Training in progress, step 12495, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 368988278
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7b8b80d9f328d9b46b771e57b9c11a0174556d172f3cffdd9f8feecd8a89c97
|
| 3 |
size 368988278
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1107079290
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bee36ad6289662fa604c98e4b82098bdb5f4ece0f63f769701b009cbd2467264
|
| 3 |
size 1107079290
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc3fd064071c0ab1e33cfa9ed205d6f6c83d6ee49646339c32b79ed5829a3484
|
| 3 |
size 1000
|
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": 2500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7039,6 +7039,1749 @@
|
|
| 7039 |
"eval_samples_per_second": 1574.473,
|
| 7040 |
"eval_steps_per_second": 49.202,
|
| 7041 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7042 |
}
|
| 7043 |
],
|
| 7044 |
"logging_steps": 10,
|
|
@@ -7053,12 +8796,12 @@
|
|
| 7053 |
"should_evaluate": false,
|
| 7054 |
"should_log": false,
|
| 7055 |
"should_save": true,
|
| 7056 |
-
"should_training_stop":
|
| 7057 |
},
|
| 7058 |
"attributes": {}
|
| 7059 |
}
|
| 7060 |
},
|
| 7061 |
-
"total_flos":
|
| 7062 |
"train_batch_size": 4,
|
| 7063 |
"trial_name": null,
|
| 7064 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9999862446870104,
|
| 5 |
"eval_steps": 2500,
|
| 6 |
+
"global_step": 12495,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7039 |
"eval_samples_per_second": 1574.473,
|
| 7040 |
"eval_steps_per_second": 49.202,
|
| 7041 |
"step": 10000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.8011094285167646,
|
| 7045 |
+
"grad_norm": 77.1875,
|
| 7046 |
+
"learning_rate": 9.984353338811487e-07,
|
| 7047 |
+
"loss": 130.3064,
|
| 7048 |
+
"step": 10010
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.8019097376361619,
|
| 7052 |
+
"grad_norm": 74.0,
|
| 7053 |
+
"learning_rate": 9.98433770778133e-07,
|
| 7054 |
+
"loss": 132.0653,
|
| 7055 |
+
"step": 10020
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.8027100467555593,
|
| 7059 |
+
"grad_norm": 84.875,
|
| 7060 |
+
"learning_rate": 9.984322076751171e-07,
|
| 7061 |
+
"loss": 132.2227,
|
| 7062 |
+
"step": 10030
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.8035103558749567,
|
| 7066 |
+
"grad_norm": 76.625,
|
| 7067 |
+
"learning_rate": 9.984306445721013e-07,
|
| 7068 |
+
"loss": 130.7247,
|
| 7069 |
+
"step": 10040
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.8043106649943541,
|
| 7073 |
+
"grad_norm": 75.8125,
|
| 7074 |
+
"learning_rate": 9.984290814690853e-07,
|
| 7075 |
+
"loss": 131.7206,
|
| 7076 |
+
"step": 10050
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.8051109741137514,
|
| 7080 |
+
"grad_norm": 73.6875,
|
| 7081 |
+
"learning_rate": 9.984275183660696e-07,
|
| 7082 |
+
"loss": 131.013,
|
| 7083 |
+
"step": 10060
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.8059112832331488,
|
| 7087 |
+
"grad_norm": 72.625,
|
| 7088 |
+
"learning_rate": 9.984259552630538e-07,
|
| 7089 |
+
"loss": 132.8077,
|
| 7090 |
+
"step": 10070
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.8067115923525462,
|
| 7094 |
+
"grad_norm": 72.375,
|
| 7095 |
+
"learning_rate": 9.98424392160038e-07,
|
| 7096 |
+
"loss": 132.2377,
|
| 7097 |
+
"step": 10080
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.8075119014719435,
|
| 7101 |
+
"grad_norm": 82.5625,
|
| 7102 |
+
"learning_rate": 9.984228290570222e-07,
|
| 7103 |
+
"loss": 131.6139,
|
| 7104 |
+
"step": 10090
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.808312210591341,
|
| 7108 |
+
"grad_norm": 75.4375,
|
| 7109 |
+
"learning_rate": 9.984212659540062e-07,
|
| 7110 |
+
"loss": 130.3295,
|
| 7111 |
+
"step": 10100
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.8091125197107383,
|
| 7115 |
+
"grad_norm": 75.5,
|
| 7116 |
+
"learning_rate": 9.984197028509904e-07,
|
| 7117 |
+
"loss": 131.7549,
|
| 7118 |
+
"step": 10110
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.8099128288301356,
|
| 7122 |
+
"grad_norm": 77.875,
|
| 7123 |
+
"learning_rate": 9.984181397479747e-07,
|
| 7124 |
+
"loss": 131.4637,
|
| 7125 |
+
"step": 10120
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.810713137949533,
|
| 7129 |
+
"grad_norm": 74.8125,
|
| 7130 |
+
"learning_rate": 9.984165766449589e-07,
|
| 7131 |
+
"loss": 131.4822,
|
| 7132 |
+
"step": 10130
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.8115134470689304,
|
| 7136 |
+
"grad_norm": 78.6875,
|
| 7137 |
+
"learning_rate": 9.984150135419429e-07,
|
| 7138 |
+
"loss": 131.1819,
|
| 7139 |
+
"step": 10140
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.8123137561883278,
|
| 7143 |
+
"grad_norm": 78.0,
|
| 7144 |
+
"learning_rate": 9.98413450438927e-07,
|
| 7145 |
+
"loss": 131.0785,
|
| 7146 |
+
"step": 10150
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.8131140653077251,
|
| 7150 |
+
"grad_norm": 78.9375,
|
| 7151 |
+
"learning_rate": 9.984118873359113e-07,
|
| 7152 |
+
"loss": 130.286,
|
| 7153 |
+
"step": 10160
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.8139143744271224,
|
| 7157 |
+
"grad_norm": 78.0625,
|
| 7158 |
+
"learning_rate": 9.984103242328953e-07,
|
| 7159 |
+
"loss": 131.6022,
|
| 7160 |
+
"step": 10170
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.8147146835465199,
|
| 7164 |
+
"grad_norm": 78.375,
|
| 7165 |
+
"learning_rate": 9.984087611298795e-07,
|
| 7166 |
+
"loss": 131.5716,
|
| 7167 |
+
"step": 10180
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.8155149926659172,
|
| 7171 |
+
"grad_norm": 74.875,
|
| 7172 |
+
"learning_rate": 9.984071980268638e-07,
|
| 7173 |
+
"loss": 131.0095,
|
| 7174 |
+
"step": 10190
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.8163153017853145,
|
| 7178 |
+
"grad_norm": 77.875,
|
| 7179 |
+
"learning_rate": 9.98405634923848e-07,
|
| 7180 |
+
"loss": 131.2664,
|
| 7181 |
+
"step": 10200
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.817115610904712,
|
| 7185 |
+
"grad_norm": 73.5625,
|
| 7186 |
+
"learning_rate": 9.98404071820832e-07,
|
| 7187 |
+
"loss": 131.8767,
|
| 7188 |
+
"step": 10210
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.8179159200241093,
|
| 7192 |
+
"grad_norm": 79.8125,
|
| 7193 |
+
"learning_rate": 9.984025087178162e-07,
|
| 7194 |
+
"loss": 128.864,
|
| 7195 |
+
"step": 10220
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.8187162291435067,
|
| 7199 |
+
"grad_norm": 77.375,
|
| 7200 |
+
"learning_rate": 9.984009456148004e-07,
|
| 7201 |
+
"loss": 130.4429,
|
| 7202 |
+
"step": 10230
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.819516538262904,
|
| 7206 |
+
"grad_norm": 70.8125,
|
| 7207 |
+
"learning_rate": 9.983993825117846e-07,
|
| 7208 |
+
"loss": 128.6023,
|
| 7209 |
+
"step": 10240
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.8203168473823014,
|
| 7213 |
+
"grad_norm": 74.4375,
|
| 7214 |
+
"learning_rate": 9.983978194087689e-07,
|
| 7215 |
+
"loss": 130.9503,
|
| 7216 |
+
"step": 10250
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.8211171565016988,
|
| 7220 |
+
"grad_norm": 80.125,
|
| 7221 |
+
"learning_rate": 9.983962563057529e-07,
|
| 7222 |
+
"loss": 132.8031,
|
| 7223 |
+
"step": 10260
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.8219174656210961,
|
| 7227 |
+
"grad_norm": 73.8125,
|
| 7228 |
+
"learning_rate": 9.98394693202737e-07,
|
| 7229 |
+
"loss": 130.115,
|
| 7230 |
+
"step": 10270
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.8227177747404936,
|
| 7234 |
+
"grad_norm": 73.75,
|
| 7235 |
+
"learning_rate": 9.983931300997213e-07,
|
| 7236 |
+
"loss": 130.5055,
|
| 7237 |
+
"step": 10280
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.8235180838598909,
|
| 7241 |
+
"grad_norm": 74.0,
|
| 7242 |
+
"learning_rate": 9.983915669967053e-07,
|
| 7243 |
+
"loss": 130.6606,
|
| 7244 |
+
"step": 10290
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.8243183929792882,
|
| 7248 |
+
"grad_norm": 78.6875,
|
| 7249 |
+
"learning_rate": 9.983900038936895e-07,
|
| 7250 |
+
"loss": 130.8658,
|
| 7251 |
+
"step": 10300
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.8251187020986857,
|
| 7255 |
+
"grad_norm": 78.0,
|
| 7256 |
+
"learning_rate": 9.983884407906737e-07,
|
| 7257 |
+
"loss": 131.8639,
|
| 7258 |
+
"step": 10310
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.825919011218083,
|
| 7262 |
+
"grad_norm": 74.0,
|
| 7263 |
+
"learning_rate": 9.98386877687658e-07,
|
| 7264 |
+
"loss": 131.2057,
|
| 7265 |
+
"step": 10320
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.8267193203374803,
|
| 7269 |
+
"grad_norm": 78.0625,
|
| 7270 |
+
"learning_rate": 9.98385314584642e-07,
|
| 7271 |
+
"loss": 131.9255,
|
| 7272 |
+
"step": 10330
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.8275196294568777,
|
| 7276 |
+
"grad_norm": 78.75,
|
| 7277 |
+
"learning_rate": 9.983837514816262e-07,
|
| 7278 |
+
"loss": 131.7983,
|
| 7279 |
+
"step": 10340
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.8283199385762751,
|
| 7283 |
+
"grad_norm": 81.9375,
|
| 7284 |
+
"learning_rate": 9.983821883786104e-07,
|
| 7285 |
+
"loss": 132.316,
|
| 7286 |
+
"step": 10350
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.8291202476956725,
|
| 7290 |
+
"grad_norm": 80.875,
|
| 7291 |
+
"learning_rate": 9.983806252755946e-07,
|
| 7292 |
+
"loss": 131.1004,
|
| 7293 |
+
"step": 10360
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.8299205568150698,
|
| 7297 |
+
"grad_norm": 71.0,
|
| 7298 |
+
"learning_rate": 9.983790621725786e-07,
|
| 7299 |
+
"loss": 132.1643,
|
| 7300 |
+
"step": 10370
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.8307208659344671,
|
| 7304 |
+
"grad_norm": 74.375,
|
| 7305 |
+
"learning_rate": 9.983774990695628e-07,
|
| 7306 |
+
"loss": 130.8654,
|
| 7307 |
+
"step": 10380
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.8315211750538646,
|
| 7311 |
+
"grad_norm": 74.875,
|
| 7312 |
+
"learning_rate": 9.98375935966547e-07,
|
| 7313 |
+
"loss": 130.205,
|
| 7314 |
+
"step": 10390
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.8323214841732619,
|
| 7318 |
+
"grad_norm": 73.375,
|
| 7319 |
+
"learning_rate": 9.983743728635313e-07,
|
| 7320 |
+
"loss": 130.6661,
|
| 7321 |
+
"step": 10400
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.8331217932926593,
|
| 7325 |
+
"grad_norm": 73.4375,
|
| 7326 |
+
"learning_rate": 9.983728097605155e-07,
|
| 7327 |
+
"loss": 131.2944,
|
| 7328 |
+
"step": 10410
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.8339221024120567,
|
| 7332 |
+
"grad_norm": 75.875,
|
| 7333 |
+
"learning_rate": 9.983712466574995e-07,
|
| 7334 |
+
"loss": 132.2854,
|
| 7335 |
+
"step": 10420
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.834722411531454,
|
| 7339 |
+
"grad_norm": 80.625,
|
| 7340 |
+
"learning_rate": 9.983696835544837e-07,
|
| 7341 |
+
"loss": 130.7651,
|
| 7342 |
+
"step": 10430
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.8355227206508514,
|
| 7346 |
+
"grad_norm": 71.6875,
|
| 7347 |
+
"learning_rate": 9.98368120451468e-07,
|
| 7348 |
+
"loss": 130.8648,
|
| 7349 |
+
"step": 10440
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.8363230297702487,
|
| 7353 |
+
"grad_norm": 77.125,
|
| 7354 |
+
"learning_rate": 9.98366557348452e-07,
|
| 7355 |
+
"loss": 130.8581,
|
| 7356 |
+
"step": 10450
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.8371233388896461,
|
| 7360 |
+
"grad_norm": 73.625,
|
| 7361 |
+
"learning_rate": 9.983649942454362e-07,
|
| 7362 |
+
"loss": 130.3991,
|
| 7363 |
+
"step": 10460
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.8379236480090435,
|
| 7367 |
+
"grad_norm": 79.375,
|
| 7368 |
+
"learning_rate": 9.983634311424204e-07,
|
| 7369 |
+
"loss": 131.2294,
|
| 7370 |
+
"step": 10470
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.8387239571284408,
|
| 7374 |
+
"grad_norm": 78.3125,
|
| 7375 |
+
"learning_rate": 9.983618680394046e-07,
|
| 7376 |
+
"loss": 129.3754,
|
| 7377 |
+
"step": 10480
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.8395242662478383,
|
| 7381 |
+
"grad_norm": 81.75,
|
| 7382 |
+
"learning_rate": 9.983603049363886e-07,
|
| 7383 |
+
"loss": 131.7653,
|
| 7384 |
+
"step": 10490
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.8403245753672356,
|
| 7388 |
+
"grad_norm": 74.9375,
|
| 7389 |
+
"learning_rate": 9.983587418333728e-07,
|
| 7390 |
+
"loss": 131.5391,
|
| 7391 |
+
"step": 10500
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.8411248844866329,
|
| 7395 |
+
"grad_norm": 76.75,
|
| 7396 |
+
"learning_rate": 9.98357178730357e-07,
|
| 7397 |
+
"loss": 131.4889,
|
| 7398 |
+
"step": 10510
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.8419251936060304,
|
| 7402 |
+
"grad_norm": 72.1875,
|
| 7403 |
+
"learning_rate": 9.98355615627341e-07,
|
| 7404 |
+
"loss": 133.0458,
|
| 7405 |
+
"step": 10520
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.8427255027254277,
|
| 7409 |
+
"grad_norm": 81.3125,
|
| 7410 |
+
"learning_rate": 9.983540525243253e-07,
|
| 7411 |
+
"loss": 130.8461,
|
| 7412 |
+
"step": 10530
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.8435258118448251,
|
| 7416 |
+
"grad_norm": 74.75,
|
| 7417 |
+
"learning_rate": 9.983524894213095e-07,
|
| 7418 |
+
"loss": 131.7515,
|
| 7419 |
+
"step": 10540
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.8443261209642224,
|
| 7423 |
+
"grad_norm": 70.8125,
|
| 7424 |
+
"learning_rate": 9.983509263182937e-07,
|
| 7425 |
+
"loss": 132.1545,
|
| 7426 |
+
"step": 10550
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.8451264300836198,
|
| 7430 |
+
"grad_norm": 76.9375,
|
| 7431 |
+
"learning_rate": 9.98349363215278e-07,
|
| 7432 |
+
"loss": 130.9946,
|
| 7433 |
+
"step": 10560
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.8459267392030172,
|
| 7437 |
+
"grad_norm": 72.25,
|
| 7438 |
+
"learning_rate": 9.983478001122621e-07,
|
| 7439 |
+
"loss": 130.338,
|
| 7440 |
+
"step": 10570
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.8467270483224145,
|
| 7444 |
+
"grad_norm": 75.1875,
|
| 7445 |
+
"learning_rate": 9.983462370092462e-07,
|
| 7446 |
+
"loss": 131.5349,
|
| 7447 |
+
"step": 10580
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.8475273574418118,
|
| 7451 |
+
"grad_norm": 70.125,
|
| 7452 |
+
"learning_rate": 9.983446739062304e-07,
|
| 7453 |
+
"loss": 130.6219,
|
| 7454 |
+
"step": 10590
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.8483276665612093,
|
| 7458 |
+
"grad_norm": 78.0625,
|
| 7459 |
+
"learning_rate": 9.983431108032146e-07,
|
| 7460 |
+
"loss": 131.3493,
|
| 7461 |
+
"step": 10600
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.8491279756806066,
|
| 7465 |
+
"grad_norm": 74.0,
|
| 7466 |
+
"learning_rate": 9.983415477001986e-07,
|
| 7467 |
+
"loss": 131.0083,
|
| 7468 |
+
"step": 10610
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.849928284800004,
|
| 7472 |
+
"grad_norm": 80.75,
|
| 7473 |
+
"learning_rate": 9.983399845971828e-07,
|
| 7474 |
+
"loss": 131.9277,
|
| 7475 |
+
"step": 10620
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.8507285939194014,
|
| 7479 |
+
"grad_norm": 79.1875,
|
| 7480 |
+
"learning_rate": 9.98338421494167e-07,
|
| 7481 |
+
"loss": 130.602,
|
| 7482 |
+
"step": 10630
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.8515289030387987,
|
| 7486 |
+
"grad_norm": 77.5625,
|
| 7487 |
+
"learning_rate": 9.98336858391151e-07,
|
| 7488 |
+
"loss": 130.4636,
|
| 7489 |
+
"step": 10640
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.8523292121581961,
|
| 7493 |
+
"grad_norm": 77.75,
|
| 7494 |
+
"learning_rate": 9.983352952881353e-07,
|
| 7495 |
+
"loss": 131.2724,
|
| 7496 |
+
"step": 10650
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.8531295212775935,
|
| 7500 |
+
"grad_norm": 78.875,
|
| 7501 |
+
"learning_rate": 9.983337321851195e-07,
|
| 7502 |
+
"loss": 130.8466,
|
| 7503 |
+
"step": 10660
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.8539298303969909,
|
| 7507 |
+
"grad_norm": 77.375,
|
| 7508 |
+
"learning_rate": 9.983321690821037e-07,
|
| 7509 |
+
"loss": 130.6045,
|
| 7510 |
+
"step": 10670
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.8547301395163882,
|
| 7514 |
+
"grad_norm": 77.875,
|
| 7515 |
+
"learning_rate": 9.983306059790877e-07,
|
| 7516 |
+
"loss": 131.8468,
|
| 7517 |
+
"step": 10680
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.8555304486357855,
|
| 7521 |
+
"grad_norm": 75.25,
|
| 7522 |
+
"learning_rate": 9.98329042876072e-07,
|
| 7523 |
+
"loss": 130.7249,
|
| 7524 |
+
"step": 10690
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.856330757755183,
|
| 7528 |
+
"grad_norm": 79.25,
|
| 7529 |
+
"learning_rate": 9.983274797730561e-07,
|
| 7530 |
+
"loss": 130.1633,
|
| 7531 |
+
"step": 10700
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.8571310668745803,
|
| 7535 |
+
"grad_norm": 74.625,
|
| 7536 |
+
"learning_rate": 9.983259166700404e-07,
|
| 7537 |
+
"loss": 130.6425,
|
| 7538 |
+
"step": 10710
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.8579313759939776,
|
| 7542 |
+
"grad_norm": 75.25,
|
| 7543 |
+
"learning_rate": 9.983243535670246e-07,
|
| 7544 |
+
"loss": 130.9092,
|
| 7545 |
+
"step": 10720
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.858731685113375,
|
| 7549 |
+
"grad_norm": 77.625,
|
| 7550 |
+
"learning_rate": 9.983227904640088e-07,
|
| 7551 |
+
"loss": 129.9958,
|
| 7552 |
+
"step": 10730
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.8595319942327724,
|
| 7556 |
+
"grad_norm": 79.5,
|
| 7557 |
+
"learning_rate": 9.983212273609928e-07,
|
| 7558 |
+
"loss": 132.0485,
|
| 7559 |
+
"step": 10740
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.8603323033521698,
|
| 7563 |
+
"grad_norm": 77.0625,
|
| 7564 |
+
"learning_rate": 9.98319664257977e-07,
|
| 7565 |
+
"loss": 131.1308,
|
| 7566 |
+
"step": 10750
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.8611326124715671,
|
| 7570 |
+
"grad_norm": 82.4375,
|
| 7571 |
+
"learning_rate": 9.983181011549612e-07,
|
| 7572 |
+
"loss": 131.665,
|
| 7573 |
+
"step": 10760
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.8619329215909645,
|
| 7577 |
+
"grad_norm": 74.375,
|
| 7578 |
+
"learning_rate": 9.983165380519452e-07,
|
| 7579 |
+
"loss": 131.7367,
|
| 7580 |
+
"step": 10770
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.8627332307103619,
|
| 7584 |
+
"grad_norm": 77.9375,
|
| 7585 |
+
"learning_rate": 9.983149749489295e-07,
|
| 7586 |
+
"loss": 130.5749,
|
| 7587 |
+
"step": 10780
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.8635335398297592,
|
| 7591 |
+
"grad_norm": 72.1875,
|
| 7592 |
+
"learning_rate": 9.983134118459137e-07,
|
| 7593 |
+
"loss": 129.4679,
|
| 7594 |
+
"step": 10790
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.8643338489491567,
|
| 7598 |
+
"grad_norm": 79.1875,
|
| 7599 |
+
"learning_rate": 9.983118487428977e-07,
|
| 7600 |
+
"loss": 132.4236,
|
| 7601 |
+
"step": 10800
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.865134158068554,
|
| 7605 |
+
"grad_norm": 72.75,
|
| 7606 |
+
"learning_rate": 9.98310285639882e-07,
|
| 7607 |
+
"loss": 129.7827,
|
| 7608 |
+
"step": 10810
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.8659344671879513,
|
| 7612 |
+
"grad_norm": 73.5625,
|
| 7613 |
+
"learning_rate": 9.983087225368661e-07,
|
| 7614 |
+
"loss": 131.0587,
|
| 7615 |
+
"step": 10820
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.8667347763073487,
|
| 7619 |
+
"grad_norm": 75.375,
|
| 7620 |
+
"learning_rate": 9.983071594338503e-07,
|
| 7621 |
+
"loss": 131.5654,
|
| 7622 |
+
"step": 10830
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.8675350854267461,
|
| 7626 |
+
"grad_norm": 79.875,
|
| 7627 |
+
"learning_rate": 9.983055963308343e-07,
|
| 7628 |
+
"loss": 131.1106,
|
| 7629 |
+
"step": 10840
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.8683353945461435,
|
| 7633 |
+
"grad_norm": 74.0625,
|
| 7634 |
+
"learning_rate": 9.983040332278186e-07,
|
| 7635 |
+
"loss": 131.5627,
|
| 7636 |
+
"step": 10850
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.8691357036655408,
|
| 7640 |
+
"grad_norm": 75.0,
|
| 7641 |
+
"learning_rate": 9.983024701248028e-07,
|
| 7642 |
+
"loss": 130.7058,
|
| 7643 |
+
"step": 10860
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.8699360127849382,
|
| 7647 |
+
"grad_norm": 77.125,
|
| 7648 |
+
"learning_rate": 9.98300907021787e-07,
|
| 7649 |
+
"loss": 131.1618,
|
| 7650 |
+
"step": 10870
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.8707363219043356,
|
| 7654 |
+
"grad_norm": 78.5,
|
| 7655 |
+
"learning_rate": 9.982993439187712e-07,
|
| 7656 |
+
"loss": 131.6918,
|
| 7657 |
+
"step": 10880
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.8715366310237329,
|
| 7661 |
+
"grad_norm": 77.0,
|
| 7662 |
+
"learning_rate": 9.982977808157552e-07,
|
| 7663 |
+
"loss": 130.2754,
|
| 7664 |
+
"step": 10890
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.8723369401431302,
|
| 7668 |
+
"grad_norm": 76.375,
|
| 7669 |
+
"learning_rate": 9.982962177127394e-07,
|
| 7670 |
+
"loss": 130.6358,
|
| 7671 |
+
"step": 10900
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.8731372492625277,
|
| 7675 |
+
"grad_norm": 76.1875,
|
| 7676 |
+
"learning_rate": 9.982946546097237e-07,
|
| 7677 |
+
"loss": 131.4433,
|
| 7678 |
+
"step": 10910
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.873937558381925,
|
| 7682 |
+
"grad_norm": 76.8125,
|
| 7683 |
+
"learning_rate": 9.982930915067079e-07,
|
| 7684 |
+
"loss": 131.1567,
|
| 7685 |
+
"step": 10920
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.8747378675013224,
|
| 7689 |
+
"grad_norm": 75.0625,
|
| 7690 |
+
"learning_rate": 9.982915284036919e-07,
|
| 7691 |
+
"loss": 130.6782,
|
| 7692 |
+
"step": 10930
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.8755381766207198,
|
| 7696 |
+
"grad_norm": 73.4375,
|
| 7697 |
+
"learning_rate": 9.982899653006761e-07,
|
| 7698 |
+
"loss": 130.3121,
|
| 7699 |
+
"step": 10940
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.8763384857401171,
|
| 7703 |
+
"grad_norm": 75.25,
|
| 7704 |
+
"learning_rate": 9.982884021976603e-07,
|
| 7705 |
+
"loss": 131.139,
|
| 7706 |
+
"step": 10950
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.8771387948595145,
|
| 7710 |
+
"grad_norm": 73.4375,
|
| 7711 |
+
"learning_rate": 9.982868390946443e-07,
|
| 7712 |
+
"loss": 130.9144,
|
| 7713 |
+
"step": 10960
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.8779391039789118,
|
| 7717 |
+
"grad_norm": 79.5,
|
| 7718 |
+
"learning_rate": 9.982852759916286e-07,
|
| 7719 |
+
"loss": 129.6913,
|
| 7720 |
+
"step": 10970
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.8787394130983093,
|
| 7724 |
+
"grad_norm": 73.3125,
|
| 7725 |
+
"learning_rate": 9.982837128886128e-07,
|
| 7726 |
+
"loss": 129.9289,
|
| 7727 |
+
"step": 10980
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.8795397222177066,
|
| 7731 |
+
"grad_norm": 73.8125,
|
| 7732 |
+
"learning_rate": 9.98282149785597e-07,
|
| 7733 |
+
"loss": 132.2781,
|
| 7734 |
+
"step": 10990
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.8803400313371039,
|
| 7738 |
+
"grad_norm": 75.9375,
|
| 7739 |
+
"learning_rate": 9.98280586682581e-07,
|
| 7740 |
+
"loss": 130.8845,
|
| 7741 |
+
"step": 11000
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.8811403404565014,
|
| 7745 |
+
"grad_norm": 76.25,
|
| 7746 |
+
"learning_rate": 9.982790235795652e-07,
|
| 7747 |
+
"loss": 129.7611,
|
| 7748 |
+
"step": 11010
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.8819406495758987,
|
| 7752 |
+
"grad_norm": 83.375,
|
| 7753 |
+
"learning_rate": 9.982774604765494e-07,
|
| 7754 |
+
"loss": 129.9167,
|
| 7755 |
+
"step": 11020
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.882740958695296,
|
| 7759 |
+
"grad_norm": 76.3125,
|
| 7760 |
+
"learning_rate": 9.982758973735337e-07,
|
| 7761 |
+
"loss": 130.6375,
|
| 7762 |
+
"step": 11030
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.8835412678146934,
|
| 7766 |
+
"grad_norm": 77.4375,
|
| 7767 |
+
"learning_rate": 9.982743342705179e-07,
|
| 7768 |
+
"loss": 130.627,
|
| 7769 |
+
"step": 11040
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.8843415769340908,
|
| 7773 |
+
"grad_norm": 71.3125,
|
| 7774 |
+
"learning_rate": 9.982727711675019e-07,
|
| 7775 |
+
"loss": 130.6759,
|
| 7776 |
+
"step": 11050
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.8851418860534882,
|
| 7780 |
+
"grad_norm": 77.875,
|
| 7781 |
+
"learning_rate": 9.98271208064486e-07,
|
| 7782 |
+
"loss": 131.0551,
|
| 7783 |
+
"step": 11060
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.8859421951728855,
|
| 7787 |
+
"grad_norm": 73.8125,
|
| 7788 |
+
"learning_rate": 9.982696449614703e-07,
|
| 7789 |
+
"loss": 130.5227,
|
| 7790 |
+
"step": 11070
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.8867425042922829,
|
| 7794 |
+
"grad_norm": 77.25,
|
| 7795 |
+
"learning_rate": 9.982680818584545e-07,
|
| 7796 |
+
"loss": 132.4977,
|
| 7797 |
+
"step": 11080
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.8875428134116803,
|
| 7801 |
+
"grad_norm": 81.375,
|
| 7802 |
+
"learning_rate": 9.982665187554385e-07,
|
| 7803 |
+
"loss": 131.0946,
|
| 7804 |
+
"step": 11090
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.8883431225310776,
|
| 7808 |
+
"grad_norm": 74.375,
|
| 7809 |
+
"learning_rate": 9.982649556524228e-07,
|
| 7810 |
+
"loss": 130.3095,
|
| 7811 |
+
"step": 11100
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.889143431650475,
|
| 7815 |
+
"grad_norm": 76.3125,
|
| 7816 |
+
"learning_rate": 9.98263392549407e-07,
|
| 7817 |
+
"loss": 131.8633,
|
| 7818 |
+
"step": 11110
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.8899437407698724,
|
| 7822 |
+
"grad_norm": 83.125,
|
| 7823 |
+
"learning_rate": 9.98261829446391e-07,
|
| 7824 |
+
"loss": 131.3789,
|
| 7825 |
+
"step": 11120
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.8907440498892697,
|
| 7829 |
+
"grad_norm": 71.375,
|
| 7830 |
+
"learning_rate": 9.982602663433752e-07,
|
| 7831 |
+
"loss": 131.0635,
|
| 7832 |
+
"step": 11130
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.8915443590086671,
|
| 7836 |
+
"grad_norm": 76.0625,
|
| 7837 |
+
"learning_rate": 9.982587032403594e-07,
|
| 7838 |
+
"loss": 131.6485,
|
| 7839 |
+
"step": 11140
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.8923446681280645,
|
| 7843 |
+
"grad_norm": 84.5625,
|
| 7844 |
+
"learning_rate": 9.982571401373434e-07,
|
| 7845 |
+
"loss": 131.5727,
|
| 7846 |
+
"step": 11150
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.8931449772474618,
|
| 7850 |
+
"grad_norm": 76.3125,
|
| 7851 |
+
"learning_rate": 9.982555770343276e-07,
|
| 7852 |
+
"loss": 131.7047,
|
| 7853 |
+
"step": 11160
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.8939452863668592,
|
| 7857 |
+
"grad_norm": 78.25,
|
| 7858 |
+
"learning_rate": 9.982540139313119e-07,
|
| 7859 |
+
"loss": 132.1118,
|
| 7860 |
+
"step": 11170
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.8947455954862565,
|
| 7864 |
+
"grad_norm": 77.5,
|
| 7865 |
+
"learning_rate": 9.98252450828296e-07,
|
| 7866 |
+
"loss": 130.6895,
|
| 7867 |
+
"step": 11180
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.895545904605654,
|
| 7871 |
+
"grad_norm": 74.75,
|
| 7872 |
+
"learning_rate": 9.982508877252803e-07,
|
| 7873 |
+
"loss": 130.8122,
|
| 7874 |
+
"step": 11190
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.8963462137250513,
|
| 7878 |
+
"grad_norm": 75.0,
|
| 7879 |
+
"learning_rate": 9.982493246222645e-07,
|
| 7880 |
+
"loss": 130.9431,
|
| 7881 |
+
"step": 11200
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.8971465228444486,
|
| 7885 |
+
"grad_norm": 75.375,
|
| 7886 |
+
"learning_rate": 9.982477615192485e-07,
|
| 7887 |
+
"loss": 131.6024,
|
| 7888 |
+
"step": 11210
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.8979468319638461,
|
| 7892 |
+
"grad_norm": 75.25,
|
| 7893 |
+
"learning_rate": 9.982461984162327e-07,
|
| 7894 |
+
"loss": 130.6127,
|
| 7895 |
+
"step": 11220
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.8987471410832434,
|
| 7899 |
+
"grad_norm": 84.375,
|
| 7900 |
+
"learning_rate": 9.98244635313217e-07,
|
| 7901 |
+
"loss": 132.7165,
|
| 7902 |
+
"step": 11230
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.8995474502026408,
|
| 7906 |
+
"grad_norm": 76.75,
|
| 7907 |
+
"learning_rate": 9.982430722102012e-07,
|
| 7908 |
+
"loss": 129.617,
|
| 7909 |
+
"step": 11240
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.9003477593220381,
|
| 7913 |
+
"grad_norm": 84.6875,
|
| 7914 |
+
"learning_rate": 9.982415091071852e-07,
|
| 7915 |
+
"loss": 130.9968,
|
| 7916 |
+
"step": 11250
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.9011480684414355,
|
| 7920 |
+
"grad_norm": 74.1875,
|
| 7921 |
+
"learning_rate": 9.982399460041694e-07,
|
| 7922 |
+
"loss": 130.6432,
|
| 7923 |
+
"step": 11260
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.9019483775608329,
|
| 7927 |
+
"grad_norm": 77.9375,
|
| 7928 |
+
"learning_rate": 9.982383829011536e-07,
|
| 7929 |
+
"loss": 131.2904,
|
| 7930 |
+
"step": 11270
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.9027486866802302,
|
| 7934 |
+
"grad_norm": 71.9375,
|
| 7935 |
+
"learning_rate": 9.982368197981376e-07,
|
| 7936 |
+
"loss": 131.8605,
|
| 7937 |
+
"step": 11280
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.9035489957996276,
|
| 7941 |
+
"grad_norm": 76.75,
|
| 7942 |
+
"learning_rate": 9.982352566951218e-07,
|
| 7943 |
+
"loss": 132.6155,
|
| 7944 |
+
"step": 11290
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.904349304919025,
|
| 7948 |
+
"grad_norm": 74.8125,
|
| 7949 |
+
"learning_rate": 9.98233693592106e-07,
|
| 7950 |
+
"loss": 130.2238,
|
| 7951 |
+
"step": 11300
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.9051496140384223,
|
| 7955 |
+
"grad_norm": 78.4375,
|
| 7956 |
+
"learning_rate": 9.9823213048909e-07,
|
| 7957 |
+
"loss": 131.5677,
|
| 7958 |
+
"step": 11310
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.9059499231578197,
|
| 7962 |
+
"grad_norm": 72.0,
|
| 7963 |
+
"learning_rate": 9.982305673860743e-07,
|
| 7964 |
+
"loss": 130.6105,
|
| 7965 |
+
"step": 11320
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.9067502322772171,
|
| 7969 |
+
"grad_norm": 74.375,
|
| 7970 |
+
"learning_rate": 9.982290042830585e-07,
|
| 7971 |
+
"loss": 130.9643,
|
| 7972 |
+
"step": 11330
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.9075505413966144,
|
| 7976 |
+
"grad_norm": 80.125,
|
| 7977 |
+
"learning_rate": 9.982274411800427e-07,
|
| 7978 |
+
"loss": 131.6026,
|
| 7979 |
+
"step": 11340
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.9083508505160118,
|
| 7983 |
+
"grad_norm": 79.1875,
|
| 7984 |
+
"learning_rate": 9.98225878077027e-07,
|
| 7985 |
+
"loss": 131.4425,
|
| 7986 |
+
"step": 11350
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.9091511596354092,
|
| 7990 |
+
"grad_norm": 76.875,
|
| 7991 |
+
"learning_rate": 9.98224314974011e-07,
|
| 7992 |
+
"loss": 131.8976,
|
| 7993 |
+
"step": 11360
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.9099514687548066,
|
| 7997 |
+
"grad_norm": 80.0625,
|
| 7998 |
+
"learning_rate": 9.982227518709952e-07,
|
| 7999 |
+
"loss": 131.6046,
|
| 8000 |
+
"step": 11370
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.9107517778742039,
|
| 8004 |
+
"grad_norm": 80.0625,
|
| 8005 |
+
"learning_rate": 9.982211887679794e-07,
|
| 8006 |
+
"loss": 130.2627,
|
| 8007 |
+
"step": 11380
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.9115520869936012,
|
| 8011 |
+
"grad_norm": 73.6875,
|
| 8012 |
+
"learning_rate": 9.982196256649636e-07,
|
| 8013 |
+
"loss": 131.1418,
|
| 8014 |
+
"step": 11390
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.9123523961129987,
|
| 8018 |
+
"grad_norm": 75.3125,
|
| 8019 |
+
"learning_rate": 9.982180625619476e-07,
|
| 8020 |
+
"loss": 130.4075,
|
| 8021 |
+
"step": 11400
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.913152705232396,
|
| 8025 |
+
"grad_norm": 81.3125,
|
| 8026 |
+
"learning_rate": 9.982164994589318e-07,
|
| 8027 |
+
"loss": 131.0209,
|
| 8028 |
+
"step": 11410
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.9139530143517933,
|
| 8032 |
+
"grad_norm": 76.0625,
|
| 8033 |
+
"learning_rate": 9.98214936355916e-07,
|
| 8034 |
+
"loss": 130.3062,
|
| 8035 |
+
"step": 11420
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.9147533234711908,
|
| 8039 |
+
"grad_norm": 73.0,
|
| 8040 |
+
"learning_rate": 9.982133732529003e-07,
|
| 8041 |
+
"loss": 131.0498,
|
| 8042 |
+
"step": 11430
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.9155536325905881,
|
| 8046 |
+
"grad_norm": 80.6875,
|
| 8047 |
+
"learning_rate": 9.982118101498843e-07,
|
| 8048 |
+
"loss": 130.2114,
|
| 8049 |
+
"step": 11440
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.9163539417099855,
|
| 8053 |
+
"grad_norm": 75.25,
|
| 8054 |
+
"learning_rate": 9.982102470468685e-07,
|
| 8055 |
+
"loss": 130.6461,
|
| 8056 |
+
"step": 11450
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.9171542508293828,
|
| 8060 |
+
"grad_norm": 82.5,
|
| 8061 |
+
"learning_rate": 9.982086839438527e-07,
|
| 8062 |
+
"loss": 131.5808,
|
| 8063 |
+
"step": 11460
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.9179545599487802,
|
| 8067 |
+
"grad_norm": 75.0625,
|
| 8068 |
+
"learning_rate": 9.982071208408367e-07,
|
| 8069 |
+
"loss": 131.6145,
|
| 8070 |
+
"step": 11470
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.9187548690681776,
|
| 8074 |
+
"grad_norm": 77.375,
|
| 8075 |
+
"learning_rate": 9.98205557737821e-07,
|
| 8076 |
+
"loss": 131.2061,
|
| 8077 |
+
"step": 11480
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.9195551781875749,
|
| 8081 |
+
"grad_norm": 82.75,
|
| 8082 |
+
"learning_rate": 9.982039946348052e-07,
|
| 8083 |
+
"loss": 130.0431,
|
| 8084 |
+
"step": 11490
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.9203554873069724,
|
| 8088 |
+
"grad_norm": 73.8125,
|
| 8089 |
+
"learning_rate": 9.982024315317894e-07,
|
| 8090 |
+
"loss": 131.8762,
|
| 8091 |
+
"step": 11500
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.9211557964263697,
|
| 8095 |
+
"grad_norm": 71.25,
|
| 8096 |
+
"learning_rate": 9.982008684287736e-07,
|
| 8097 |
+
"loss": 131.6012,
|
| 8098 |
+
"step": 11510
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.921956105545767,
|
| 8102 |
+
"grad_norm": 77.5,
|
| 8103 |
+
"learning_rate": 9.981993053257576e-07,
|
| 8104 |
+
"loss": 131.3987,
|
| 8105 |
+
"step": 11520
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.9227564146651644,
|
| 8109 |
+
"grad_norm": 79.3125,
|
| 8110 |
+
"learning_rate": 9.981977422227418e-07,
|
| 8111 |
+
"loss": 132.0234,
|
| 8112 |
+
"step": 11530
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.9235567237845618,
|
| 8116 |
+
"grad_norm": 81.0,
|
| 8117 |
+
"learning_rate": 9.98196179119726e-07,
|
| 8118 |
+
"loss": 132.7055,
|
| 8119 |
+
"step": 11540
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.9243570329039592,
|
| 8123 |
+
"grad_norm": 81.5,
|
| 8124 |
+
"learning_rate": 9.981946160167102e-07,
|
| 8125 |
+
"loss": 131.5491,
|
| 8126 |
+
"step": 11550
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.9251573420233565,
|
| 8130 |
+
"grad_norm": 81.3125,
|
| 8131 |
+
"learning_rate": 9.981930529136943e-07,
|
| 8132 |
+
"loss": 129.9155,
|
| 8133 |
+
"step": 11560
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.9259576511427539,
|
| 8137 |
+
"grad_norm": 77.5,
|
| 8138 |
+
"learning_rate": 9.981914898106785e-07,
|
| 8139 |
+
"loss": 131.0387,
|
| 8140 |
+
"step": 11570
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.9267579602621513,
|
| 8144 |
+
"grad_norm": 74.5625,
|
| 8145 |
+
"learning_rate": 9.981899267076627e-07,
|
| 8146 |
+
"loss": 129.5762,
|
| 8147 |
+
"step": 11580
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.9275582693815486,
|
| 8151 |
+
"grad_norm": 72.5,
|
| 8152 |
+
"learning_rate": 9.981883636046467e-07,
|
| 8153 |
+
"loss": 129.8288,
|
| 8154 |
+
"step": 11590
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.9283585785009459,
|
| 8158 |
+
"grad_norm": 74.6875,
|
| 8159 |
+
"learning_rate": 9.98186800501631e-07,
|
| 8160 |
+
"loss": 130.526,
|
| 8161 |
+
"step": 11600
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.9291588876203434,
|
| 8165 |
+
"grad_norm": 79.3125,
|
| 8166 |
+
"learning_rate": 9.981852373986151e-07,
|
| 8167 |
+
"loss": 130.7928,
|
| 8168 |
+
"step": 11610
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.9299591967397407,
|
| 8172 |
+
"grad_norm": 74.5625,
|
| 8173 |
+
"learning_rate": 9.981836742955994e-07,
|
| 8174 |
+
"loss": 130.5095,
|
| 8175 |
+
"step": 11620
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.9307595058591381,
|
| 8179 |
+
"grad_norm": 78.6875,
|
| 8180 |
+
"learning_rate": 9.981821111925834e-07,
|
| 8181 |
+
"loss": 130.1216,
|
| 8182 |
+
"step": 11630
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.9315598149785355,
|
| 8186 |
+
"grad_norm": 73.9375,
|
| 8187 |
+
"learning_rate": 9.981805480895676e-07,
|
| 8188 |
+
"loss": 130.3142,
|
| 8189 |
+
"step": 11640
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.9323601240979328,
|
| 8193 |
+
"grad_norm": 80.25,
|
| 8194 |
+
"learning_rate": 9.981789849865518e-07,
|
| 8195 |
+
"loss": 129.8451,
|
| 8196 |
+
"step": 11650
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.9331604332173302,
|
| 8200 |
+
"grad_norm": 77.75,
|
| 8201 |
+
"learning_rate": 9.98177421883536e-07,
|
| 8202 |
+
"loss": 131.5875,
|
| 8203 |
+
"step": 11660
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.9339607423367275,
|
| 8207 |
+
"grad_norm": 75.9375,
|
| 8208 |
+
"learning_rate": 9.981758587805202e-07,
|
| 8209 |
+
"loss": 131.1091,
|
| 8210 |
+
"step": 11670
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.934761051456125,
|
| 8214 |
+
"grad_norm": 79.4375,
|
| 8215 |
+
"learning_rate": 9.981742956775042e-07,
|
| 8216 |
+
"loss": 129.5825,
|
| 8217 |
+
"step": 11680
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.9355613605755223,
|
| 8221 |
+
"grad_norm": 78.375,
|
| 8222 |
+
"learning_rate": 9.981727325744885e-07,
|
| 8223 |
+
"loss": 131.5865,
|
| 8224 |
+
"step": 11690
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.9363616696949196,
|
| 8228 |
+
"grad_norm": 74.25,
|
| 8229 |
+
"learning_rate": 9.981711694714727e-07,
|
| 8230 |
+
"loss": 131.3969,
|
| 8231 |
+
"step": 11700
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.9371619788143171,
|
| 8235 |
+
"grad_norm": 82.5625,
|
| 8236 |
+
"learning_rate": 9.98169606368457e-07,
|
| 8237 |
+
"loss": 131.5427,
|
| 8238 |
+
"step": 11710
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.9379622879337144,
|
| 8242 |
+
"grad_norm": 81.125,
|
| 8243 |
+
"learning_rate": 9.98168043265441e-07,
|
| 8244 |
+
"loss": 131.2085,
|
| 8245 |
+
"step": 11720
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.9387625970531117,
|
| 8249 |
+
"grad_norm": 76.25,
|
| 8250 |
+
"learning_rate": 9.981664801624251e-07,
|
| 8251 |
+
"loss": 130.8618,
|
| 8252 |
+
"step": 11730
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.9395629061725091,
|
| 8256 |
+
"grad_norm": 75.0625,
|
| 8257 |
+
"learning_rate": 9.981649170594093e-07,
|
| 8258 |
+
"loss": 130.0715,
|
| 8259 |
+
"step": 11740
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.9403632152919065,
|
| 8263 |
+
"grad_norm": 76.4375,
|
| 8264 |
+
"learning_rate": 9.981633539563933e-07,
|
| 8265 |
+
"loss": 131.8718,
|
| 8266 |
+
"step": 11750
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.9411635244113039,
|
| 8270 |
+
"grad_norm": 78.1875,
|
| 8271 |
+
"learning_rate": 9.981617908533776e-07,
|
| 8272 |
+
"loss": 131.3843,
|
| 8273 |
+
"step": 11760
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.9419638335307012,
|
| 8277 |
+
"grad_norm": 75.3125,
|
| 8278 |
+
"learning_rate": 9.981602277503618e-07,
|
| 8279 |
+
"loss": 130.1011,
|
| 8280 |
+
"step": 11770
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.9427641426500986,
|
| 8284 |
+
"grad_norm": 78.0625,
|
| 8285 |
+
"learning_rate": 9.98158664647346e-07,
|
| 8286 |
+
"loss": 131.4609,
|
| 8287 |
+
"step": 11780
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.943564451769496,
|
| 8291 |
+
"grad_norm": 73.75,
|
| 8292 |
+
"learning_rate": 9.9815710154433e-07,
|
| 8293 |
+
"loss": 130.214,
|
| 8294 |
+
"step": 11790
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.9443647608888933,
|
| 8298 |
+
"grad_norm": 81.3125,
|
| 8299 |
+
"learning_rate": 9.981555384413142e-07,
|
| 8300 |
+
"loss": 132.272,
|
| 8301 |
+
"step": 11800
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.9451650700082908,
|
| 8305 |
+
"grad_norm": 76.0625,
|
| 8306 |
+
"learning_rate": 9.981539753382984e-07,
|
| 8307 |
+
"loss": 130.5801,
|
| 8308 |
+
"step": 11810
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.9459653791276881,
|
| 8312 |
+
"grad_norm": 71.3125,
|
| 8313 |
+
"learning_rate": 9.981524122352827e-07,
|
| 8314 |
+
"loss": 130.7302,
|
| 8315 |
+
"step": 11820
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.9467656882470854,
|
| 8319 |
+
"grad_norm": 84.0625,
|
| 8320 |
+
"learning_rate": 9.981508491322667e-07,
|
| 8321 |
+
"loss": 130.839,
|
| 8322 |
+
"step": 11830
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.9475659973664828,
|
| 8326 |
+
"grad_norm": 78.5,
|
| 8327 |
+
"learning_rate": 9.981492860292509e-07,
|
| 8328 |
+
"loss": 130.5891,
|
| 8329 |
+
"step": 11840
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.9483663064858802,
|
| 8333 |
+
"grad_norm": 78.4375,
|
| 8334 |
+
"learning_rate": 9.98147722926235e-07,
|
| 8335 |
+
"loss": 130.0701,
|
| 8336 |
+
"step": 11850
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.9491666156052775,
|
| 8340 |
+
"grad_norm": 76.1875,
|
| 8341 |
+
"learning_rate": 9.981461598232193e-07,
|
| 8342 |
+
"loss": 130.6237,
|
| 8343 |
+
"step": 11860
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.9499669247246749,
|
| 8347 |
+
"grad_norm": 79.625,
|
| 8348 |
+
"learning_rate": 9.981445967202035e-07,
|
| 8349 |
+
"loss": 131.0294,
|
| 8350 |
+
"step": 11870
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.9507672338440722,
|
| 8354 |
+
"grad_norm": 79.3125,
|
| 8355 |
+
"learning_rate": 9.981430336171875e-07,
|
| 8356 |
+
"loss": 132.2408,
|
| 8357 |
+
"step": 11880
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.9515675429634697,
|
| 8361 |
+
"grad_norm": 72.0625,
|
| 8362 |
+
"learning_rate": 9.981414705141718e-07,
|
| 8363 |
+
"loss": 131.3878,
|
| 8364 |
+
"step": 11890
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.952367852082867,
|
| 8368 |
+
"grad_norm": 75.5625,
|
| 8369 |
+
"learning_rate": 9.98139907411156e-07,
|
| 8370 |
+
"loss": 131.9279,
|
| 8371 |
+
"step": 11900
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.9531681612022643,
|
| 8375 |
+
"grad_norm": 86.5,
|
| 8376 |
+
"learning_rate": 9.9813834430814e-07,
|
| 8377 |
+
"loss": 130.3583,
|
| 8378 |
+
"step": 11910
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.9539684703216618,
|
| 8382 |
+
"grad_norm": 74.4375,
|
| 8383 |
+
"learning_rate": 9.981367812051242e-07,
|
| 8384 |
+
"loss": 130.4393,
|
| 8385 |
+
"step": 11920
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.9547687794410591,
|
| 8389 |
+
"grad_norm": 74.6875,
|
| 8390 |
+
"learning_rate": 9.981352181021084e-07,
|
| 8391 |
+
"loss": 129.8773,
|
| 8392 |
+
"step": 11930
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.9555690885604565,
|
| 8396 |
+
"grad_norm": 77.6875,
|
| 8397 |
+
"learning_rate": 9.981336549990924e-07,
|
| 8398 |
+
"loss": 130.8676,
|
| 8399 |
+
"step": 11940
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.9563693976798538,
|
| 8403 |
+
"grad_norm": 80.5,
|
| 8404 |
+
"learning_rate": 9.981320918960767e-07,
|
| 8405 |
+
"loss": 131.1644,
|
| 8406 |
+
"step": 11950
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.9571697067992512,
|
| 8410 |
+
"grad_norm": 81.125,
|
| 8411 |
+
"learning_rate": 9.981305287930609e-07,
|
| 8412 |
+
"loss": 131.0869,
|
| 8413 |
+
"step": 11960
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.9579700159186486,
|
| 8417 |
+
"grad_norm": 85.0,
|
| 8418 |
+
"learning_rate": 9.98128965690045e-07,
|
| 8419 |
+
"loss": 130.5896,
|
| 8420 |
+
"step": 11970
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.9587703250380459,
|
| 8424 |
+
"grad_norm": 72.625,
|
| 8425 |
+
"learning_rate": 9.981274025870293e-07,
|
| 8426 |
+
"loss": 131.133,
|
| 8427 |
+
"step": 11980
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.9595706341574433,
|
| 8431 |
+
"grad_norm": 77.625,
|
| 8432 |
+
"learning_rate": 9.981258394840133e-07,
|
| 8433 |
+
"loss": 131.9948,
|
| 8434 |
+
"step": 11990
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.9603709432768407,
|
| 8438 |
+
"grad_norm": 78.25,
|
| 8439 |
+
"learning_rate": 9.981242763809975e-07,
|
| 8440 |
+
"loss": 130.3504,
|
| 8441 |
+
"step": 12000
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.961171252396238,
|
| 8445 |
+
"grad_norm": 69.0625,
|
| 8446 |
+
"learning_rate": 9.981227132779817e-07,
|
| 8447 |
+
"loss": 131.8812,
|
| 8448 |
+
"step": 12010
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.9619715615156355,
|
| 8452 |
+
"grad_norm": 76.125,
|
| 8453 |
+
"learning_rate": 9.98121150174966e-07,
|
| 8454 |
+
"loss": 131.7717,
|
| 8455 |
+
"step": 12020
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.9627718706350328,
|
| 8459 |
+
"grad_norm": 76.6875,
|
| 8460 |
+
"learning_rate": 9.981195870719502e-07,
|
| 8461 |
+
"loss": 131.0739,
|
| 8462 |
+
"step": 12030
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.9635721797544301,
|
| 8466 |
+
"grad_norm": 79.5,
|
| 8467 |
+
"learning_rate": 9.981180239689342e-07,
|
| 8468 |
+
"loss": 132.3484,
|
| 8469 |
+
"step": 12040
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.9643724888738275,
|
| 8473 |
+
"grad_norm": 81.25,
|
| 8474 |
+
"learning_rate": 9.981164608659184e-07,
|
| 8475 |
+
"loss": 131.6032,
|
| 8476 |
+
"step": 12050
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.9651727979932249,
|
| 8480 |
+
"grad_norm": 80.375,
|
| 8481 |
+
"learning_rate": 9.981148977629026e-07,
|
| 8482 |
+
"loss": 131.8261,
|
| 8483 |
+
"step": 12060
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.9659731071126223,
|
| 8487 |
+
"grad_norm": 80.9375,
|
| 8488 |
+
"learning_rate": 9.981133346598866e-07,
|
| 8489 |
+
"loss": 129.7895,
|
| 8490 |
+
"step": 12070
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.9667734162320196,
|
| 8494 |
+
"grad_norm": 77.75,
|
| 8495 |
+
"learning_rate": 9.981117715568709e-07,
|
| 8496 |
+
"loss": 129.8964,
|
| 8497 |
+
"step": 12080
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.967573725351417,
|
| 8501 |
+
"grad_norm": 72.5,
|
| 8502 |
+
"learning_rate": 9.98110208453855e-07,
|
| 8503 |
+
"loss": 130.2092,
|
| 8504 |
+
"step": 12090
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.9683740344708144,
|
| 8508 |
+
"grad_norm": 70.75,
|
| 8509 |
+
"learning_rate": 9.98108645350839e-07,
|
| 8510 |
+
"loss": 130.064,
|
| 8511 |
+
"step": 12100
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.9691743435902117,
|
| 8515 |
+
"grad_norm": 82.3125,
|
| 8516 |
+
"learning_rate": 9.981070822478233e-07,
|
| 8517 |
+
"loss": 133.1195,
|
| 8518 |
+
"step": 12110
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.9699746527096091,
|
| 8522 |
+
"grad_norm": 73.375,
|
| 8523 |
+
"learning_rate": 9.981055191448075e-07,
|
| 8524 |
+
"loss": 131.6047,
|
| 8525 |
+
"step": 12120
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.9707749618290065,
|
| 8529 |
+
"grad_norm": 78.0625,
|
| 8530 |
+
"learning_rate": 9.981039560417917e-07,
|
| 8531 |
+
"loss": 131.0189,
|
| 8532 |
+
"step": 12130
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.9715752709484038,
|
| 8536 |
+
"grad_norm": 80.375,
|
| 8537 |
+
"learning_rate": 9.98102392938776e-07,
|
| 8538 |
+
"loss": 131.1126,
|
| 8539 |
+
"step": 12140
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.9723755800678012,
|
| 8543 |
+
"grad_norm": 74.4375,
|
| 8544 |
+
"learning_rate": 9.9810082983576e-07,
|
| 8545 |
+
"loss": 131.885,
|
| 8546 |
+
"step": 12150
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.9731758891871986,
|
| 8550 |
+
"grad_norm": 78.25,
|
| 8551 |
+
"learning_rate": 9.980992667327442e-07,
|
| 8552 |
+
"loss": 130.9126,
|
| 8553 |
+
"step": 12160
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.9739761983065959,
|
| 8557 |
+
"grad_norm": 83.875,
|
| 8558 |
+
"learning_rate": 9.980977036297284e-07,
|
| 8559 |
+
"loss": 131.2764,
|
| 8560 |
+
"step": 12170
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.9747765074259933,
|
| 8564 |
+
"grad_norm": 72.875,
|
| 8565 |
+
"learning_rate": 9.980961405267126e-07,
|
| 8566 |
+
"loss": 132.7158,
|
| 8567 |
+
"step": 12180
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.9755768165453906,
|
| 8571 |
+
"grad_norm": 77.25,
|
| 8572 |
+
"learning_rate": 9.980945774236968e-07,
|
| 8573 |
+
"loss": 131.1642,
|
| 8574 |
+
"step": 12190
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.9763771256647881,
|
| 8578 |
+
"grad_norm": 77.125,
|
| 8579 |
+
"learning_rate": 9.980930143206808e-07,
|
| 8580 |
+
"loss": 130.3661,
|
| 8581 |
+
"step": 12200
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.9771774347841854,
|
| 8585 |
+
"grad_norm": 81.25,
|
| 8586 |
+
"learning_rate": 9.98091451217665e-07,
|
| 8587 |
+
"loss": 132.4058,
|
| 8588 |
+
"step": 12210
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.9779777439035827,
|
| 8592 |
+
"grad_norm": 77.1875,
|
| 8593 |
+
"learning_rate": 9.980898881146493e-07,
|
| 8594 |
+
"loss": 131.1993,
|
| 8595 |
+
"step": 12220
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.9787780530229802,
|
| 8599 |
+
"grad_norm": 78.375,
|
| 8600 |
+
"learning_rate": 9.980883250116333e-07,
|
| 8601 |
+
"loss": 129.8341,
|
| 8602 |
+
"step": 12230
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.9795783621423775,
|
| 8606 |
+
"grad_norm": 82.3125,
|
| 8607 |
+
"learning_rate": 9.980867619086175e-07,
|
| 8608 |
+
"loss": 130.408,
|
| 8609 |
+
"step": 12240
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.9803786712617749,
|
| 8613 |
+
"grad_norm": 79.375,
|
| 8614 |
+
"learning_rate": 9.980851988056017e-07,
|
| 8615 |
+
"loss": 132.0676,
|
| 8616 |
+
"step": 12250
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.9811789803811722,
|
| 8620 |
+
"grad_norm": 75.4375,
|
| 8621 |
+
"learning_rate": 9.980836357025857e-07,
|
| 8622 |
+
"loss": 130.5135,
|
| 8623 |
+
"step": 12260
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.9819792895005696,
|
| 8627 |
+
"grad_norm": 79.75,
|
| 8628 |
+
"learning_rate": 9.9808207259957e-07,
|
| 8629 |
+
"loss": 129.7958,
|
| 8630 |
+
"step": 12270
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.982779598619967,
|
| 8634 |
+
"grad_norm": 78.8125,
|
| 8635 |
+
"learning_rate": 9.980805094965542e-07,
|
| 8636 |
+
"loss": 131.8359,
|
| 8637 |
+
"step": 12280
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.9835799077393643,
|
| 8641 |
+
"grad_norm": 73.75,
|
| 8642 |
+
"learning_rate": 9.980789463935384e-07,
|
| 8643 |
+
"loss": 130.6134,
|
| 8644 |
+
"step": 12290
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.9843802168587616,
|
| 8648 |
+
"grad_norm": 79.0,
|
| 8649 |
+
"learning_rate": 9.980773832905224e-07,
|
| 8650 |
+
"loss": 130.2782,
|
| 8651 |
+
"step": 12300
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.9851805259781591,
|
| 8655 |
+
"grad_norm": 74.0,
|
| 8656 |
+
"learning_rate": 9.980758201875066e-07,
|
| 8657 |
+
"loss": 130.808,
|
| 8658 |
+
"step": 12310
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.9859808350975564,
|
| 8662 |
+
"grad_norm": 73.0,
|
| 8663 |
+
"learning_rate": 9.980742570844908e-07,
|
| 8664 |
+
"loss": 130.4844,
|
| 8665 |
+
"step": 12320
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.9867811442169538,
|
| 8669 |
+
"grad_norm": 72.4375,
|
| 8670 |
+
"learning_rate": 9.98072693981475e-07,
|
| 8671 |
+
"loss": 131.225,
|
| 8672 |
+
"step": 12330
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.9875814533363512,
|
| 8676 |
+
"grad_norm": 78.875,
|
| 8677 |
+
"learning_rate": 9.980711308784593e-07,
|
| 8678 |
+
"loss": 132.9991,
|
| 8679 |
+
"step": 12340
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.9883817624557485,
|
| 8683 |
+
"grad_norm": 75.9375,
|
| 8684 |
+
"learning_rate": 9.980695677754435e-07,
|
| 8685 |
+
"loss": 131.7026,
|
| 8686 |
+
"step": 12350
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.9891820715751459,
|
| 8690 |
+
"grad_norm": 74.125,
|
| 8691 |
+
"learning_rate": 9.980680046724275e-07,
|
| 8692 |
+
"loss": 130.6302,
|
| 8693 |
+
"step": 12360
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.9899823806945433,
|
| 8697 |
+
"grad_norm": 80.75,
|
| 8698 |
+
"learning_rate": 9.980664415694117e-07,
|
| 8699 |
+
"loss": 129.955,
|
| 8700 |
+
"step": 12370
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.9907826898139407,
|
| 8704 |
+
"grad_norm": 70.125,
|
| 8705 |
+
"learning_rate": 9.98064878466396e-07,
|
| 8706 |
+
"loss": 130.9879,
|
| 8707 |
+
"step": 12380
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.991582998933338,
|
| 8711 |
+
"grad_norm": 79.75,
|
| 8712 |
+
"learning_rate": 9.9806331536338e-07,
|
| 8713 |
+
"loss": 130.061,
|
| 8714 |
+
"step": 12390
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.9923833080527353,
|
| 8718 |
+
"grad_norm": 75.375,
|
| 8719 |
+
"learning_rate": 9.980617522603641e-07,
|
| 8720 |
+
"loss": 129.8558,
|
| 8721 |
+
"step": 12400
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.9931836171721328,
|
| 8725 |
+
"grad_norm": 78.4375,
|
| 8726 |
+
"learning_rate": 9.980601891573484e-07,
|
| 8727 |
+
"loss": 129.8251,
|
| 8728 |
+
"step": 12410
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.9939839262915301,
|
| 8732 |
+
"grad_norm": 70.1875,
|
| 8733 |
+
"learning_rate": 9.980586260543324e-07,
|
| 8734 |
+
"loss": 130.4507,
|
| 8735 |
+
"step": 12420
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.9947842354109274,
|
| 8739 |
+
"grad_norm": 74.4375,
|
| 8740 |
+
"learning_rate": 9.980570629513166e-07,
|
| 8741 |
+
"loss": 131.6167,
|
| 8742 |
+
"step": 12430
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.9955845445303249,
|
| 8746 |
+
"grad_norm": 76.1875,
|
| 8747 |
+
"learning_rate": 9.980554998483008e-07,
|
| 8748 |
+
"loss": 132.0188,
|
| 8749 |
+
"step": 12440
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.9963848536497222,
|
| 8753 |
+
"grad_norm": 79.6875,
|
| 8754 |
+
"learning_rate": 9.98053936745285e-07,
|
| 8755 |
+
"loss": 130.7771,
|
| 8756 |
+
"step": 12450
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.9971851627691196,
|
| 8760 |
+
"grad_norm": 74.25,
|
| 8761 |
+
"learning_rate": 9.98052373642269e-07,
|
| 8762 |
+
"loss": 130.4612,
|
| 8763 |
+
"step": 12460
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.9979854718885169,
|
| 8767 |
+
"grad_norm": 78.0625,
|
| 8768 |
+
"learning_rate": 9.980508105392532e-07,
|
| 8769 |
+
"loss": 129.6428,
|
| 8770 |
+
"step": 12470
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.9987857810079143,
|
| 8774 |
+
"grad_norm": 77.6875,
|
| 8775 |
+
"learning_rate": 9.980492474362375e-07,
|
| 8776 |
+
"loss": 130.6133,
|
| 8777 |
+
"step": 12480
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.9995860901273117,
|
| 8781 |
+
"grad_norm": 77.75,
|
| 8782 |
+
"learning_rate": 9.980476843332217e-07,
|
| 8783 |
+
"loss": 131.8606,
|
| 8784 |
+
"step": 12490
|
| 8785 |
}
|
| 8786 |
],
|
| 8787 |
"logging_steps": 10,
|
|
|
|
| 8796 |
"should_evaluate": false,
|
| 8797 |
"should_log": false,
|
| 8798 |
"should_save": true,
|
| 8799 |
+
"should_training_stop": true
|
| 8800 |
},
|
| 8801 |
"attributes": {}
|
| 8802 |
}
|
| 8803 |
},
|
| 8804 |
+
"total_flos": 3.450015163893783e+19,
|
| 8805 |
"train_batch_size": 4,
|
| 8806 |
"trial_name": null,
|
| 8807 |
"trial_params": null
|