7514 |
+
"grad_norm": 6.313105583190918,
|
| 7515 |
+
"learning_rate": 1.6836285876027335e-05,
|
| 7516 |
+
"loss": 11.0838,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.051110420501541916,
|
| 7521 |
+
"grad_norm": 7.8725056648254395,
|
| 7522 |
+
"learning_rate": 1.6702535208417346e-05,
|
| 7523 |
+
"loss": 12.5174,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.05115823193325524,
|
| 7528 |
+
"grad_norm": 7.930119514465332,
|
| 7529 |
+
"learning_rate": 1.6569269504638397e-05,
|
| 7530 |
+
"loss": 11.488,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.05120604336496856,
|
| 7535 |
+
"grad_norm": 6.53126335144043,
|
| 7536 |
+
"learning_rate": 1.6436489540572996e-05,
|
| 7537 |
+
"loss": 10.2127,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.05125385479668189,
|
| 7542 |
+
"grad_norm": 5.600967884063721,
|
| 7543 |
+
"learning_rate": 1.6304196089275658e-05,
|
| 7544 |
+
"loss": 12.0328,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.05130166622839521,
|
| 7549 |
+
"grad_norm": 6.714117050170898,
|
| 7550 |
+
"learning_rate": 1.6172389920968357e-05,
|
| 7551 |
+
"loss": 12.2817,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.051349477660108535,
|
| 7556 |
+
"grad_norm": 6.806501388549805,
|
| 7557 |
+
"learning_rate": 1.60410718030361e-05,
|
| 7558 |
+
"loss": 12.2648,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.051397289091821854,
|
| 7563 |
+
"grad_norm": 7.509237766265869,
|
| 7564 |
+
"learning_rate": 1.591024250002243e-05,
|
| 7565 |
+
"loss": 12.5679,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.051445100523535174,
|
| 7570 |
+
"grad_norm": 7.574392795562744,
|
| 7571 |
+
"learning_rate": 1.577990277362491e-05,
|
| 7572 |
+
"loss": 12.0793,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.0514929119552485,
|
| 7577 |
+
"grad_norm": 7.31062126159668,
|
| 7578 |
+
"learning_rate": 1.565005338269082e-05,
|
| 7579 |
+
"loss": 14.0407,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.05154072338696182,
|
| 7584 |
+
"grad_norm": 6.899569988250732,
|
| 7585 |
+
"learning_rate": 1.5520695083212678e-05,
|
| 7586 |
+
"loss": 10.8106,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.05158853481867515,
|
| 7591 |
+
"grad_norm": 7.219051837921143,
|
| 7592 |
+
"learning_rate": 1.539182862832378e-05,
|
| 7593 |
+
"loss": 13.0134,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.05163634625038847,
|
| 7598 |
+
"grad_norm": 7.442750453948975,
|
| 7599 |
+
"learning_rate": 1.5263454768293904e-05,
|
| 7600 |
+
"loss": 11.3706,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.05168415768210179,
|
| 7605 |
+
"grad_norm": 6.3968329429626465,
|
| 7606 |
+
"learning_rate": 1.5135574250524897e-05,
|
| 7607 |
+
"loss": 10.042,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.05173196911381511,
|
| 7612 |
+
"grad_norm": 7.059074401855469,
|
| 7613 |
+
"learning_rate": 1.500818781954635e-05,
|
| 7614 |
+
"loss": 11.7948,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.05177978054552843,
|
| 7619 |
+
"grad_norm": 7.279366493225098,
|
| 7620 |
+
"learning_rate": 1.4881296217011198e-05,
|
| 7621 |
+
"loss": 10.2265,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.05182759197724176,
|
| 7626 |
+
"grad_norm": 7.711565971374512,
|
| 7627 |
+
"learning_rate": 1.4754900181691467e-05,
|
| 7628 |
+
"loss": 12.3845,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.05187540340895508,
|
| 7633 |
+
"grad_norm": 7.382181644439697,
|
| 7634 |
+
"learning_rate": 1.4629000449474007e-05,
|
| 7635 |
+
"loss": 9.8954,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.051923214840668405,
|
| 7640 |
+
"grad_norm": 7.441666126251221,
|
| 7641 |
+
"learning_rate": 1.450359775335608e-05,
|
| 7642 |
+
"loss": 12.4277,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.051971026272381725,
|
| 7647 |
+
"grad_norm": 7.462728977203369,
|
| 7648 |
+
"learning_rate": 1.437869282344121e-05,
|
| 7649 |
+
"loss": 13.3137,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.05201883770409505,
|
| 7654 |
+
"grad_norm": 6.264698505401611,
|
| 7655 |
+
"learning_rate": 1.425428638693489e-05,
|
| 7656 |
+
"loss": 7.785,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.05206664913580837,
|
| 7661 |
+
"grad_norm": 6.829832077026367,
|
| 7662 |
+
"learning_rate": 1.4130379168140362e-05,
|
| 7663 |
+
"loss": 9.5492,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.0521144605675217,
|
| 7668 |
+
"grad_norm": 8.121318817138672,
|
| 7669 |
+
"learning_rate": 1.4006971888454323e-05,
|
| 7670 |
+
"loss": 11.7454,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.05216227199923502,
|
| 7675 |
+
"grad_norm": 5.979797840118408,
|
| 7676 |
+
"learning_rate": 1.3884065266362845e-05,
|
| 7677 |
+
"loss": 9.4902,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.05221008343094834,
|
| 7682 |
+
"grad_norm": 6.876123428344727,
|
| 7683 |
+
"learning_rate": 1.3761660017437128e-05,
|
| 7684 |
+
"loss": 12.9726,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.052257894862661664,
|
| 7689 |
+
"grad_norm": 6.709342956542969,
|
| 7690 |
+
"learning_rate": 1.3639756854329334e-05,
|
| 7691 |
+
"loss": 12.6137,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.05230570629437498,
|
| 7696 |
+
"grad_norm": 6.490479469299316,
|
| 7697 |
+
"learning_rate": 1.3518356486768446e-05,
|
| 7698 |
+
"loss": 11.4913,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.05235351772608831,
|
| 7703 |
+
"grad_norm": 8.935108184814453,
|
| 7704 |
+
"learning_rate": 1.339745962155613e-05,
|
| 7705 |
+
"loss": 13.1655,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.05240132915780163,
|
| 7710 |
+
"grad_norm": 6.405674934387207,
|
| 7711 |
+
"learning_rate": 1.3277066962562645e-05,
|
| 7712 |
+
"loss": 13.5308,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.052449140589514956,
|
| 7717 |
+
"grad_norm": 7.632229328155518,
|
| 7718 |
+
"learning_rate": 1.3157179210722715e-05,
|
| 7719 |
+
"loss": 11.0844,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.052496952021228276,
|
| 7724 |
+
"grad_norm": 7.769608497619629,
|
| 7725 |
+
"learning_rate": 1.3037797064031421e-05,
|
| 7726 |
+
"loss": 11.1969,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.052544763452941595,
|
| 7731 |
+
"grad_norm": 7.831042289733887,
|
| 7732 |
+
"learning_rate": 1.2918921217540226e-05,
|
| 7733 |
+
"loss": 11.9603,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.05259257488465492,
|
| 7738 |
+
"grad_norm": 6.1776838302612305,
|
| 7739 |
+
"learning_rate": 1.280055236335287e-05,
|
| 7740 |
+
"loss": 10.3584,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.05264038631636824,
|
| 7745 |
+
"grad_norm": 8.177087783813477,
|
| 7746 |
+
"learning_rate": 1.2682691190621309e-05,
|
| 7747 |
+
"loss": 14.1836,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.05268819774808157,
|
| 7752 |
+
"grad_norm": 7.492428779602051,
|
| 7753 |
+
"learning_rate": 1.2565338385541792e-05,
|
| 7754 |
+
"loss": 12.9438,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.05273600917979489,
|
| 7759 |
+
"grad_norm": 7.132182598114014,
|
| 7760 |
+
"learning_rate": 1.2448494631350783e-05,
|
| 7761 |
+
"loss": 11.3015,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.052783820611508214,
|
| 7766 |
+
"grad_norm": 8.338322639465332,
|
| 7767 |
+
"learning_rate": 1.233216060832103e-05,
|
| 7768 |
+
"loss": 13.59,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.052831632043221534,
|
| 7773 |
+
"grad_norm": 8.018399238586426,
|
| 7774 |
+
"learning_rate": 1.221633699375756e-05,
|
| 7775 |
+
"loss": 13.9362,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.052879443474934854,
|
| 7780 |
+
"grad_norm": 6.557053089141846,
|
| 7781 |
+
"learning_rate": 1.2101024461993805e-05,
|
| 7782 |
+
"loss": 12.7612,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.05292725490664818,
|
| 7787 |
+
"grad_norm": 6.3985514640808105,
|
| 7788 |
+
"learning_rate": 1.198622368438761e-05,
|
| 7789 |
+
"loss": 12.1431,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.0529750663383615,
|
| 7794 |
+
"grad_norm": 7.745150089263916,
|
| 7795 |
+
"learning_rate": 1.1871935329317363e-05,
|
| 7796 |
+
"loss": 11.0428,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.053022877770074826,
|
| 7801 |
+
"grad_norm": 8.713289260864258,
|
| 7802 |
+
"learning_rate": 1.1758160062178093e-05,
|
| 7803 |
+
"loss": 13.541,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.053070689201788146,
|
| 7808 |
+
"grad_norm": 6.27971076965332,
|
| 7809 |
+
"learning_rate": 1.1644898545377581e-05,
|
| 7810 |
+
"loss": 10.5678,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.05311850063350147,
|
| 7815 |
+
"grad_norm": 7.161699295043945,
|
| 7816 |
+
"learning_rate": 1.153215143833255e-05,
|
| 7817 |
+
"loss": 12.641,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.05316631206521479,
|
| 7822 |
+
"grad_norm": 6.462125778198242,
|
| 7823 |
+
"learning_rate": 1.1419919397464718e-05,
|
| 7824 |
+
"loss": 10.5856,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.05321412349692812,
|
| 7829 |
+
"grad_norm": 6.3692240715026855,
|
| 7830 |
+
"learning_rate": 1.1308203076197122e-05,
|
| 7831 |
+
"loss": 12.8644,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.05326193492864144,
|
| 7836 |
+
"grad_norm": 6.57526159286499,
|
| 7837 |
+
"learning_rate": 1.1197003124950222e-05,
|
| 7838 |
+
"loss": 11.6179,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.05330974636035476,
|
| 7843 |
+
"grad_norm": 7.567281723022461,
|
| 7844 |
+
"learning_rate": 1.1086320191138133e-05,
|
| 7845 |
+
"loss": 11.221,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.053357557792068085,
|
| 7850 |
+
"grad_norm": 5.894258499145508,
|
| 7851 |
+
"learning_rate": 1.097615491916485e-05,
|
| 7852 |
+
"loss": 10.6669,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.053405369223781404,
|
| 7857 |
+
"grad_norm": 7.11670446395874,
|
| 7858 |
+
"learning_rate": 1.0866507950420524e-05,
|
| 7859 |
+
"loss": 11.9137,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.05345318065549473,
|
| 7864 |
+
"grad_norm": 6.696661949157715,
|
| 7865 |
+
"learning_rate": 1.0757379923277667e-05,
|
| 7866 |
+
"loss": 11.6216,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.05350099208720805,
|
| 7871 |
+
"grad_norm": 7.777387619018555,
|
| 7872 |
+
"learning_rate": 1.0648771473087538e-05,
|
| 7873 |
+
"loss": 15.0296,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.05354880351892138,
|
| 7878 |
+
"grad_norm": 7.215236186981201,
|
| 7879 |
+
"learning_rate": 1.0540683232176307e-05,
|
| 7880 |
+
"loss": 11.3117,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.0535966149506347,
|
| 7885 |
+
"grad_norm": 7.303989887237549,
|
| 7886 |
+
"learning_rate": 1.0433115829841522e-05,
|
| 7887 |
+
"loss": 9.7025,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.053644426382348016,
|
| 7892 |
+
"grad_norm": 6.581008434295654,
|
| 7893 |
+
"learning_rate": 1.0326069892348322e-05,
|
| 7894 |
+
"loss": 12.8498,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.05369223781406134,
|
| 7899 |
+
"grad_norm": 5.351386547088623,
|
| 7900 |
+
"learning_rate": 1.0219546042925843e-05,
|
| 7901 |
+
"loss": 9.6386,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.05374004924577466,
|
| 7906 |
+
"grad_norm": 7.279900074005127,
|
| 7907 |
+
"learning_rate": 1.0113544901763606e-05,
|
| 7908 |
+
"loss": 11.4363,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.05378786067748799,
|
| 7913 |
+
"grad_norm": 6.933126449584961,
|
| 7914 |
+
"learning_rate": 1.0008067086007878e-05,
|
| 7915 |
+
"loss": 9.6372,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.05383567210920131,
|
| 7920 |
+
"grad_norm": 6.844083309173584,
|
| 7921 |
+
"learning_rate": 9.903113209758096e-06,
|
| 7922 |
+
"loss": 11.0777,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.053883483540914635,
|
| 7927 |
+
"grad_norm": 6.126784801483154,
|
| 7928 |
+
"learning_rate": 9.798683884063264e-06,
|
| 7929 |
+
"loss": 9.2428,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.053931294972627955,
|
| 7934 |
+
"grad_norm": 6.594939231872559,
|
| 7935 |
+
"learning_rate": 9.694779716918434e-06,
|
| 7936 |
+
"loss": 11.9669,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.053979106404341275,
|
| 7941 |
+
"grad_norm": 6.806014537811279,
|
| 7942 |
+
"learning_rate": 9.59140131326114e-06,
|
| 7943 |
+
"loss": 11.3165,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.0540269178360546,
|
| 7948 |
+
"grad_norm": 6.542304515838623,
|
| 7949 |
+
"learning_rate": 9.488549274967872e-06,
|
| 7950 |
+
"loss": 12.3538,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.05407472926776792,
|
| 7955 |
+
"grad_norm": 7.642573833465576,
|
| 7956 |
+
"learning_rate": 9.386224200850623e-06,
|
| 7957 |
+
"loss": 12.2293,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.05412254069948125,
|
| 7962 |
+
"grad_norm": 7.607949733734131,
|
| 7963 |
+
"learning_rate": 9.284426686653303e-06,
|
| 7964 |
+
"loss": 12.8654,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.05417035213119457,
|
| 7969 |
+
"grad_norm": 6.807147979736328,
|
| 7970 |
+
"learning_rate": 9.183157325048397e-06,
|
| 7971 |
+
"loss": 10.16,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.054218163562907894,
|
| 7976 |
+
"grad_norm": 7.009206771850586,
|
| 7977 |
+
"learning_rate": 9.082416705633379e-06,
|
| 7978 |
+
"loss": 12.3162,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.05426597499462121,
|
| 7983 |
+
"grad_norm": 6.721374034881592,
|
| 7984 |
+
"learning_rate": 8.98220541492738e-06,
|
| 7985 |
+
"loss": 11.8486,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.05431378642633454,
|
| 7990 |
+
"grad_norm": 6.294310569763184,
|
| 7991 |
+
"learning_rate": 8.882524036367757e-06,
|
| 7992 |
+
"loss": 9.6196,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.05436159785804786,
|
| 7997 |
+
"grad_norm": 7.5386962890625,
|
| 7998 |
+
"learning_rate": 8.783373150306661e-06,
|
| 7999 |
+
"loss": 13.269,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.05440940928976118,
|
| 8004 |
+
"grad_norm": 8.121341705322266,
|
| 8005 |
+
"learning_rate": 8.68475333400769e-06,
|
| 8006 |
+
"loss": 16.3141,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.054457220721474506,
|
| 8011 |
+
"grad_norm": 7.0010504722595215,
|
| 8012 |
+
"learning_rate": 8.586665161642494e-06,
|
| 8013 |
+
"loss": 11.5196,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.054505032153187825,
|
| 8018 |
+
"grad_norm": 8.508298873901367,
|
| 8019 |
+
"learning_rate": 8.489109204287472e-06,
|
| 8020 |
+
"loss": 11.4205,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.05455284358490115,
|
| 8025 |
+
"grad_norm": 6.388782024383545,
|
| 8026 |
+
"learning_rate": 8.392086029920443e-06,
|
| 8027 |
+
"loss": 11.5982,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.05460065501661447,
|
| 8032 |
+
"grad_norm": 7.11677885055542,
|
| 8033 |
+
"learning_rate": 8.295596203417278e-06,
|
| 8034 |
+
"loss": 10.2629,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.0546484664483278,
|
| 8039 |
+
"grad_norm": 6.895127296447754,
|
| 8040 |
+
"learning_rate": 8.199640286548693e-06,
|
| 8041 |
+
"loss": 11.1813,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.05469627788004112,
|
| 8046 |
+
"grad_norm": 6.778639793395996,
|
| 8047 |
+
"learning_rate": 8.10421883797694e-06,
|
| 8048 |
+
"loss": 9.8393,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.05474408931175444,
|
| 8053 |
+
"grad_norm": 7.860418796539307,
|
| 8054 |
+
"learning_rate": 8.009332413252558e-06,
|
| 8055 |
+
"loss": 11.9863,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.054791900743467764,
|
| 8060 |
+
"grad_norm": 7.845333576202393,
|
| 8061 |
+
"learning_rate": 7.914981564811142e-06,
|
| 8062 |
+
"loss": 12.8243,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.054839712175181084,
|
| 8067 |
+
"grad_norm": 6.43133020401001,
|
| 8068 |
+
"learning_rate": 7.821166841970107e-06,
|
| 8069 |
+
"loss": 10.4886,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.05488752360689441,
|
| 8074 |
+
"grad_norm": 6.970253944396973,
|
| 8075 |
+
"learning_rate": 7.727888790925542e-06,
|
| 8076 |
+
"loss": 11.5634,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.05493533503860773,
|
| 8081 |
+
"grad_norm": 7.549833297729492,
|
| 8082 |
+
"learning_rate": 7.635147954748934e-06,
|
| 8083 |
+
"loss": 10.6819,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.054983146470321057,
|
| 8088 |
+
"grad_norm": 9.84428596496582,
|
| 8089 |
+
"learning_rate": 7.542944873384106e-06,
|
| 8090 |
+
"loss": 13.0569,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.055030957902034376,
|
| 8095 |
+
"grad_norm": 6.6666460037231445,
|
| 8096 |
+
"learning_rate": 7.4512800836440525e-06,
|
| 8097 |
+
"loss": 10.906,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.055078769333747696,
|
| 8102 |
+
"grad_norm": 6.872984409332275,
|
| 8103 |
+
"learning_rate": 7.360154119207751e-06,
|
| 8104 |
+
"loss": 9.7305,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.05512658076546102,
|
| 8109 |
+
"grad_norm": 8.127420425415039,
|
| 8110 |
+
"learning_rate": 7.269567510617126e-06,
|
| 8111 |
+
"loss": 13.443,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.05517439219717434,
|
| 8116 |
+
"grad_norm": 6.642383098602295,
|
| 8117 |
+
"learning_rate": 7.179520785273941e-06,
|
| 8118 |
+
"loss": 10.8743,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.05522220362888767,
|
| 8123 |
+
"grad_norm": 6.925289154052734,
|
| 8124 |
+
"learning_rate": 7.0900144674367005e-06,
|
| 8125 |
+
"loss": 11.6619,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.05527001506060099,
|
| 8130 |
+
"grad_norm": 6.449171543121338,
|
| 8131 |
+
"learning_rate": 7.0010490782176145e-06,
|
| 8132 |
+
"loss": 12.4408,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.055317826492314315,
|
| 8137 |
+
"grad_norm": 7.721719741821289,
|
| 8138 |
+
"learning_rate": 6.9126251355795864e-06,
|
| 8139 |
+
"loss": 12.5158,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.055365637924027634,
|
| 8144 |
+
"grad_norm": 7.007317543029785,
|
| 8145 |
+
"learning_rate": 6.824743154333157e-06,
|
| 8146 |
+
"loss": 10.3344,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.05541344935574096,
|
| 8151 |
+
"grad_norm": 9.989958763122559,
|
| 8152 |
+
"learning_rate": 6.737403646133567e-06,
|
| 8153 |
+
"loss": 12.409,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.05546126078745428,
|
| 8158 |
+
"grad_norm": 6.475602149963379,
|
| 8159 |
+
"learning_rate": 6.650607119477692e-06,
|
| 8160 |
+
"loss": 12.2925,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.0555090722191676,
|
| 8165 |
+
"grad_norm": 7.471884250640869,
|
| 8166 |
+
"learning_rate": 6.564354079701163e-06,
|
| 8167 |
+
"loss": 12.3177,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.05555688365088093,
|
| 8172 |
+
"grad_norm": 7.412891387939453,
|
| 8173 |
+
"learning_rate": 6.4786450289753715e-06,
|
| 8174 |
+
"loss": 10.5241,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.05560469508259425,
|
| 8179 |
+
"grad_norm": 7.155189514160156,
|
| 8180 |
+
"learning_rate": 6.393480466304613e-06,
|
| 8181 |
+
"loss": 11.2839,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.05565250651430757,
|
| 8186 |
+
"grad_norm": 6.134313106536865,
|
| 8187 |
+
"learning_rate": 6.30886088752305e-06,
|
| 8188 |
+
"loss": 9.5243,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.05570031794602089,
|
| 8193 |
+
"grad_norm": 6.45501184463501,
|
| 8194 |
+
"learning_rate": 6.22478678529197e-06,
|
| 8195 |
+
"loss": 11.4822,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.05574812937773422,
|
| 8200 |
+
"grad_norm": 6.3346452713012695,
|
| 8201 |
+
"learning_rate": 6.141258649096837e-06,
|
| 8202 |
+
"loss": 9.9668,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.05579594080944754,
|
| 8207 |
+
"grad_norm": 6.544136047363281,
|
| 8208 |
+
"learning_rate": 6.058276965244491e-06,
|
| 8209 |
+
"loss": 13.252,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.05584375224116086,
|
| 8214 |
+
"grad_norm": 8.531256675720215,
|
| 8215 |
+
"learning_rate": 5.975842216860239e-06,
|
| 8216 |
+
"loss": 13.8205,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.055891563672874185,
|
| 8221 |
+
"grad_norm": 7.049127578735352,
|
| 8222 |
+
"learning_rate": 5.893954883885133e-06,
|
| 8223 |
+
"loss": 10.6642,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.055939375104587505,
|
| 8228 |
+
"grad_norm": 6.844118595123291,
|
| 8229 |
+
"learning_rate": 5.812615443073122e-06,
|
| 8230 |
+
"loss": 12.2604,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.05598718653630083,
|
| 8235 |
+
"grad_norm": 6.81496524810791,
|
| 8236 |
+
"learning_rate": 5.731824367988259e-06,
|
| 8237 |
+
"loss": 12.3213,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.05603499796801415,
|
| 8242 |
+
"grad_norm": 8.04981517791748,
|
| 8243 |
+
"learning_rate": 5.651582129001986e-06,
|
| 8244 |
+
"loss": 9.9006,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.05608280939972748,
|
| 8249 |
+
"grad_norm": 6.623665809631348,
|
| 8250 |
+
"learning_rate": 5.571889193290414e-06,
|
| 8251 |
+
"loss": 9.4621,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.0561306208314408,
|
| 8256 |
+
"grad_norm": 6.761471271514893,
|
| 8257 |
+
"learning_rate": 5.492746024831541e-06,
|
| 8258 |
+
"loss": 11.1984,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.05617843226315412,
|
| 8263 |
+
"grad_norm": 5.669954776763916,
|
| 8264 |
+
"learning_rate": 5.414153084402573e-06,
|
| 8265 |
+
"loss": 11.2231,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.056226243694867444,
|
| 8270 |
+
"grad_norm": 7.645977973937988,
|
| 8271 |
+
"learning_rate": 5.336110829577279e-06,
|
| 8272 |
+
"loss": 13.0729,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.05627405512658076,
|
| 8277 |
+
"grad_norm": 7.997268199920654,
|
| 8278 |
+
"learning_rate": 5.258619714723278e-06,
|
| 8279 |
+
"loss": 10.0103,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.05632186655829409,
|
| 8284 |
+
"grad_norm": 7.1278977394104,
|
| 8285 |
+
"learning_rate": 5.181680190999394e-06,
|
| 8286 |
+
"loss": 12.6017,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.05636967799000741,
|
| 8291 |
+
"grad_norm": 6.510688304901123,
|
| 8292 |
+
"learning_rate": 5.105292706353093e-06,
|
| 8293 |
+
"loss": 11.9529,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.056417489421720736,
|
| 8298 |
+
"grad_norm": 7.182712554931641,
|
| 8299 |
+
"learning_rate": 5.029457705517793e-06,
|
| 8300 |
+
"loss": 10.1696,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.056465300853434056,
|
| 8305 |
+
"grad_norm": 6.3885908126831055,
|
| 8306 |
+
"learning_rate": 4.954175630010316e-06,
|
| 8307 |
+
"loss": 9.6611,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.05651311228514738,
|
| 8312 |
+
"grad_norm": 7.285274028778076,
|
| 8313 |
+
"learning_rate": 4.879446918128339e-06,
|
| 8314 |
+
"loss": 11.5422,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.0565609237168607,
|
| 8319 |
+
"grad_norm": 6.955239295959473,
|
| 8320 |
+
"learning_rate": 4.805272004947781e-06,
|
| 8321 |
+
"loss": 9.3422,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.05660873514857402,
|
| 8326 |
+
"grad_norm": 6.375068664550781,
|
| 8327 |
+
"learning_rate": 4.73165132232034e-06,
|
| 8328 |
+
"loss": 9.9667,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.05665654658028735,
|
| 8333 |
+
"grad_norm": 7.190155982971191,
|
| 8334 |
+
"learning_rate": 4.6585852988708985e-06,
|
| 8335 |
+
"loss": 11.2559,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.05670435801200067,
|
| 8340 |
+
"grad_norm": 7.193498611450195,
|
| 8341 |
+
"learning_rate": 4.586074359995119e-06,
|
| 8342 |
+
"loss": 10.39,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.056752169443713994,
|
| 8347 |
+
"grad_norm": 6.857291221618652,
|
| 8348 |
+
"learning_rate": 4.514118927856892e-06,
|
| 8349 |
+
"loss": 12.3133,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.056799980875427314,
|
| 8354 |
+
"grad_norm": 7.048860549926758,
|
| 8355 |
+
"learning_rate": 4.442719421385922e-06,
|
| 8356 |
+
"loss": 12.2941,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.05684779230714064,
|
| 8361 |
+
"grad_norm": 7.470992565155029,
|
| 8362 |
+
"learning_rate": 4.371876256275287e-06,
|
| 8363 |
+
"loss": 13.0183,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.05689560373885396,
|
| 8368 |
+
"grad_norm": 8.032382011413574,
|
| 8369 |
+
"learning_rate": 4.30158984497896e-06,
|
| 8370 |
+
"loss": 13.0273,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.05694341517056728,
|
| 8375 |
+
"grad_norm": 6.280318260192871,
|
| 8376 |
+
"learning_rate": 4.231860596709503e-06,
|
| 8377 |
+
"loss": 11.835,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.056991226602280606,
|
| 8382 |
+
"grad_norm": 6.940912246704102,
|
| 8383 |
+
"learning_rate": 4.162688917435631e-06,
|
| 8384 |
+
"loss": 12.0494,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.057039038033993926,
|
| 8389 |
+
"grad_norm": 7.232605934143066,
|
| 8390 |
+
"learning_rate": 4.094075209879788e-06,
|
| 8391 |
+
"loss": 12.8787,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.05708684946570725,
|
| 8396 |
+
"grad_norm": 7.259566307067871,
|
| 8397 |
+
"learning_rate": 4.026019873515952e-06,
|
| 8398 |
+
"loss": 11.6557,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.05713466089742057,
|
| 8403 |
+
"grad_norm": 7.399826526641846,
|
| 8404 |
+
"learning_rate": 3.958523304567174e-06,
|
| 8405 |
+
"loss": 13.6205,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.0571824723291339,
|
| 8410 |
+
"grad_norm": 7.567840576171875,
|
| 8411 |
+
"learning_rate": 3.891585896003335e-06,
|
| 8412 |
+
"loss": 13.1901,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.05723028376084722,
|
| 8417 |
+
"grad_norm": 6.9567341804504395,
|
| 8418 |
+
"learning_rate": 3.825208037538841e-06,
|
| 8419 |
+
"loss": 12.7935,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.05727809519256054,
|
| 8424 |
+
"grad_norm": 7.210753917694092,
|
| 8425 |
+
"learning_rate": 3.7593901156303566e-06,
|
| 8426 |
+
"loss": 10.5421,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.057325906624273865,
|
| 8431 |
+
"grad_norm": 7.2676825523376465,
|
| 8432 |
+
"learning_rate": 3.69413251347458e-06,
|
| 8433 |
+
"loss": 10.3655,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.057373718055987184,
|
| 8438 |
+
"grad_norm": 6.523344993591309,
|
| 8439 |
+
"learning_rate": 3.6294356110059157e-06,
|
| 8440 |
+
"loss": 11.2613,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.05742152948770051,
|
| 8445 |
+
"grad_norm": 9.02381706237793,
|
| 8446 |
+
"learning_rate": 3.565299784894427e-06,
|
| 8447 |
+
"loss": 11.9912,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.05746934091941383,
|
| 8452 |
+
"grad_norm": 6.536443710327148,
|
| 8453 |
+
"learning_rate": 3.501725408543477e-06,
|
| 8454 |
+
"loss": 11.7479,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.05751715235112716,
|
| 8459 |
+
"grad_norm": 7.091581344604492,
|
| 8460 |
+
"learning_rate": 3.4387128520876775e-06,
|
| 8461 |
+
"loss": 10.0515,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.05756496378284048,
|
| 8466 |
+
"grad_norm": 6.528881072998047,
|
| 8467 |
+
"learning_rate": 3.3762624823906573e-06,
|
| 8468 |
+
"loss": 11.7824,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.057612775214553796,
|
| 8473 |
+
"grad_norm": 6.467398643493652,
|
| 8474 |
+
"learning_rate": 3.314374663042963e-06,
|
| 8475 |
+
"loss": 10.3721,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.05766058664626712,
|
| 8480 |
+
"grad_norm": 10.72854232788086,
|
| 8481 |
+
"learning_rate": 3.253049754359927e-06,
|
| 8482 |
+
"loss": 15.8296,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.05770839807798044,
|
| 8487 |
+
"grad_norm": 6.468982696533203,
|
| 8488 |
+
"learning_rate": 3.1922881133795825e-06,
|
| 8489 |
+
"loss": 10.5848,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.05775620950969377,
|
| 8494 |
+
"grad_norm": 6.516100883483887,
|
| 8495 |
+
"learning_rate": 3.1320900938605626e-06,
|
| 8496 |
+
"loss": 10.7587,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.05780402094140709,
|
| 8501 |
+
"grad_norm": 6.87551212310791,
|
| 8502 |
+
"learning_rate": 3.07245604628007e-06,
|
| 8503 |
+
"loss": 11.3465,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.057851832373120415,
|
| 8508 |
+
"grad_norm": 6.65150260925293,
|
| 8509 |
+
"learning_rate": 3.0133863178318232e-06,
|
| 8510 |
+
"loss": 10.4935,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.057899643804833735,
|
| 8515 |
+
"grad_norm": 6.728663444519043,
|
| 8516 |
+
"learning_rate": 2.9548812524240353e-06,
|
| 8517 |
+
"loss": 10.9992,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.05794745523654706,
|
| 8522 |
+
"grad_norm": 5.60107421875,
|
| 8523 |
+
"learning_rate": 2.8969411906773937e-06,
|
| 8524 |
+
"loss": 9.7708,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.05799526666826038,
|
| 8529 |
+
"grad_norm": 6.653834342956543,
|
| 8530 |
+
"learning_rate": 2.839566469923105e-06,
|
| 8531 |
+
"loss": 12.5889,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.0580430780999737,
|
| 8536 |
+
"grad_norm": 6.426825046539307,
|
| 8537 |
+
"learning_rate": 2.7827574242009437e-06,
|
| 8538 |
+
"loss": 10.1172,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.05809088953168703,
|
| 8543 |
+
"grad_norm": 6.510440349578857,
|
| 8544 |
+
"learning_rate": 2.7265143842572284e-06,
|
| 8545 |
+
"loss": 11.1428,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.05813870096340035,
|
| 8550 |
+
"grad_norm": 6.843871593475342,
|
| 8551 |
+
"learning_rate": 2.6708376775430033e-06,
|
| 8552 |
+
"loss": 11.4051,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.058186512395113674,
|
| 8557 |
+
"grad_norm": 8.29066276550293,
|
| 8558 |
+
"learning_rate": 2.615727628212039e-06,
|
| 8559 |
+
"loss": 12.0261,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.05823432382682699,
|
| 8564 |
+
"grad_norm": 5.986488342285156,
|
| 8565 |
+
"learning_rate": 2.561184557119023e-06,
|
| 8566 |
+
"loss": 12.2707,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.05828213525854032,
|
| 8571 |
+
"grad_norm": 6.171065807342529,
|
| 8572 |
+
"learning_rate": 2.5072087818176382e-06,
|
| 8573 |
+
"loss": 7.9814,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.05832994669025364,
|
| 8578 |
+
"grad_norm": 7.454899311065674,
|
| 8579 |
+
"learning_rate": 2.4538006165587324e-06,
|
| 8580 |
+
"loss": 13.3756,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.05837775812196696,
|
| 8585 |
+
"grad_norm": 7.085451126098633,
|
| 8586 |
+
"learning_rate": 2.4009603722884742e-06,
|
| 8587 |
+
"loss": 13.5912,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.058425569553680286,
|
| 8592 |
+
"grad_norm": 7.177112579345703,
|
| 8593 |
+
"learning_rate": 2.3486883566465777e-06,
|
| 8594 |
+
"loss": 13.3228,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.058473380985393605,
|
| 8599 |
+
"grad_norm": 6.507716178894043,
|
| 8600 |
+
"learning_rate": 2.2969848739644582e-06,
|
| 8601 |
+
"loss": 10.533,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.05852119241710693,
|
| 8606 |
+
"grad_norm": 7.587296009063721,
|
| 8607 |
+
"learning_rate": 2.245850225263524e-06,
|
| 8608 |
+
"loss": 12.835,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.05856900384882025,
|
| 8613 |
+
"grad_norm": 8.082344055175781,
|
| 8614 |
+
"learning_rate": 2.195284708253387e-06,
|
| 8615 |
+
"loss": 10.1423,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.05861681528053358,
|
| 8620 |
+
"grad_norm": 7.659343719482422,
|
| 8621 |
+
"learning_rate": 2.145288617330099e-06,
|
| 8622 |
+
"loss": 12.5473,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.0586646267122469,
|
| 8627 |
+
"grad_norm": 6.5159807205200195,
|
| 8628 |
+
"learning_rate": 2.095862243574531e-06,
|
| 8629 |
+
"loss": 11.7112,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.05871243814396022,
|
| 8634 |
+
"grad_norm": 7.886244773864746,
|
| 8635 |
+
"learning_rate": 2.0470058747505516e-06,
|
| 8636 |
+
"loss": 13.4634,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.058760249575673544,
|
| 8641 |
+
"grad_norm": 8.021929740905762,
|
| 8642 |
+
"learning_rate": 1.998719795303472e-06,
|
| 8643 |
+
"loss": 13.5485,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.058808061007386864,
|
| 8648 |
+
"grad_norm": 7.092513084411621,
|
| 8649 |
+
"learning_rate": 1.951004286358327e-06,
|
| 8650 |
+
"loss": 13.0203,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.05885587243910019,
|
| 8655 |
+
"grad_norm": 6.523121356964111,
|
| 8656 |
+
"learning_rate": 1.9038596257182184e-06,
|
| 8657 |
+
"loss": 10.1733,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.05890368387081351,
|
| 8662 |
+
"grad_norm": 7.202290058135986,
|
| 8663 |
+
"learning_rate": 1.857286087862775e-06,
|
| 8664 |
+
"loss": 11.7106,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.058951495302526837,
|
| 8669 |
+
"grad_norm": 6.51995849609375,
|
| 8670 |
+
"learning_rate": 1.8112839439464736e-06,
|
| 8671 |
+
"loss": 10.0029,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.058999306734240156,
|
| 8676 |
+
"grad_norm": 6.796818733215332,
|
| 8677 |
+
"learning_rate": 1.7658534617971067e-06,
|
| 8678 |
+
"loss": 10.7521,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.05904711816595348,
|
| 8683 |
+
"grad_norm": 6.953072547912598,
|
| 8684 |
+
"learning_rate": 1.7209949059142083e-06,
|
| 8685 |
+
"loss": 11.2363,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.0590949295976668,
|
| 8690 |
+
"grad_norm": 7.390401363372803,
|
| 8691 |
+
"learning_rate": 1.6767085374675085e-06,
|
| 8692 |
+
"loss": 11.4966,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.05914274102938012,
|
| 8697 |
+
"grad_norm": 6.77098274230957,
|
| 8698 |
+
"learning_rate": 1.6329946142954356e-06,
|
| 8699 |
+
"loss": 9.9186,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.05919055246109345,
|
| 8704 |
+
"grad_norm": 7.101603984832764,
|
| 8705 |
+
"learning_rate": 1.5898533909035952e-06,
|
| 8706 |
+
"loss": 11.5049,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.05923836389280677,
|
| 8711 |
+
"grad_norm": 6.689188003540039,
|
| 8712 |
+
"learning_rate": 1.5472851184632931e-06,
|
| 8713 |
+
"loss": 11.6284,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.059286175324520095,
|
| 8718 |
+
"grad_norm": 7.310275554656982,
|
| 8719 |
+
"learning_rate": 1.5052900448100815e-06,
|
| 8720 |
+
"loss": 12.954,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.059333986756233414,
|
| 8725 |
+
"grad_norm": 7.711053848266602,
|
| 8726 |
+
"learning_rate": 1.4638684144422821e-06,
|
| 8727 |
+
"loss": 12.1924,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.05938179818794674,
|
| 8732 |
+
"grad_norm": 7.312699317932129,
|
| 8733 |
+
"learning_rate": 1.4230204685196203e-06,
|
| 8734 |
+
"loss": 13.7446,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.05942960961966006,
|
| 8739 |
+
"grad_norm": 6.659768104553223,
|
| 8740 |
+
"learning_rate": 1.3827464448617711e-06,
|
| 8741 |
+
"loss": 13.6394,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.05947742105137338,
|
| 8746 |
+
"grad_norm": 7.438172817230225,
|
| 8747 |
+
"learning_rate": 1.3430465779469936e-06,
|
| 8748 |
+
"loss": 12.6777,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.05952523248308671,
|
| 8753 |
+
"grad_norm": 6.835502624511719,
|
| 8754 |
+
"learning_rate": 1.3039210989107764e-06,
|
| 8755 |
+
"loss": 10.0316,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.05957304391480003,
|
| 8760 |
+
"grad_norm": 6.754507064819336,
|
| 8761 |
+
"learning_rate": 1.2653702355444608e-06,
|
| 8762 |
+
"loss": 10.1478,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.05962085534651335,
|
| 8767 |
+
"grad_norm": 7.723879337310791,
|
| 8768 |
+
"learning_rate": 1.2273942122939531e-06,
|
| 8769 |
+
"loss": 12.8445,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.05966866677822667,
|
| 8774 |
+
"grad_norm": 6.9790496826171875,
|
| 8775 |
+
"learning_rate": 1.189993250258381e-06,
|
| 8776 |
+
"loss": 13.2241,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.05971647820994,
|
| 8781 |
+
"grad_norm": 6.446763038635254,
|
| 8782 |
+
"learning_rate": 1.1531675671888619e-06,
|
| 8783 |
+
"loss": 11.5262,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.05976428964165332,
|
| 8788 |
+
"grad_norm": 7.207417964935303,
|
| 8789 |
+
"learning_rate": 1.1169173774871478e-06,
|
| 8790 |
+
"loss": 11.1738,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.05981210107336664,
|
| 8795 |
+
"grad_norm": 7.113643169403076,
|
| 8796 |
+
"learning_rate": 1.0812428922044592e-06,
|
| 8797 |
+
"loss": 11.5882,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.059859912505079965,
|
| 8802 |
+
"grad_norm": 7.8461689949035645,
|
| 8803 |
+
"learning_rate": 1.0461443190402099e-06,
|
| 8804 |
+
"loss": 13.6024,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.059907723936793285,
|
| 8809 |
+
"grad_norm": 6.816285133361816,
|
| 8810 |
+
"learning_rate": 1.011621862340817e-06,
|
| 8811 |
+
"loss": 10.8082,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.05995553536850661,
|
| 8816 |
+
"grad_norm": 8.05375862121582,
|
| 8817 |
+
"learning_rate": 9.776757230985145e-07,
|
| 8818 |
+
"loss": 13.7541,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.06000334680021993,
|
| 8823 |
+
"grad_norm": 6.277020454406738,
|
| 8824 |
+
"learning_rate": 9.443060989501651e-07,
|
| 8825 |
+
"loss": 12.6768,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.06005115823193326,
|
| 8830 |
+
"grad_norm": 6.9199628829956055,
|
| 8831 |
+
"learning_rate": 9.11513184176116e-07,
|
| 8832 |
+
"loss": 10.5309,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.06009896966364658,
|
| 8837 |
+
"grad_norm": 7.061563491821289,
|
| 8838 |
+
"learning_rate": 8.792971696990671e-07,
|
| 8839 |
+
"loss": 13.3549,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.060146781095359904,
|
| 8844 |
+
"grad_norm": 6.853957176208496,
|
| 8845 |
+
"learning_rate": 8.476582430830049e-07,
|
| 8846 |
+
"loss": 10.8686,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.060194592527073223,
|
| 8851 |
+
"grad_norm": 6.672121524810791,
|
| 8852 |
+
"learning_rate": 8.165965885319927e-07,
|
| 8853 |
+
"loss": 11.9522,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.06024240395878654,
|
| 8858 |
+
"grad_norm": 6.61238956451416,
|
| 8859 |
+
"learning_rate": 7.861123868892483e-07,
|
| 8860 |
+
"loss": 11.737,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.06029021539049987,
|
| 8865 |
+
"grad_norm": 6.505923748016357,
|
| 8866 |
+
"learning_rate": 7.562058156359686e-07,
|
| 8867 |
+
"loss": 11.4601,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.06033802682221319,
|
| 8872 |
+
"grad_norm": 7.02233362197876,
|
| 8873 |
+
"learning_rate": 7.26877048890362e-07,
|
| 8874 |
+
"loss": 10.8346,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.060385838253926516,
|
| 8879 |
+
"grad_norm": 6.840794563293457,
|
| 8880 |
+
"learning_rate": 6.981262574066394e-07,
|
| 8881 |
+
"loss": 12.5558,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.060433649685639836,
|
| 8886 |
+
"grad_norm": 7.369102954864502,
|
| 8887 |
+
"learning_rate": 6.699536085739588e-07,
|
| 8888 |
+
"loss": 13.166,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.06048146111735316,
|
| 8893 |
+
"grad_norm": 7.65620231628418,
|
| 8894 |
+
"learning_rate": 6.423592664155043e-07,
|
| 8895 |
+
"loss": 12.9833,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.06052927254906648,
|
| 8900 |
+
"grad_norm": 7.955878734588623,
|
| 8901 |
+
"learning_rate": 6.153433915875418e-07,
|
| 8902 |
+
"loss": 14.9006,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.0605770839807798,
|
| 8907 |
+
"grad_norm": 7.511544704437256,
|
| 8908 |
+
"learning_rate": 5.889061413784091e-07,
|
| 8909 |
+
"loss": 12.0573,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.06062489541249313,
|
| 8914 |
+
"grad_norm": 8.05403995513916,
|
| 8915 |
+
"learning_rate": 5.630476697076947e-07,
|
| 8916 |
+
"loss": 12.9865,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.06067270684420645,
|
| 8921 |
+
"grad_norm": 7.220417499542236,
|
| 8922 |
+
"learning_rate": 5.377681271252822e-07,
|
| 8923 |
+
"loss": 11.141,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.060720518275919774,
|
| 8928 |
+
"grad_norm": 6.97089147567749,
|
| 8929 |
+
"learning_rate": 5.130676608104845e-07,
|
| 8930 |
+
"loss": 11.2041,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.060768329707633094,
|
| 8935 |
+
"grad_norm": 7.240792274475098,
|
| 8936 |
+
"learning_rate": 4.889464145711897e-07,
|
| 8937 |
+
"loss": 12.4409,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.06081614113934642,
|
| 8942 |
+
"grad_norm": 7.993712902069092,
|
| 8943 |
+
"learning_rate": 4.6540452884303864e-07,
|
| 8944 |
+
"loss": 11.321,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.06086395257105974,
|
| 8949 |
+
"grad_norm": 6.494480609893799,
|
| 8950 |
+
"learning_rate": 4.4244214068857036e-07,
|
| 8951 |
+
"loss": 10.7806,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.06091176400277306,
|
| 8956 |
+
"grad_norm": 6.773935317993164,
|
| 8957 |
+
"learning_rate": 4.2005938379645614e-07,
|
| 8958 |
+
"loss": 12.2017,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.060959575434486386,
|
| 8963 |
+
"grad_norm": 7.454319953918457,
|
| 8964 |
+
"learning_rate": 3.9825638848073333e-07,
|
| 8965 |
+
"loss": 13.1307,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.061007386866199706,
|
| 8970 |
+
"grad_norm": 7.689677715301514,
|
| 8971 |
+
"learning_rate": 3.7703328167999485e-07,
|
| 8972 |
+
"loss": 11.5839,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.06105519829791303,
|
| 8977 |
+
"grad_norm": 8.754450798034668,
|
| 8978 |
+
"learning_rate": 3.56390186956701e-07,
|
| 8979 |
+
"loss": 13.1473,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.06110300972962635,
|
| 8984 |
+
"grad_norm": 6.511455535888672,
|
| 8985 |
+
"learning_rate": 3.363272244964355e-07,
|
| 8986 |
+
"loss": 10.383,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.06115082116133968,
|
| 8991 |
+
"grad_norm": 8.072748184204102,
|
| 8992 |
+
"learning_rate": 3.16844511107206e-07,
|
| 8993 |
+
"loss": 16.1493,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.061198632593053,
|
| 8998 |
+
"grad_norm": 6.8627166748046875,
|
| 8999 |
+
"learning_rate": 2.979421602187782e-07,
|
| 9000 |
+
"loss": 10.3464,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.061246444024766325,
|
| 9005 |
+
"grad_norm": 6.9087748527526855,
|
| 9006 |
+
"learning_rate": 2.7962028188198706e-07,
|
| 9007 |
+
"loss": 11.6314,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.061294255456479645,
|
| 9012 |
+
"grad_norm": 7.146842956542969,
|
| 9013 |
+
"learning_rate": 2.6187898276813784e-07,
|
| 9014 |
+
"loss": 10.5916,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.061342066888192964,
|
| 9019 |
+
"grad_norm": 8.035828590393066,
|
| 9020 |
+
"learning_rate": 2.447183661683283e-07,
|
| 9021 |
+
"loss": 13.8099,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.06138987831990629,
|
| 9026 |
+
"grad_norm": 6.372766017913818,
|
| 9027 |
+
"learning_rate": 2.2813853199292746e-07,
|
| 9028 |
+
"loss": 11.8733,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.06143768975161961,
|
| 9033 |
+
"grad_norm": 7.708188533782959,
|
| 9034 |
+
"learning_rate": 2.121395767709089e-07,
|
| 9035 |
+
"loss": 10.1432,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.06148550118333294,
|
| 9040 |
+
"grad_norm": 6.371932029724121,
|
| 9041 |
+
"learning_rate": 1.9672159364935162e-07,
|
| 9042 |
+
"loss": 10.7602,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.06153331261504626,
|
| 9047 |
+
"grad_norm": 7.205697059631348,
|
| 9048 |
+
"learning_rate": 1.8188467239285133e-07,
|
| 9049 |
+
"loss": 13.4996,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.06158112404675958,
|
| 9054 |
+
"grad_norm": 6.705113887786865,
|
| 9055 |
+
"learning_rate": 1.6762889938303217e-07,
|
| 9056 |
+
"loss": 11.523,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.0616289354784729,
|
| 9061 |
+
"grad_norm": 6.464982986450195,
|
| 9062 |
+
"learning_rate": 1.539543576180358e-07,
|
| 9063 |
+
"loss": 10.8957,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.06167674691018622,
|
| 9068 |
+
"grad_norm": 6.801351547241211,
|
| 9069 |
+
"learning_rate": 1.408611267120219e-07,
|
| 9070 |
+
"loss": 10.9815,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.06172455834189955,
|
| 9075 |
+
"grad_norm": 8.094093322753906,
|
| 9076 |
+
"learning_rate": 1.2834928289472416e-07,
|
| 9077 |
+
"loss": 12.1063,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.06177236977361287,
|
| 9082 |
+
"grad_norm": 8.686542510986328,
|
| 9083 |
+
"learning_rate": 1.164188990109949e-07,
|
| 9084 |
+
"loss": 12.2343,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.061820181205326195,
|
| 9089 |
+
"grad_norm": 7.394459247589111,
|
| 9090 |
+
"learning_rate": 1.0507004452039448e-07,
|
| 9091 |
+
"loss": 12.3694,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.061867992637039515,
|
| 9096 |
+
"grad_norm": 6.753342628479004,
|
| 9097 |
+
"learning_rate": 9.430278549675819e-08,
|
| 9098 |
+
"loss": 10.8457,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.06191580406875284,
|
| 9103 |
+
"grad_norm": 7.126917362213135,
|
| 9104 |
+
"learning_rate": 8.411718462786322e-08,
|
| 9105 |
+
"loss": 12.1192,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.06196361550046616,
|
| 9110 |
+
"grad_norm": 9.326420783996582,
|
| 9111 |
+
"learning_rate": 7.451330121498457e-08,
|
| 9112 |
+
"loss": 13.2389,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.06201142693217948,
|
| 9117 |
+
"grad_norm": 7.877248287200928,
|
| 9118 |
+
"learning_rate": 6.549119117263969e-08,
|
| 9119 |
+
"loss": 12.5016,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.06205923836389281,
|
| 9124 |
+
"grad_norm": 8.499676704406738,
|
| 9125 |
+
"learning_rate": 5.705090702819993e-08,
|
| 9126 |
+
"loss": 10.6485,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.06210704979560613,
|
| 9131 |
+
"grad_norm": 6.845163345336914,
|
| 9132 |
+
"learning_rate": 4.919249792159075e-08,
|
| 9133 |
+
"loss": 10.7636,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.062154861227319454,
|
| 9138 |
+
"grad_norm": 6.525848865509033,
|
| 9139 |
+
"learning_rate": 4.191600960505859e-08,
|
| 9140 |
+
"loss": 9.5123,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.06220267265903277,
|
| 9145 |
+
"grad_norm": 6.789247989654541,
|
| 9146 |
+
"learning_rate": 3.5221484442837795e-08,
|
| 9147 |
+
"loss": 12.1668,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.0622504840907461,
|
| 9152 |
+
"grad_norm": 7.4512152671813965,
|
| 9153 |
+
"learning_rate": 2.9108961410917458e-08,
|
| 9154 |
+
"loss": 14.4092,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.06229829552245942,
|
| 9159 |
+
"grad_norm": 7.412023544311523,
|
| 9160 |
+
"learning_rate": 2.357847609686381e-08,
|
| 9161 |
+
"loss": 13.492,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.062346106954172746,
|
| 9166 |
+
"grad_norm": 7.283682823181152,
|
| 9167 |
+
"learning_rate": 1.8630060699553752e-08,
|
| 9168 |
+
"loss": 11.885,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.062393918385886066,
|
| 9173 |
+
"grad_norm": 6.971307754516602,
|
| 9174 |
+
"learning_rate": 1.426374402901942e-08,
|
| 9175 |
+
"loss": 11.7913,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.062441729817599385,
|
| 9180 |
+
"grad_norm": 6.387993812561035,
|
| 9181 |
+
"learning_rate": 1.0479551506259456e-08,
|
| 9182 |
+
"loss": 10.0542,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.06248954124931271,
|
| 9187 |
+
"grad_norm": 7.178387641906738,
|
| 9188 |
+
"learning_rate": 7.277505163139076e-09,
|
| 9189 |
+
"loss": 15.0374,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.06253735268102603,
|
| 9194 |
+
"grad_norm": 7.890112400054932,
|
| 9195 |
+
"learning_rate": 4.657623642179143e-09,
|
| 9196 |
+
"loss": 13.1065,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.06258516411273936,
|
| 9201 |
+
"grad_norm": 6.9360527992248535,
|
| 9202 |
+
"learning_rate": 2.6199221965228504e-09,
|
| 9203 |
+
"loss": 13.3427,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.06263297554445268,
|
| 9208 |
+
"grad_norm": 7.331573009490967,
|
| 9209 |
+
"learning_rate": 1.1644126898247009e-09,
|
| 9210 |
+
"loss": 11.5051,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.062680786976166,
|
| 9215 |
+
"grad_norm": 6.696556568145752,
|
| 9216 |
+
"learning_rate": 2.911035961616904e-10,
|
| 9217 |
+
"loss": 13.0456,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.06272859840787932,
|
| 9222 |
+
"grad_norm": 7.385967254638672,
|
| 9223 |
+
"learning_rate": 0.0,
|
| 9224 |
+
"loss": 12.7968,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.06272859840787932,
|
| 9229 |
+
"eval_loss": 2.930758476257324,
|
| 9230 |
+
"eval_runtime": 80.645,
|
| 9231 |
+
"eval_samples_per_second": 109.207,
|
| 9232 |
+
"eval_steps_per_second": 54.61,
|
| 9233 |
+
"step": 1312
|
| 9234 |
}
|
| 9235 |
],
|
| 9236 |
"logging_steps": 1,
|
|
|
|
| 9245 |
"should_evaluate": false,
|
| 9246 |
"should_log": false,
|
| 9247 |
"should_save": true,
|
| 9248 |
+
"should_training_stop": true
|
| 9249 |
},
|
| 9250 |
"attributes": {}
|
| 9251 |
}
|
| 9252 |
},
|
| 9253 |
+
"total_flos": 2784951020814336.0,
|
| 9254 |
"train_batch_size": 2,
|
| 9255 |
"trial_name": null,
|
| 9256 |
"trial_params": null
|