Training in progress, step 1074, 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 327040
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5f4d52450587f4fa8f738d530dc02f74bedc8aa67258375566da1b71480db5f
|
| 3 |
size 327040
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 739578
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:080422a5ebc8a5201dbdbaae1098e13469b38bd3698b73429fd225dcbb6518a7
|
| 3 |
size 739578
|
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:fdca2119bcae4b80ee49c3f4f99fa78f0d111539f3150b42fe3915937204f0a5
|
| 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 sha256:85fea812480d166bed0e9c88b3d70d26dfa1e02f1b479a6ccde4090471653360
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 358,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5043,6 +5043,2520 @@
|
|
| 5043 |
"eval_samples_per_second": 79.172,
|
| 5044 |
"eval_steps_per_second": 39.586,
|
| 5045 |
"step": 716
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5046 |
}
|
| 5047 |
],
|
| 5048 |
"logging_steps": 1,
|
|
@@ -5062,7 +7576,7 @@
|
|
| 5062 |
"attributes": {}
|
| 5063 |
}
|
| 5064 |
},
|
| 5065 |
-
"total_flos":
|
| 5066 |
"train_batch_size": 2,
|
| 5067 |
"trial_name": null,
|
| 5068 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.04177241037309297,
|
| 5 |
"eval_steps": 358,
|
| 6 |
+
"global_step": 1074,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5043 |
"eval_samples_per_second": 79.172,
|
| 5044 |
"eval_steps_per_second": 39.586,
|
| 5045 |
"step": 716
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.02788716781890843,
|
| 5049 |
+
"grad_norm": 0.08253801614046097,
|
| 5050 |
+
"learning_rate": 0.00010077378357218021,
|
| 5051 |
+
"loss": 11.4232,
|
| 5052 |
+
"step": 717
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.02792606205575489,
|
| 5056 |
+
"grad_norm": 0.13256096839904785,
|
| 5057 |
+
"learning_rate": 0.00010055270525306016,
|
| 5058 |
+
"loss": 11.4315,
|
| 5059 |
+
"step": 718
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.027964956292601345,
|
| 5063 |
+
"grad_norm": 0.17937391996383667,
|
| 5064 |
+
"learning_rate": 0.00010033162423243987,
|
| 5065 |
+
"loss": 11.4851,
|
| 5066 |
+
"step": 719
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.0280038505294478,
|
| 5070 |
+
"grad_norm": 0.20057813823223114,
|
| 5071 |
+
"learning_rate": 0.0001001105415909142,
|
| 5072 |
+
"loss": 11.5311,
|
| 5073 |
+
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.028042744766294255,
|
| 5077 |
+
"grad_norm": 0.08090359717607498,
|
| 5078 |
+
"learning_rate": 9.988945840908583e-05,
|
| 5079 |
+
"loss": 11.4295,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.02808163900314071,
|
| 5084 |
+
"grad_norm": 0.082366943359375,
|
| 5085 |
+
"learning_rate": 9.966837576756016e-05,
|
| 5086 |
+
"loss": 11.4418,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.028120533239987165,
|
| 5091 |
+
"grad_norm": 0.15174053609371185,
|
| 5092 |
+
"learning_rate": 9.944729474693987e-05,
|
| 5093 |
+
"loss": 11.4378,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.02815942747683362,
|
| 5098 |
+
"grad_norm": 0.09956763684749603,
|
| 5099 |
+
"learning_rate": 9.92262164278198e-05,
|
| 5100 |
+
"loss": 11.4256,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.028198321713680075,
|
| 5105 |
+
"grad_norm": 0.06715458631515503,
|
| 5106 |
+
"learning_rate": 9.900514189078155e-05,
|
| 5107 |
+
"loss": 11.444,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.028237215950526532,
|
| 5112 |
+
"grad_norm": 0.10695479065179825,
|
| 5113 |
+
"learning_rate": 9.878407221638816e-05,
|
| 5114 |
+
"loss": 11.4519,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.028276110187372985,
|
| 5119 |
+
"grad_norm": 0.08852813392877579,
|
| 5120 |
+
"learning_rate": 9.85630084851791e-05,
|
| 5121 |
+
"loss": 11.4474,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.028315004424219442,
|
| 5126 |
+
"grad_norm": 0.1369630992412567,
|
| 5127 |
+
"learning_rate": 9.834195177766452e-05,
|
| 5128 |
+
"loss": 11.4339,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.028353898661065895,
|
| 5133 |
+
"grad_norm": 0.09969980269670486,
|
| 5134 |
+
"learning_rate": 9.81209031743205e-05,
|
| 5135 |
+
"loss": 11.4579,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.028392792897912352,
|
| 5140 |
+
"grad_norm": 0.15188638865947723,
|
| 5141 |
+
"learning_rate": 9.789986375558331e-05,
|
| 5142 |
+
"loss": 11.5502,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.028431687134758805,
|
| 5147 |
+
"grad_norm": 0.08629871159791946,
|
| 5148 |
+
"learning_rate": 9.767883460184443e-05,
|
| 5149 |
+
"loss": 11.4381,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.028470581371605262,
|
| 5154 |
+
"grad_norm": 0.19648370146751404,
|
| 5155 |
+
"learning_rate": 9.74578167934452e-05,
|
| 5156 |
+
"loss": 11.4772,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.02850947560845172,
|
| 5161 |
+
"grad_norm": 0.10891472548246384,
|
| 5162 |
+
"learning_rate": 9.723681141067139e-05,
|
| 5163 |
+
"loss": 11.4388,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.028548369845298172,
|
| 5168 |
+
"grad_norm": 0.1772705316543579,
|
| 5169 |
+
"learning_rate": 9.701581953374815e-05,
|
| 5170 |
+
"loss": 11.4936,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.02858726408214463,
|
| 5175 |
+
"grad_norm": 0.29078036546707153,
|
| 5176 |
+
"learning_rate": 9.679484224283449e-05,
|
| 5177 |
+
"loss": 11.5355,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.028626158318991082,
|
| 5182 |
+
"grad_norm": 0.10268472880125046,
|
| 5183 |
+
"learning_rate": 9.657388061801828e-05,
|
| 5184 |
+
"loss": 11.4682,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.02866505255583754,
|
| 5189 |
+
"grad_norm": 0.11252088099718094,
|
| 5190 |
+
"learning_rate": 9.635293573931072e-05,
|
| 5191 |
+
"loss": 11.4544,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.028703946792683992,
|
| 5196 |
+
"grad_norm": 0.0970296636223793,
|
| 5197 |
+
"learning_rate": 9.613200868664112e-05,
|
| 5198 |
+
"loss": 11.4352,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.02874284102953045,
|
| 5203 |
+
"grad_norm": 0.13067781925201416,
|
| 5204 |
+
"learning_rate": 9.591110053985182e-05,
|
| 5205 |
+
"loss": 11.4347,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.028781735266376906,
|
| 5210 |
+
"grad_norm": 0.17787784337997437,
|
| 5211 |
+
"learning_rate": 9.569021237869258e-05,
|
| 5212 |
+
"loss": 11.4483,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.02882062950322336,
|
| 5217 |
+
"grad_norm": 0.15255478024482727,
|
| 5218 |
+
"learning_rate": 9.54693452828156e-05,
|
| 5219 |
+
"loss": 11.4333,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.028859523740069816,
|
| 5224 |
+
"grad_norm": 0.09174659103155136,
|
| 5225 |
+
"learning_rate": 9.524850033177e-05,
|
| 5226 |
+
"loss": 11.4265,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.02889841797691627,
|
| 5231 |
+
"grad_norm": 0.1348842978477478,
|
| 5232 |
+
"learning_rate": 9.502767860499672e-05,
|
| 5233 |
+
"loss": 11.4743,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.028937312213762726,
|
| 5238 |
+
"grad_norm": 0.17099758982658386,
|
| 5239 |
+
"learning_rate": 9.48068811818233e-05,
|
| 5240 |
+
"loss": 11.4782,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.02897620645060918,
|
| 5245 |
+
"grad_norm": 0.0950162261724472,
|
| 5246 |
+
"learning_rate": 9.458610914145826e-05,
|
| 5247 |
+
"loss": 11.4753,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.029015100687455636,
|
| 5252 |
+
"grad_norm": 0.11666002869606018,
|
| 5253 |
+
"learning_rate": 9.436536356298624e-05,
|
| 5254 |
+
"loss": 11.4198,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.029053994924302093,
|
| 5259 |
+
"grad_norm": 0.1315421611070633,
|
| 5260 |
+
"learning_rate": 9.414464552536242e-05,
|
| 5261 |
+
"loss": 11.4877,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.029092889161148546,
|
| 5266 |
+
"grad_norm": 0.15513554215431213,
|
| 5267 |
+
"learning_rate": 9.39239561074075e-05,
|
| 5268 |
+
"loss": 11.4861,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.029131783397995003,
|
| 5273 |
+
"grad_norm": 0.1369619071483612,
|
| 5274 |
+
"learning_rate": 9.370329638780213e-05,
|
| 5275 |
+
"loss": 11.5185,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.029170677634841456,
|
| 5280 |
+
"grad_norm": 0.10951586067676544,
|
| 5281 |
+
"learning_rate": 9.348266744508191e-05,
|
| 5282 |
+
"loss": 11.4876,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.029209571871687913,
|
| 5287 |
+
"grad_norm": 0.0937129557132721,
|
| 5288 |
+
"learning_rate": 9.326207035763202e-05,
|
| 5289 |
+
"loss": 11.4312,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.029248466108534366,
|
| 5294 |
+
"grad_norm": 0.1251746565103531,
|
| 5295 |
+
"learning_rate": 9.304150620368188e-05,
|
| 5296 |
+
"loss": 11.4397,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.029287360345380823,
|
| 5301 |
+
"grad_norm": 0.13234980404376984,
|
| 5302 |
+
"learning_rate": 9.282097606129998e-05,
|
| 5303 |
+
"loss": 11.4528,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.02932625458222728,
|
| 5308 |
+
"grad_norm": 0.16654668748378754,
|
| 5309 |
+
"learning_rate": 9.260048100838848e-05,
|
| 5310 |
+
"loss": 11.4857,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.029365148819073733,
|
| 5315 |
+
"grad_norm": 0.40939077734947205,
|
| 5316 |
+
"learning_rate": 9.238002212267821e-05,
|
| 5317 |
+
"loss": 11.4327,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.02940404305592019,
|
| 5322 |
+
"grad_norm": 0.14264138042926788,
|
| 5323 |
+
"learning_rate": 9.215960048172299e-05,
|
| 5324 |
+
"loss": 11.4855,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.029442937292766643,
|
| 5329 |
+
"grad_norm": 0.14503702521324158,
|
| 5330 |
+
"learning_rate": 9.193921716289482e-05,
|
| 5331 |
+
"loss": 11.4595,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.0294818315296131,
|
| 5336 |
+
"grad_norm": 0.11121979355812073,
|
| 5337 |
+
"learning_rate": 9.171887324337826e-05,
|
| 5338 |
+
"loss": 11.4433,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.029520725766459553,
|
| 5343 |
+
"grad_norm": 0.11435031145811081,
|
| 5344 |
+
"learning_rate": 9.149856980016529e-05,
|
| 5345 |
+
"loss": 11.4518,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.02955962000330601,
|
| 5350 |
+
"grad_norm": 0.21833956241607666,
|
| 5351 |
+
"learning_rate": 9.127830791005015e-05,
|
| 5352 |
+
"loss": 11.452,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.029598514240152467,
|
| 5357 |
+
"grad_norm": 0.12180177867412567,
|
| 5358 |
+
"learning_rate": 9.105808864962384e-05,
|
| 5359 |
+
"loss": 11.4448,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.02963740847699892,
|
| 5364 |
+
"grad_norm": 0.13487689197063446,
|
| 5365 |
+
"learning_rate": 9.083791309526908e-05,
|
| 5366 |
+
"loss": 11.4377,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.029676302713845377,
|
| 5371 |
+
"grad_norm": 0.17410726845264435,
|
| 5372 |
+
"learning_rate": 9.061778232315505e-05,
|
| 5373 |
+
"loss": 11.441,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.02971519695069183,
|
| 5378 |
+
"grad_norm": 0.12577226758003235,
|
| 5379 |
+
"learning_rate": 9.039769740923183e-05,
|
| 5380 |
+
"loss": 11.4545,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.029754091187538287,
|
| 5385 |
+
"grad_norm": 0.10234140604734421,
|
| 5386 |
+
"learning_rate": 9.017765942922554e-05,
|
| 5387 |
+
"loss": 11.4476,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.02979298542438474,
|
| 5392 |
+
"grad_norm": 0.17257264256477356,
|
| 5393 |
+
"learning_rate": 8.995766945863277e-05,
|
| 5394 |
+
"loss": 11.4764,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.029831879661231197,
|
| 5399 |
+
"grad_norm": 0.1410018652677536,
|
| 5400 |
+
"learning_rate": 8.973772857271558e-05,
|
| 5401 |
+
"loss": 11.4282,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.029870773898077654,
|
| 5406 |
+
"grad_norm": 0.11423639208078384,
|
| 5407 |
+
"learning_rate": 8.951783784649602e-05,
|
| 5408 |
+
"loss": 11.4617,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.029909668134924107,
|
| 5413 |
+
"grad_norm": 0.10588809102773666,
|
| 5414 |
+
"learning_rate": 8.929799835475093e-05,
|
| 5415 |
+
"loss": 11.4645,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.029948562371770564,
|
| 5420 |
+
"grad_norm": 0.18268363177776337,
|
| 5421 |
+
"learning_rate": 8.907821117200694e-05,
|
| 5422 |
+
"loss": 11.4531,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.029987456608617017,
|
| 5427 |
+
"grad_norm": 0.10537099093198776,
|
| 5428 |
+
"learning_rate": 8.88584773725347e-05,
|
| 5429 |
+
"loss": 11.4033,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.030026350845463474,
|
| 5434 |
+
"grad_norm": 0.13354121148586273,
|
| 5435 |
+
"learning_rate": 8.863879803034421e-05,
|
| 5436 |
+
"loss": 11.4562,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.030065245082309927,
|
| 5441 |
+
"grad_norm": 0.09870657324790955,
|
| 5442 |
+
"learning_rate": 8.841917421917912e-05,
|
| 5443 |
+
"loss": 11.4425,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.030104139319156384,
|
| 5448 |
+
"grad_norm": 0.08798401057720184,
|
| 5449 |
+
"learning_rate": 8.819960701251175e-05,
|
| 5450 |
+
"loss": 11.4367,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.03014303355600284,
|
| 5455 |
+
"grad_norm": 0.139079287648201,
|
| 5456 |
+
"learning_rate": 8.798009748353765e-05,
|
| 5457 |
+
"loss": 11.4877,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.030181927792849294,
|
| 5462 |
+
"grad_norm": 0.1492471843957901,
|
| 5463 |
+
"learning_rate": 8.776064670517059e-05,
|
| 5464 |
+
"loss": 11.4751,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.03022082202969575,
|
| 5469 |
+
"grad_norm": 0.15318681299686432,
|
| 5470 |
+
"learning_rate": 8.754125575003708e-05,
|
| 5471 |
+
"loss": 11.4705,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.030259716266542204,
|
| 5476 |
+
"grad_norm": 0.14461193978786469,
|
| 5477 |
+
"learning_rate": 8.732192569047126e-05,
|
| 5478 |
+
"loss": 11.4745,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.03029861050338866,
|
| 5483 |
+
"grad_norm": 0.14277328550815582,
|
| 5484 |
+
"learning_rate": 8.710265759850963e-05,
|
| 5485 |
+
"loss": 11.4334,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.030337504740235114,
|
| 5490 |
+
"grad_norm": 0.09315398335456848,
|
| 5491 |
+
"learning_rate": 8.688345254588578e-05,
|
| 5492 |
+
"loss": 11.4297,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.03037639897708157,
|
| 5497 |
+
"grad_norm": 0.08005926758050919,
|
| 5498 |
+
"learning_rate": 8.666431160402518e-05,
|
| 5499 |
+
"loss": 11.4431,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.030415293213928028,
|
| 5504 |
+
"grad_norm": 0.10046279430389404,
|
| 5505 |
+
"learning_rate": 8.644523584404003e-05,
|
| 5506 |
+
"loss": 11.4787,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.03045418745077448,
|
| 5511 |
+
"grad_norm": 0.14663808047771454,
|
| 5512 |
+
"learning_rate": 8.62262263367238e-05,
|
| 5513 |
+
"loss": 11.4878,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.030493081687620938,
|
| 5518 |
+
"grad_norm": 0.17167599499225616,
|
| 5519 |
+
"learning_rate": 8.600728415254624e-05,
|
| 5520 |
+
"loss": 11.5243,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.03053197592446739,
|
| 5525 |
+
"grad_norm": 0.11615641415119171,
|
| 5526 |
+
"learning_rate": 8.578841036164794e-05,
|
| 5527 |
+
"loss": 11.4355,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.030570870161313848,
|
| 5532 |
+
"grad_norm": 0.08393886685371399,
|
| 5533 |
+
"learning_rate": 8.55696060338353e-05,
|
| 5534 |
+
"loss": 11.4553,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.0306097643981603,
|
| 5539 |
+
"grad_norm": 0.16598129272460938,
|
| 5540 |
+
"learning_rate": 8.535087223857508e-05,
|
| 5541 |
+
"loss": 11.446,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.030648658635006758,
|
| 5546 |
+
"grad_norm": 0.1524801105260849,
|
| 5547 |
+
"learning_rate": 8.513221004498946e-05,
|
| 5548 |
+
"loss": 11.4439,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.030687552871853215,
|
| 5553 |
+
"grad_norm": 0.1081952229142189,
|
| 5554 |
+
"learning_rate": 8.491362052185053e-05,
|
| 5555 |
+
"loss": 11.4627,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.030726447108699668,
|
| 5560 |
+
"grad_norm": 0.13826072216033936,
|
| 5561 |
+
"learning_rate": 8.469510473757513e-05,
|
| 5562 |
+
"loss": 11.4835,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.030765341345546125,
|
| 5567 |
+
"grad_norm": 0.1568002849817276,
|
| 5568 |
+
"learning_rate": 8.447666376021985e-05,
|
| 5569 |
+
"loss": 11.4659,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.030804235582392578,
|
| 5574 |
+
"grad_norm": 0.09875816106796265,
|
| 5575 |
+
"learning_rate": 8.425829865747549e-05,
|
| 5576 |
+
"loss": 11.4209,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.030843129819239035,
|
| 5581 |
+
"grad_norm": 0.18653692305088043,
|
| 5582 |
+
"learning_rate": 8.404001049666211e-05,
|
| 5583 |
+
"loss": 11.4606,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.030882024056085488,
|
| 5588 |
+
"grad_norm": 0.19002626836299896,
|
| 5589 |
+
"learning_rate": 8.382180034472353e-05,
|
| 5590 |
+
"loss": 11.4729,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.030920918292931945,
|
| 5595 |
+
"grad_norm": 0.09468155354261398,
|
| 5596 |
+
"learning_rate": 8.360366926822247e-05,
|
| 5597 |
+
"loss": 11.4465,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.0309598125297784,
|
| 5602 |
+
"grad_norm": 0.12331503629684448,
|
| 5603 |
+
"learning_rate": 8.338561833333506e-05,
|
| 5604 |
+
"loss": 11.4442,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.030998706766624855,
|
| 5609 |
+
"grad_norm": 0.1488187164068222,
|
| 5610 |
+
"learning_rate": 8.316764860584567e-05,
|
| 5611 |
+
"loss": 11.4704,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.03103760100347131,
|
| 5616 |
+
"grad_norm": 0.08257905393838882,
|
| 5617 |
+
"learning_rate": 8.294976115114184e-05,
|
| 5618 |
+
"loss": 11.4504,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.031076495240317765,
|
| 5623 |
+
"grad_norm": 0.11658480763435364,
|
| 5624 |
+
"learning_rate": 8.273195703420884e-05,
|
| 5625 |
+
"loss": 11.4836,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.03111538947716422,
|
| 5630 |
+
"grad_norm": 0.11987362056970596,
|
| 5631 |
+
"learning_rate": 8.25142373196248e-05,
|
| 5632 |
+
"loss": 11.4406,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.031154283714010675,
|
| 5637 |
+
"grad_norm": 0.1509653776884079,
|
| 5638 |
+
"learning_rate": 8.229660307155518e-05,
|
| 5639 |
+
"loss": 11.497,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.03119317795085713,
|
| 5644 |
+
"grad_norm": 0.11979538947343826,
|
| 5645 |
+
"learning_rate": 8.207905535374767e-05,
|
| 5646 |
+
"loss": 11.4391,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.03123207218770359,
|
| 5651 |
+
"grad_norm": 0.18861103057861328,
|
| 5652 |
+
"learning_rate": 8.186159522952716e-05,
|
| 5653 |
+
"loss": 11.4934,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.031270966424550045,
|
| 5658 |
+
"grad_norm": 0.06771805882453918,
|
| 5659 |
+
"learning_rate": 8.164422376179023e-05,
|
| 5660 |
+
"loss": 11.442,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.031309860661396495,
|
| 5665 |
+
"grad_norm": 0.28224071860313416,
|
| 5666 |
+
"learning_rate": 8.142694201300027e-05,
|
| 5667 |
+
"loss": 11.4551,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.03134875489824295,
|
| 5672 |
+
"grad_norm": 0.15878629684448242,
|
| 5673 |
+
"learning_rate": 8.120975104518203e-05,
|
| 5674 |
+
"loss": 11.4591,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.03138764913508941,
|
| 5679 |
+
"grad_norm": 0.16484294831752777,
|
| 5680 |
+
"learning_rate": 8.099265191991665e-05,
|
| 5681 |
+
"loss": 11.5227,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.031426543371935865,
|
| 5686 |
+
"grad_norm": 0.19311536848545074,
|
| 5687 |
+
"learning_rate": 8.077564569833632e-05,
|
| 5688 |
+
"loss": 11.4931,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.03146543760878232,
|
| 5693 |
+
"grad_norm": 0.11146111786365509,
|
| 5694 |
+
"learning_rate": 8.055873344111906e-05,
|
| 5695 |
+
"loss": 11.4418,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.03150433184562877,
|
| 5700 |
+
"grad_norm": 0.12124677747488022,
|
| 5701 |
+
"learning_rate": 8.03419162084838e-05,
|
| 5702 |
+
"loss": 11.4164,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.03154322608247523,
|
| 5707 |
+
"grad_norm": 0.09567411988973618,
|
| 5708 |
+
"learning_rate": 8.012519506018476e-05,
|
| 5709 |
+
"loss": 11.4232,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.031582120319321685,
|
| 5714 |
+
"grad_norm": 0.0929986760020256,
|
| 5715 |
+
"learning_rate": 7.990857105550675e-05,
|
| 5716 |
+
"loss": 11.4187,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.03162101455616814,
|
| 5721 |
+
"grad_norm": 0.13243938982486725,
|
| 5722 |
+
"learning_rate": 7.969204525325962e-05,
|
| 5723 |
+
"loss": 11.4515,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.03165990879301459,
|
| 5728 |
+
"grad_norm": 0.1833588033914566,
|
| 5729 |
+
"learning_rate": 7.947561871177327e-05,
|
| 5730 |
+
"loss": 11.4609,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.03169880302986105,
|
| 5735 |
+
"grad_norm": 0.1955696940422058,
|
| 5736 |
+
"learning_rate": 7.92592924888925e-05,
|
| 5737 |
+
"loss": 11.5276,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.031737697266707506,
|
| 5742 |
+
"grad_norm": 0.12613818049430847,
|
| 5743 |
+
"learning_rate": 7.904306764197168e-05,
|
| 5744 |
+
"loss": 11.4316,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.03177659150355396,
|
| 5749 |
+
"grad_norm": 0.13009190559387207,
|
| 5750 |
+
"learning_rate": 7.882694522786974e-05,
|
| 5751 |
+
"loss": 11.4452,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.03181548574040042,
|
| 5756 |
+
"grad_norm": 0.10820070654153824,
|
| 5757 |
+
"learning_rate": 7.861092630294484e-05,
|
| 5758 |
+
"loss": 11.4772,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.03185437997724687,
|
| 5763 |
+
"grad_norm": 0.15169255435466766,
|
| 5764 |
+
"learning_rate": 7.839501192304947e-05,
|
| 5765 |
+
"loss": 11.4619,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.031893274214093326,
|
| 5770 |
+
"grad_norm": 0.13594582676887512,
|
| 5771 |
+
"learning_rate": 7.817920314352503e-05,
|
| 5772 |
+
"loss": 11.4355,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.03193216845093978,
|
| 5777 |
+
"grad_norm": 0.1923629492521286,
|
| 5778 |
+
"learning_rate": 7.796350101919671e-05,
|
| 5779 |
+
"loss": 11.4255,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.03197106268778624,
|
| 5784 |
+
"grad_norm": 0.10615510493516922,
|
| 5785 |
+
"learning_rate": 7.774790660436858e-05,
|
| 5786 |
+
"loss": 11.4286,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.032009956924632696,
|
| 5791 |
+
"grad_norm": 0.08784908056259155,
|
| 5792 |
+
"learning_rate": 7.753242095281802e-05,
|
| 5793 |
+
"loss": 11.4347,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.032048851161479146,
|
| 5798 |
+
"grad_norm": 0.11678597331047058,
|
| 5799 |
+
"learning_rate": 7.731704511779099e-05,
|
| 5800 |
+
"loss": 11.4353,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.0320877453983256,
|
| 5805 |
+
"grad_norm": 0.14838770031929016,
|
| 5806 |
+
"learning_rate": 7.710178015199655e-05,
|
| 5807 |
+
"loss": 11.465,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.03212663963517206,
|
| 5812 |
+
"grad_norm": 0.09852113574743271,
|
| 5813 |
+
"learning_rate": 7.688662710760194e-05,
|
| 5814 |
+
"loss": 11.4075,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.032165533872018516,
|
| 5819 |
+
"grad_norm": 0.09276288747787476,
|
| 5820 |
+
"learning_rate": 7.667158703622739e-05,
|
| 5821 |
+
"loss": 11.4557,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.032204428108864966,
|
| 5826 |
+
"grad_norm": 0.10213444381952286,
|
| 5827 |
+
"learning_rate": 7.645666098894082e-05,
|
| 5828 |
+
"loss": 11.4035,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.03224332234571142,
|
| 5833 |
+
"grad_norm": 0.1934029459953308,
|
| 5834 |
+
"learning_rate": 7.624185001625292e-05,
|
| 5835 |
+
"loss": 11.4614,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.03228221658255788,
|
| 5840 |
+
"grad_norm": 0.09062288701534271,
|
| 5841 |
+
"learning_rate": 7.602715516811183e-05,
|
| 5842 |
+
"loss": 11.4192,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.032321110819404336,
|
| 5847 |
+
"grad_norm": 0.12071729451417923,
|
| 5848 |
+
"learning_rate": 7.581257749389828e-05,
|
| 5849 |
+
"loss": 11.4517,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.03236000505625079,
|
| 5854 |
+
"grad_norm": 0.18437331914901733,
|
| 5855 |
+
"learning_rate": 7.559811804242008e-05,
|
| 5856 |
+
"loss": 11.5739,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.03239889929309724,
|
| 5861 |
+
"grad_norm": 0.1351187527179718,
|
| 5862 |
+
"learning_rate": 7.538377786190724e-05,
|
| 5863 |
+
"loss": 11.467,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.0324377935299437,
|
| 5868 |
+
"grad_norm": 0.11380963027477264,
|
| 5869 |
+
"learning_rate": 7.516955800000696e-05,
|
| 5870 |
+
"loss": 11.4142,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.032476687766790156,
|
| 5875 |
+
"grad_norm": 0.20516403019428253,
|
| 5876 |
+
"learning_rate": 7.49554595037781e-05,
|
| 5877 |
+
"loss": 11.4732,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.03251558200363661,
|
| 5882 |
+
"grad_norm": 0.11636844277381897,
|
| 5883 |
+
"learning_rate": 7.474148341968652e-05,
|
| 5884 |
+
"loss": 11.4619,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.03255447624048307,
|
| 5889 |
+
"grad_norm": 0.10400319844484329,
|
| 5890 |
+
"learning_rate": 7.45276307935996e-05,
|
| 5891 |
+
"loss": 11.4187,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.03259337047732952,
|
| 5896 |
+
"grad_norm": 0.13016699254512787,
|
| 5897 |
+
"learning_rate": 7.431390267078142e-05,
|
| 5898 |
+
"loss": 11.4434,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.032632264714175976,
|
| 5903 |
+
"grad_norm": 0.07082528620958328,
|
| 5904 |
+
"learning_rate": 7.410030009588744e-05,
|
| 5905 |
+
"loss": 11.4225,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.03267115895102243,
|
| 5910 |
+
"grad_norm": 0.13742923736572266,
|
| 5911 |
+
"learning_rate": 7.388682411295946e-05,
|
| 5912 |
+
"loss": 11.4645,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.03271005318786889,
|
| 5917 |
+
"grad_norm": 0.11265579611063004,
|
| 5918 |
+
"learning_rate": 7.367347576542059e-05,
|
| 5919 |
+
"loss": 11.4307,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.03274894742471534,
|
| 5924 |
+
"grad_norm": 0.1302904635667801,
|
| 5925 |
+
"learning_rate": 7.346025609606996e-05,
|
| 5926 |
+
"loss": 11.4813,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.032787841661561797,
|
| 5931 |
+
"grad_norm": 0.12135647982358932,
|
| 5932 |
+
"learning_rate": 7.324716614707793e-05,
|
| 5933 |
+
"loss": 11.4464,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.03282673589840825,
|
| 5938 |
+
"grad_norm": 0.11820655316114426,
|
| 5939 |
+
"learning_rate": 7.30342069599807e-05,
|
| 5940 |
+
"loss": 11.4554,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.03286563013525471,
|
| 5945 |
+
"grad_norm": 0.10697997361421585,
|
| 5946 |
+
"learning_rate": 7.282137957567528e-05,
|
| 5947 |
+
"loss": 11.4417,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.03290452437210117,
|
| 5952 |
+
"grad_norm": 0.09324019402265549,
|
| 5953 |
+
"learning_rate": 7.260868503441466e-05,
|
| 5954 |
+
"loss": 11.4145,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.03294341860894762,
|
| 5959 |
+
"grad_norm": 0.09825508296489716,
|
| 5960 |
+
"learning_rate": 7.23961243758023e-05,
|
| 5961 |
+
"loss": 11.4253,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.03298231284579407,
|
| 5966 |
+
"grad_norm": 0.1215229257941246,
|
| 5967 |
+
"learning_rate": 7.218369863878744e-05,
|
| 5968 |
+
"loss": 11.4605,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.03302120708264053,
|
| 5973 |
+
"grad_norm": 0.15956827998161316,
|
| 5974 |
+
"learning_rate": 7.197140886165969e-05,
|
| 5975 |
+
"loss": 11.4586,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.03306010131948699,
|
| 5980 |
+
"grad_norm": 0.11994804441928864,
|
| 5981 |
+
"learning_rate": 7.175925608204428e-05,
|
| 5982 |
+
"loss": 11.4953,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.033098995556333444,
|
| 5987 |
+
"grad_norm": 0.12891148030757904,
|
| 5988 |
+
"learning_rate": 7.154724133689677e-05,
|
| 5989 |
+
"loss": 11.4618,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.033137889793179894,
|
| 5994 |
+
"grad_norm": 0.16294585168361664,
|
| 5995 |
+
"learning_rate": 7.133536566249794e-05,
|
| 5996 |
+
"loss": 11.4927,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.03317678403002635,
|
| 6001 |
+
"grad_norm": 0.1092517077922821,
|
| 6002 |
+
"learning_rate": 7.112363009444903e-05,
|
| 6003 |
+
"loss": 11.4409,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.03321567826687281,
|
| 6008 |
+
"grad_norm": 0.12596727907657623,
|
| 6009 |
+
"learning_rate": 7.091203566766622e-05,
|
| 6010 |
+
"loss": 11.4381,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.033254572503719264,
|
| 6015 |
+
"grad_norm": 0.22151672840118408,
|
| 6016 |
+
"learning_rate": 7.070058341637605e-05,
|
| 6017 |
+
"loss": 11.4834,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.033293466740565714,
|
| 6022 |
+
"grad_norm": 0.16800400614738464,
|
| 6023 |
+
"learning_rate": 7.048927437410999e-05,
|
| 6024 |
+
"loss": 11.4446,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.03333236097741217,
|
| 6029 |
+
"grad_norm": 0.11195861548185349,
|
| 6030 |
+
"learning_rate": 7.027810957369957e-05,
|
| 6031 |
+
"loss": 11.4279,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.03337125521425863,
|
| 6036 |
+
"grad_norm": 0.08265417069196701,
|
| 6037 |
+
"learning_rate": 7.006709004727139e-05,
|
| 6038 |
+
"loss": 11.4501,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.033410149451105084,
|
| 6043 |
+
"grad_norm": 0.12703652679920197,
|
| 6044 |
+
"learning_rate": 6.985621682624183e-05,
|
| 6045 |
+
"loss": 11.48,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.03344904368795154,
|
| 6050 |
+
"grad_norm": 0.14403915405273438,
|
| 6051 |
+
"learning_rate": 6.964549094131229e-05,
|
| 6052 |
+
"loss": 11.4881,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.03348793792479799,
|
| 6057 |
+
"grad_norm": 0.07923027873039246,
|
| 6058 |
+
"learning_rate": 6.943491342246393e-05,
|
| 6059 |
+
"loss": 11.4331,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.03352683216164445,
|
| 6064 |
+
"grad_norm": 0.16718092560768127,
|
| 6065 |
+
"learning_rate": 6.922448529895282e-05,
|
| 6066 |
+
"loss": 11.481,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.033565726398490904,
|
| 6071 |
+
"grad_norm": 0.12445984780788422,
|
| 6072 |
+
"learning_rate": 6.901420759930473e-05,
|
| 6073 |
+
"loss": 11.4598,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.03360462063533736,
|
| 6078 |
+
"grad_norm": 0.10626041144132614,
|
| 6079 |
+
"learning_rate": 6.880408135131022e-05,
|
| 6080 |
+
"loss": 11.4551,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.03364351487218382,
|
| 6085 |
+
"grad_norm": 0.13476891815662384,
|
| 6086 |
+
"learning_rate": 6.85941075820197e-05,
|
| 6087 |
+
"loss": 11.4829,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.03368240910903027,
|
| 6092 |
+
"grad_norm": 0.13352851569652557,
|
| 6093 |
+
"learning_rate": 6.838428731773806e-05,
|
| 6094 |
+
"loss": 11.485,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.033721303345876724,
|
| 6099 |
+
"grad_norm": 0.11239789426326752,
|
| 6100 |
+
"learning_rate": 6.817462158402015e-05,
|
| 6101 |
+
"loss": 11.4483,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.03376019758272318,
|
| 6106 |
+
"grad_norm": 0.08315054327249527,
|
| 6107 |
+
"learning_rate": 6.796511140566531e-05,
|
| 6108 |
+
"loss": 11.4452,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.03379909181956964,
|
| 6113 |
+
"grad_norm": 0.22786268591880798,
|
| 6114 |
+
"learning_rate": 6.775575780671266e-05,
|
| 6115 |
+
"loss": 11.5424,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.03383798605641609,
|
| 6120 |
+
"grad_norm": 0.17151309549808502,
|
| 6121 |
+
"learning_rate": 6.754656181043602e-05,
|
| 6122 |
+
"loss": 11.4625,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.033876880293262544,
|
| 6127 |
+
"grad_norm": 0.13574554026126862,
|
| 6128 |
+
"learning_rate": 6.733752443933878e-05,
|
| 6129 |
+
"loss": 11.4441,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.033915774530109,
|
| 6134 |
+
"grad_norm": 0.13274209201335907,
|
| 6135 |
+
"learning_rate": 6.712864671514911e-05,
|
| 6136 |
+
"loss": 11.4399,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.03395466876695546,
|
| 6141 |
+
"grad_norm": 0.13931576907634735,
|
| 6142 |
+
"learning_rate": 6.691992965881475e-05,
|
| 6143 |
+
"loss": 11.4427,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.033993563003801915,
|
| 6148 |
+
"grad_norm": 0.1982814520597458,
|
| 6149 |
+
"learning_rate": 6.671137429049827e-05,
|
| 6150 |
+
"loss": 11.4683,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.034032457240648364,
|
| 6155 |
+
"grad_norm": 0.09911420941352844,
|
| 6156 |
+
"learning_rate": 6.650298162957183e-05,
|
| 6157 |
+
"loss": 11.4214,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.03407135147749482,
|
| 6162 |
+
"grad_norm": 0.17978575825691223,
|
| 6163 |
+
"learning_rate": 6.629475269461234e-05,
|
| 6164 |
+
"loss": 11.4612,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.03411024571434128,
|
| 6169 |
+
"grad_norm": 0.16942349076271057,
|
| 6170 |
+
"learning_rate": 6.608668850339652e-05,
|
| 6171 |
+
"loss": 11.4971,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.034149139951187735,
|
| 6176 |
+
"grad_norm": 0.14842084050178528,
|
| 6177 |
+
"learning_rate": 6.587879007289576e-05,
|
| 6178 |
+
"loss": 11.4384,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.03418803418803419,
|
| 6183 |
+
"grad_norm": 0.12057257443666458,
|
| 6184 |
+
"learning_rate": 6.567105841927132e-05,
|
| 6185 |
+
"loss": 11.4452,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.03422692842488064,
|
| 6190 |
+
"grad_norm": 0.12537352740764618,
|
| 6191 |
+
"learning_rate": 6.546349455786926e-05,
|
| 6192 |
+
"loss": 11.4557,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.0342658226617271,
|
| 6197 |
+
"grad_norm": 0.18904034793376923,
|
| 6198 |
+
"learning_rate": 6.525609950321552e-05,
|
| 6199 |
+
"loss": 11.4493,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.034304716898573555,
|
| 6204 |
+
"grad_norm": 0.18003982305526733,
|
| 6205 |
+
"learning_rate": 6.50488742690109e-05,
|
| 6206 |
+
"loss": 11.4539,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.03434361113542001,
|
| 6211 |
+
"grad_norm": 0.15979625284671783,
|
| 6212 |
+
"learning_rate": 6.484181986812625e-05,
|
| 6213 |
+
"loss": 11.553,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.03438250537226646,
|
| 6218 |
+
"grad_norm": 0.10028652101755142,
|
| 6219 |
+
"learning_rate": 6.463493731259742e-05,
|
| 6220 |
+
"loss": 11.4449,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.03442139960911292,
|
| 6225 |
+
"grad_norm": 0.22036820650100708,
|
| 6226 |
+
"learning_rate": 6.442822761362015e-05,
|
| 6227 |
+
"loss": 11.5304,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.034460293845959375,
|
| 6232 |
+
"grad_norm": 0.1596294641494751,
|
| 6233 |
+
"learning_rate": 6.422169178154556e-05,
|
| 6234 |
+
"loss": 11.4617,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.03449918808280583,
|
| 6239 |
+
"grad_norm": 0.23585407435894012,
|
| 6240 |
+
"learning_rate": 6.40153308258747e-05,
|
| 6241 |
+
"loss": 11.5596,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.03453808231965229,
|
| 6246 |
+
"grad_norm": 0.134441077709198,
|
| 6247 |
+
"learning_rate": 6.3809145755254e-05,
|
| 6248 |
+
"loss": 11.4394,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.03457697655649874,
|
| 6253 |
+
"grad_norm": 0.08511587977409363,
|
| 6254 |
+
"learning_rate": 6.360313757747022e-05,
|
| 6255 |
+
"loss": 11.4562,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.034615870793345195,
|
| 6260 |
+
"grad_norm": 0.17673659324645996,
|
| 6261 |
+
"learning_rate": 6.33973072994454e-05,
|
| 6262 |
+
"loss": 11.4896,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.03465476503019165,
|
| 6267 |
+
"grad_norm": 0.0889359638094902,
|
| 6268 |
+
"learning_rate": 6.319165592723218e-05,
|
| 6269 |
+
"loss": 11.4523,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.03469365926703811,
|
| 6274 |
+
"grad_norm": 0.15608401596546173,
|
| 6275 |
+
"learning_rate": 6.298618446600856e-05,
|
| 6276 |
+
"loss": 11.4688,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.034732553503884565,
|
| 6281 |
+
"grad_norm": 0.17957602441310883,
|
| 6282 |
+
"learning_rate": 6.278089392007343e-05,
|
| 6283 |
+
"loss": 11.4594,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.034771447740731015,
|
| 6288 |
+
"grad_norm": 0.23120824992656708,
|
| 6289 |
+
"learning_rate": 6.257578529284113e-05,
|
| 6290 |
+
"loss": 11.5103,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.03481034197757747,
|
| 6295 |
+
"grad_norm": 0.10850831866264343,
|
| 6296 |
+
"learning_rate": 6.237085958683704e-05,
|
| 6297 |
+
"loss": 11.467,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.03484923621442393,
|
| 6302 |
+
"grad_norm": 0.06931223720312119,
|
| 6303 |
+
"learning_rate": 6.216611780369238e-05,
|
| 6304 |
+
"loss": 11.4337,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.034888130451270385,
|
| 6309 |
+
"grad_norm": 0.1407434195280075,
|
| 6310 |
+
"learning_rate": 6.196156094413934e-05,
|
| 6311 |
+
"loss": 11.4687,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.034927024688116835,
|
| 6316 |
+
"grad_norm": 0.17006301879882812,
|
| 6317 |
+
"learning_rate": 6.175719000800637e-05,
|
| 6318 |
+
"loss": 11.4503,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.03496591892496329,
|
| 6323 |
+
"grad_norm": 0.10757336020469666,
|
| 6324 |
+
"learning_rate": 6.155300599421306e-05,
|
| 6325 |
+
"loss": 11.4638,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.03500481316180975,
|
| 6330 |
+
"grad_norm": 0.10327316075563431,
|
| 6331 |
+
"learning_rate": 6.134900990076541e-05,
|
| 6332 |
+
"loss": 11.4722,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.035043707398656206,
|
| 6337 |
+
"grad_norm": 0.10495606809854507,
|
| 6338 |
+
"learning_rate": 6.114520272475088e-05,
|
| 6339 |
+
"loss": 11.4596,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.03508260163550266,
|
| 6344 |
+
"grad_norm": 0.17221322655677795,
|
| 6345 |
+
"learning_rate": 6.094158546233359e-05,
|
| 6346 |
+
"loss": 11.4826,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.03512149587234911,
|
| 6351 |
+
"grad_norm": 0.16179487109184265,
|
| 6352 |
+
"learning_rate": 6.073815910874942e-05,
|
| 6353 |
+
"loss": 11.4734,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.03516039010919557,
|
| 6358 |
+
"grad_norm": 0.1274636834859848,
|
| 6359 |
+
"learning_rate": 6.053492465830097e-05,
|
| 6360 |
+
"loss": 11.4439,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.035199284346042026,
|
| 6365 |
+
"grad_norm": 0.11228005588054657,
|
| 6366 |
+
"learning_rate": 6.0331883104353115e-05,
|
| 6367 |
+
"loss": 11.4514,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.03523817858288848,
|
| 6372 |
+
"grad_norm": 0.17040938138961792,
|
| 6373 |
+
"learning_rate": 6.012903543932766e-05,
|
| 6374 |
+
"loss": 11.4617,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.03527707281973493,
|
| 6379 |
+
"grad_norm": 0.15520794689655304,
|
| 6380 |
+
"learning_rate": 5.9926382654698857e-05,
|
| 6381 |
+
"loss": 11.4857,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.03531596705658139,
|
| 6386 |
+
"grad_norm": 0.12189356982707977,
|
| 6387 |
+
"learning_rate": 5.972392574098844e-05,
|
| 6388 |
+
"loss": 11.4562,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.035354861293427846,
|
| 6393 |
+
"grad_norm": 0.0995606780052185,
|
| 6394 |
+
"learning_rate": 5.952166568776062e-05,
|
| 6395 |
+
"loss": 11.4251,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.0353937555302743,
|
| 6400 |
+
"grad_norm": 0.1529882550239563,
|
| 6401 |
+
"learning_rate": 5.931960348361759e-05,
|
| 6402 |
+
"loss": 11.4374,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.03543264976712076,
|
| 6407 |
+
"grad_norm": 0.12336838990449905,
|
| 6408 |
+
"learning_rate": 5.9117740116194375e-05,
|
| 6409 |
+
"loss": 11.4488,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.03547154400396721,
|
| 6414 |
+
"grad_norm": 0.11592377722263336,
|
| 6415 |
+
"learning_rate": 5.8916076572154165e-05,
|
| 6416 |
+
"loss": 11.4422,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.035510438240813666,
|
| 6421 |
+
"grad_norm": 0.15878193080425262,
|
| 6422 |
+
"learning_rate": 5.871461383718344e-05,
|
| 6423 |
+
"loss": 11.4852,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.03554933247766012,
|
| 6428 |
+
"grad_norm": 0.14143121242523193,
|
| 6429 |
+
"learning_rate": 5.851335289598722e-05,
|
| 6430 |
+
"loss": 11.4936,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.03558822671450658,
|
| 6435 |
+
"grad_norm": 0.20466415584087372,
|
| 6436 |
+
"learning_rate": 5.831229473228418e-05,
|
| 6437 |
+
"loss": 11.5021,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.035627120951353036,
|
| 6442 |
+
"grad_norm": 0.1679922640323639,
|
| 6443 |
+
"learning_rate": 5.811144032880182e-05,
|
| 6444 |
+
"loss": 11.4565,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.035666015188199486,
|
| 6449 |
+
"grad_norm": 0.11905571073293686,
|
| 6450 |
+
"learning_rate": 5.791079066727174e-05,
|
| 6451 |
+
"loss": 11.4342,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.03570490942504594,
|
| 6456 |
+
"grad_norm": 0.08226253092288971,
|
| 6457 |
+
"learning_rate": 5.7710346728424836e-05,
|
| 6458 |
+
"loss": 11.4062,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.0357438036618924,
|
| 6463 |
+
"grad_norm": 0.1482539027929306,
|
| 6464 |
+
"learning_rate": 5.751010949198643e-05,
|
| 6465 |
+
"loss": 11.4366,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.035782697898738856,
|
| 6470 |
+
"grad_norm": 0.10693041235208511,
|
| 6471 |
+
"learning_rate": 5.7310079936671545e-05,
|
| 6472 |
+
"loss": 11.4555,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.035821592135585306,
|
| 6477 |
+
"grad_norm": 0.16853336989879608,
|
| 6478 |
+
"learning_rate": 5.711025904018013e-05,
|
| 6479 |
+
"loss": 11.4451,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.03586048637243176,
|
| 6484 |
+
"grad_norm": 0.16642871499061584,
|
| 6485 |
+
"learning_rate": 5.691064777919223e-05,
|
| 6486 |
+
"loss": 11.4352,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.03589938060927822,
|
| 6491 |
+
"grad_norm": 0.11272871494293213,
|
| 6492 |
+
"learning_rate": 5.6711247129363156e-05,
|
| 6493 |
+
"loss": 11.4952,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.035938274846124676,
|
| 6498 |
+
"grad_norm": 0.13608944416046143,
|
| 6499 |
+
"learning_rate": 5.651205806531903e-05,
|
| 6500 |
+
"loss": 11.4288,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.03597716908297113,
|
| 6505 |
+
"grad_norm": 0.14277823269367218,
|
| 6506 |
+
"learning_rate": 5.631308156065152e-05,
|
| 6507 |
+
"loss": 11.4076,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.03601606331981758,
|
| 6512 |
+
"grad_norm": 0.10945692658424377,
|
| 6513 |
+
"learning_rate": 5.611431858791348e-05,
|
| 6514 |
+
"loss": 11.4533,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.03605495755666404,
|
| 6519 |
+
"grad_norm": 0.10961952060461044,
|
| 6520 |
+
"learning_rate": 5.59157701186142e-05,
|
| 6521 |
+
"loss": 11.473,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.0360938517935105,
|
| 6526 |
+
"grad_norm": 0.08197829127311707,
|
| 6527 |
+
"learning_rate": 5.571743712321422e-05,
|
| 6528 |
+
"loss": 11.4289,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.03613274603035695,
|
| 6533 |
+
"grad_norm": 0.11842351406812668,
|
| 6534 |
+
"learning_rate": 5.551932057112115e-05,
|
| 6535 |
+
"loss": 11.4473,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.03617164026720341,
|
| 6540 |
+
"grad_norm": 0.14072228968143463,
|
| 6541 |
+
"learning_rate": 5.532142143068455e-05,
|
| 6542 |
+
"loss": 11.4301,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.03621053450404986,
|
| 6547 |
+
"grad_norm": 0.1937941163778305,
|
| 6548 |
+
"learning_rate": 5.512374066919137e-05,
|
| 6549 |
+
"loss": 11.4606,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.03624942874089632,
|
| 6554 |
+
"grad_norm": 0.13401161134243011,
|
| 6555 |
+
"learning_rate": 5.492627925286113e-05,
|
| 6556 |
+
"loss": 11.4766,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.036288322977742773,
|
| 6561 |
+
"grad_norm": 0.11659156531095505,
|
| 6562 |
+
"learning_rate": 5.4729038146841294e-05,
|
| 6563 |
+
"loss": 11.4341,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.03632721721458923,
|
| 6568 |
+
"grad_norm": 0.14816398918628693,
|
| 6569 |
+
"learning_rate": 5.453201831520245e-05,
|
| 6570 |
+
"loss": 11.4842,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.03636611145143568,
|
| 6575 |
+
"grad_norm": 0.15731480717658997,
|
| 6576 |
+
"learning_rate": 5.4335220720933664e-05,
|
| 6577 |
+
"loss": 11.4512,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.03640500568828214,
|
| 6582 |
+
"grad_norm": 0.10614141076803207,
|
| 6583 |
+
"learning_rate": 5.4138646325937813e-05,
|
| 6584 |
+
"loss": 11.4329,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.036443899925128594,
|
| 6589 |
+
"grad_norm": 0.13888613879680634,
|
| 6590 |
+
"learning_rate": 5.3942296091026656e-05,
|
| 6591 |
+
"loss": 11.4717,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.03648279416197505,
|
| 6596 |
+
"grad_norm": 0.14382483065128326,
|
| 6597 |
+
"learning_rate": 5.37461709759165e-05,
|
| 6598 |
+
"loss": 11.4275,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.03652168839882151,
|
| 6603 |
+
"grad_norm": 0.11295495927333832,
|
| 6604 |
+
"learning_rate": 5.3550271939223195e-05,
|
| 6605 |
+
"loss": 11.4522,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.03656058263566796,
|
| 6610 |
+
"grad_norm": 0.28914937376976013,
|
| 6611 |
+
"learning_rate": 5.335459993845764e-05,
|
| 6612 |
+
"loss": 11.5041,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.036599476872514414,
|
| 6617 |
+
"grad_norm": 0.16712334752082825,
|
| 6618 |
+
"learning_rate": 5.3159155930021e-05,
|
| 6619 |
+
"loss": 11.4382,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.03663837110936087,
|
| 6624 |
+
"grad_norm": 0.22522681951522827,
|
| 6625 |
+
"learning_rate": 5.2963940869200056e-05,
|
| 6626 |
+
"loss": 11.5194,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.03667726534620733,
|
| 6631 |
+
"grad_norm": 0.10965728759765625,
|
| 6632 |
+
"learning_rate": 5.276895571016257e-05,
|
| 6633 |
+
"loss": 11.4299,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.036716159583053784,
|
| 6638 |
+
"grad_norm": 0.11578682065010071,
|
| 6639 |
+
"learning_rate": 5.257420140595257e-05,
|
| 6640 |
+
"loss": 11.4538,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.036755053819900234,
|
| 6645 |
+
"grad_norm": 0.10459594428539276,
|
| 6646 |
+
"learning_rate": 5.237967890848574e-05,
|
| 6647 |
+
"loss": 11.4419,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.03679394805674669,
|
| 6652 |
+
"grad_norm": 0.1656852513551712,
|
| 6653 |
+
"learning_rate": 5.218538916854473e-05,
|
| 6654 |
+
"loss": 11.4236,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.03683284229359315,
|
| 6659 |
+
"grad_norm": 0.1385713666677475,
|
| 6660 |
+
"learning_rate": 5.1991333135774525e-05,
|
| 6661 |
+
"loss": 11.4504,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.036871736530439604,
|
| 6666 |
+
"grad_norm": 0.15063555538654327,
|
| 6667 |
+
"learning_rate": 5.179751175867784e-05,
|
| 6668 |
+
"loss": 11.4632,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.036910630767286054,
|
| 6673 |
+
"grad_norm": 0.13096244633197784,
|
| 6674 |
+
"learning_rate": 5.1603925984610326e-05,
|
| 6675 |
+
"loss": 11.4364,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.03694952500413251,
|
| 6680 |
+
"grad_norm": 0.11817897111177444,
|
| 6681 |
+
"learning_rate": 5.141057675977619e-05,
|
| 6682 |
+
"loss": 11.4361,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.03698841924097897,
|
| 6687 |
+
"grad_norm": 0.10221054404973984,
|
| 6688 |
+
"learning_rate": 5.1217465029223375e-05,
|
| 6689 |
+
"loss": 11.4402,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.037027313477825424,
|
| 6694 |
+
"grad_norm": 0.13744771480560303,
|
| 6695 |
+
"learning_rate": 5.102459173683903e-05,
|
| 6696 |
+
"loss": 11.5,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.03706620771467188,
|
| 6701 |
+
"grad_norm": 0.19383054971694946,
|
| 6702 |
+
"learning_rate": 5.0831957825344865e-05,
|
| 6703 |
+
"loss": 11.5256,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.03710510195151833,
|
| 6708 |
+
"grad_norm": 0.1514645218849182,
|
| 6709 |
+
"learning_rate": 5.063956423629255e-05,
|
| 6710 |
+
"loss": 11.4661,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.03714399618836479,
|
| 6715 |
+
"grad_norm": 0.11683838069438934,
|
| 6716 |
+
"learning_rate": 5.044741191005908e-05,
|
| 6717 |
+
"loss": 11.4504,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.037182890425211244,
|
| 6722 |
+
"grad_norm": 0.11084909737110138,
|
| 6723 |
+
"learning_rate": 5.025550178584226e-05,
|
| 6724 |
+
"loss": 11.4211,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.0372217846620577,
|
| 6729 |
+
"grad_norm": 0.20357000827789307,
|
| 6730 |
+
"learning_rate": 5.0063834801656084e-05,
|
| 6731 |
+
"loss": 11.4345,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.03726067889890416,
|
| 6736 |
+
"grad_norm": 0.1357165426015854,
|
| 6737 |
+
"learning_rate": 4.9872411894325965e-05,
|
| 6738 |
+
"loss": 11.4705,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.03729957313575061,
|
| 6743 |
+
"grad_norm": 0.15270249545574188,
|
| 6744 |
+
"learning_rate": 4.9681233999484564e-05,
|
| 6745 |
+
"loss": 11.4765,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.037338467372597064,
|
| 6750 |
+
"grad_norm": 0.097058966755867,
|
| 6751 |
+
"learning_rate": 4.9490302051566886e-05,
|
| 6752 |
+
"loss": 11.4642,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.03737736160944352,
|
| 6757 |
+
"grad_norm": 0.11222375929355621,
|
| 6758 |
+
"learning_rate": 4.929961698380564e-05,
|
| 6759 |
+
"loss": 11.4392,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.03741625584628998,
|
| 6764 |
+
"grad_norm": 0.11173541843891144,
|
| 6765 |
+
"learning_rate": 4.910917972822713e-05,
|
| 6766 |
+
"loss": 11.4275,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.03745515008313643,
|
| 6771 |
+
"grad_norm": 0.21483619511127472,
|
| 6772 |
+
"learning_rate": 4.891899121564615e-05,
|
| 6773 |
+
"loss": 11.4803,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.037494044319982885,
|
| 6778 |
+
"grad_norm": 0.10671492666006088,
|
| 6779 |
+
"learning_rate": 4.872905237566183e-05,
|
| 6780 |
+
"loss": 11.4381,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.03753293855682934,
|
| 6785 |
+
"grad_norm": 0.12410018593072891,
|
| 6786 |
+
"learning_rate": 4.853936413665294e-05,
|
| 6787 |
+
"loss": 11.4423,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.0375718327936758,
|
| 6792 |
+
"grad_norm": 0.13656941056251526,
|
| 6793 |
+
"learning_rate": 4.8349927425773345e-05,
|
| 6794 |
+
"loss": 11.4282,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.037610727030522255,
|
| 6799 |
+
"grad_norm": 0.1547286957502365,
|
| 6800 |
+
"learning_rate": 4.8160743168947496e-05,
|
| 6801 |
+
"loss": 11.4462,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.037649621267368705,
|
| 6806 |
+
"grad_norm": 0.10143474489450455,
|
| 6807 |
+
"learning_rate": 4.797181229086594e-05,
|
| 6808 |
+
"loss": 11.4454,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.03768851550421516,
|
| 6813 |
+
"grad_norm": 0.13370105624198914,
|
| 6814 |
+
"learning_rate": 4.7783135714980744e-05,
|
| 6815 |
+
"loss": 11.4257,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.03772740974106162,
|
| 6820 |
+
"grad_norm": 0.16253124177455902,
|
| 6821 |
+
"learning_rate": 4.7594714363500915e-05,
|
| 6822 |
+
"loss": 11.4393,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.037766303977908075,
|
| 6827 |
+
"grad_norm": 0.14151079952716827,
|
| 6828 |
+
"learning_rate": 4.7406549157388156e-05,
|
| 6829 |
+
"loss": 11.4231,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.03780519821475453,
|
| 6834 |
+
"grad_norm": 0.13136602938175201,
|
| 6835 |
+
"learning_rate": 4.721864101635211e-05,
|
| 6836 |
+
"loss": 11.4503,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.03784409245160098,
|
| 6841 |
+
"grad_norm": 0.11752691119909286,
|
| 6842 |
+
"learning_rate": 4.70309908588458e-05,
|
| 6843 |
+
"loss": 11.431,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.03788298668844744,
|
| 6848 |
+
"grad_norm": 0.13689513504505157,
|
| 6849 |
+
"learning_rate": 4.6843599602061583e-05,
|
| 6850 |
+
"loss": 11.4344,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.037921880925293895,
|
| 6855 |
+
"grad_norm": 0.1035035029053688,
|
| 6856 |
+
"learning_rate": 4.665646816192606e-05,
|
| 6857 |
+
"loss": 11.4391,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.03796077516214035,
|
| 6862 |
+
"grad_norm": 0.1925671100616455,
|
| 6863 |
+
"learning_rate": 4.646959745309609e-05,
|
| 6864 |
+
"loss": 11.459,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.0379996693989868,
|
| 6869 |
+
"grad_norm": 0.17556695640087128,
|
| 6870 |
+
"learning_rate": 4.62829883889541e-05,
|
| 6871 |
+
"loss": 11.4547,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.03803856363583326,
|
| 6876 |
+
"grad_norm": 0.15916649997234344,
|
| 6877 |
+
"learning_rate": 4.609664188160362e-05,
|
| 6878 |
+
"loss": 11.4699,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.038077457872679715,
|
| 6883 |
+
"grad_norm": 0.15165077149868011,
|
| 6884 |
+
"learning_rate": 4.591055884186489e-05,
|
| 6885 |
+
"loss": 11.4279,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.03811635210952617,
|
| 6890 |
+
"grad_norm": 0.13335153460502625,
|
| 6891 |
+
"learning_rate": 4.572474017927038e-05,
|
| 6892 |
+
"loss": 11.4491,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.03815524634637263,
|
| 6897 |
+
"grad_norm": 0.12981079518795013,
|
| 6898 |
+
"learning_rate": 4.553918680206037e-05,
|
| 6899 |
+
"loss": 11.4268,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.03819414058321908,
|
| 6904 |
+
"grad_norm": 0.17829671502113342,
|
| 6905 |
+
"learning_rate": 4.535389961717834e-05,
|
| 6906 |
+
"loss": 11.4539,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.038233034820065535,
|
| 6911 |
+
"grad_norm": 0.17136840522289276,
|
| 6912 |
+
"learning_rate": 4.516887953026691e-05,
|
| 6913 |
+
"loss": 11.5136,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.03827192905691199,
|
| 6918 |
+
"grad_norm": 0.13877429068088531,
|
| 6919 |
+
"learning_rate": 4.498412744566305e-05,
|
| 6920 |
+
"loss": 11.4558,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.03831082329375845,
|
| 6925 |
+
"grad_norm": 0.16534513235092163,
|
| 6926 |
+
"learning_rate": 4.47996442663937e-05,
|
| 6927 |
+
"loss": 11.4867,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.038349717530604906,
|
| 6932 |
+
"grad_norm": 0.1574074625968933,
|
| 6933 |
+
"learning_rate": 4.461543089417173e-05,
|
| 6934 |
+
"loss": 11.4415,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.038388611767451356,
|
| 6939 |
+
"grad_norm": 0.2112518548965454,
|
| 6940 |
+
"learning_rate": 4.443148822939095e-05,
|
| 6941 |
+
"loss": 11.4709,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.03842750600429781,
|
| 6946 |
+
"grad_norm": 0.18481174111366272,
|
| 6947 |
+
"learning_rate": 4.42478171711222e-05,
|
| 6948 |
+
"loss": 11.4723,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.03846640024114427,
|
| 6953 |
+
"grad_norm": 0.1394161880016327,
|
| 6954 |
+
"learning_rate": 4.406441861710871e-05,
|
| 6955 |
+
"loss": 11.4503,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.038505294477990726,
|
| 6960 |
+
"grad_norm": 0.11776059120893478,
|
| 6961 |
+
"learning_rate": 4.388129346376178e-05,
|
| 6962 |
+
"loss": 11.4752,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.038544188714837176,
|
| 6967 |
+
"grad_norm": 0.1907920092344284,
|
| 6968 |
+
"learning_rate": 4.369844260615635e-05,
|
| 6969 |
+
"loss": 11.4634,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.03858308295168363,
|
| 6974 |
+
"grad_norm": 0.16755948960781097,
|
| 6975 |
+
"learning_rate": 4.351586693802674e-05,
|
| 6976 |
+
"loss": 11.4734,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.03862197718853009,
|
| 6981 |
+
"grad_norm": 0.10538428276777267,
|
| 6982 |
+
"learning_rate": 4.333356735176218e-05,
|
| 6983 |
+
"loss": 11.431,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.038660871425376546,
|
| 6988 |
+
"grad_norm": 0.18529963493347168,
|
| 6989 |
+
"learning_rate": 4.315154473840235e-05,
|
| 6990 |
+
"loss": 11.5198,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.038699765662223,
|
| 6995 |
+
"grad_norm": 0.1970740407705307,
|
| 6996 |
+
"learning_rate": 4.296979998763338e-05,
|
| 6997 |
+
"loss": 11.4301,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.03873865989906945,
|
| 7002 |
+
"grad_norm": 0.14860780537128448,
|
| 7003 |
+
"learning_rate": 4.278833398778306e-05,
|
| 7004 |
+
"loss": 11.4689,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.03877755413591591,
|
| 7009 |
+
"grad_norm": 0.16582021117210388,
|
| 7010 |
+
"learning_rate": 4.260714762581677e-05,
|
| 7011 |
+
"loss": 11.4262,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.038816448372762366,
|
| 7016 |
+
"grad_norm": 0.11632680147886276,
|
| 7017 |
+
"learning_rate": 4.242624178733322e-05,
|
| 7018 |
+
"loss": 11.4374,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.03885534260960882,
|
| 7023 |
+
"grad_norm": 0.20491375029087067,
|
| 7024 |
+
"learning_rate": 4.224561735655977e-05,
|
| 7025 |
+
"loss": 11.4342,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.03889423684645528,
|
| 7030 |
+
"grad_norm": 0.10806465148925781,
|
| 7031 |
+
"learning_rate": 4.206527521634842e-05,
|
| 7032 |
+
"loss": 11.4343,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.03893313108330173,
|
| 7037 |
+
"grad_norm": 0.1917140781879425,
|
| 7038 |
+
"learning_rate": 4.1885216248171425e-05,
|
| 7039 |
+
"loss": 11.4662,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.038972025320148186,
|
| 7044 |
+
"grad_norm": 0.1264260858297348,
|
| 7045 |
+
"learning_rate": 4.17054413321169e-05,
|
| 7046 |
+
"loss": 11.4654,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.03901091955699464,
|
| 7051 |
+
"grad_norm": 0.10900475829839706,
|
| 7052 |
+
"learning_rate": 4.152595134688464e-05,
|
| 7053 |
+
"loss": 11.4422,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.0390498137938411,
|
| 7058 |
+
"grad_norm": 0.10250162333250046,
|
| 7059 |
+
"learning_rate": 4.13467471697817e-05,
|
| 7060 |
+
"loss": 11.4218,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.03908870803068755,
|
| 7065 |
+
"grad_norm": 0.11095335334539413,
|
| 7066 |
+
"learning_rate": 4.1167829676718225e-05,
|
| 7067 |
+
"loss": 11.4274,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.039127602267534006,
|
| 7072 |
+
"grad_norm": 0.12225431948900223,
|
| 7073 |
+
"learning_rate": 4.0989199742203e-05,
|
| 7074 |
+
"loss": 11.4403,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.03916649650438046,
|
| 7079 |
+
"grad_norm": 0.10829450935125351,
|
| 7080 |
+
"learning_rate": 4.081085823933949e-05,
|
| 7081 |
+
"loss": 11.4363,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.03920539074122692,
|
| 7086 |
+
"grad_norm": 0.15065525472164154,
|
| 7087 |
+
"learning_rate": 4.0632806039821145e-05,
|
| 7088 |
+
"loss": 11.4778,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.03924428497807338,
|
| 7093 |
+
"grad_norm": 0.13061833381652832,
|
| 7094 |
+
"learning_rate": 4.045504401392749e-05,
|
| 7095 |
+
"loss": 11.4449,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.039283179214919826,
|
| 7100 |
+
"grad_norm": 0.13270919024944305,
|
| 7101 |
+
"learning_rate": 4.02775730305198e-05,
|
| 7102 |
+
"loss": 11.4269,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.03932207345176628,
|
| 7107 |
+
"grad_norm": 0.11030741780996323,
|
| 7108 |
+
"learning_rate": 4.010039395703664e-05,
|
| 7109 |
+
"loss": 11.4468,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.03936096768861274,
|
| 7114 |
+
"grad_norm": 0.17667675018310547,
|
| 7115 |
+
"learning_rate": 3.99235076594899e-05,
|
| 7116 |
+
"loss": 11.4739,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.0393998619254592,
|
| 7121 |
+
"grad_norm": 0.17058613896369934,
|
| 7122 |
+
"learning_rate": 3.9746915002460405e-05,
|
| 7123 |
+
"loss": 11.4725,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.03943875616230565,
|
| 7128 |
+
"grad_norm": 0.1783858984708786,
|
| 7129 |
+
"learning_rate": 3.9570616849093745e-05,
|
| 7130 |
+
"loss": 11.4287,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.0394776503991521,
|
| 7135 |
+
"grad_norm": 0.13282090425491333,
|
| 7136 |
+
"learning_rate": 3.939461406109605e-05,
|
| 7137 |
+
"loss": 11.443,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.03951654463599856,
|
| 7142 |
+
"grad_norm": 0.1711062490940094,
|
| 7143 |
+
"learning_rate": 3.921890749872973e-05,
|
| 7144 |
+
"loss": 11.4178,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.03955543887284502,
|
| 7149 |
+
"grad_norm": 0.10078589618206024,
|
| 7150 |
+
"learning_rate": 3.90434980208094e-05,
|
| 7151 |
+
"loss": 11.438,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.039594333109691474,
|
| 7156 |
+
"grad_norm": 0.11982541531324387,
|
| 7157 |
+
"learning_rate": 3.8868386484697417e-05,
|
| 7158 |
+
"loss": 11.4521,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.03963322734653792,
|
| 7163 |
+
"grad_norm": 0.13653124868869781,
|
| 7164 |
+
"learning_rate": 3.869357374630011e-05,
|
| 7165 |
+
"loss": 11.4497,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.03967212158338438,
|
| 7170 |
+
"grad_norm": 0.15596039593219757,
|
| 7171 |
+
"learning_rate": 3.851906066006311e-05,
|
| 7172 |
+
"loss": 11.4674,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.03971101582023084,
|
| 7177 |
+
"grad_norm": 0.13571912050247192,
|
| 7178 |
+
"learning_rate": 3.834484807896753e-05,
|
| 7179 |
+
"loss": 11.4646,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.039749910057077294,
|
| 7184 |
+
"grad_norm": 0.14839304983615875,
|
| 7185 |
+
"learning_rate": 3.817093685452578e-05,
|
| 7186 |
+
"loss": 11.4354,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.03978880429392375,
|
| 7191 |
+
"grad_norm": 0.11423840373754501,
|
| 7192 |
+
"learning_rate": 3.799732783677711e-05,
|
| 7193 |
+
"loss": 11.4233,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.0398276985307702,
|
| 7198 |
+
"grad_norm": 0.12816986441612244,
|
| 7199 |
+
"learning_rate": 3.7824021874283745e-05,
|
| 7200 |
+
"loss": 11.4109,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.03986659276761666,
|
| 7205 |
+
"grad_norm": 0.1389741748571396,
|
| 7206 |
+
"learning_rate": 3.7651019814126654e-05,
|
| 7207 |
+
"loss": 11.4411,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.039905487004463114,
|
| 7212 |
+
"grad_norm": 0.2206103503704071,
|
| 7213 |
+
"learning_rate": 3.747832250190139e-05,
|
| 7214 |
+
"loss": 11.484,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.03994438124130957,
|
| 7219 |
+
"grad_norm": 0.1348106563091278,
|
| 7220 |
+
"learning_rate": 3.730593078171396e-05,
|
| 7221 |
+
"loss": 11.4859,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.03998327547815603,
|
| 7226 |
+
"grad_norm": 0.1366932988166809,
|
| 7227 |
+
"learning_rate": 3.713384549617669e-05,
|
| 7228 |
+
"loss": 11.4695,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.04002216971500248,
|
| 7233 |
+
"grad_norm": 0.13749776780605316,
|
| 7234 |
+
"learning_rate": 3.696206748640416e-05,
|
| 7235 |
+
"loss": 11.4675,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.040061063951848934,
|
| 7240 |
+
"grad_norm": 0.11645928025245667,
|
| 7241 |
+
"learning_rate": 3.679059759200901e-05,
|
| 7242 |
+
"loss": 11.4201,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.04009995818869539,
|
| 7247 |
+
"grad_norm": 0.20807428658008575,
|
| 7248 |
+
"learning_rate": 3.661943665109796e-05,
|
| 7249 |
+
"loss": 11.4443,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.04013885242554185,
|
| 7254 |
+
"grad_norm": 0.09833139926195145,
|
| 7255 |
+
"learning_rate": 3.6448585500267485e-05,
|
| 7256 |
+
"loss": 11.4419,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.0401777466623883,
|
| 7261 |
+
"grad_norm": 0.1817658692598343,
|
| 7262 |
+
"learning_rate": 3.627804497460001e-05,
|
| 7263 |
+
"loss": 11.4636,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.040216640899234754,
|
| 7268 |
+
"grad_norm": 0.13411729037761688,
|
| 7269 |
+
"learning_rate": 3.610781590765966e-05,
|
| 7270 |
+
"loss": 11.4606,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.04025553513608121,
|
| 7275 |
+
"grad_norm": 0.0920453742146492,
|
| 7276 |
+
"learning_rate": 3.593789913148818e-05,
|
| 7277 |
+
"loss": 11.4563,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.04029442937292767,
|
| 7282 |
+
"grad_norm": 0.1194140836596489,
|
| 7283 |
+
"learning_rate": 3.576829547660098e-05,
|
| 7284 |
+
"loss": 11.4411,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.040333323609774124,
|
| 7289 |
+
"grad_norm": 0.16156767308712006,
|
| 7290 |
+
"learning_rate": 3.559900577198292e-05,
|
| 7291 |
+
"loss": 11.4888,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.040372217846620574,
|
| 7296 |
+
"grad_norm": 0.25374388694763184,
|
| 7297 |
+
"learning_rate": 3.5430030845084406e-05,
|
| 7298 |
+
"loss": 11.5209,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.04041111208346703,
|
| 7303 |
+
"grad_norm": 0.12839582562446594,
|
| 7304 |
+
"learning_rate": 3.5261371521817244e-05,
|
| 7305 |
+
"loss": 11.4657,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.04045000632031349,
|
| 7310 |
+
"grad_norm": 0.1393585056066513,
|
| 7311 |
+
"learning_rate": 3.509302862655064e-05,
|
| 7312 |
+
"loss": 11.4392,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.040488900557159944,
|
| 7317 |
+
"grad_norm": 0.17082121968269348,
|
| 7318 |
+
"learning_rate": 3.4925002982107205e-05,
|
| 7319 |
+
"loss": 11.4479,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.0405277947940064,
|
| 7324 |
+
"grad_norm": 0.11696211993694305,
|
| 7325 |
+
"learning_rate": 3.475729540975885e-05,
|
| 7326 |
+
"loss": 11.4272,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.04056668903085285,
|
| 7331 |
+
"grad_norm": 0.21628519892692566,
|
| 7332 |
+
"learning_rate": 3.45899067292229e-05,
|
| 7333 |
+
"loss": 11.4565,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.04060558326769931,
|
| 7338 |
+
"grad_norm": 0.2241363525390625,
|
| 7339 |
+
"learning_rate": 3.442283775865783e-05,
|
| 7340 |
+
"loss": 11.436,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.040644477504545765,
|
| 7345 |
+
"grad_norm": 0.16380998492240906,
|
| 7346 |
+
"learning_rate": 3.4256089314659725e-05,
|
| 7347 |
+
"loss": 11.4788,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.04068337174139222,
|
| 7352 |
+
"grad_norm": 0.12565840780735016,
|
| 7353 |
+
"learning_rate": 3.408966221225773e-05,
|
| 7354 |
+
"loss": 11.4718,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.04072226597823867,
|
| 7359 |
+
"grad_norm": 0.1578199714422226,
|
| 7360 |
+
"learning_rate": 3.3923557264910514e-05,
|
| 7361 |
+
"loss": 11.4613,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.04076116021508513,
|
| 7366 |
+
"grad_norm": 0.11667878925800323,
|
| 7367 |
+
"learning_rate": 3.3757775284502045e-05,
|
| 7368 |
+
"loss": 11.4321,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.040800054451931585,
|
| 7373 |
+
"grad_norm": 0.1548505276441574,
|
| 7374 |
+
"learning_rate": 3.3592317081337755e-05,
|
| 7375 |
+
"loss": 11.4568,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.04083894868877804,
|
| 7380 |
+
"grad_norm": 0.14588265120983124,
|
| 7381 |
+
"learning_rate": 3.342718346414049e-05,
|
| 7382 |
+
"loss": 11.4962,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.0408778429256245,
|
| 7387 |
+
"grad_norm": 0.12974365055561066,
|
| 7388 |
+
"learning_rate": 3.3262375240046596e-05,
|
| 7389 |
+
"loss": 11.4519,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.04091673716247095,
|
| 7394 |
+
"grad_norm": 0.1877010613679886,
|
| 7395 |
+
"learning_rate": 3.309789321460202e-05,
|
| 7396 |
+
"loss": 11.4317,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.040955631399317405,
|
| 7401 |
+
"grad_norm": 0.1269165575504303,
|
| 7402 |
+
"learning_rate": 3.293373819175816e-05,
|
| 7403 |
+
"loss": 11.4635,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.04099452563616386,
|
| 7408 |
+
"grad_norm": 0.12434379011392593,
|
| 7409 |
+
"learning_rate": 3.276991097386831e-05,
|
| 7410 |
+
"loss": 11.429,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.04103341987301032,
|
| 7415 |
+
"grad_norm": 0.12698589265346527,
|
| 7416 |
+
"learning_rate": 3.260641236168346e-05,
|
| 7417 |
+
"loss": 11.4503,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.041072314109856775,
|
| 7422 |
+
"grad_norm": 0.18467244505882263,
|
| 7423 |
+
"learning_rate": 3.24432431543483e-05,
|
| 7424 |
+
"loss": 11.4486,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.041111208346703225,
|
| 7429 |
+
"grad_norm": 0.14477095007896423,
|
| 7430 |
+
"learning_rate": 3.228040414939773e-05,
|
| 7431 |
+
"loss": 11.4645,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.04115010258354968,
|
| 7436 |
+
"grad_norm": 0.1388678103685379,
|
| 7437 |
+
"learning_rate": 3.211789614275241e-05,
|
| 7438 |
+
"loss": 11.4697,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.04118899682039614,
|
| 7443 |
+
"grad_norm": 0.14132258296012878,
|
| 7444 |
+
"learning_rate": 3.1955719928715345e-05,
|
| 7445 |
+
"loss": 11.4368,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.041227891057242595,
|
| 7450 |
+
"grad_norm": 0.18010205030441284,
|
| 7451 |
+
"learning_rate": 3.1793876299967816e-05,
|
| 7452 |
+
"loss": 11.4437,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.041266785294089045,
|
| 7457 |
+
"grad_norm": 0.10433457046747208,
|
| 7458 |
+
"learning_rate": 3.163236604756535e-05,
|
| 7459 |
+
"loss": 11.4442,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.0413056795309355,
|
| 7464 |
+
"grad_norm": 0.16652584075927734,
|
| 7465 |
+
"learning_rate": 3.1471189960934146e-05,
|
| 7466 |
+
"loss": 11.4379,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.04134457376778196,
|
| 7471 |
+
"grad_norm": 0.12524031102657318,
|
| 7472 |
+
"learning_rate": 3.1310348827867006e-05,
|
| 7473 |
+
"loss": 11.4358,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.041383468004628415,
|
| 7478 |
+
"grad_norm": 0.12246005982160568,
|
| 7479 |
+
"learning_rate": 3.114984343451963e-05,
|
| 7480 |
+
"loss": 11.4099,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.04142236224147487,
|
| 7485 |
+
"grad_norm": 0.13616223633289337,
|
| 7486 |
+
"learning_rate": 3.098967456540652e-05,
|
| 7487 |
+
"loss": 11.4322,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.04146125647832132,
|
| 7492 |
+
"grad_norm": 0.145111083984375,
|
| 7493 |
+
"learning_rate": 3.082984300339756e-05,
|
| 7494 |
+
"loss": 11.5065,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.04150015071516778,
|
| 7499 |
+
"grad_norm": 0.13896699249744415,
|
| 7500 |
+
"learning_rate": 3.0670349529713816e-05,
|
| 7501 |
+
"loss": 11.4505,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.041539044952014235,
|
| 7506 |
+
"grad_norm": 0.11143442243337631,
|
| 7507 |
+
"learning_rate": 3.051119492392379e-05,
|
| 7508 |
+
"loss": 11.4411,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.04157793918886069,
|
| 7513 |
+
"grad_norm": 0.15392428636550903,
|
| 7514 |
+
"learning_rate": 3.0352379963939883e-05,
|
| 7515 |
+
"loss": 11.4361,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.04161683342570715,
|
| 7520 |
+
"grad_norm": 0.16626837849617004,
|
| 7521 |
+
"learning_rate": 3.0193905426014146e-05,
|
| 7522 |
+
"loss": 11.4083,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.0416557276625536,
|
| 7527 |
+
"grad_norm": 0.2432129830121994,
|
| 7528 |
+
"learning_rate": 3.003577208473488e-05,
|
| 7529 |
+
"loss": 11.5145,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.041694621899400056,
|
| 7534 |
+
"grad_norm": 0.10259351879358292,
|
| 7535 |
+
"learning_rate": 2.9877980713022648e-05,
|
| 7536 |
+
"loss": 11.4465,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.04173351613624651,
|
| 7541 |
+
"grad_norm": 0.22118544578552246,
|
| 7542 |
+
"learning_rate": 2.9720532082126518e-05,
|
| 7543 |
+
"loss": 11.4629,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.04177241037309297,
|
| 7548 |
+
"grad_norm": 0.21239879727363586,
|
| 7549 |
+
"learning_rate": 2.9563426961620367e-05,
|
| 7550 |
+
"loss": 11.4424,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.04177241037309297,
|
| 7555 |
+
"eval_loss": 11.48007583618164,
|
| 7556 |
+
"eval_runtime": 139.1239,
|
| 7557 |
+
"eval_samples_per_second": 77.816,
|
| 7558 |
+
"eval_steps_per_second": 38.908,
|
| 7559 |
+
"step": 1074
|
| 7560 |
}
|
| 7561 |
],
|
| 7562 |
"logging_steps": 1,
|
|
|
|
| 7576 |
"attributes": {}
|
| 7577 |
}
|
| 7578 |
},
|
| 7579 |
+
"total_flos": 134417473339392.0,
|
| 7580 |
"train_batch_size": 2,
|
| 7581 |
"trial_name": null,
|
| 7582 |
"trial_params": null
|