Training in progress, step 1203, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 100059752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5766c9076aa65820bd20114ead0fddc573d82a8d44c1c7266ebce7fed9751ce
|
| 3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 51244404
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79c1191b1d61fd2fc498d2424137ea0f51d2cfa6d721e5df1fc14f2eee9e6403
|
| 3 |
size 51244404
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cdc455725f19e73c6431fdb94cdc19a0e5e952d9bce5b35d5315287d5e0ef7e
|
| 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 sha256:9b4f7b1adcf3f1039fcdf81d8121c4a3574bf90804ee42c6615ea449229e14ac
|
| 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":
|
| 5 |
"eval_steps": 301,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6360,6 +6360,2106 @@
|
|
| 6360 |
"eval_samples_per_second": 15.319,
|
| 6361 |
"eval_steps_per_second": 7.674,
|
| 6362 |
"step": 903
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6363 |
}
|
| 6364 |
],
|
| 6365 |
"logging_steps": 1,
|
|
@@ -6374,12 +8474,12 @@
|
|
| 6374 |
"should_evaluate": false,
|
| 6375 |
"should_log": false,
|
| 6376 |
"should_save": true,
|
| 6377 |
-
"should_training_stop":
|
| 6378 |
},
|
| 6379 |
"attributes": {}
|
| 6380 |
}
|
| 6381 |
},
|
| 6382 |
-
"total_flos":
|
| 6383 |
"train_batch_size": 2,
|
| 6384 |
"trial_name": null,
|
| 6385 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0006238303181534,
|
| 5 |
"eval_steps": 301,
|
| 6 |
+
"global_step": 1203,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6360 |
"eval_samples_per_second": 15.319,
|
| 6361 |
"eval_steps_per_second": 7.674,
|
| 6362 |
"step": 903
|
| 6363 |
+
},
|
| 6364 |
+
{
|
| 6365 |
+
"epoch": 0.7519234768143065,
|
| 6366 |
+
"grad_norm": 0.45154818892478943,
|
| 6367 |
+
"learning_rate": 2.942911444520342e-05,
|
| 6368 |
+
"loss": 0.4398,
|
| 6369 |
+
"step": 904
|
| 6370 |
+
},
|
| 6371 |
+
{
|
| 6372 |
+
"epoch": 0.7527552505718444,
|
| 6373 |
+
"grad_norm": 0.4417174160480499,
|
| 6374 |
+
"learning_rate": 2.9242785626565138e-05,
|
| 6375 |
+
"loss": 0.4324,
|
| 6376 |
+
"step": 905
|
| 6377 |
+
},
|
| 6378 |
+
{
|
| 6379 |
+
"epoch": 0.7535870243293824,
|
| 6380 |
+
"grad_norm": 0.48358669877052307,
|
| 6381 |
+
"learning_rate": 2.90569474777329e-05,
|
| 6382 |
+
"loss": 0.4895,
|
| 6383 |
+
"step": 906
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 0.7544187980869204,
|
| 6387 |
+
"grad_norm": 0.4476510286331177,
|
| 6388 |
+
"learning_rate": 2.887160128741163e-05,
|
| 6389 |
+
"loss": 0.4196,
|
| 6390 |
+
"step": 907
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.7552505718444583,
|
| 6394 |
+
"grad_norm": 0.46220117807388306,
|
| 6395 |
+
"learning_rate": 2.868674834089474e-05,
|
| 6396 |
+
"loss": 0.4528,
|
| 6397 |
+
"step": 908
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.7560823456019963,
|
| 6401 |
+
"grad_norm": 0.4371815323829651,
|
| 6402 |
+
"learning_rate": 2.850238992005514e-05,
|
| 6403 |
+
"loss": 0.4295,
|
| 6404 |
+
"step": 909
|
| 6405 |
+
},
|
| 6406 |
+
{
|
| 6407 |
+
"epoch": 0.7569141193595342,
|
| 6408 |
+
"grad_norm": 0.4548187255859375,
|
| 6409 |
+
"learning_rate": 2.831852730333646e-05,
|
| 6410 |
+
"loss": 0.4048,
|
| 6411 |
+
"step": 910
|
| 6412 |
+
},
|
| 6413 |
+
{
|
| 6414 |
+
"epoch": 0.7577458931170722,
|
| 6415 |
+
"grad_norm": 0.4508820176124573,
|
| 6416 |
+
"learning_rate": 2.81351617657442e-05,
|
| 6417 |
+
"loss": 0.4449,
|
| 6418 |
+
"step": 911
|
| 6419 |
+
},
|
| 6420 |
+
{
|
| 6421 |
+
"epoch": 0.7585776668746101,
|
| 6422 |
+
"grad_norm": 0.45161473751068115,
|
| 6423 |
+
"learning_rate": 2.795229457883678e-05,
|
| 6424 |
+
"loss": 0.4065,
|
| 6425 |
+
"step": 912
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 0.7594094406321481,
|
| 6429 |
+
"grad_norm": 0.46211326122283936,
|
| 6430 |
+
"learning_rate": 2.776992701071681e-05,
|
| 6431 |
+
"loss": 0.4616,
|
| 6432 |
+
"step": 913
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.7602412143896861,
|
| 6436 |
+
"grad_norm": 0.4814813435077667,
|
| 6437 |
+
"learning_rate": 2.7588060326022204e-05,
|
| 6438 |
+
"loss": 0.4786,
|
| 6439 |
+
"step": 914
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.7610729881472239,
|
| 6443 |
+
"grad_norm": 0.4523828327655792,
|
| 6444 |
+
"learning_rate": 2.7406695785917548e-05,
|
| 6445 |
+
"loss": 0.4583,
|
| 6446 |
+
"step": 915
|
| 6447 |
+
},
|
| 6448 |
+
{
|
| 6449 |
+
"epoch": 0.7619047619047619,
|
| 6450 |
+
"grad_norm": 0.4711090624332428,
|
| 6451 |
+
"learning_rate": 2.722583464808528e-05,
|
| 6452 |
+
"loss": 0.4626,
|
| 6453 |
+
"step": 916
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.7627365356622998,
|
| 6457 |
+
"grad_norm": 0.4674610197544098,
|
| 6458 |
+
"learning_rate": 2.7045478166716843e-05,
|
| 6459 |
+
"loss": 0.4564,
|
| 6460 |
+
"step": 917
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.7635683094198378,
|
| 6464 |
+
"grad_norm": 0.4593535363674164,
|
| 6465 |
+
"learning_rate": 2.6865627592504293e-05,
|
| 6466 |
+
"loss": 0.4415,
|
| 6467 |
+
"step": 918
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.7644000831773757,
|
| 6471 |
+
"grad_norm": 0.48565688729286194,
|
| 6472 |
+
"learning_rate": 2.668628417263137e-05,
|
| 6473 |
+
"loss": 0.4568,
|
| 6474 |
+
"step": 919
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.7652318569349137,
|
| 6478 |
+
"grad_norm": 0.45922335982322693,
|
| 6479 |
+
"learning_rate": 2.650744915076485e-05,
|
| 6480 |
+
"loss": 0.4595,
|
| 6481 |
+
"step": 920
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.7660636306924516,
|
| 6485 |
+
"grad_norm": 0.4446061849594116,
|
| 6486 |
+
"learning_rate": 2.6329123767046072e-05,
|
| 6487 |
+
"loss": 0.4567,
|
| 6488 |
+
"step": 921
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.7668954044499896,
|
| 6492 |
+
"grad_norm": 0.4568280279636383,
|
| 6493 |
+
"learning_rate": 2.615130925808228e-05,
|
| 6494 |
+
"loss": 0.4301,
|
| 6495 |
+
"step": 922
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.7677271782075276,
|
| 6499 |
+
"grad_norm": 0.437994122505188,
|
| 6500 |
+
"learning_rate": 2.597400685693795e-05,
|
| 6501 |
+
"loss": 0.4077,
|
| 6502 |
+
"step": 923
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.7685589519650655,
|
| 6506 |
+
"grad_norm": 0.47482725977897644,
|
| 6507 |
+
"learning_rate": 2.5797217793126372e-05,
|
| 6508 |
+
"loss": 0.4414,
|
| 6509 |
+
"step": 924
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.7693907257226035,
|
| 6513 |
+
"grad_norm": 0.4981728196144104,
|
| 6514 |
+
"learning_rate": 2.5620943292601075e-05,
|
| 6515 |
+
"loss": 0.5072,
|
| 6516 |
+
"step": 925
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.7702224994801414,
|
| 6520 |
+
"grad_norm": 0.4595918357372284,
|
| 6521 |
+
"learning_rate": 2.5445184577747307e-05,
|
| 6522 |
+
"loss": 0.4303,
|
| 6523 |
+
"step": 926
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.7710542732376794,
|
| 6527 |
+
"grad_norm": 0.4366806447505951,
|
| 6528 |
+
"learning_rate": 2.5269942867373598e-05,
|
| 6529 |
+
"loss": 0.4047,
|
| 6530 |
+
"step": 927
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.7718860469952173,
|
| 6534 |
+
"grad_norm": 0.42185187339782715,
|
| 6535 |
+
"learning_rate": 2.5095219376703184e-05,
|
| 6536 |
+
"loss": 0.394,
|
| 6537 |
+
"step": 928
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.7727178207527553,
|
| 6541 |
+
"grad_norm": 0.42919063568115234,
|
| 6542 |
+
"learning_rate": 2.492101531736579e-05,
|
| 6543 |
+
"loss": 0.3702,
|
| 6544 |
+
"step": 929
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.7735495945102931,
|
| 6548 |
+
"grad_norm": 0.4393722414970398,
|
| 6549 |
+
"learning_rate": 2.4747331897389104e-05,
|
| 6550 |
+
"loss": 0.4539,
|
| 6551 |
+
"step": 930
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 0.7743813682678311,
|
| 6555 |
+
"grad_norm": 0.4793844521045685,
|
| 6556 |
+
"learning_rate": 2.4574170321190303e-05,
|
| 6557 |
+
"loss": 0.4688,
|
| 6558 |
+
"step": 931
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.7752131420253691,
|
| 6562 |
+
"grad_norm": 0.4290564954280853,
|
| 6563 |
+
"learning_rate": 2.4401531789567977e-05,
|
| 6564 |
+
"loss": 0.4374,
|
| 6565 |
+
"step": 932
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.776044915782907,
|
| 6569 |
+
"grad_norm": 0.4846777617931366,
|
| 6570 |
+
"learning_rate": 2.4229417499693496e-05,
|
| 6571 |
+
"loss": 0.4552,
|
| 6572 |
+
"step": 933
|
| 6573 |
+
},
|
| 6574 |
+
{
|
| 6575 |
+
"epoch": 0.776876689540445,
|
| 6576 |
+
"grad_norm": 0.43358123302459717,
|
| 6577 |
+
"learning_rate": 2.4057828645102897e-05,
|
| 6578 |
+
"loss": 0.4127,
|
| 6579 |
+
"step": 934
|
| 6580 |
+
},
|
| 6581 |
+
{
|
| 6582 |
+
"epoch": 0.7777084632979829,
|
| 6583 |
+
"grad_norm": 0.43994030356407166,
|
| 6584 |
+
"learning_rate": 2.38867664156886e-05,
|
| 6585 |
+
"loss": 0.3671,
|
| 6586 |
+
"step": 935
|
| 6587 |
+
},
|
| 6588 |
+
{
|
| 6589 |
+
"epoch": 0.7785402370555209,
|
| 6590 |
+
"grad_norm": 0.49586546421051025,
|
| 6591 |
+
"learning_rate": 2.3716231997691008e-05,
|
| 6592 |
+
"loss": 0.4594,
|
| 6593 |
+
"step": 936
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 0.7793720108130588,
|
| 6597 |
+
"grad_norm": 0.4775140583515167,
|
| 6598 |
+
"learning_rate": 2.3546226573690444e-05,
|
| 6599 |
+
"loss": 0.4929,
|
| 6600 |
+
"step": 937
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.7802037845705968,
|
| 6604 |
+
"grad_norm": 0.451071560382843,
|
| 6605 |
+
"learning_rate": 2.3376751322599e-05,
|
| 6606 |
+
"loss": 0.432,
|
| 6607 |
+
"step": 938
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.7810355583281348,
|
| 6611 |
+
"grad_norm": 0.4588756859302521,
|
| 6612 |
+
"learning_rate": 2.3207807419652063e-05,
|
| 6613 |
+
"loss": 0.4233,
|
| 6614 |
+
"step": 939
|
| 6615 |
+
},
|
| 6616 |
+
{
|
| 6617 |
+
"epoch": 0.7818673320856727,
|
| 6618 |
+
"grad_norm": 0.4052916169166565,
|
| 6619 |
+
"learning_rate": 2.3039396036400462e-05,
|
| 6620 |
+
"loss": 0.3442,
|
| 6621 |
+
"step": 940
|
| 6622 |
+
},
|
| 6623 |
+
{
|
| 6624 |
+
"epoch": 0.7826991058432107,
|
| 6625 |
+
"grad_norm": 0.4676628112792969,
|
| 6626 |
+
"learning_rate": 2.2871518340702258e-05,
|
| 6627 |
+
"loss": 0.4591,
|
| 6628 |
+
"step": 941
|
| 6629 |
+
},
|
| 6630 |
+
{
|
| 6631 |
+
"epoch": 0.7835308796007486,
|
| 6632 |
+
"grad_norm": 0.45338648557662964,
|
| 6633 |
+
"learning_rate": 2.2704175496714554e-05,
|
| 6634 |
+
"loss": 0.4201,
|
| 6635 |
+
"step": 942
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 0.7843626533582866,
|
| 6639 |
+
"grad_norm": 0.49033913016319275,
|
| 6640 |
+
"learning_rate": 2.2537368664885527e-05,
|
| 6641 |
+
"loss": 0.4297,
|
| 6642 |
+
"step": 943
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.7851944271158245,
|
| 6646 |
+
"grad_norm": 0.5046612620353699,
|
| 6647 |
+
"learning_rate": 2.2371099001946384e-05,
|
| 6648 |
+
"loss": 0.4662,
|
| 6649 |
+
"step": 944
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.7860262008733624,
|
| 6653 |
+
"grad_norm": 0.4667772650718689,
|
| 6654 |
+
"learning_rate": 2.2205367660903263e-05,
|
| 6655 |
+
"loss": 0.4556,
|
| 6656 |
+
"step": 945
|
| 6657 |
+
},
|
| 6658 |
+
{
|
| 6659 |
+
"epoch": 0.7868579746309003,
|
| 6660 |
+
"grad_norm": 0.4427892863750458,
|
| 6661 |
+
"learning_rate": 2.2040175791029305e-05,
|
| 6662 |
+
"loss": 0.4268,
|
| 6663 |
+
"step": 946
|
| 6664 |
+
},
|
| 6665 |
+
{
|
| 6666 |
+
"epoch": 0.7876897483884383,
|
| 6667 |
+
"grad_norm": 0.4500638246536255,
|
| 6668 |
+
"learning_rate": 2.1875524537856618e-05,
|
| 6669 |
+
"loss": 0.4634,
|
| 6670 |
+
"step": 947
|
| 6671 |
+
},
|
| 6672 |
+
{
|
| 6673 |
+
"epoch": 0.7885215221459763,
|
| 6674 |
+
"grad_norm": 0.4674094021320343,
|
| 6675 |
+
"learning_rate": 2.171141504316839e-05,
|
| 6676 |
+
"loss": 0.4786,
|
| 6677 |
+
"step": 948
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 0.7893532959035142,
|
| 6681 |
+
"grad_norm": 0.44143038988113403,
|
| 6682 |
+
"learning_rate": 2.1547848444991024e-05,
|
| 6683 |
+
"loss": 0.4095,
|
| 6684 |
+
"step": 949
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.7901850696610522,
|
| 6688 |
+
"grad_norm": 0.4612862765789032,
|
| 6689 |
+
"learning_rate": 2.1384825877586046e-05,
|
| 6690 |
+
"loss": 0.4616,
|
| 6691 |
+
"step": 950
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.7910168434185901,
|
| 6695 |
+
"grad_norm": 0.4669111669063568,
|
| 6696 |
+
"learning_rate": 2.122234847144248e-05,
|
| 6697 |
+
"loss": 0.4714,
|
| 6698 |
+
"step": 951
|
| 6699 |
+
},
|
| 6700 |
+
{
|
| 6701 |
+
"epoch": 0.7918486171761281,
|
| 6702 |
+
"grad_norm": 0.4334806203842163,
|
| 6703 |
+
"learning_rate": 2.1060417353268845e-05,
|
| 6704 |
+
"loss": 0.4137,
|
| 6705 |
+
"step": 952
|
| 6706 |
+
},
|
| 6707 |
+
{
|
| 6708 |
+
"epoch": 0.792680390933666,
|
| 6709 |
+
"grad_norm": 0.4584311246871948,
|
| 6710 |
+
"learning_rate": 2.0899033645985423e-05,
|
| 6711 |
+
"loss": 0.4528,
|
| 6712 |
+
"step": 953
|
| 6713 |
+
},
|
| 6714 |
+
{
|
| 6715 |
+
"epoch": 0.793512164691204,
|
| 6716 |
+
"grad_norm": 0.4858227074146271,
|
| 6717 |
+
"learning_rate": 2.0738198468716464e-05,
|
| 6718 |
+
"loss": 0.4774,
|
| 6719 |
+
"step": 954
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 0.794343938448742,
|
| 6723 |
+
"grad_norm": 0.45899567008018494,
|
| 6724 |
+
"learning_rate": 2.0577912936782317e-05,
|
| 6725 |
+
"loss": 0.4579,
|
| 6726 |
+
"step": 955
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.7951757122062799,
|
| 6730 |
+
"grad_norm": 0.4413417875766754,
|
| 6731 |
+
"learning_rate": 2.041817816169187e-05,
|
| 6732 |
+
"loss": 0.3861,
|
| 6733 |
+
"step": 956
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.7960074859638179,
|
| 6737 |
+
"grad_norm": 0.4700511395931244,
|
| 6738 |
+
"learning_rate": 2.025899525113474e-05,
|
| 6739 |
+
"loss": 0.4471,
|
| 6740 |
+
"step": 957
|
| 6741 |
+
},
|
| 6742 |
+
{
|
| 6743 |
+
"epoch": 0.7968392597213558,
|
| 6744 |
+
"grad_norm": 0.4677594304084778,
|
| 6745 |
+
"learning_rate": 2.010036530897359e-05,
|
| 6746 |
+
"loss": 0.4574,
|
| 6747 |
+
"step": 958
|
| 6748 |
+
},
|
| 6749 |
+
{
|
| 6750 |
+
"epoch": 0.7976710334788938,
|
| 6751 |
+
"grad_norm": 0.4514382779598236,
|
| 6752 |
+
"learning_rate": 1.9942289435236506e-05,
|
| 6753 |
+
"loss": 0.4355,
|
| 6754 |
+
"step": 959
|
| 6755 |
+
},
|
| 6756 |
+
{
|
| 6757 |
+
"epoch": 0.7985028072364317,
|
| 6758 |
+
"grad_norm": 0.45568642020225525,
|
| 6759 |
+
"learning_rate": 1.9784768726109392e-05,
|
| 6760 |
+
"loss": 0.4268,
|
| 6761 |
+
"step": 960
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 0.7993345809939696,
|
| 6765 |
+
"grad_norm": 0.47598856687545776,
|
| 6766 |
+
"learning_rate": 1.962780427392823e-05,
|
| 6767 |
+
"loss": 0.4575,
|
| 6768 |
+
"step": 961
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.8001663547515075,
|
| 6772 |
+
"grad_norm": 0.4754261076450348,
|
| 6773 |
+
"learning_rate": 1.9471397167171713e-05,
|
| 6774 |
+
"loss": 0.4829,
|
| 6775 |
+
"step": 962
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.8009981285090455,
|
| 6779 |
+
"grad_norm": 0.4439312219619751,
|
| 6780 |
+
"learning_rate": 1.931554849045355e-05,
|
| 6781 |
+
"loss": 0.4202,
|
| 6782 |
+
"step": 963
|
| 6783 |
+
},
|
| 6784 |
+
{
|
| 6785 |
+
"epoch": 0.8018299022665835,
|
| 6786 |
+
"grad_norm": 0.443600594997406,
|
| 6787 |
+
"learning_rate": 1.916025932451493e-05,
|
| 6788 |
+
"loss": 0.3892,
|
| 6789 |
+
"step": 964
|
| 6790 |
+
},
|
| 6791 |
+
{
|
| 6792 |
+
"epoch": 0.8026616760241214,
|
| 6793 |
+
"grad_norm": 0.4861088693141937,
|
| 6794 |
+
"learning_rate": 1.9005530746217238e-05,
|
| 6795 |
+
"loss": 0.4411,
|
| 6796 |
+
"step": 965
|
| 6797 |
+
},
|
| 6798 |
+
{
|
| 6799 |
+
"epoch": 0.8034934497816594,
|
| 6800 |
+
"grad_norm": 0.42934009432792664,
|
| 6801 |
+
"learning_rate": 1.8851363828534252e-05,
|
| 6802 |
+
"loss": 0.3886,
|
| 6803 |
+
"step": 966
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 0.8043252235391973,
|
| 6807 |
+
"grad_norm": 0.4442944824695587,
|
| 6808 |
+
"learning_rate": 1.869775964054501e-05,
|
| 6809 |
+
"loss": 0.4428,
|
| 6810 |
+
"step": 967
|
| 6811 |
+
},
|
| 6812 |
+
{
|
| 6813 |
+
"epoch": 0.8051569972967353,
|
| 6814 |
+
"grad_norm": 0.49083849787712097,
|
| 6815 |
+
"learning_rate": 1.8544719247426222e-05,
|
| 6816 |
+
"loss": 0.4401,
|
| 6817 |
+
"step": 968
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 0.8059887710542732,
|
| 6821 |
+
"grad_norm": 0.48380911350250244,
|
| 6822 |
+
"learning_rate": 1.839224371044491e-05,
|
| 6823 |
+
"loss": 0.4609,
|
| 6824 |
+
"step": 969
|
| 6825 |
+
},
|
| 6826 |
+
{
|
| 6827 |
+
"epoch": 0.8068205448118112,
|
| 6828 |
+
"grad_norm": 0.47577401995658875,
|
| 6829 |
+
"learning_rate": 1.8240334086951115e-05,
|
| 6830 |
+
"loss": 0.4557,
|
| 6831 |
+
"step": 970
|
| 6832 |
+
},
|
| 6833 |
+
{
|
| 6834 |
+
"epoch": 0.8076523185693492,
|
| 6835 |
+
"grad_norm": 0.4672863781452179,
|
| 6836 |
+
"learning_rate": 1.8088991430370506e-05,
|
| 6837 |
+
"loss": 0.456,
|
| 6838 |
+
"step": 971
|
| 6839 |
+
},
|
| 6840 |
+
{
|
| 6841 |
+
"epoch": 0.8084840923268871,
|
| 6842 |
+
"grad_norm": 0.46177253127098083,
|
| 6843 |
+
"learning_rate": 1.793821679019707e-05,
|
| 6844 |
+
"loss": 0.4676,
|
| 6845 |
+
"step": 972
|
| 6846 |
+
},
|
| 6847 |
+
{
|
| 6848 |
+
"epoch": 0.8093158660844251,
|
| 6849 |
+
"grad_norm": 0.4453592896461487,
|
| 6850 |
+
"learning_rate": 1.77880112119859e-05,
|
| 6851 |
+
"loss": 0.4122,
|
| 6852 |
+
"step": 973
|
| 6853 |
+
},
|
| 6854 |
+
{
|
| 6855 |
+
"epoch": 0.810147639841963,
|
| 6856 |
+
"grad_norm": 0.46347111463546753,
|
| 6857 |
+
"learning_rate": 1.76383757373458e-05,
|
| 6858 |
+
"loss": 0.4048,
|
| 6859 |
+
"step": 974
|
| 6860 |
+
},
|
| 6861 |
+
{
|
| 6862 |
+
"epoch": 0.810979413599501,
|
| 6863 |
+
"grad_norm": 0.4728952944278717,
|
| 6864 |
+
"learning_rate": 1.7489311403932274e-05,
|
| 6865 |
+
"loss": 0.4938,
|
| 6866 |
+
"step": 975
|
| 6867 |
+
},
|
| 6868 |
+
{
|
| 6869 |
+
"epoch": 0.8118111873570388,
|
| 6870 |
+
"grad_norm": 0.44490066170692444,
|
| 6871 |
+
"learning_rate": 1.7340819245440167e-05,
|
| 6872 |
+
"loss": 0.4376,
|
| 6873 |
+
"step": 976
|
| 6874 |
+
},
|
| 6875 |
+
{
|
| 6876 |
+
"epoch": 0.8126429611145768,
|
| 6877 |
+
"grad_norm": 0.42778635025024414,
|
| 6878 |
+
"learning_rate": 1.7192900291596492e-05,
|
| 6879 |
+
"loss": 0.3666,
|
| 6880 |
+
"step": 977
|
| 6881 |
+
},
|
| 6882 |
+
{
|
| 6883 |
+
"epoch": 0.8134747348721147,
|
| 6884 |
+
"grad_norm": 0.46628859639167786,
|
| 6885 |
+
"learning_rate": 1.7045555568153416e-05,
|
| 6886 |
+
"loss": 0.4607,
|
| 6887 |
+
"step": 978
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 0.8143065086296527,
|
| 6891 |
+
"grad_norm": 0.48081284761428833,
|
| 6892 |
+
"learning_rate": 1.6898786096881104e-05,
|
| 6893 |
+
"loss": 0.4264,
|
| 6894 |
+
"step": 979
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.8151382823871907,
|
| 6898 |
+
"grad_norm": 0.4593854546546936,
|
| 6899 |
+
"learning_rate": 1.6752592895560492e-05,
|
| 6900 |
+
"loss": 0.426,
|
| 6901 |
+
"step": 980
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.8159700561447286,
|
| 6905 |
+
"grad_norm": 0.462982714176178,
|
| 6906 |
+
"learning_rate": 1.6606976977976407e-05,
|
| 6907 |
+
"loss": 0.409,
|
| 6908 |
+
"step": 981
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 0.8168018299022666,
|
| 6912 |
+
"grad_norm": 0.48566728830337524,
|
| 6913 |
+
"learning_rate": 1.6461939353910493e-05,
|
| 6914 |
+
"loss": 0.4496,
|
| 6915 |
+
"step": 982
|
| 6916 |
+
},
|
| 6917 |
+
{
|
| 6918 |
+
"epoch": 0.8176336036598045,
|
| 6919 |
+
"grad_norm": 0.4559323489665985,
|
| 6920 |
+
"learning_rate": 1.6317481029134117e-05,
|
| 6921 |
+
"loss": 0.4085,
|
| 6922 |
+
"step": 983
|
| 6923 |
+
},
|
| 6924 |
+
{
|
| 6925 |
+
"epoch": 0.8184653774173425,
|
| 6926 |
+
"grad_norm": 0.4498133957386017,
|
| 6927 |
+
"learning_rate": 1.6173603005401504e-05,
|
| 6928 |
+
"loss": 0.396,
|
| 6929 |
+
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.8192971511748804,
|
| 6933 |
+
"grad_norm": 0.46261173486709595,
|
| 6934 |
+
"learning_rate": 1.6030306280442764e-05,
|
| 6935 |
+
"loss": 0.4115,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.8201289249324184,
|
| 6940 |
+
"grad_norm": 0.5084740519523621,
|
| 6941 |
+
"learning_rate": 1.588759184795694e-05,
|
| 6942 |
+
"loss": 0.5119,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.8209606986899564,
|
| 6947 |
+
"grad_norm": 0.48162010312080383,
|
| 6948 |
+
"learning_rate": 1.5745460697605142e-05,
|
| 6949 |
+
"loss": 0.4707,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.8217924724474943,
|
| 6954 |
+
"grad_norm": 0.40404409170150757,
|
| 6955 |
+
"learning_rate": 1.5603913815003633e-05,
|
| 6956 |
+
"loss": 0.3163,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.8226242462050323,
|
| 6961 |
+
"grad_norm": 0.43817463517189026,
|
| 6962 |
+
"learning_rate": 1.5462952181717115e-05,
|
| 6963 |
+
"loss": 0.3779,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.8234560199625702,
|
| 6968 |
+
"grad_norm": 0.44949933886528015,
|
| 6969 |
+
"learning_rate": 1.532257677525183e-05,
|
| 6970 |
+
"loss": 0.4216,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.8242877937201081,
|
| 6975 |
+
"grad_norm": 0.4588971734046936,
|
| 6976 |
+
"learning_rate": 1.518278856904869e-05,
|
| 6977 |
+
"loss": 0.4182,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.825119567477646,
|
| 6982 |
+
"grad_norm": 0.46600207686424255,
|
| 6983 |
+
"learning_rate": 1.5043588532476827e-05,
|
| 6984 |
+
"loss": 0.3981,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.825951341235184,
|
| 6989 |
+
"grad_norm": 0.4738927185535431,
|
| 6990 |
+
"learning_rate": 1.4904977630826499e-05,
|
| 6991 |
+
"loss": 0.4327,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.826783114992722,
|
| 6996 |
+
"grad_norm": 0.45611268281936646,
|
| 6997 |
+
"learning_rate": 1.476695682530268e-05,
|
| 6998 |
+
"loss": 0.4133,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.8276148887502599,
|
| 7003 |
+
"grad_norm": 0.4523008465766907,
|
| 7004 |
+
"learning_rate": 1.4629527073018267e-05,
|
| 7005 |
+
"loss": 0.4163,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.8284466625077979,
|
| 7010 |
+
"grad_norm": 0.4965008795261383,
|
| 7011 |
+
"learning_rate": 1.449268932698743e-05,
|
| 7012 |
+
"loss": 0.4769,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.8292784362653358,
|
| 7017 |
+
"grad_norm": 0.4614514112472534,
|
| 7018 |
+
"learning_rate": 1.4356444536119084e-05,
|
| 7019 |
+
"loss": 0.4124,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.8301102100228738,
|
| 7024 |
+
"grad_norm": 0.4713315963745117,
|
| 7025 |
+
"learning_rate": 1.422079364521024e-05,
|
| 7026 |
+
"loss": 0.4393,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.8309419837804117,
|
| 7031 |
+
"grad_norm": 0.4365687668323517,
|
| 7032 |
+
"learning_rate": 1.4085737594939497e-05,
|
| 7033 |
+
"loss": 0.3662,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.8317737575379497,
|
| 7038 |
+
"grad_norm": 0.44617828726768494,
|
| 7039 |
+
"learning_rate": 1.3951277321860467e-05,
|
| 7040 |
+
"loss": 0.391,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.8326055312954876,
|
| 7045 |
+
"grad_norm": 0.4467557966709137,
|
| 7046 |
+
"learning_rate": 1.381741375839537e-05,
|
| 7047 |
+
"loss": 0.3977,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.8334373050530256,
|
| 7052 |
+
"grad_norm": 0.4606059789657593,
|
| 7053 |
+
"learning_rate": 1.3684147832828408e-05,
|
| 7054 |
+
"loss": 0.4407,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.8342690788105636,
|
| 7059 |
+
"grad_norm": 0.47593462467193604,
|
| 7060 |
+
"learning_rate": 1.355148046929956e-05,
|
| 7061 |
+
"loss": 0.4656,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.8351008525681015,
|
| 7066 |
+
"grad_norm": 0.42734506726264954,
|
| 7067 |
+
"learning_rate": 1.3419412587797907e-05,
|
| 7068 |
+
"loss": 0.3723,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.8359326263256395,
|
| 7073 |
+
"grad_norm": 0.4908114969730377,
|
| 7074 |
+
"learning_rate": 1.3287945104155486e-05,
|
| 7075 |
+
"loss": 0.4673,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.8367644000831774,
|
| 7080 |
+
"grad_norm": 0.4308733642101288,
|
| 7081 |
+
"learning_rate": 1.3157078930040856e-05,
|
| 7082 |
+
"loss": 0.4019,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.8375961738407153,
|
| 7087 |
+
"grad_norm": 0.4562091529369354,
|
| 7088 |
+
"learning_rate": 1.3026814972952672e-05,
|
| 7089 |
+
"loss": 0.3943,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.8384279475982532,
|
| 7094 |
+
"grad_norm": 0.43687310814857483,
|
| 7095 |
+
"learning_rate": 1.289715413621354e-05,
|
| 7096 |
+
"loss": 0.4678,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.8392597213557912,
|
| 7101 |
+
"grad_norm": 0.430271178483963,
|
| 7102 |
+
"learning_rate": 1.2768097318963701e-05,
|
| 7103 |
+
"loss": 0.3959,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.8400914951133291,
|
| 7108 |
+
"grad_norm": 0.46327558159828186,
|
| 7109 |
+
"learning_rate": 1.2639645416154744e-05,
|
| 7110 |
+
"loss": 0.4285,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.8409232688708671,
|
| 7115 |
+
"grad_norm": 0.4796364903450012,
|
| 7116 |
+
"learning_rate": 1.2511799318543494e-05,
|
| 7117 |
+
"loss": 0.4121,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.8417550426284051,
|
| 7122 |
+
"grad_norm": 0.4678579866886139,
|
| 7123 |
+
"learning_rate": 1.2384559912685767e-05,
|
| 7124 |
+
"loss": 0.4179,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.842586816385943,
|
| 7129 |
+
"grad_norm": 0.4598294794559479,
|
| 7130 |
+
"learning_rate": 1.2257928080930237e-05,
|
| 7131 |
+
"loss": 0.4069,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.843418590143481,
|
| 7136 |
+
"grad_norm": 0.46210917830467224,
|
| 7137 |
+
"learning_rate": 1.2131904701412345e-05,
|
| 7138 |
+
"loss": 0.4178,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.8442503639010189,
|
| 7143 |
+
"grad_norm": 0.4848369061946869,
|
| 7144 |
+
"learning_rate": 1.2006490648048118e-05,
|
| 7145 |
+
"loss": 0.4427,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.8450821376585569,
|
| 7150 |
+
"grad_norm": 0.47031518816947937,
|
| 7151 |
+
"learning_rate": 1.1881686790528279e-05,
|
| 7152 |
+
"loss": 0.4483,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.8459139114160948,
|
| 7157 |
+
"grad_norm": 0.4915429949760437,
|
| 7158 |
+
"learning_rate": 1.1757493994312053e-05,
|
| 7159 |
+
"loss": 0.4175,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.8467456851736328,
|
| 7164 |
+
"grad_norm": 0.507524311542511,
|
| 7165 |
+
"learning_rate": 1.1633913120621188e-05,
|
| 7166 |
+
"loss": 0.4322,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.8475774589311708,
|
| 7171 |
+
"grad_norm": 0.5011469125747681,
|
| 7172 |
+
"learning_rate": 1.151094502643414e-05,
|
| 7173 |
+
"loss": 0.4903,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.8484092326887087,
|
| 7178 |
+
"grad_norm": 0.49545034766197205,
|
| 7179 |
+
"learning_rate": 1.1388590564479896e-05,
|
| 7180 |
+
"loss": 0.4745,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.8492410064462467,
|
| 7185 |
+
"grad_norm": 0.43036097288131714,
|
| 7186 |
+
"learning_rate": 1.1266850583232224e-05,
|
| 7187 |
+
"loss": 0.3641,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.8500727802037845,
|
| 7192 |
+
"grad_norm": 0.48114851117134094,
|
| 7193 |
+
"learning_rate": 1.1145725926903772e-05,
|
| 7194 |
+
"loss": 0.4766,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.8509045539613225,
|
| 7199 |
+
"grad_norm": 0.4713135063648224,
|
| 7200 |
+
"learning_rate": 1.1025217435440116e-05,
|
| 7201 |
+
"loss": 0.467,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.8517363277188604,
|
| 7206 |
+
"grad_norm": 0.4947219789028168,
|
| 7207 |
+
"learning_rate": 1.0905325944514033e-05,
|
| 7208 |
+
"loss": 0.4558,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.8525681014763984,
|
| 7213 |
+
"grad_norm": 0.5086840987205505,
|
| 7214 |
+
"learning_rate": 1.078605228551971e-05,
|
| 7215 |
+
"loss": 0.4464,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.8533998752339363,
|
| 7220 |
+
"grad_norm": 0.46704235672950745,
|
| 7221 |
+
"learning_rate": 1.0667397285566893e-05,
|
| 7222 |
+
"loss": 0.4556,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.8542316489914743,
|
| 7227 |
+
"grad_norm": 0.4636327624320984,
|
| 7228 |
+
"learning_rate": 1.0549361767475241e-05,
|
| 7229 |
+
"loss": 0.4259,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.8550634227490123,
|
| 7234 |
+
"grad_norm": 0.44666600227355957,
|
| 7235 |
+
"learning_rate": 1.0431946549768567e-05,
|
| 7236 |
+
"loss": 0.4069,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.8558951965065502,
|
| 7241 |
+
"grad_norm": 0.4543648660182953,
|
| 7242 |
+
"learning_rate": 1.0315152446669141e-05,
|
| 7243 |
+
"loss": 0.4679,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.8567269702640882,
|
| 7248 |
+
"grad_norm": 0.48564526438713074,
|
| 7249 |
+
"learning_rate": 1.019898026809214e-05,
|
| 7250 |
+
"loss": 0.4505,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.8575587440216261,
|
| 7255 |
+
"grad_norm": 0.44043809175491333,
|
| 7256 |
+
"learning_rate": 1.008343081963996e-05,
|
| 7257 |
+
"loss": 0.3891,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.8583905177791641,
|
| 7262 |
+
"grad_norm": 0.467408686876297,
|
| 7263 |
+
"learning_rate": 9.968504902596564e-06,
|
| 7264 |
+
"loss": 0.4349,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.859222291536702,
|
| 7269 |
+
"grad_norm": 0.45551759004592896,
|
| 7270 |
+
"learning_rate": 9.85420331392214e-06,
|
| 7271 |
+
"loss": 0.4104,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.86005406529424,
|
| 7276 |
+
"grad_norm": 0.45863786339759827,
|
| 7277 |
+
"learning_rate": 9.740526846247311e-06,
|
| 7278 |
+
"loss": 0.4216,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.860885839051778,
|
| 7283 |
+
"grad_norm": 0.4453166127204895,
|
| 7284 |
+
"learning_rate": 9.62747628786782e-06,
|
| 7285 |
+
"loss": 0.3981,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.8617176128093159,
|
| 7290 |
+
"grad_norm": 0.47664642333984375,
|
| 7291 |
+
"learning_rate": 9.515052422739035e-06,
|
| 7292 |
+
"loss": 0.4761,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.8625493865668539,
|
| 7297 |
+
"grad_norm": 0.4840780794620514,
|
| 7298 |
+
"learning_rate": 9.403256030470387e-06,
|
| 7299 |
+
"loss": 0.4519,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.8633811603243917,
|
| 7304 |
+
"grad_norm": 0.5594334006309509,
|
| 7305 |
+
"learning_rate": 9.292087886320166e-06,
|
| 7306 |
+
"loss": 0.4863,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.8642129340819297,
|
| 7311 |
+
"grad_norm": 0.47458481788635254,
|
| 7312 |
+
"learning_rate": 9.181548761189995e-06,
|
| 7313 |
+
"loss": 0.4448,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.8650447078394676,
|
| 7318 |
+
"grad_norm": 0.4778815507888794,
|
| 7319 |
+
"learning_rate": 9.071639421619526e-06,
|
| 7320 |
+
"loss": 0.4443,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.8658764815970056,
|
| 7325 |
+
"grad_norm": 0.47068047523498535,
|
| 7326 |
+
"learning_rate": 8.962360629781163e-06,
|
| 7327 |
+
"loss": 0.4098,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.8667082553545435,
|
| 7332 |
+
"grad_norm": 0.46489661931991577,
|
| 7333 |
+
"learning_rate": 8.853713143474684e-06,
|
| 7334 |
+
"loss": 0.4351,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.8675400291120815,
|
| 7339 |
+
"grad_norm": 0.4436318278312683,
|
| 7340 |
+
"learning_rate": 8.745697716122081e-06,
|
| 7341 |
+
"loss": 0.4292,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.8683718028696195,
|
| 7346 |
+
"grad_norm": 0.4547750949859619,
|
| 7347 |
+
"learning_rate": 8.638315096762317e-06,
|
| 7348 |
+
"loss": 0.4378,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.8692035766271574,
|
| 7353 |
+
"grad_norm": 0.44218558073043823,
|
| 7354 |
+
"learning_rate": 8.531566030046034e-06,
|
| 7355 |
+
"loss": 0.4036,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.8700353503846954,
|
| 7360 |
+
"grad_norm": 0.4511972665786743,
|
| 7361 |
+
"learning_rate": 8.425451256230587e-06,
|
| 7362 |
+
"loss": 0.411,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.8708671241422333,
|
| 7367 |
+
"grad_norm": 0.45339876413345337,
|
| 7368 |
+
"learning_rate": 8.319971511174718e-06,
|
| 7369 |
+
"loss": 0.4316,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.8716988978997713,
|
| 7374 |
+
"grad_norm": 0.45976632833480835,
|
| 7375 |
+
"learning_rate": 8.2151275263335e-06,
|
| 7376 |
+
"loss": 0.4238,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.8725306716573092,
|
| 7381 |
+
"grad_norm": 0.47895899415016174,
|
| 7382 |
+
"learning_rate": 8.110920028753355e-06,
|
| 7383 |
+
"loss": 0.4596,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.8733624454148472,
|
| 7388 |
+
"grad_norm": 0.452475368976593,
|
| 7389 |
+
"learning_rate": 8.007349741066938e-06,
|
| 7390 |
+
"loss": 0.4124,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.8741942191723852,
|
| 7395 |
+
"grad_norm": 0.46902817487716675,
|
| 7396 |
+
"learning_rate": 7.904417381488083e-06,
|
| 7397 |
+
"loss": 0.4208,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.8750259929299231,
|
| 7402 |
+
"grad_norm": 0.48741471767425537,
|
| 7403 |
+
"learning_rate": 7.802123663806938e-06,
|
| 7404 |
+
"loss": 0.4514,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.875857766687461,
|
| 7409 |
+
"grad_norm": 0.4875292479991913,
|
| 7410 |
+
"learning_rate": 7.700469297384927e-06,
|
| 7411 |
+
"loss": 0.4575,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.8766895404449989,
|
| 7416 |
+
"grad_norm": 0.46032848954200745,
|
| 7417 |
+
"learning_rate": 7.599454987149867e-06,
|
| 7418 |
+
"loss": 0.4457,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.8775213142025369,
|
| 7423 |
+
"grad_norm": 0.4430507719516754,
|
| 7424 |
+
"learning_rate": 7.499081433591071e-06,
|
| 7425 |
+
"loss": 0.3952,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.8783530879600748,
|
| 7430 |
+
"grad_norm": 0.4799799919128418,
|
| 7431 |
+
"learning_rate": 7.399349332754458e-06,
|
| 7432 |
+
"loss": 0.4549,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.8791848617176128,
|
| 7437 |
+
"grad_norm": 0.4532495439052582,
|
| 7438 |
+
"learning_rate": 7.300259376237795e-06,
|
| 7439 |
+
"loss": 0.4187,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.8800166354751507,
|
| 7444 |
+
"grad_norm": 0.45743227005004883,
|
| 7445 |
+
"learning_rate": 7.2018122511858686e-06,
|
| 7446 |
+
"loss": 0.4541,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.8808484092326887,
|
| 7451 |
+
"grad_norm": 0.4854099452495575,
|
| 7452 |
+
"learning_rate": 7.1040086402856425e-06,
|
| 7453 |
+
"loss": 0.4634,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.8816801829902267,
|
| 7458 |
+
"grad_norm": 0.445450097322464,
|
| 7459 |
+
"learning_rate": 7.006849221761735e-06,
|
| 7460 |
+
"loss": 0.4417,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.8825119567477646,
|
| 7465 |
+
"grad_norm": 0.47935259342193604,
|
| 7466 |
+
"learning_rate": 6.9103346693714324e-06,
|
| 7467 |
+
"loss": 0.4476,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.8833437305053026,
|
| 7472 |
+
"grad_norm": 0.4244360625743866,
|
| 7473 |
+
"learning_rate": 6.814465652400237e-06,
|
| 7474 |
+
"loss": 0.3806,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.8841755042628405,
|
| 7479 |
+
"grad_norm": 0.47703632712364197,
|
| 7480 |
+
"learning_rate": 6.719242835657147e-06,
|
| 7481 |
+
"loss": 0.4705,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.8850072780203785,
|
| 7486 |
+
"grad_norm": 0.49910375475883484,
|
| 7487 |
+
"learning_rate": 6.624666879470009e-06,
|
| 7488 |
+
"loss": 0.4734,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.8858390517779164,
|
| 7493 |
+
"grad_norm": 0.47923290729522705,
|
| 7494 |
+
"learning_rate": 6.530738439681016e-06,
|
| 7495 |
+
"loss": 0.4531,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.8866708255354544,
|
| 7500 |
+
"grad_norm": 0.47322043776512146,
|
| 7501 |
+
"learning_rate": 6.437458167642163e-06,
|
| 7502 |
+
"loss": 0.4317,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.8875025992929924,
|
| 7507 |
+
"grad_norm": 0.4898202419281006,
|
| 7508 |
+
"learning_rate": 6.344826710210583e-06,
|
| 7509 |
+
"loss": 0.4535,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.8883343730505302,
|
| 7514 |
+
"grad_norm": 0.46324723958969116,
|
| 7515 |
+
"learning_rate": 6.252844709744255e-06,
|
| 7516 |
+
"loss": 0.4363,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.8891661468080682,
|
| 7521 |
+
"grad_norm": 0.4827021360397339,
|
| 7522 |
+
"learning_rate": 6.1615128040974355e-06,
|
| 7523 |
+
"loss": 0.5059,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.8899979205656061,
|
| 7528 |
+
"grad_norm": 0.4697610139846802,
|
| 7529 |
+
"learning_rate": 6.070831626616236e-06,
|
| 7530 |
+
"loss": 0.4423,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.8908296943231441,
|
| 7535 |
+
"grad_norm": 0.46442911028862,
|
| 7536 |
+
"learning_rate": 5.980801806134318e-06,
|
| 7537 |
+
"loss": 0.4639,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.891661468080682,
|
| 7542 |
+
"grad_norm": 0.46676331758499146,
|
| 7543 |
+
"learning_rate": 5.891423966968413e-06,
|
| 7544 |
+
"loss": 0.4359,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.89249324183822,
|
| 7549 |
+
"grad_norm": 0.4345012903213501,
|
| 7550 |
+
"learning_rate": 5.80269872891408e-06,
|
| 7551 |
+
"loss": 0.3871,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.8933250155957579,
|
| 7556 |
+
"grad_norm": 0.4475749731063843,
|
| 7557 |
+
"learning_rate": 5.71462670724141e-06,
|
| 7558 |
+
"loss": 0.4113,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.8941567893532959,
|
| 7563 |
+
"grad_norm": 0.4608396589756012,
|
| 7564 |
+
"learning_rate": 5.627208512690641e-06,
|
| 7565 |
+
"loss": 0.4331,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.8949885631108339,
|
| 7570 |
+
"grad_norm": 0.4600012004375458,
|
| 7571 |
+
"learning_rate": 5.540444751468099e-06,
|
| 7572 |
+
"loss": 0.4451,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.8958203368683718,
|
| 7577 |
+
"grad_norm": 0.4469583332538605,
|
| 7578 |
+
"learning_rate": 5.45433602524188e-06,
|
| 7579 |
+
"loss": 0.4097,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.8966521106259098,
|
| 7584 |
+
"grad_norm": 0.4726197123527527,
|
| 7585 |
+
"learning_rate": 5.3688829311376754e-06,
|
| 7586 |
+
"loss": 0.4048,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.8974838843834477,
|
| 7591 |
+
"grad_norm": 0.48056867718696594,
|
| 7592 |
+
"learning_rate": 5.2840860617346724e-06,
|
| 7593 |
+
"loss": 0.4246,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.8983156581409857,
|
| 7598 |
+
"grad_norm": 0.4238966405391693,
|
| 7599 |
+
"learning_rate": 5.199946005061462e-06,
|
| 7600 |
+
"loss": 0.3802,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.8991474318985236,
|
| 7605 |
+
"grad_norm": 0.4405467212200165,
|
| 7606 |
+
"learning_rate": 5.116463344591893e-06,
|
| 7607 |
+
"loss": 0.4105,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.8999792056560616,
|
| 7612 |
+
"grad_norm": 0.4323815405368805,
|
| 7613 |
+
"learning_rate": 5.033638659241102e-06,
|
| 7614 |
+
"loss": 0.4078,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.9008109794135996,
|
| 7619 |
+
"grad_norm": 0.4602968394756317,
|
| 7620 |
+
"learning_rate": 4.9514725233614005e-06,
|
| 7621 |
+
"loss": 0.4464,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.9016427531711374,
|
| 7626 |
+
"grad_norm": 0.5263029932975769,
|
| 7627 |
+
"learning_rate": 4.869965506738416e-06,
|
| 7628 |
+
"loss": 0.4628,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.9024745269286754,
|
| 7633 |
+
"grad_norm": 0.48673149943351746,
|
| 7634 |
+
"learning_rate": 4.789118174587071e-06,
|
| 7635 |
+
"loss": 0.4498,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.9033063006862133,
|
| 7640 |
+
"grad_norm": 0.48753026127815247,
|
| 7641 |
+
"learning_rate": 4.708931087547585e-06,
|
| 7642 |
+
"loss": 0.4805,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.9041380744437513,
|
| 7647 |
+
"grad_norm": 0.47595274448394775,
|
| 7648 |
+
"learning_rate": 4.629404801681791e-06,
|
| 7649 |
+
"loss": 0.4401,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.9049698482012892,
|
| 7654 |
+
"grad_norm": 0.4522855877876282,
|
| 7655 |
+
"learning_rate": 4.5505398684691056e-06,
|
| 7656 |
+
"loss": 0.3991,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.9058016219588272,
|
| 7661 |
+
"grad_norm": 0.43736204504966736,
|
| 7662 |
+
"learning_rate": 4.472336834802737e-06,
|
| 7663 |
+
"loss": 0.4245,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.9066333957163651,
|
| 7668 |
+
"grad_norm": 0.4781038761138916,
|
| 7669 |
+
"learning_rate": 4.394796242985932e-06,
|
| 7670 |
+
"loss": 0.4392,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.9074651694739031,
|
| 7675 |
+
"grad_norm": 0.48694223165512085,
|
| 7676 |
+
"learning_rate": 4.317918630728235e-06,
|
| 7677 |
+
"loss": 0.4516,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.9082969432314411,
|
| 7682 |
+
"grad_norm": 0.4531137943267822,
|
| 7683 |
+
"learning_rate": 4.241704531141632e-06,
|
| 7684 |
+
"loss": 0.4547,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.909128716988979,
|
| 7689 |
+
"grad_norm": 0.4477342367172241,
|
| 7690 |
+
"learning_rate": 4.16615447273706e-06,
|
| 7691 |
+
"loss": 0.4185,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.909960490746517,
|
| 7696 |
+
"grad_norm": 0.41774898767471313,
|
| 7697 |
+
"learning_rate": 4.091268979420537e-06,
|
| 7698 |
+
"loss": 0.3805,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.9107922645040549,
|
| 7703 |
+
"grad_norm": 0.458509236574173,
|
| 7704 |
+
"learning_rate": 4.017048570489646e-06,
|
| 7705 |
+
"loss": 0.4202,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.9116240382615929,
|
| 7710 |
+
"grad_norm": 0.46096915006637573,
|
| 7711 |
+
"learning_rate": 3.943493760629924e-06,
|
| 7712 |
+
"loss": 0.4468,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.9124558120191308,
|
| 7717 |
+
"grad_norm": 0.42885664105415344,
|
| 7718 |
+
"learning_rate": 3.870605059911237e-06,
|
| 7719 |
+
"loss": 0.3855,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.9132875857766688,
|
| 7724 |
+
"grad_norm": 0.4437970519065857,
|
| 7725 |
+
"learning_rate": 3.798382973784298e-06,
|
| 7726 |
+
"loss": 0.4203,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.9141193595342066,
|
| 7731 |
+
"grad_norm": 0.4737008213996887,
|
| 7732 |
+
"learning_rate": 3.7268280030771652e-06,
|
| 7733 |
+
"loss": 0.4431,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.9149511332917446,
|
| 7738 |
+
"grad_norm": 0.43085286021232605,
|
| 7739 |
+
"learning_rate": 3.6559406439917178e-06,
|
| 7740 |
+
"loss": 0.4002,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.9157829070492826,
|
| 7745 |
+
"grad_norm": 0.4739695191383362,
|
| 7746 |
+
"learning_rate": 3.5857213881002827e-06,
|
| 7747 |
+
"loss": 0.4255,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.9166146808068205,
|
| 7752 |
+
"grad_norm": 0.458615243434906,
|
| 7753 |
+
"learning_rate": 3.516170722342127e-06,
|
| 7754 |
+
"loss": 0.4576,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.9174464545643585,
|
| 7759 |
+
"grad_norm": 0.4438440203666687,
|
| 7760 |
+
"learning_rate": 3.4472891290201927e-06,
|
| 7761 |
+
"loss": 0.4244,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.9182782283218964,
|
| 7766 |
+
"grad_norm": 0.4436255395412445,
|
| 7767 |
+
"learning_rate": 3.3790770857976994e-06,
|
| 7768 |
+
"loss": 0.4183,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.9191100020794344,
|
| 7773 |
+
"grad_norm": 0.45590856671333313,
|
| 7774 |
+
"learning_rate": 3.311535065694804e-06,
|
| 7775 |
+
"loss": 0.424,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.9199417758369723,
|
| 7780 |
+
"grad_norm": 0.458316445350647,
|
| 7781 |
+
"learning_rate": 3.2446635370853684e-06,
|
| 7782 |
+
"loss": 0.4382,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.9207735495945103,
|
| 7787 |
+
"grad_norm": 0.4662122130393982,
|
| 7788 |
+
"learning_rate": 3.17846296369374e-06,
|
| 7789 |
+
"loss": 0.475,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.9216053233520483,
|
| 7794 |
+
"grad_norm": 0.44484686851501465,
|
| 7795 |
+
"learning_rate": 3.1129338045914e-06,
|
| 7796 |
+
"loss": 0.4014,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.9224370971095862,
|
| 7801 |
+
"grad_norm": 0.46727603673934937,
|
| 7802 |
+
"learning_rate": 3.0480765141939314e-06,
|
| 7803 |
+
"loss": 0.4509,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.9232688708671242,
|
| 7808 |
+
"grad_norm": 0.49041593074798584,
|
| 7809 |
+
"learning_rate": 2.9838915422577995e-06,
|
| 7810 |
+
"loss": 0.4275,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.9241006446246621,
|
| 7815 |
+
"grad_norm": 0.4922788739204407,
|
| 7816 |
+
"learning_rate": 2.920379333877221e-06,
|
| 7817 |
+
"loss": 0.425,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.9249324183822001,
|
| 7822 |
+
"grad_norm": 0.44843536615371704,
|
| 7823 |
+
"learning_rate": 2.857540329481112e-06,
|
| 7824 |
+
"loss": 0.4224,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.925764192139738,
|
| 7829 |
+
"grad_norm": 0.4896073341369629,
|
| 7830 |
+
"learning_rate": 2.795374964830022e-06,
|
| 7831 |
+
"loss": 0.4124,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.9265959658972759,
|
| 7836 |
+
"grad_norm": 0.4726916253566742,
|
| 7837 |
+
"learning_rate": 2.7338836710130824e-06,
|
| 7838 |
+
"loss": 0.4593,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.9274277396548138,
|
| 7843 |
+
"grad_norm": 0.4574068784713745,
|
| 7844 |
+
"learning_rate": 2.673066874445096e-06,
|
| 7845 |
+
"loss": 0.4232,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.9282595134123518,
|
| 7850 |
+
"grad_norm": 0.4518861174583435,
|
| 7851 |
+
"learning_rate": 2.6129249968634527e-06,
|
| 7852 |
+
"loss": 0.4438,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.9290912871698898,
|
| 7857 |
+
"grad_norm": 0.4974750876426697,
|
| 7858 |
+
"learning_rate": 2.5534584553253525e-06,
|
| 7859 |
+
"loss": 0.4648,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.9299230609274277,
|
| 7864 |
+
"grad_norm": 0.4564787745475769,
|
| 7865 |
+
"learning_rate": 2.494667662204797e-06,
|
| 7866 |
+
"loss": 0.4413,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.9307548346849657,
|
| 7871 |
+
"grad_norm": 0.4412038028240204,
|
| 7872 |
+
"learning_rate": 2.4365530251897585e-06,
|
| 7873 |
+
"loss": 0.406,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.9315866084425036,
|
| 7878 |
+
"grad_norm": 0.4330199658870697,
|
| 7879 |
+
"learning_rate": 2.3791149472794374e-06,
|
| 7880 |
+
"loss": 0.3749,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.9324183822000416,
|
| 7885 |
+
"grad_norm": 0.4556577503681183,
|
| 7886 |
+
"learning_rate": 2.3223538267813316e-06,
|
| 7887 |
+
"loss": 0.4138,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.9332501559575795,
|
| 7892 |
+
"grad_norm": 0.4702088236808777,
|
| 7893 |
+
"learning_rate": 2.2662700573085503e-06,
|
| 7894 |
+
"loss": 0.4574,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.9340819297151175,
|
| 7899 |
+
"grad_norm": 0.47332215309143066,
|
| 7900 |
+
"learning_rate": 2.2108640277771154e-06,
|
| 7901 |
+
"loss": 0.4662,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.9349137034726555,
|
| 7906 |
+
"grad_norm": 0.4619726538658142,
|
| 7907 |
+
"learning_rate": 2.156136122403174e-06,
|
| 7908 |
+
"loss": 0.4262,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.9357454772301934,
|
| 7913 |
+
"grad_norm": 0.4245651662349701,
|
| 7914 |
+
"learning_rate": 2.1020867207004025e-06,
|
| 7915 |
+
"loss": 0.3824,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.9365772509877314,
|
| 7920 |
+
"grad_norm": 0.43925681710243225,
|
| 7921 |
+
"learning_rate": 2.048716197477374e-06,
|
| 7922 |
+
"loss": 0.4017,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.9374090247452693,
|
| 7927 |
+
"grad_norm": 0.4655887484550476,
|
| 7928 |
+
"learning_rate": 1.9960249228349047e-06,
|
| 7929 |
+
"loss": 0.4691,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.9382407985028073,
|
| 7934 |
+
"grad_norm": 0.4686010777950287,
|
| 7935 |
+
"learning_rate": 1.944013262163569e-06,
|
| 7936 |
+
"loss": 0.4289,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.9390725722603452,
|
| 7941 |
+
"grad_norm": 0.4429991543292999,
|
| 7942 |
+
"learning_rate": 1.8926815761410864e-06,
|
| 7943 |
+
"loss": 0.3923,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.9399043460178831,
|
| 7948 |
+
"grad_norm": 0.4607723653316498,
|
| 7949 |
+
"learning_rate": 1.8420302207298623e-06,
|
| 7950 |
+
"loss": 0.4203,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.940736119775421,
|
| 7955 |
+
"grad_norm": 0.4541813135147095,
|
| 7956 |
+
"learning_rate": 1.7920595471745073e-06,
|
| 7957 |
+
"loss": 0.4336,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.941567893532959,
|
| 7962 |
+
"grad_norm": 0.4974818825721741,
|
| 7963 |
+
"learning_rate": 1.7427699019994415e-06,
|
| 7964 |
+
"loss": 0.4327,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.942399667290497,
|
| 7969 |
+
"grad_norm": 0.46558845043182373,
|
| 7970 |
+
"learning_rate": 1.6941616270063854e-06,
|
| 7971 |
+
"loss": 0.4299,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.9432314410480349,
|
| 7976 |
+
"grad_norm": 0.47151467204093933,
|
| 7977 |
+
"learning_rate": 1.6462350592721498e-06,
|
| 7978 |
+
"loss": 0.4253,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.9440632148055729,
|
| 7983 |
+
"grad_norm": 0.4254817068576813,
|
| 7984 |
+
"learning_rate": 1.5989905311461273e-06,
|
| 7985 |
+
"loss": 0.3865,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.9448949885631108,
|
| 7990 |
+
"grad_norm": 0.43250134587287903,
|
| 7991 |
+
"learning_rate": 1.5524283702481158e-06,
|
| 7992 |
+
"loss": 0.3956,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.9457267623206488,
|
| 7997 |
+
"grad_norm": 0.477190226316452,
|
| 7998 |
+
"learning_rate": 1.5065488994659983e-06,
|
| 7999 |
+
"loss": 0.4346,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.9465585360781867,
|
| 8004 |
+
"grad_norm": 0.45829352736473083,
|
| 8005 |
+
"learning_rate": 1.461352436953478e-06,
|
| 8006 |
+
"loss": 0.4406,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.9473903098357247,
|
| 8011 |
+
"grad_norm": 0.47035565972328186,
|
| 8012 |
+
"learning_rate": 1.4168392961279252e-06,
|
| 8013 |
+
"loss": 0.4279,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.9482220835932627,
|
| 8018 |
+
"grad_norm": 0.4768435060977936,
|
| 8019 |
+
"learning_rate": 1.3730097856681667e-06,
|
| 8020 |
+
"loss": 0.4548,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.9490538573508006,
|
| 8025 |
+
"grad_norm": 0.45586657524108887,
|
| 8026 |
+
"learning_rate": 1.329864209512377e-06,
|
| 8027 |
+
"loss": 0.4571,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.9498856311083386,
|
| 8032 |
+
"grad_norm": 0.475225031375885,
|
| 8033 |
+
"learning_rate": 1.2874028668559246e-06,
|
| 8034 |
+
"loss": 0.4173,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.9507174048658765,
|
| 8039 |
+
"grad_norm": 0.4526824355125427,
|
| 8040 |
+
"learning_rate": 1.245626052149318e-06,
|
| 8041 |
+
"loss": 0.403,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.9515491786234145,
|
| 8046 |
+
"grad_norm": 0.47225192189216614,
|
| 8047 |
+
"learning_rate": 1.2045340550961959e-06,
|
| 8048 |
+
"loss": 0.4694,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.9523809523809523,
|
| 8053 |
+
"grad_norm": 0.4445044696331024,
|
| 8054 |
+
"learning_rate": 1.164127160651285e-06,
|
| 8055 |
+
"loss": 0.4257,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.9532127261384903,
|
| 8060 |
+
"grad_norm": 0.4514501690864563,
|
| 8061 |
+
"learning_rate": 1.1244056490184008e-06,
|
| 8062 |
+
"loss": 0.4636,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.9540444998960282,
|
| 8067 |
+
"grad_norm": 0.4854033887386322,
|
| 8068 |
+
"learning_rate": 1.0853697956485941e-06,
|
| 8069 |
+
"loss": 0.4298,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.9548762736535662,
|
| 8074 |
+
"grad_norm": 0.494552880525589,
|
| 8075 |
+
"learning_rate": 1.0470198712381084e-06,
|
| 8076 |
+
"loss": 0.4852,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.9557080474111042,
|
| 8081 |
+
"grad_norm": 0.4838801622390747,
|
| 8082 |
+
"learning_rate": 1.009356141726614e-06,
|
| 8083 |
+
"loss": 0.431,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.9565398211686421,
|
| 8088 |
+
"grad_norm": 0.4265715479850769,
|
| 8089 |
+
"learning_rate": 9.723788682953538e-07,
|
| 8090 |
+
"loss": 0.3796,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.9573715949261801,
|
| 8095 |
+
"grad_norm": 0.45620259642601013,
|
| 8096 |
+
"learning_rate": 9.360883073652238e-07,
|
| 8097 |
+
"loss": 0.4008,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.958203368683718,
|
| 8102 |
+
"grad_norm": 0.4679751992225647,
|
| 8103 |
+
"learning_rate": 9.004847105951509e-07,
|
| 8104 |
+
"loss": 0.4653,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.959035142441256,
|
| 8109 |
+
"grad_norm": 0.4542837142944336,
|
| 8110 |
+
"learning_rate": 8.655683248802282e-07,
|
| 8111 |
+
"loss": 0.4187,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.9598669161987939,
|
| 8116 |
+
"grad_norm": 0.46689337491989136,
|
| 8117 |
+
"learning_rate": 8.313393923500612e-07,
|
| 8118 |
+
"loss": 0.4679,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.9606986899563319,
|
| 8123 |
+
"grad_norm": 0.45560774207115173,
|
| 8124 |
+
"learning_rate": 7.977981503670795e-07,
|
| 8125 |
+
"loss": 0.4445,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.9615304637138699,
|
| 8130 |
+
"grad_norm": 0.4644782543182373,
|
| 8131 |
+
"learning_rate": 7.64944831524872e-07,
|
| 8132 |
+
"loss": 0.4626,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.9623622374714078,
|
| 8137 |
+
"grad_norm": 0.4683065116405487,
|
| 8138 |
+
"learning_rate": 7.327796636465767e-07,
|
| 8139 |
+
"loss": 0.4704,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.9631940112289458,
|
| 8144 |
+
"grad_norm": 0.4686606228351593,
|
| 8145 |
+
"learning_rate": 7.01302869783338e-07,
|
| 8146 |
+
"loss": 0.4245,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.9640257849864837,
|
| 8151 |
+
"grad_norm": 0.47688302397727966,
|
| 8152 |
+
"learning_rate": 6.705146682127184e-07,
|
| 8153 |
+
"loss": 0.4646,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.9648575587440216,
|
| 8158 |
+
"grad_norm": 0.4900341033935547,
|
| 8159 |
+
"learning_rate": 6.404152724371892e-07,
|
| 8160 |
+
"loss": 0.4657,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.9656893325015595,
|
| 8165 |
+
"grad_norm": 0.4813731014728546,
|
| 8166 |
+
"learning_rate": 6.110048911826871e-07,
|
| 8167 |
+
"loss": 0.4603,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.9665211062590975,
|
| 8172 |
+
"grad_norm": 0.43915826082229614,
|
| 8173 |
+
"learning_rate": 5.82283728397115e-07,
|
| 8174 |
+
"loss": 0.3793,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.9673528800166354,
|
| 8179 |
+
"grad_norm": 0.4620192050933838,
|
| 8180 |
+
"learning_rate": 5.542519832489545e-07,
|
| 8181 |
+
"loss": 0.4506,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.9681846537741734,
|
| 8186 |
+
"grad_norm": 0.4768211543560028,
|
| 8187 |
+
"learning_rate": 5.269098501259007e-07,
|
| 8188 |
+
"loss": 0.4329,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.9690164275317114,
|
| 8193 |
+
"grad_norm": 0.5138925909996033,
|
| 8194 |
+
"learning_rate": 5.002575186334734e-07,
|
| 8195 |
+
"loss": 0.4742,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.9698482012892493,
|
| 8200 |
+
"grad_norm": 0.4895179569721222,
|
| 8201 |
+
"learning_rate": 4.7429517359374174e-07,
|
| 8202 |
+
"loss": 0.4397,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.9706799750467873,
|
| 8207 |
+
"grad_norm": 0.46533793210983276,
|
| 8208 |
+
"learning_rate": 4.490229950440239e-07,
|
| 8209 |
+
"loss": 0.4477,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.9715117488043252,
|
| 8214 |
+
"grad_norm": 0.4750311076641083,
|
| 8215 |
+
"learning_rate": 4.2444115823562225e-07,
|
| 8216 |
+
"loss": 0.4647,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.9723435225618632,
|
| 8221 |
+
"grad_norm": 0.48082664608955383,
|
| 8222 |
+
"learning_rate": 4.005498336326463e-07,
|
| 8223 |
+
"loss": 0.4107,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.9731752963194011,
|
| 8228 |
+
"grad_norm": 0.46984878182411194,
|
| 8229 |
+
"learning_rate": 3.7734918691081367e-07,
|
| 8230 |
+
"loss": 0.4366,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.9740070700769391,
|
| 8235 |
+
"grad_norm": 0.49155181646347046,
|
| 8236 |
+
"learning_rate": 3.548393789562732e-07,
|
| 8237 |
+
"loss": 0.4726,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.974838843834477,
|
| 8242 |
+
"grad_norm": 0.48586782813072205,
|
| 8243 |
+
"learning_rate": 3.330205658645391e-07,
|
| 8244 |
+
"loss": 0.4504,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.975670617592015,
|
| 8249 |
+
"grad_norm": 0.43636637926101685,
|
| 8250 |
+
"learning_rate": 3.118928989393699e-07,
|
| 8251 |
+
"loss": 0.414,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.976502391349553,
|
| 8256 |
+
"grad_norm": 0.4744214713573456,
|
| 8257 |
+
"learning_rate": 2.9145652469174664e-07,
|
| 8258 |
+
"loss": 0.4395,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.9773341651070909,
|
| 8263 |
+
"grad_norm": 0.4799196422100067,
|
| 8264 |
+
"learning_rate": 2.717115848388296e-07,
|
| 8265 |
+
"loss": 0.3952,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.9781659388646288,
|
| 8270 |
+
"grad_norm": 0.4748809039592743,
|
| 8271 |
+
"learning_rate": 2.526582163029811e-07,
|
| 8272 |
+
"loss": 0.4411,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.9789977126221667,
|
| 8277 |
+
"grad_norm": 0.4506809711456299,
|
| 8278 |
+
"learning_rate": 2.3429655121085526e-07,
|
| 8279 |
+
"loss": 0.4154,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.9798294863797047,
|
| 8284 |
+
"grad_norm": 0.46106207370758057,
|
| 8285 |
+
"learning_rate": 2.1662671689242076e-07,
|
| 8286 |
+
"loss": 0.4242,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.9806612601372426,
|
| 8291 |
+
"grad_norm": 0.46232593059539795,
|
| 8292 |
+
"learning_rate": 1.996488358801174e-07,
|
| 8293 |
+
"loss": 0.4076,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.9814930338947806,
|
| 8298 |
+
"grad_norm": 0.4545399844646454,
|
| 8299 |
+
"learning_rate": 1.8336302590798992e-07,
|
| 8300 |
+
"loss": 0.4014,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.9823248076523186,
|
| 8305 |
+
"grad_norm": 0.4233172833919525,
|
| 8306 |
+
"learning_rate": 1.677693999109109e-07,
|
| 8307 |
+
"loss": 0.3912,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.9831565814098565,
|
| 8312 |
+
"grad_norm": 0.4664671719074249,
|
| 8313 |
+
"learning_rate": 1.5286806602372583e-07,
|
| 8314 |
+
"loss": 0.4249,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.9839883551673945,
|
| 8319 |
+
"grad_norm": 0.511982262134552,
|
| 8320 |
+
"learning_rate": 1.3865912758054268e-07,
|
| 8321 |
+
"loss": 0.5448,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.9848201289249324,
|
| 8326 |
+
"grad_norm": 0.4627574682235718,
|
| 8327 |
+
"learning_rate": 1.2514268311405453e-07,
|
| 8328 |
+
"loss": 0.4176,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.9856519026824704,
|
| 8333 |
+
"grad_norm": 0.47234347462654114,
|
| 8334 |
+
"learning_rate": 1.1231882635477364e-07,
|
| 8335 |
+
"loss": 0.4579,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.9864836764400083,
|
| 8340 |
+
"grad_norm": 0.4667273163795471,
|
| 8341 |
+
"learning_rate": 1.0018764623045406e-07,
|
| 8342 |
+
"loss": 0.4705,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.9873154501975463,
|
| 8347 |
+
"grad_norm": 0.44828054308891296,
|
| 8348 |
+
"learning_rate": 8.874922686541443e-08,
|
| 8349 |
+
"loss": 0.401,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.9881472239550843,
|
| 8354 |
+
"grad_norm": 0.5001073479652405,
|
| 8355 |
+
"learning_rate": 7.80036475800272e-08,
|
| 8356 |
+
"loss": 0.4742,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.9889789977126222,
|
| 8361 |
+
"grad_norm": 0.5340840816497803,
|
| 8362 |
+
"learning_rate": 6.795098289008595e-08,
|
| 8363 |
+
"loss": 0.4985,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.9898107714701602,
|
| 8368 |
+
"grad_norm": 0.45380574464797974,
|
| 8369 |
+
"learning_rate": 5.8591302506361135e-08,
|
| 8370 |
+
"loss": 0.4504,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.990642545227698,
|
| 8375 |
+
"grad_norm": 0.4703597128391266,
|
| 8376 |
+
"learning_rate": 4.992467133406731e-08,
|
| 8377 |
+
"loss": 0.4435,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.991474318985236,
|
| 8382 |
+
"grad_norm": 0.46635764837265015,
|
| 8383 |
+
"learning_rate": 4.1951149472441164e-08,
|
| 8384 |
+
"loss": 0.4138,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.9923060927427739,
|
| 8389 |
+
"grad_norm": 0.43708378076553345,
|
| 8390 |
+
"learning_rate": 3.4670792214297476e-08,
|
| 8391 |
+
"loss": 0.3964,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.9931378665003119,
|
| 8396 |
+
"grad_norm": 0.45950600504875183,
|
| 8397 |
+
"learning_rate": 2.808365004569602e-08,
|
| 8398 |
+
"loss": 0.4067,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.9939696402578498,
|
| 8403 |
+
"grad_norm": 0.43873485922813416,
|
| 8404 |
+
"learning_rate": 2.2189768645519694e-08,
|
| 8405 |
+
"loss": 0.4141,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.9948014140153878,
|
| 8410 |
+
"grad_norm": 0.46396079659461975,
|
| 8411 |
+
"learning_rate": 1.6989188885219166e-08,
|
| 8412 |
+
"loss": 0.3859,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.9956331877729258,
|
| 8417 |
+
"grad_norm": 0.4713720977306366,
|
| 8418 |
+
"learning_rate": 1.2481946828502012e-08,
|
| 8419 |
+
"loss": 0.4257,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.9964649615304637,
|
| 8424 |
+
"grad_norm": 0.5061950087547302,
|
| 8425 |
+
"learning_rate": 8.668073731088466e-09,
|
| 8426 |
+
"loss": 0.4706,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.9972967352880017,
|
| 8431 |
+
"grad_norm": 0.4641589820384979,
|
| 8432 |
+
"learning_rate": 5.547596040489378e-09,
|
| 8433 |
+
"loss": 0.4415,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.9981285090455396,
|
| 8438 |
+
"grad_norm": 0.45247629284858704,
|
| 8439 |
+
"learning_rate": 3.1205353958285723e-09,
|
| 8440 |
+
"loss": 0.3908,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.9989602828030776,
|
| 8445 |
+
"grad_norm": 0.4349338114261627,
|
| 8446 |
+
"learning_rate": 1.3869086276985244e-09,
|
| 8447 |
+
"loss": 0.4466,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.9997920565606155,
|
| 8452 |
+
"grad_norm": 0.4767285883426666,
|
| 8453 |
+
"learning_rate": 3.467277580271322e-10,
|
| 8454 |
+
"loss": 0.475,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 1.0006238303181534,
|
| 8459 |
+
"grad_norm": 1.3944224119186401,
|
| 8460 |
+
"learning_rate": 0.0,
|
| 8461 |
+
"loss": 0.6621,
|
| 8462 |
+
"step": 1203
|
| 8463 |
}
|
| 8464 |
],
|
| 8465 |
"logging_steps": 1,
|
|
|
|
| 8474 |
"should_evaluate": false,
|
| 8475 |
"should_log": false,
|
| 8476 |
"should_save": true,
|
| 8477 |
+
"should_training_stop": true
|
| 8478 |
},
|
| 8479 |
"attributes": {}
|
| 8480 |
}
|
| 8481 |
},
|
| 8482 |
+
"total_flos": 4.5976081746886656e+17,
|
| 8483 |
"train_batch_size": 2,
|
| 8484 |
"trial_name": null,
|
| 8485 |
"trial_params": null
|