Training in progress, step 738, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3582214344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95792ae8ded0f33efb25a96aabbf4c312850393ce1082a0f917a64823378f02b
|
| 3 |
size 3582214344
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5116251049
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07f63198ca14d6240b593c6ef2aa68da636bb260b294b206a3ce74687629bd0c
|
| 3 |
size 5116251049
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:181c5f0270cf39930062ddfa3767a2481d0c360f120b11f8e25dbf533a1cdaba
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66e045378572462b16d6cd3a7e04fcd559ab9fef53367627c4abe98bf2d141e7
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 369,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2599,6 +2599,2597 @@
|
|
| 2599 |
"eval_samples_per_second": 11.938,
|
| 2600 |
"eval_steps_per_second": 2.985,
|
| 2601 |
"step": 369
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2602 |
}
|
| 2603 |
],
|
| 2604 |
"logging_steps": 1,
|
|
@@ -2618,7 +5209,7 @@
|
|
| 2618 |
"attributes": {}
|
| 2619 |
}
|
| 2620 |
},
|
| 2621 |
-
"total_flos":
|
| 2622 |
"train_batch_size": 8,
|
| 2623 |
"trial_name": null,
|
| 2624 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.42105263157894735,
|
| 6 |
"eval_steps": 369,
|
| 7 |
+
"global_step": 738,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2599 |
"eval_samples_per_second": 11.938,
|
| 2600 |
"eval_steps_per_second": 2.985,
|
| 2601 |
"step": 369
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 0.21109684781058338,
|
| 2605 |
+
"grad_norm": 1.0078125,
|
| 2606 |
+
"learning_rate": 4.0798226164079826e-05,
|
| 2607 |
+
"loss": 1.203234076499939,
|
| 2608 |
+
"step": 370
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 0.21166737983169306,
|
| 2612 |
+
"grad_norm": 1.0390625,
|
| 2613 |
+
"learning_rate": 4.0770509977827055e-05,
|
| 2614 |
+
"loss": 1.2333259582519531,
|
| 2615 |
+
"step": 371
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 0.21223791185280275,
|
| 2619 |
+
"grad_norm": 1.046875,
|
| 2620 |
+
"learning_rate": 4.074279379157428e-05,
|
| 2621 |
+
"loss": 1.2060984373092651,
|
| 2622 |
+
"step": 372
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 0.21280844387391243,
|
| 2626 |
+
"grad_norm": 1.0234375,
|
| 2627 |
+
"learning_rate": 4.0715077605321514e-05,
|
| 2628 |
+
"loss": 1.1909129619598389,
|
| 2629 |
+
"step": 373
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 0.2133789758950221,
|
| 2633 |
+
"grad_norm": 1.046875,
|
| 2634 |
+
"learning_rate": 4.068736141906874e-05,
|
| 2635 |
+
"loss": 1.2396963834762573,
|
| 2636 |
+
"step": 374
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 0.2139495079161318,
|
| 2640 |
+
"grad_norm": 1.0078125,
|
| 2641 |
+
"learning_rate": 4.0659645232815966e-05,
|
| 2642 |
+
"loss": 1.1830250024795532,
|
| 2643 |
+
"step": 375
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 0.21452003993724147,
|
| 2647 |
+
"grad_norm": 1.03125,
|
| 2648 |
+
"learning_rate": 4.0631929046563196e-05,
|
| 2649 |
+
"loss": 1.207044005393982,
|
| 2650 |
+
"step": 376
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 0.21509057195835116,
|
| 2654 |
+
"grad_norm": 1.203125,
|
| 2655 |
+
"learning_rate": 4.0604212860310425e-05,
|
| 2656 |
+
"loss": 1.2795757055282593,
|
| 2657 |
+
"step": 377
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 0.21566110397946084,
|
| 2661 |
+
"grad_norm": 1.0546875,
|
| 2662 |
+
"learning_rate": 4.057649667405765e-05,
|
| 2663 |
+
"loss": 1.2492969036102295,
|
| 2664 |
+
"step": 378
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 0.21623163600057052,
|
| 2668 |
+
"grad_norm": 1.0546875,
|
| 2669 |
+
"learning_rate": 4.0548780487804884e-05,
|
| 2670 |
+
"loss": 1.3094936609268188,
|
| 2671 |
+
"step": 379
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 0.21680216802168023,
|
| 2675 |
+
"grad_norm": 1.0390625,
|
| 2676 |
+
"learning_rate": 4.052106430155211e-05,
|
| 2677 |
+
"loss": 1.2260823249816895,
|
| 2678 |
+
"step": 380
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 0.2173727000427899,
|
| 2682 |
+
"grad_norm": 1.078125,
|
| 2683 |
+
"learning_rate": 4.0493348115299336e-05,
|
| 2684 |
+
"loss": 1.2405587434768677,
|
| 2685 |
+
"step": 381
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 0.2179432320638996,
|
| 2689 |
+
"grad_norm": 1.09375,
|
| 2690 |
+
"learning_rate": 4.0465631929046566e-05,
|
| 2691 |
+
"loss": 1.1963216066360474,
|
| 2692 |
+
"step": 382
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.21851376408500928,
|
| 2696 |
+
"grad_norm": 1.1484375,
|
| 2697 |
+
"learning_rate": 4.043791574279379e-05,
|
| 2698 |
+
"loss": 1.2458081245422363,
|
| 2699 |
+
"step": 383
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 0.21908429610611896,
|
| 2703 |
+
"grad_norm": 1.0859375,
|
| 2704 |
+
"learning_rate": 4.0410199556541025e-05,
|
| 2705 |
+
"loss": 1.1974573135375977,
|
| 2706 |
+
"step": 384
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 0.21965482812722864,
|
| 2710 |
+
"grad_norm": 1.0390625,
|
| 2711 |
+
"learning_rate": 4.038248337028825e-05,
|
| 2712 |
+
"loss": 1.2237815856933594,
|
| 2713 |
+
"step": 385
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 0.22022536014833832,
|
| 2717 |
+
"grad_norm": 1.03125,
|
| 2718 |
+
"learning_rate": 4.035476718403548e-05,
|
| 2719 |
+
"loss": 1.2369771003723145,
|
| 2720 |
+
"step": 386
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 0.220795892169448,
|
| 2724 |
+
"grad_norm": 1.0859375,
|
| 2725 |
+
"learning_rate": 4.0327050997782706e-05,
|
| 2726 |
+
"loss": 1.2545832395553589,
|
| 2727 |
+
"step": 387
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 0.2213664241905577,
|
| 2731 |
+
"grad_norm": 1.03125,
|
| 2732 |
+
"learning_rate": 4.0299334811529936e-05,
|
| 2733 |
+
"loss": 1.2126426696777344,
|
| 2734 |
+
"step": 388
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 0.22193695621166737,
|
| 2738 |
+
"grad_norm": 1.0625,
|
| 2739 |
+
"learning_rate": 4.0271618625277165e-05,
|
| 2740 |
+
"loss": 1.2321901321411133,
|
| 2741 |
+
"step": 389
|
| 2742 |
+
},
|
| 2743 |
+
{
|
| 2744 |
+
"epoch": 0.22250748823277705,
|
| 2745 |
+
"grad_norm": 1.03125,
|
| 2746 |
+
"learning_rate": 4.0243902439024395e-05,
|
| 2747 |
+
"loss": 1.2315490245819092,
|
| 2748 |
+
"step": 390
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 0.22307802025388676,
|
| 2752 |
+
"grad_norm": 1.0390625,
|
| 2753 |
+
"learning_rate": 4.021618625277162e-05,
|
| 2754 |
+
"loss": 1.1859689950942993,
|
| 2755 |
+
"step": 391
|
| 2756 |
+
},
|
| 2757 |
+
{
|
| 2758 |
+
"epoch": 0.22364855227499644,
|
| 2759 |
+
"grad_norm": 1.03125,
|
| 2760 |
+
"learning_rate": 4.018847006651885e-05,
|
| 2761 |
+
"loss": 1.2416760921478271,
|
| 2762 |
+
"step": 392
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.22421908429610612,
|
| 2766 |
+
"grad_norm": 1.1015625,
|
| 2767 |
+
"learning_rate": 4.0160753880266076e-05,
|
| 2768 |
+
"loss": 1.3080382347106934,
|
| 2769 |
+
"step": 393
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 0.2247896163172158,
|
| 2773 |
+
"grad_norm": 1.0625,
|
| 2774 |
+
"learning_rate": 4.0133037694013306e-05,
|
| 2775 |
+
"loss": 1.2275526523590088,
|
| 2776 |
+
"step": 394
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 0.2253601483383255,
|
| 2780 |
+
"grad_norm": 1.03125,
|
| 2781 |
+
"learning_rate": 4.0105321507760535e-05,
|
| 2782 |
+
"loss": 1.2734044790267944,
|
| 2783 |
+
"step": 395
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 0.22593068035943517,
|
| 2787 |
+
"grad_norm": 1.09375,
|
| 2788 |
+
"learning_rate": 4.007760532150776e-05,
|
| 2789 |
+
"loss": 1.2480955123901367,
|
| 2790 |
+
"step": 396
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 0.22650121238054485,
|
| 2794 |
+
"grad_norm": 1.078125,
|
| 2795 |
+
"learning_rate": 4.0049889135254994e-05,
|
| 2796 |
+
"loss": 1.2629410028457642,
|
| 2797 |
+
"step": 397
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 0.22707174440165453,
|
| 2801 |
+
"grad_norm": 1.046875,
|
| 2802 |
+
"learning_rate": 4.002217294900222e-05,
|
| 2803 |
+
"loss": 1.190090537071228,
|
| 2804 |
+
"step": 398
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.22764227642276422,
|
| 2808 |
+
"grad_norm": 1.0859375,
|
| 2809 |
+
"learning_rate": 3.9994456762749446e-05,
|
| 2810 |
+
"loss": 1.2843146324157715,
|
| 2811 |
+
"step": 399
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 0.2282128084438739,
|
| 2815 |
+
"grad_norm": 1.0546875,
|
| 2816 |
+
"learning_rate": 3.9966740576496676e-05,
|
| 2817 |
+
"loss": 1.2836047410964966,
|
| 2818 |
+
"step": 400
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 0.2287833404649836,
|
| 2822 |
+
"grad_norm": 1.0234375,
|
| 2823 |
+
"learning_rate": 3.9939024390243905e-05,
|
| 2824 |
+
"loss": 1.1873021125793457,
|
| 2825 |
+
"step": 401
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 0.2293538724860933,
|
| 2829 |
+
"grad_norm": 1.03125,
|
| 2830 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 2831 |
+
"loss": 1.228004813194275,
|
| 2832 |
+
"step": 402
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 0.22992440450720297,
|
| 2836 |
+
"grad_norm": 1.0078125,
|
| 2837 |
+
"learning_rate": 3.9883592017738364e-05,
|
| 2838 |
+
"loss": 1.2318588495254517,
|
| 2839 |
+
"step": 403
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 0.23049493652831265,
|
| 2843 |
+
"grad_norm": 1.046875,
|
| 2844 |
+
"learning_rate": 3.985587583148559e-05,
|
| 2845 |
+
"loss": 1.218421220779419,
|
| 2846 |
+
"step": 404
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.23106546854942234,
|
| 2850 |
+
"grad_norm": 1.0625,
|
| 2851 |
+
"learning_rate": 3.9828159645232816e-05,
|
| 2852 |
+
"loss": 1.3068960905075073,
|
| 2853 |
+
"step": 405
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 0.23163600057053202,
|
| 2857 |
+
"grad_norm": 1.0234375,
|
| 2858 |
+
"learning_rate": 3.9800443458980046e-05,
|
| 2859 |
+
"loss": 1.2189011573791504,
|
| 2860 |
+
"step": 406
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 0.2322065325916417,
|
| 2864 |
+
"grad_norm": 1.046875,
|
| 2865 |
+
"learning_rate": 3.9772727272727275e-05,
|
| 2866 |
+
"loss": 1.2019367218017578,
|
| 2867 |
+
"step": 407
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 0.23277706461275138,
|
| 2871 |
+
"grad_norm": 1.046875,
|
| 2872 |
+
"learning_rate": 3.9745011086474505e-05,
|
| 2873 |
+
"loss": 1.2285387516021729,
|
| 2874 |
+
"step": 408
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.23334759663386107,
|
| 2878 |
+
"grad_norm": 1.015625,
|
| 2879 |
+
"learning_rate": 3.971729490022173e-05,
|
| 2880 |
+
"loss": 1.1963067054748535,
|
| 2881 |
+
"step": 409
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 0.23391812865497075,
|
| 2885 |
+
"grad_norm": 1.078125,
|
| 2886 |
+
"learning_rate": 3.9689578713968964e-05,
|
| 2887 |
+
"loss": 1.3005050420761108,
|
| 2888 |
+
"step": 410
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 0.23448866067608046,
|
| 2892 |
+
"grad_norm": 1.0234375,
|
| 2893 |
+
"learning_rate": 3.9661862527716186e-05,
|
| 2894 |
+
"loss": 1.2429478168487549,
|
| 2895 |
+
"step": 411
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 0.23505919269719014,
|
| 2899 |
+
"grad_norm": 1.046875,
|
| 2900 |
+
"learning_rate": 3.9634146341463416e-05,
|
| 2901 |
+
"loss": 1.2445229291915894,
|
| 2902 |
+
"step": 412
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 0.23562972471829982,
|
| 2906 |
+
"grad_norm": 1.046875,
|
| 2907 |
+
"learning_rate": 3.9606430155210645e-05,
|
| 2908 |
+
"loss": 1.2569499015808105,
|
| 2909 |
+
"step": 413
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 0.2362002567394095,
|
| 2913 |
+
"grad_norm": 1.0,
|
| 2914 |
+
"learning_rate": 3.9578713968957875e-05,
|
| 2915 |
+
"loss": 1.232776165008545,
|
| 2916 |
+
"step": 414
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 0.23677078876051919,
|
| 2920 |
+
"grad_norm": 1.03125,
|
| 2921 |
+
"learning_rate": 3.9550997782705104e-05,
|
| 2922 |
+
"loss": 1.2104380130767822,
|
| 2923 |
+
"step": 415
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 0.23734132078162887,
|
| 2927 |
+
"grad_norm": 1.1015625,
|
| 2928 |
+
"learning_rate": 3.952328159645233e-05,
|
| 2929 |
+
"loss": 1.2908308506011963,
|
| 2930 |
+
"step": 416
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.23791185280273855,
|
| 2934 |
+
"grad_norm": 1.0625,
|
| 2935 |
+
"learning_rate": 3.9495565410199557e-05,
|
| 2936 |
+
"loss": 1.1678047180175781,
|
| 2937 |
+
"step": 417
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 0.23848238482384823,
|
| 2941 |
+
"grad_norm": 1.0859375,
|
| 2942 |
+
"learning_rate": 3.9467849223946786e-05,
|
| 2943 |
+
"loss": 1.310725212097168,
|
| 2944 |
+
"step": 418
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 0.2390529168449579,
|
| 2948 |
+
"grad_norm": 1.0546875,
|
| 2949 |
+
"learning_rate": 3.9440133037694015e-05,
|
| 2950 |
+
"loss": 1.2618491649627686,
|
| 2951 |
+
"step": 419
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 0.2396234488660676,
|
| 2955 |
+
"grad_norm": 1.0234375,
|
| 2956 |
+
"learning_rate": 3.9412416851441245e-05,
|
| 2957 |
+
"loss": 1.1795238256454468,
|
| 2958 |
+
"step": 420
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 0.2401939808871773,
|
| 2962 |
+
"grad_norm": 1.0546875,
|
| 2963 |
+
"learning_rate": 3.9384700665188474e-05,
|
| 2964 |
+
"loss": 1.2187573909759521,
|
| 2965 |
+
"step": 421
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 0.240764512908287,
|
| 2969 |
+
"grad_norm": 1.03125,
|
| 2970 |
+
"learning_rate": 3.93569844789357e-05,
|
| 2971 |
+
"loss": 1.2171461582183838,
|
| 2972 |
+
"step": 422
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 0.24133504492939667,
|
| 2976 |
+
"grad_norm": 1.046875,
|
| 2977 |
+
"learning_rate": 3.932926829268293e-05,
|
| 2978 |
+
"loss": 1.2295634746551514,
|
| 2979 |
+
"step": 423
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 0.24190557695050635,
|
| 2983 |
+
"grad_norm": 1.0859375,
|
| 2984 |
+
"learning_rate": 3.9301552106430156e-05,
|
| 2985 |
+
"loss": 1.2483271360397339,
|
| 2986 |
+
"step": 424
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 0.24247610897161603,
|
| 2990 |
+
"grad_norm": 1.0546875,
|
| 2991 |
+
"learning_rate": 3.9273835920177385e-05,
|
| 2992 |
+
"loss": 1.1881691217422485,
|
| 2993 |
+
"step": 425
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 0.24304664099272572,
|
| 2997 |
+
"grad_norm": 1.03125,
|
| 2998 |
+
"learning_rate": 3.9246119733924615e-05,
|
| 2999 |
+
"loss": 1.1997624635696411,
|
| 3000 |
+
"step": 426
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 0.2436171730138354,
|
| 3004 |
+
"grad_norm": 1.1015625,
|
| 3005 |
+
"learning_rate": 3.9218403547671844e-05,
|
| 3006 |
+
"loss": 1.2510207891464233,
|
| 3007 |
+
"step": 427
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 0.24418770503494508,
|
| 3011 |
+
"grad_norm": 1.046875,
|
| 3012 |
+
"learning_rate": 3.9190687361419074e-05,
|
| 3013 |
+
"loss": 1.2188156843185425,
|
| 3014 |
+
"step": 428
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 0.24475823705605476,
|
| 3018 |
+
"grad_norm": 1.0234375,
|
| 3019 |
+
"learning_rate": 3.9162971175166297e-05,
|
| 3020 |
+
"loss": 1.228477954864502,
|
| 3021 |
+
"step": 429
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 0.24532876907716444,
|
| 3025 |
+
"grad_norm": 1.1171875,
|
| 3026 |
+
"learning_rate": 3.9135254988913526e-05,
|
| 3027 |
+
"loss": 1.3039709329605103,
|
| 3028 |
+
"step": 430
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 0.24589930109827415,
|
| 3032 |
+
"grad_norm": 1.0703125,
|
| 3033 |
+
"learning_rate": 3.9107538802660755e-05,
|
| 3034 |
+
"loss": 1.2193942070007324,
|
| 3035 |
+
"step": 431
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 0.24646983311938384,
|
| 3039 |
+
"grad_norm": 1.0859375,
|
| 3040 |
+
"learning_rate": 3.9079822616407985e-05,
|
| 3041 |
+
"loss": 1.2380352020263672,
|
| 3042 |
+
"step": 432
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 0.24704036514049352,
|
| 3046 |
+
"grad_norm": 1.046875,
|
| 3047 |
+
"learning_rate": 3.905210643015521e-05,
|
| 3048 |
+
"loss": 1.1670141220092773,
|
| 3049 |
+
"step": 433
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 0.2476108971616032,
|
| 3053 |
+
"grad_norm": 1.0625,
|
| 3054 |
+
"learning_rate": 3.9024390243902444e-05,
|
| 3055 |
+
"loss": 1.2406682968139648,
|
| 3056 |
+
"step": 434
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 0.24818142918271288,
|
| 3060 |
+
"grad_norm": 1.0625,
|
| 3061 |
+
"learning_rate": 3.8996674057649667e-05,
|
| 3062 |
+
"loss": 1.200782060623169,
|
| 3063 |
+
"step": 435
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 0.24875196120382256,
|
| 3067 |
+
"grad_norm": 1.046875,
|
| 3068 |
+
"learning_rate": 3.89689578713969e-05,
|
| 3069 |
+
"loss": 1.1442952156066895,
|
| 3070 |
+
"step": 436
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 0.24932249322493225,
|
| 3074 |
+
"grad_norm": 1.03125,
|
| 3075 |
+
"learning_rate": 3.8941241685144125e-05,
|
| 3076 |
+
"loss": 1.15338134765625,
|
| 3077 |
+
"step": 437
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 0.24989302524604193,
|
| 3081 |
+
"grad_norm": 0.98828125,
|
| 3082 |
+
"learning_rate": 3.8913525498891355e-05,
|
| 3083 |
+
"loss": 1.1609077453613281,
|
| 3084 |
+
"step": 438
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 0.25046355726715164,
|
| 3088 |
+
"grad_norm": 1.1015625,
|
| 3089 |
+
"learning_rate": 3.8885809312638584e-05,
|
| 3090 |
+
"loss": 1.257835030555725,
|
| 3091 |
+
"step": 439
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 0.2510340892882613,
|
| 3095 |
+
"grad_norm": 1.0859375,
|
| 3096 |
+
"learning_rate": 3.8858093126385814e-05,
|
| 3097 |
+
"loss": 1.2244375944137573,
|
| 3098 |
+
"step": 440
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.251604621309371,
|
| 3102 |
+
"grad_norm": 1.0859375,
|
| 3103 |
+
"learning_rate": 3.8830376940133037e-05,
|
| 3104 |
+
"loss": 1.2138961553573608,
|
| 3105 |
+
"step": 441
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 0.2521751533304807,
|
| 3109 |
+
"grad_norm": 1.0546875,
|
| 3110 |
+
"learning_rate": 3.8802660753880266e-05,
|
| 3111 |
+
"loss": 1.240128755569458,
|
| 3112 |
+
"step": 442
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 0.25274568535159037,
|
| 3116 |
+
"grad_norm": 1.0,
|
| 3117 |
+
"learning_rate": 3.8774944567627496e-05,
|
| 3118 |
+
"loss": 1.2070982456207275,
|
| 3119 |
+
"step": 443
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 0.25331621737270005,
|
| 3123 |
+
"grad_norm": 1.0546875,
|
| 3124 |
+
"learning_rate": 3.8747228381374725e-05,
|
| 3125 |
+
"loss": 1.2733830213546753,
|
| 3126 |
+
"step": 444
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 0.25388674939380973,
|
| 3130 |
+
"grad_norm": 1.046875,
|
| 3131 |
+
"learning_rate": 3.8719512195121954e-05,
|
| 3132 |
+
"loss": 1.1820507049560547,
|
| 3133 |
+
"step": 445
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 0.2544572814149194,
|
| 3137 |
+
"grad_norm": 1.0078125,
|
| 3138 |
+
"learning_rate": 3.869179600886918e-05,
|
| 3139 |
+
"loss": 1.196885108947754,
|
| 3140 |
+
"step": 446
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 0.2550278134360291,
|
| 3144 |
+
"grad_norm": 1.03125,
|
| 3145 |
+
"learning_rate": 3.866407982261641e-05,
|
| 3146 |
+
"loss": 1.1905972957611084,
|
| 3147 |
+
"step": 447
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 0.2555983454571388,
|
| 3151 |
+
"grad_norm": 1.09375,
|
| 3152 |
+
"learning_rate": 3.8636363636363636e-05,
|
| 3153 |
+
"loss": 1.2579684257507324,
|
| 3154 |
+
"step": 448
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 0.25616887747824846,
|
| 3158 |
+
"grad_norm": 1.0,
|
| 3159 |
+
"learning_rate": 3.8608647450110866e-05,
|
| 3160 |
+
"loss": 1.1727596521377563,
|
| 3161 |
+
"step": 449
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 0.25673940949935814,
|
| 3165 |
+
"grad_norm": 1.015625,
|
| 3166 |
+
"learning_rate": 3.8580931263858095e-05,
|
| 3167 |
+
"loss": 1.1504234075546265,
|
| 3168 |
+
"step": 450
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 0.2573099415204678,
|
| 3172 |
+
"grad_norm": 1.015625,
|
| 3173 |
+
"learning_rate": 3.8553215077605324e-05,
|
| 3174 |
+
"loss": 1.1405715942382812,
|
| 3175 |
+
"step": 451
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 0.2578804735415775,
|
| 3179 |
+
"grad_norm": 0.98828125,
|
| 3180 |
+
"learning_rate": 3.8525498891352554e-05,
|
| 3181 |
+
"loss": 1.220837116241455,
|
| 3182 |
+
"step": 452
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 0.2584510055626872,
|
| 3186 |
+
"grad_norm": 1.0234375,
|
| 3187 |
+
"learning_rate": 3.8497782705099777e-05,
|
| 3188 |
+
"loss": 1.1962711811065674,
|
| 3189 |
+
"step": 453
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 0.25902153758379687,
|
| 3193 |
+
"grad_norm": 1.015625,
|
| 3194 |
+
"learning_rate": 3.8470066518847006e-05,
|
| 3195 |
+
"loss": 1.1877164840698242,
|
| 3196 |
+
"step": 454
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 0.25959206960490655,
|
| 3200 |
+
"grad_norm": 1.046875,
|
| 3201 |
+
"learning_rate": 3.8442350332594236e-05,
|
| 3202 |
+
"loss": 1.2504132986068726,
|
| 3203 |
+
"step": 455
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 0.2601626016260163,
|
| 3207 |
+
"grad_norm": 1.015625,
|
| 3208 |
+
"learning_rate": 3.8414634146341465e-05,
|
| 3209 |
+
"loss": 1.1902315616607666,
|
| 3210 |
+
"step": 456
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 0.26073313364712597,
|
| 3214 |
+
"grad_norm": 1.03125,
|
| 3215 |
+
"learning_rate": 3.8386917960088694e-05,
|
| 3216 |
+
"loss": 1.2856203317642212,
|
| 3217 |
+
"step": 457
|
| 3218 |
+
},
|
| 3219 |
+
{
|
| 3220 |
+
"epoch": 0.26130366566823565,
|
| 3221 |
+
"grad_norm": 1.0625,
|
| 3222 |
+
"learning_rate": 3.8359201773835924e-05,
|
| 3223 |
+
"loss": 1.2528060674667358,
|
| 3224 |
+
"step": 458
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 0.26187419768934533,
|
| 3228 |
+
"grad_norm": 1.078125,
|
| 3229 |
+
"learning_rate": 3.833148558758315e-05,
|
| 3230 |
+
"loss": 1.1831871271133423,
|
| 3231 |
+
"step": 459
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 0.262444729710455,
|
| 3235 |
+
"grad_norm": 1.015625,
|
| 3236 |
+
"learning_rate": 3.830376940133038e-05,
|
| 3237 |
+
"loss": 1.1781988143920898,
|
| 3238 |
+
"step": 460
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 0.2630152617315647,
|
| 3242 |
+
"grad_norm": 1.015625,
|
| 3243 |
+
"learning_rate": 3.8276053215077606e-05,
|
| 3244 |
+
"loss": 1.193709373474121,
|
| 3245 |
+
"step": 461
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 0.2635857937526744,
|
| 3249 |
+
"grad_norm": 1.078125,
|
| 3250 |
+
"learning_rate": 3.8248337028824835e-05,
|
| 3251 |
+
"loss": 1.1997225284576416,
|
| 3252 |
+
"step": 462
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 0.26415632577378406,
|
| 3256 |
+
"grad_norm": 1.0078125,
|
| 3257 |
+
"learning_rate": 3.8220620842572064e-05,
|
| 3258 |
+
"loss": 1.159136176109314,
|
| 3259 |
+
"step": 463
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 0.26472685779489374,
|
| 3263 |
+
"grad_norm": 1.015625,
|
| 3264 |
+
"learning_rate": 3.8192904656319294e-05,
|
| 3265 |
+
"loss": 1.242883324623108,
|
| 3266 |
+
"step": 464
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 0.2652973898160034,
|
| 3270 |
+
"grad_norm": 1.0703125,
|
| 3271 |
+
"learning_rate": 3.8165188470066523e-05,
|
| 3272 |
+
"loss": 1.2907770872116089,
|
| 3273 |
+
"step": 465
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 0.2658679218371131,
|
| 3277 |
+
"grad_norm": 1.09375,
|
| 3278 |
+
"learning_rate": 3.8137472283813746e-05,
|
| 3279 |
+
"loss": 1.2596560716629028,
|
| 3280 |
+
"step": 466
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 0.2664384538582228,
|
| 3284 |
+
"grad_norm": 1.1171875,
|
| 3285 |
+
"learning_rate": 3.8109756097560976e-05,
|
| 3286 |
+
"loss": 1.2509888410568237,
|
| 3287 |
+
"step": 467
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 0.26700898587933247,
|
| 3291 |
+
"grad_norm": 0.97265625,
|
| 3292 |
+
"learning_rate": 3.8082039911308205e-05,
|
| 3293 |
+
"loss": 1.2029120922088623,
|
| 3294 |
+
"step": 468
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 0.26757951790044215,
|
| 3298 |
+
"grad_norm": 1.015625,
|
| 3299 |
+
"learning_rate": 3.8054323725055435e-05,
|
| 3300 |
+
"loss": 1.210568904876709,
|
| 3301 |
+
"step": 469
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 0.26815004992155184,
|
| 3305 |
+
"grad_norm": 0.9921875,
|
| 3306 |
+
"learning_rate": 3.8026607538802664e-05,
|
| 3307 |
+
"loss": 1.1661216020584106,
|
| 3308 |
+
"step": 470
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 0.2687205819426615,
|
| 3312 |
+
"grad_norm": 1.03125,
|
| 3313 |
+
"learning_rate": 3.7998891352549893e-05,
|
| 3314 |
+
"loss": 1.229252576828003,
|
| 3315 |
+
"step": 471
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 0.2692911139637712,
|
| 3319 |
+
"grad_norm": 1.0546875,
|
| 3320 |
+
"learning_rate": 3.7971175166297116e-05,
|
| 3321 |
+
"loss": 1.209242343902588,
|
| 3322 |
+
"step": 472
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 0.2698616459848809,
|
| 3326 |
+
"grad_norm": 1.0390625,
|
| 3327 |
+
"learning_rate": 3.794345898004435e-05,
|
| 3328 |
+
"loss": 1.2709503173828125,
|
| 3329 |
+
"step": 473
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 0.27043217800599056,
|
| 3333 |
+
"grad_norm": 1.078125,
|
| 3334 |
+
"learning_rate": 3.7915742793791575e-05,
|
| 3335 |
+
"loss": 1.2316001653671265,
|
| 3336 |
+
"step": 474
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 0.27100271002710025,
|
| 3340 |
+
"grad_norm": 1.03125,
|
| 3341 |
+
"learning_rate": 3.7888026607538805e-05,
|
| 3342 |
+
"loss": 1.2138065099716187,
|
| 3343 |
+
"step": 475
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 0.27157324204820993,
|
| 3347 |
+
"grad_norm": 1.0,
|
| 3348 |
+
"learning_rate": 3.7860310421286034e-05,
|
| 3349 |
+
"loss": 1.1936984062194824,
|
| 3350 |
+
"step": 476
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 0.27214377406931967,
|
| 3354 |
+
"grad_norm": 1.015625,
|
| 3355 |
+
"learning_rate": 3.783259423503326e-05,
|
| 3356 |
+
"loss": 1.2338573932647705,
|
| 3357 |
+
"step": 477
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 0.27271430609042935,
|
| 3361 |
+
"grad_norm": 1.046875,
|
| 3362 |
+
"learning_rate": 3.780487804878049e-05,
|
| 3363 |
+
"loss": 1.2421263456344604,
|
| 3364 |
+
"step": 478
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 0.27328483811153903,
|
| 3368 |
+
"grad_norm": 1.03125,
|
| 3369 |
+
"learning_rate": 3.7777161862527716e-05,
|
| 3370 |
+
"loss": 1.2414464950561523,
|
| 3371 |
+
"step": 479
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 0.2738553701326487,
|
| 3375 |
+
"grad_norm": 0.99609375,
|
| 3376 |
+
"learning_rate": 3.7749445676274945e-05,
|
| 3377 |
+
"loss": 1.2261340618133545,
|
| 3378 |
+
"step": 480
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 0.2744259021537584,
|
| 3382 |
+
"grad_norm": 1.0234375,
|
| 3383 |
+
"learning_rate": 3.7721729490022175e-05,
|
| 3384 |
+
"loss": 1.208221435546875,
|
| 3385 |
+
"step": 481
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 0.2749964341748681,
|
| 3389 |
+
"grad_norm": 1.0234375,
|
| 3390 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 3391 |
+
"loss": 1.2820276021957397,
|
| 3392 |
+
"step": 482
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 0.27556696619597776,
|
| 3396 |
+
"grad_norm": 1.0703125,
|
| 3397 |
+
"learning_rate": 3.7666297117516633e-05,
|
| 3398 |
+
"loss": 1.262161374092102,
|
| 3399 |
+
"step": 483
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 0.27613749821708744,
|
| 3403 |
+
"grad_norm": 0.9921875,
|
| 3404 |
+
"learning_rate": 3.763858093126386e-05,
|
| 3405 |
+
"loss": 1.2242916822433472,
|
| 3406 |
+
"step": 484
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 0.2767080302381971,
|
| 3410 |
+
"grad_norm": 1.0078125,
|
| 3411 |
+
"learning_rate": 3.7610864745011086e-05,
|
| 3412 |
+
"loss": 1.1797833442687988,
|
| 3413 |
+
"step": 485
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 0.2772785622593068,
|
| 3417 |
+
"grad_norm": 1.0546875,
|
| 3418 |
+
"learning_rate": 3.758314855875832e-05,
|
| 3419 |
+
"loss": 1.2725660800933838,
|
| 3420 |
+
"step": 486
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 0.2778490942804165,
|
| 3424 |
+
"grad_norm": 1.015625,
|
| 3425 |
+
"learning_rate": 3.7555432372505545e-05,
|
| 3426 |
+
"loss": 1.195313572883606,
|
| 3427 |
+
"step": 487
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 0.27841962630152617,
|
| 3431 |
+
"grad_norm": 0.94921875,
|
| 3432 |
+
"learning_rate": 3.7527716186252774e-05,
|
| 3433 |
+
"loss": 1.1661468744277954,
|
| 3434 |
+
"step": 488
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 0.27899015832263585,
|
| 3438 |
+
"grad_norm": 1.0390625,
|
| 3439 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 3440 |
+
"loss": 1.2072978019714355,
|
| 3441 |
+
"step": 489
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 0.27956069034374553,
|
| 3445 |
+
"grad_norm": 0.98828125,
|
| 3446 |
+
"learning_rate": 3.7472283813747226e-05,
|
| 3447 |
+
"loss": 1.203414797782898,
|
| 3448 |
+
"step": 490
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 0.2801312223648552,
|
| 3452 |
+
"grad_norm": 1.0390625,
|
| 3453 |
+
"learning_rate": 3.7444567627494456e-05,
|
| 3454 |
+
"loss": 1.2426180839538574,
|
| 3455 |
+
"step": 491
|
| 3456 |
+
},
|
| 3457 |
+
{
|
| 3458 |
+
"epoch": 0.2807017543859649,
|
| 3459 |
+
"grad_norm": 1.015625,
|
| 3460 |
+
"learning_rate": 3.7416851441241685e-05,
|
| 3461 |
+
"loss": 1.232536792755127,
|
| 3462 |
+
"step": 492
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 0.2812722864070746,
|
| 3466 |
+
"grad_norm": 1.03125,
|
| 3467 |
+
"learning_rate": 3.7389135254988915e-05,
|
| 3468 |
+
"loss": 1.266850471496582,
|
| 3469 |
+
"step": 493
|
| 3470 |
+
},
|
| 3471 |
+
{
|
| 3472 |
+
"epoch": 0.28184281842818426,
|
| 3473 |
+
"grad_norm": 1.046875,
|
| 3474 |
+
"learning_rate": 3.7361419068736144e-05,
|
| 3475 |
+
"loss": 1.2585172653198242,
|
| 3476 |
+
"step": 494
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 0.28241335044929394,
|
| 3480 |
+
"grad_norm": 0.99609375,
|
| 3481 |
+
"learning_rate": 3.7333702882483374e-05,
|
| 3482 |
+
"loss": 1.2028322219848633,
|
| 3483 |
+
"step": 495
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 0.2829838824704036,
|
| 3487 |
+
"grad_norm": 1.0546875,
|
| 3488 |
+
"learning_rate": 3.7305986696230596e-05,
|
| 3489 |
+
"loss": 1.2268320322036743,
|
| 3490 |
+
"step": 496
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 0.28355441449151336,
|
| 3494 |
+
"grad_norm": 1.03125,
|
| 3495 |
+
"learning_rate": 3.727827050997783e-05,
|
| 3496 |
+
"loss": 1.2339527606964111,
|
| 3497 |
+
"step": 497
|
| 3498 |
+
},
|
| 3499 |
+
{
|
| 3500 |
+
"epoch": 0.28412494651262304,
|
| 3501 |
+
"grad_norm": 1.046875,
|
| 3502 |
+
"learning_rate": 3.7250554323725055e-05,
|
| 3503 |
+
"loss": 1.2072274684906006,
|
| 3504 |
+
"step": 498
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 0.2846954785337327,
|
| 3508 |
+
"grad_norm": 0.99609375,
|
| 3509 |
+
"learning_rate": 3.7222838137472285e-05,
|
| 3510 |
+
"loss": 1.235311508178711,
|
| 3511 |
+
"step": 499
|
| 3512 |
+
},
|
| 3513 |
+
{
|
| 3514 |
+
"epoch": 0.2852660105548424,
|
| 3515 |
+
"grad_norm": 1.0390625,
|
| 3516 |
+
"learning_rate": 3.7195121951219514e-05,
|
| 3517 |
+
"loss": 1.2435599565505981,
|
| 3518 |
+
"step": 500
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 0.2858365425759521,
|
| 3522 |
+
"grad_norm": 1.0390625,
|
| 3523 |
+
"learning_rate": 3.7167405764966744e-05,
|
| 3524 |
+
"loss": 1.2234078645706177,
|
| 3525 |
+
"step": 501
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 0.2864070745970618,
|
| 3529 |
+
"grad_norm": 1.0703125,
|
| 3530 |
+
"learning_rate": 3.713968957871397e-05,
|
| 3531 |
+
"loss": 1.2654131650924683,
|
| 3532 |
+
"step": 502
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 0.28697760661817145,
|
| 3536 |
+
"grad_norm": 1.0625,
|
| 3537 |
+
"learning_rate": 3.7111973392461196e-05,
|
| 3538 |
+
"loss": 1.226614236831665,
|
| 3539 |
+
"step": 503
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 0.28754813863928114,
|
| 3543 |
+
"grad_norm": 1.109375,
|
| 3544 |
+
"learning_rate": 3.7084257206208425e-05,
|
| 3545 |
+
"loss": 1.2334555387496948,
|
| 3546 |
+
"step": 504
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 0.2881186706603908,
|
| 3550 |
+
"grad_norm": 1.0546875,
|
| 3551 |
+
"learning_rate": 3.7056541019955655e-05,
|
| 3552 |
+
"loss": 1.2169506549835205,
|
| 3553 |
+
"step": 505
|
| 3554 |
+
},
|
| 3555 |
+
{
|
| 3556 |
+
"epoch": 0.2886892026815005,
|
| 3557 |
+
"grad_norm": 1.078125,
|
| 3558 |
+
"learning_rate": 3.7028824833702884e-05,
|
| 3559 |
+
"loss": 1.2664920091629028,
|
| 3560 |
+
"step": 506
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 0.2892597347026102,
|
| 3564 |
+
"grad_norm": 1.2109375,
|
| 3565 |
+
"learning_rate": 3.7001108647450114e-05,
|
| 3566 |
+
"loss": 1.2238786220550537,
|
| 3567 |
+
"step": 507
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 0.28983026672371986,
|
| 3571 |
+
"grad_norm": 1.03125,
|
| 3572 |
+
"learning_rate": 3.697339246119734e-05,
|
| 3573 |
+
"loss": 1.179901361465454,
|
| 3574 |
+
"step": 508
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 0.29040079874482955,
|
| 3578 |
+
"grad_norm": 1.078125,
|
| 3579 |
+
"learning_rate": 3.6945676274944566e-05,
|
| 3580 |
+
"loss": 1.2527443170547485,
|
| 3581 |
+
"step": 509
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 0.29097133076593923,
|
| 3585 |
+
"grad_norm": 1.0234375,
|
| 3586 |
+
"learning_rate": 3.69179600886918e-05,
|
| 3587 |
+
"loss": 1.2478464841842651,
|
| 3588 |
+
"step": 510
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 0.2915418627870489,
|
| 3592 |
+
"grad_norm": 1.1015625,
|
| 3593 |
+
"learning_rate": 3.6890243902439025e-05,
|
| 3594 |
+
"loss": 1.2006577253341675,
|
| 3595 |
+
"step": 511
|
| 3596 |
+
},
|
| 3597 |
+
{
|
| 3598 |
+
"epoch": 0.2921123948081586,
|
| 3599 |
+
"grad_norm": 1.1015625,
|
| 3600 |
+
"learning_rate": 3.6862527716186254e-05,
|
| 3601 |
+
"loss": 1.283043384552002,
|
| 3602 |
+
"step": 512
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 0.2926829268292683,
|
| 3606 |
+
"grad_norm": 1.0234375,
|
| 3607 |
+
"learning_rate": 3.6834811529933484e-05,
|
| 3608 |
+
"loss": 1.223816156387329,
|
| 3609 |
+
"step": 513
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 0.29325345885037796,
|
| 3613 |
+
"grad_norm": 1.1171875,
|
| 3614 |
+
"learning_rate": 3.6807095343680706e-05,
|
| 3615 |
+
"loss": 1.2357165813446045,
|
| 3616 |
+
"step": 514
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 0.29382399087148764,
|
| 3620 |
+
"grad_norm": 1.078125,
|
| 3621 |
+
"learning_rate": 3.677937915742794e-05,
|
| 3622 |
+
"loss": 1.2494802474975586,
|
| 3623 |
+
"step": 515
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 0.2943945228925973,
|
| 3627 |
+
"grad_norm": 1.046875,
|
| 3628 |
+
"learning_rate": 3.6751662971175165e-05,
|
| 3629 |
+
"loss": 1.2093576192855835,
|
| 3630 |
+
"step": 516
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 0.29496505491370706,
|
| 3634 |
+
"grad_norm": 1.0234375,
|
| 3635 |
+
"learning_rate": 3.6723946784922395e-05,
|
| 3636 |
+
"loss": 1.192871332168579,
|
| 3637 |
+
"step": 517
|
| 3638 |
+
},
|
| 3639 |
+
{
|
| 3640 |
+
"epoch": 0.29553558693481674,
|
| 3641 |
+
"grad_norm": 0.93359375,
|
| 3642 |
+
"learning_rate": 3.6696230598669624e-05,
|
| 3643 |
+
"loss": 1.1430253982543945,
|
| 3644 |
+
"step": 518
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 0.2961061189559264,
|
| 3648 |
+
"grad_norm": 1.03125,
|
| 3649 |
+
"learning_rate": 3.6668514412416854e-05,
|
| 3650 |
+
"loss": 1.2123762369155884,
|
| 3651 |
+
"step": 519
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 0.2966766509770361,
|
| 3655 |
+
"grad_norm": 1.0703125,
|
| 3656 |
+
"learning_rate": 3.664079822616408e-05,
|
| 3657 |
+
"loss": 1.2201260328292847,
|
| 3658 |
+
"step": 520
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 0.2972471829981458,
|
| 3662 |
+
"grad_norm": 1.0546875,
|
| 3663 |
+
"learning_rate": 3.661308203991131e-05,
|
| 3664 |
+
"loss": 1.1812068223953247,
|
| 3665 |
+
"step": 521
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 0.29781771501925547,
|
| 3669 |
+
"grad_norm": 1.078125,
|
| 3670 |
+
"learning_rate": 3.6585365853658535e-05,
|
| 3671 |
+
"loss": 1.2447538375854492,
|
| 3672 |
+
"step": 522
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 0.29838824704036515,
|
| 3676 |
+
"grad_norm": 1.0625,
|
| 3677 |
+
"learning_rate": 3.655764966740577e-05,
|
| 3678 |
+
"loss": 1.2636268138885498,
|
| 3679 |
+
"step": 523
|
| 3680 |
+
},
|
| 3681 |
+
{
|
| 3682 |
+
"epoch": 0.29895877906147483,
|
| 3683 |
+
"grad_norm": 1.0546875,
|
| 3684 |
+
"learning_rate": 3.6529933481152994e-05,
|
| 3685 |
+
"loss": 1.2320729494094849,
|
| 3686 |
+
"step": 524
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 0.2995293110825845,
|
| 3690 |
+
"grad_norm": 1.0859375,
|
| 3691 |
+
"learning_rate": 3.6502217294900224e-05,
|
| 3692 |
+
"loss": 1.2655476331710815,
|
| 3693 |
+
"step": 525
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 0.3000998431036942,
|
| 3697 |
+
"grad_norm": 1.0390625,
|
| 3698 |
+
"learning_rate": 3.647450110864745e-05,
|
| 3699 |
+
"loss": 1.2109198570251465,
|
| 3700 |
+
"step": 526
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 0.3006703751248039,
|
| 3704 |
+
"grad_norm": 1.046875,
|
| 3705 |
+
"learning_rate": 3.6446784922394676e-05,
|
| 3706 |
+
"loss": 1.2380175590515137,
|
| 3707 |
+
"step": 527
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 0.30124090714591356,
|
| 3711 |
+
"grad_norm": 1.0234375,
|
| 3712 |
+
"learning_rate": 3.641906873614191e-05,
|
| 3713 |
+
"loss": 1.2023993730545044,
|
| 3714 |
+
"step": 528
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 0.30181143916702324,
|
| 3718 |
+
"grad_norm": 1.0234375,
|
| 3719 |
+
"learning_rate": 3.6391352549889135e-05,
|
| 3720 |
+
"loss": 1.239518404006958,
|
| 3721 |
+
"step": 529
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 0.3023819711881329,
|
| 3725 |
+
"grad_norm": 0.9921875,
|
| 3726 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 3727 |
+
"loss": 1.2405352592468262,
|
| 3728 |
+
"step": 530
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 0.3029525032092426,
|
| 3732 |
+
"grad_norm": 1.0390625,
|
| 3733 |
+
"learning_rate": 3.6335920177383594e-05,
|
| 3734 |
+
"loss": 1.269554853439331,
|
| 3735 |
+
"step": 531
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 0.3035230352303523,
|
| 3739 |
+
"grad_norm": 1.0078125,
|
| 3740 |
+
"learning_rate": 3.630820399113082e-05,
|
| 3741 |
+
"loss": 1.256522297859192,
|
| 3742 |
+
"step": 532
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 0.30409356725146197,
|
| 3746 |
+
"grad_norm": 0.99609375,
|
| 3747 |
+
"learning_rate": 3.628048780487805e-05,
|
| 3748 |
+
"loss": 1.2245392799377441,
|
| 3749 |
+
"step": 533
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 0.30466409927257165,
|
| 3753 |
+
"grad_norm": 0.99609375,
|
| 3754 |
+
"learning_rate": 3.625277161862528e-05,
|
| 3755 |
+
"loss": 1.2256156206130981,
|
| 3756 |
+
"step": 534
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 0.30523463129368134,
|
| 3760 |
+
"grad_norm": 1.078125,
|
| 3761 |
+
"learning_rate": 3.6225055432372505e-05,
|
| 3762 |
+
"loss": 1.2551851272583008,
|
| 3763 |
+
"step": 535
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 0.305805163314791,
|
| 3767 |
+
"grad_norm": 1.0234375,
|
| 3768 |
+
"learning_rate": 3.619733924611974e-05,
|
| 3769 |
+
"loss": 1.1682400703430176,
|
| 3770 |
+
"step": 536
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 0.30637569533590076,
|
| 3774 |
+
"grad_norm": 1.0703125,
|
| 3775 |
+
"learning_rate": 3.6169623059866964e-05,
|
| 3776 |
+
"loss": 1.2278921604156494,
|
| 3777 |
+
"step": 537
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 0.30694622735701044,
|
| 3781 |
+
"grad_norm": 1.03125,
|
| 3782 |
+
"learning_rate": 3.6141906873614186e-05,
|
| 3783 |
+
"loss": 1.2167140245437622,
|
| 3784 |
+
"step": 538
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 0.3075167593781201,
|
| 3788 |
+
"grad_norm": 0.99609375,
|
| 3789 |
+
"learning_rate": 3.611419068736142e-05,
|
| 3790 |
+
"loss": 1.2471628189086914,
|
| 3791 |
+
"step": 539
|
| 3792 |
+
},
|
| 3793 |
+
{
|
| 3794 |
+
"epoch": 0.3080872913992298,
|
| 3795 |
+
"grad_norm": 1.0546875,
|
| 3796 |
+
"learning_rate": 3.6086474501108645e-05,
|
| 3797 |
+
"loss": 1.2300347089767456,
|
| 3798 |
+
"step": 540
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 0.3086578234203395,
|
| 3802 |
+
"grad_norm": 1.015625,
|
| 3803 |
+
"learning_rate": 3.605875831485588e-05,
|
| 3804 |
+
"loss": 1.1582870483398438,
|
| 3805 |
+
"step": 541
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 0.30922835544144917,
|
| 3809 |
+
"grad_norm": 1.03125,
|
| 3810 |
+
"learning_rate": 3.6031042128603104e-05,
|
| 3811 |
+
"loss": 1.2606914043426514,
|
| 3812 |
+
"step": 542
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 0.30979888746255885,
|
| 3816 |
+
"grad_norm": 1.0,
|
| 3817 |
+
"learning_rate": 3.6003325942350334e-05,
|
| 3818 |
+
"loss": 1.2054803371429443,
|
| 3819 |
+
"step": 543
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 0.31036941948366853,
|
| 3823 |
+
"grad_norm": 1.0078125,
|
| 3824 |
+
"learning_rate": 3.597560975609756e-05,
|
| 3825 |
+
"loss": 1.1797690391540527,
|
| 3826 |
+
"step": 544
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 0.3109399515047782,
|
| 3830 |
+
"grad_norm": 1.03125,
|
| 3831 |
+
"learning_rate": 3.594789356984479e-05,
|
| 3832 |
+
"loss": 1.1780451536178589,
|
| 3833 |
+
"step": 545
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 0.3115104835258879,
|
| 3837 |
+
"grad_norm": 1.0546875,
|
| 3838 |
+
"learning_rate": 3.5920177383592015e-05,
|
| 3839 |
+
"loss": 1.2812529802322388,
|
| 3840 |
+
"step": 546
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 0.3120810155469976,
|
| 3844 |
+
"grad_norm": 1.0625,
|
| 3845 |
+
"learning_rate": 3.589246119733925e-05,
|
| 3846 |
+
"loss": 1.3007402420043945,
|
| 3847 |
+
"step": 547
|
| 3848 |
+
},
|
| 3849 |
+
{
|
| 3850 |
+
"epoch": 0.31265154756810726,
|
| 3851 |
+
"grad_norm": 0.9921875,
|
| 3852 |
+
"learning_rate": 3.5864745011086474e-05,
|
| 3853 |
+
"loss": 1.1987743377685547,
|
| 3854 |
+
"step": 548
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 0.31322207958921694,
|
| 3858 |
+
"grad_norm": 1.078125,
|
| 3859 |
+
"learning_rate": 3.583702882483371e-05,
|
| 3860 |
+
"loss": 1.2217564582824707,
|
| 3861 |
+
"step": 549
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 0.3137926116103266,
|
| 3865 |
+
"grad_norm": 0.99609375,
|
| 3866 |
+
"learning_rate": 3.580931263858093e-05,
|
| 3867 |
+
"loss": 1.211827039718628,
|
| 3868 |
+
"step": 550
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 0.3143631436314363,
|
| 3872 |
+
"grad_norm": 1.03125,
|
| 3873 |
+
"learning_rate": 3.5781596452328156e-05,
|
| 3874 |
+
"loss": 1.2164710760116577,
|
| 3875 |
+
"step": 551
|
| 3876 |
+
},
|
| 3877 |
+
{
|
| 3878 |
+
"epoch": 0.314933675652546,
|
| 3879 |
+
"grad_norm": 0.98828125,
|
| 3880 |
+
"learning_rate": 3.575388026607539e-05,
|
| 3881 |
+
"loss": 1.2393014430999756,
|
| 3882 |
+
"step": 552
|
| 3883 |
+
},
|
| 3884 |
+
{
|
| 3885 |
+
"epoch": 0.31550420767365567,
|
| 3886 |
+
"grad_norm": 0.96875,
|
| 3887 |
+
"learning_rate": 3.5726164079822615e-05,
|
| 3888 |
+
"loss": 1.1759617328643799,
|
| 3889 |
+
"step": 553
|
| 3890 |
+
},
|
| 3891 |
+
{
|
| 3892 |
+
"epoch": 0.31607473969476535,
|
| 3893 |
+
"grad_norm": 1.015625,
|
| 3894 |
+
"learning_rate": 3.5698447893569844e-05,
|
| 3895 |
+
"loss": 1.2184211015701294,
|
| 3896 |
+
"step": 554
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 0.31664527171587503,
|
| 3900 |
+
"grad_norm": 1.046875,
|
| 3901 |
+
"learning_rate": 3.5670731707317074e-05,
|
| 3902 |
+
"loss": 1.248216152191162,
|
| 3903 |
+
"step": 555
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 0.3172158037369847,
|
| 3907 |
+
"grad_norm": 1.0078125,
|
| 3908 |
+
"learning_rate": 3.56430155210643e-05,
|
| 3909 |
+
"loss": 1.2055684328079224,
|
| 3910 |
+
"step": 556
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 0.3177863357580944,
|
| 3914 |
+
"grad_norm": 1.0546875,
|
| 3915 |
+
"learning_rate": 3.561529933481153e-05,
|
| 3916 |
+
"loss": 1.19916832447052,
|
| 3917 |
+
"step": 557
|
| 3918 |
+
},
|
| 3919 |
+
{
|
| 3920 |
+
"epoch": 0.31835686777920413,
|
| 3921 |
+
"grad_norm": 0.95703125,
|
| 3922 |
+
"learning_rate": 3.558758314855876e-05,
|
| 3923 |
+
"loss": 1.151750087738037,
|
| 3924 |
+
"step": 558
|
| 3925 |
+
},
|
| 3926 |
+
{
|
| 3927 |
+
"epoch": 0.3189273998003138,
|
| 3928 |
+
"grad_norm": 0.99609375,
|
| 3929 |
+
"learning_rate": 3.5559866962305985e-05,
|
| 3930 |
+
"loss": 1.254964828491211,
|
| 3931 |
+
"step": 559
|
| 3932 |
+
},
|
| 3933 |
+
{
|
| 3934 |
+
"epoch": 0.3194979318214235,
|
| 3935 |
+
"grad_norm": 1.0546875,
|
| 3936 |
+
"learning_rate": 3.553215077605322e-05,
|
| 3937 |
+
"loss": 1.251706600189209,
|
| 3938 |
+
"step": 560
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 0.3200684638425332,
|
| 3942 |
+
"grad_norm": 1.2265625,
|
| 3943 |
+
"learning_rate": 3.5504434589800444e-05,
|
| 3944 |
+
"loss": 1.1918596029281616,
|
| 3945 |
+
"step": 561
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 0.32063899586364286,
|
| 3949 |
+
"grad_norm": 1.0625,
|
| 3950 |
+
"learning_rate": 3.547671840354767e-05,
|
| 3951 |
+
"loss": 1.2538777589797974,
|
| 3952 |
+
"step": 562
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 0.32120952788475254,
|
| 3956 |
+
"grad_norm": 1.078125,
|
| 3957 |
+
"learning_rate": 3.54490022172949e-05,
|
| 3958 |
+
"loss": 1.227068543434143,
|
| 3959 |
+
"step": 563
|
| 3960 |
+
},
|
| 3961 |
+
{
|
| 3962 |
+
"epoch": 0.3217800599058622,
|
| 3963 |
+
"grad_norm": 1.015625,
|
| 3964 |
+
"learning_rate": 3.5421286031042125e-05,
|
| 3965 |
+
"loss": 1.1811244487762451,
|
| 3966 |
+
"step": 564
|
| 3967 |
+
},
|
| 3968 |
+
{
|
| 3969 |
+
"epoch": 0.3223505919269719,
|
| 3970 |
+
"grad_norm": 1.015625,
|
| 3971 |
+
"learning_rate": 3.539356984478936e-05,
|
| 3972 |
+
"loss": 1.162517786026001,
|
| 3973 |
+
"step": 565
|
| 3974 |
+
},
|
| 3975 |
+
{
|
| 3976 |
+
"epoch": 0.3229211239480816,
|
| 3977 |
+
"grad_norm": 0.9921875,
|
| 3978 |
+
"learning_rate": 3.5365853658536584e-05,
|
| 3979 |
+
"loss": 1.1981290578842163,
|
| 3980 |
+
"step": 566
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 0.32349165596919127,
|
| 3984 |
+
"grad_norm": 0.9765625,
|
| 3985 |
+
"learning_rate": 3.5338137472283814e-05,
|
| 3986 |
+
"loss": 1.1930001974105835,
|
| 3987 |
+
"step": 567
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 0.32406218799030095,
|
| 3991 |
+
"grad_norm": 1.0546875,
|
| 3992 |
+
"learning_rate": 3.531042128603104e-05,
|
| 3993 |
+
"loss": 1.2397738695144653,
|
| 3994 |
+
"step": 568
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 0.32463272001141064,
|
| 3998 |
+
"grad_norm": 1.078125,
|
| 3999 |
+
"learning_rate": 3.528270509977827e-05,
|
| 4000 |
+
"loss": 1.273198127746582,
|
| 4001 |
+
"step": 569
|
| 4002 |
+
},
|
| 4003 |
+
{
|
| 4004 |
+
"epoch": 0.3252032520325203,
|
| 4005 |
+
"grad_norm": 0.96875,
|
| 4006 |
+
"learning_rate": 3.52549889135255e-05,
|
| 4007 |
+
"loss": 1.1873741149902344,
|
| 4008 |
+
"step": 570
|
| 4009 |
+
},
|
| 4010 |
+
{
|
| 4011 |
+
"epoch": 0.32577378405363,
|
| 4012 |
+
"grad_norm": 0.9921875,
|
| 4013 |
+
"learning_rate": 3.522727272727273e-05,
|
| 4014 |
+
"loss": 1.2132840156555176,
|
| 4015 |
+
"step": 571
|
| 4016 |
+
},
|
| 4017 |
+
{
|
| 4018 |
+
"epoch": 0.3263443160747397,
|
| 4019 |
+
"grad_norm": 0.984375,
|
| 4020 |
+
"learning_rate": 3.5199556541019954e-05,
|
| 4021 |
+
"loss": 1.1881725788116455,
|
| 4022 |
+
"step": 572
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 0.32691484809584936,
|
| 4026 |
+
"grad_norm": 1.0546875,
|
| 4027 |
+
"learning_rate": 3.517184035476719e-05,
|
| 4028 |
+
"loss": 1.2296414375305176,
|
| 4029 |
+
"step": 573
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 0.32748538011695905,
|
| 4033 |
+
"grad_norm": 0.98828125,
|
| 4034 |
+
"learning_rate": 3.514412416851441e-05,
|
| 4035 |
+
"loss": 1.2116769552230835,
|
| 4036 |
+
"step": 574
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 0.32805591213806873,
|
| 4040 |
+
"grad_norm": 1.0,
|
| 4041 |
+
"learning_rate": 3.511640798226164e-05,
|
| 4042 |
+
"loss": 1.194542646408081,
|
| 4043 |
+
"step": 575
|
| 4044 |
+
},
|
| 4045 |
+
{
|
| 4046 |
+
"epoch": 0.3286264441591784,
|
| 4047 |
+
"grad_norm": 1.03125,
|
| 4048 |
+
"learning_rate": 3.508869179600887e-05,
|
| 4049 |
+
"loss": 1.2189078330993652,
|
| 4050 |
+
"step": 576
|
| 4051 |
+
},
|
| 4052 |
+
{
|
| 4053 |
+
"epoch": 0.3291969761802881,
|
| 4054 |
+
"grad_norm": 0.96484375,
|
| 4055 |
+
"learning_rate": 3.5060975609756095e-05,
|
| 4056 |
+
"loss": 1.1380560398101807,
|
| 4057 |
+
"step": 577
|
| 4058 |
+
},
|
| 4059 |
+
{
|
| 4060 |
+
"epoch": 0.32976750820139783,
|
| 4061 |
+
"grad_norm": 1.015625,
|
| 4062 |
+
"learning_rate": 3.503325942350333e-05,
|
| 4063 |
+
"loss": 1.1995842456817627,
|
| 4064 |
+
"step": 578
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 0.3303380402225075,
|
| 4068 |
+
"grad_norm": 1.0625,
|
| 4069 |
+
"learning_rate": 3.5005543237250554e-05,
|
| 4070 |
+
"loss": 1.254304051399231,
|
| 4071 |
+
"step": 579
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 0.3309085722436172,
|
| 4075 |
+
"grad_norm": 1.0,
|
| 4076 |
+
"learning_rate": 3.497782705099778e-05,
|
| 4077 |
+
"loss": 1.201616883277893,
|
| 4078 |
+
"step": 580
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 0.3314791042647269,
|
| 4082 |
+
"grad_norm": 0.98828125,
|
| 4083 |
+
"learning_rate": 3.495011086474501e-05,
|
| 4084 |
+
"loss": 1.1772336959838867,
|
| 4085 |
+
"step": 581
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 0.33204963628583656,
|
| 4089 |
+
"grad_norm": 1.0234375,
|
| 4090 |
+
"learning_rate": 3.492239467849224e-05,
|
| 4091 |
+
"loss": 1.1937668323516846,
|
| 4092 |
+
"step": 582
|
| 4093 |
+
},
|
| 4094 |
+
{
|
| 4095 |
+
"epoch": 0.33262016830694624,
|
| 4096 |
+
"grad_norm": 0.9765625,
|
| 4097 |
+
"learning_rate": 3.489467849223947e-05,
|
| 4098 |
+
"loss": 1.186886191368103,
|
| 4099 |
+
"step": 583
|
| 4100 |
+
},
|
| 4101 |
+
{
|
| 4102 |
+
"epoch": 0.3331907003280559,
|
| 4103 |
+
"grad_norm": 1.0390625,
|
| 4104 |
+
"learning_rate": 3.48669623059867e-05,
|
| 4105 |
+
"loss": 1.2187786102294922,
|
| 4106 |
+
"step": 584
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 0.3337612323491656,
|
| 4110 |
+
"grad_norm": 1.0234375,
|
| 4111 |
+
"learning_rate": 3.4839246119733924e-05,
|
| 4112 |
+
"loss": 1.1842401027679443,
|
| 4113 |
+
"step": 585
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 0.3343317643702753,
|
| 4117 |
+
"grad_norm": 1.0,
|
| 4118 |
+
"learning_rate": 3.481152993348116e-05,
|
| 4119 |
+
"loss": 1.1953545808792114,
|
| 4120 |
+
"step": 586
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 0.33490229639138497,
|
| 4124 |
+
"grad_norm": 1.0078125,
|
| 4125 |
+
"learning_rate": 3.478381374722838e-05,
|
| 4126 |
+
"loss": 1.1909786462783813,
|
| 4127 |
+
"step": 587
|
| 4128 |
+
},
|
| 4129 |
+
{
|
| 4130 |
+
"epoch": 0.33547282841249465,
|
| 4131 |
+
"grad_norm": 0.96484375,
|
| 4132 |
+
"learning_rate": 3.475609756097561e-05,
|
| 4133 |
+
"loss": 1.201062798500061,
|
| 4134 |
+
"step": 588
|
| 4135 |
+
},
|
| 4136 |
+
{
|
| 4137 |
+
"epoch": 0.33604336043360433,
|
| 4138 |
+
"grad_norm": 0.9921875,
|
| 4139 |
+
"learning_rate": 3.472838137472284e-05,
|
| 4140 |
+
"loss": 1.2262158393859863,
|
| 4141 |
+
"step": 589
|
| 4142 |
+
},
|
| 4143 |
+
{
|
| 4144 |
+
"epoch": 0.336613892454714,
|
| 4145 |
+
"grad_norm": 1.0390625,
|
| 4146 |
+
"learning_rate": 3.4700665188470064e-05,
|
| 4147 |
+
"loss": 1.255564570426941,
|
| 4148 |
+
"step": 590
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 0.3371844244758237,
|
| 4152 |
+
"grad_norm": 0.984375,
|
| 4153 |
+
"learning_rate": 3.46729490022173e-05,
|
| 4154 |
+
"loss": 1.1916460990905762,
|
| 4155 |
+
"step": 591
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 0.3377549564969334,
|
| 4159 |
+
"grad_norm": 1.03125,
|
| 4160 |
+
"learning_rate": 3.464523281596452e-05,
|
| 4161 |
+
"loss": 1.1728994846343994,
|
| 4162 |
+
"step": 592
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 0.33832548851804306,
|
| 4166 |
+
"grad_norm": 1.0234375,
|
| 4167 |
+
"learning_rate": 3.461751662971175e-05,
|
| 4168 |
+
"loss": 1.2145668268203735,
|
| 4169 |
+
"step": 593
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 0.33889602053915274,
|
| 4173 |
+
"grad_norm": 0.9921875,
|
| 4174 |
+
"learning_rate": 3.458980044345898e-05,
|
| 4175 |
+
"loss": 1.2174324989318848,
|
| 4176 |
+
"step": 594
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 0.3394665525602624,
|
| 4180 |
+
"grad_norm": 1.0390625,
|
| 4181 |
+
"learning_rate": 3.456208425720621e-05,
|
| 4182 |
+
"loss": 1.1968474388122559,
|
| 4183 |
+
"step": 595
|
| 4184 |
+
},
|
| 4185 |
+
{
|
| 4186 |
+
"epoch": 0.3400370845813721,
|
| 4187 |
+
"grad_norm": 1.0078125,
|
| 4188 |
+
"learning_rate": 3.453436807095344e-05,
|
| 4189 |
+
"loss": 1.1793067455291748,
|
| 4190 |
+
"step": 596
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 0.3406076166024818,
|
| 4194 |
+
"grad_norm": 1.03125,
|
| 4195 |
+
"learning_rate": 3.450665188470067e-05,
|
| 4196 |
+
"loss": 1.2109010219573975,
|
| 4197 |
+
"step": 597
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 0.3411781486235915,
|
| 4201 |
+
"grad_norm": 1.0546875,
|
| 4202 |
+
"learning_rate": 3.447893569844789e-05,
|
| 4203 |
+
"loss": 1.2412149906158447,
|
| 4204 |
+
"step": 598
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 0.3417486806447012,
|
| 4208 |
+
"grad_norm": 1.0078125,
|
| 4209 |
+
"learning_rate": 3.445121951219512e-05,
|
| 4210 |
+
"loss": 1.1886482238769531,
|
| 4211 |
+
"step": 599
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 0.3423192126658109,
|
| 4215 |
+
"grad_norm": 1.015625,
|
| 4216 |
+
"learning_rate": 3.442350332594235e-05,
|
| 4217 |
+
"loss": 1.1711212396621704,
|
| 4218 |
+
"step": 600
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 0.3428897446869206,
|
| 4222 |
+
"grad_norm": 0.98046875,
|
| 4223 |
+
"learning_rate": 3.4395787139689575e-05,
|
| 4224 |
+
"loss": 1.1890015602111816,
|
| 4225 |
+
"step": 601
|
| 4226 |
+
},
|
| 4227 |
+
{
|
| 4228 |
+
"epoch": 0.34346027670803025,
|
| 4229 |
+
"grad_norm": 0.95703125,
|
| 4230 |
+
"learning_rate": 3.436807095343681e-05,
|
| 4231 |
+
"loss": 1.1860285997390747,
|
| 4232 |
+
"step": 602
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 0.34403080872913994,
|
| 4236 |
+
"grad_norm": 0.98046875,
|
| 4237 |
+
"learning_rate": 3.4340354767184034e-05,
|
| 4238 |
+
"loss": 1.2001878023147583,
|
| 4239 |
+
"step": 603
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 0.3446013407502496,
|
| 4243 |
+
"grad_norm": 1.0234375,
|
| 4244 |
+
"learning_rate": 3.431263858093127e-05,
|
| 4245 |
+
"loss": 1.1815104484558105,
|
| 4246 |
+
"step": 604
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 0.3451718727713593,
|
| 4250 |
+
"grad_norm": 1.015625,
|
| 4251 |
+
"learning_rate": 3.428492239467849e-05,
|
| 4252 |
+
"loss": 1.1652307510375977,
|
| 4253 |
+
"step": 605
|
| 4254 |
+
},
|
| 4255 |
+
{
|
| 4256 |
+
"epoch": 0.345742404792469,
|
| 4257 |
+
"grad_norm": 1.0234375,
|
| 4258 |
+
"learning_rate": 3.425720620842572e-05,
|
| 4259 |
+
"loss": 1.1888481378555298,
|
| 4260 |
+
"step": 606
|
| 4261 |
+
},
|
| 4262 |
+
{
|
| 4263 |
+
"epoch": 0.34631293681357866,
|
| 4264 |
+
"grad_norm": 1.0390625,
|
| 4265 |
+
"learning_rate": 3.422949002217295e-05,
|
| 4266 |
+
"loss": 1.2198981046676636,
|
| 4267 |
+
"step": 607
|
| 4268 |
+
},
|
| 4269 |
+
{
|
| 4270 |
+
"epoch": 0.34688346883468835,
|
| 4271 |
+
"grad_norm": 1.03125,
|
| 4272 |
+
"learning_rate": 3.420177383592018e-05,
|
| 4273 |
+
"loss": 1.2088303565979004,
|
| 4274 |
+
"step": 608
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 0.34745400085579803,
|
| 4278 |
+
"grad_norm": 1.046875,
|
| 4279 |
+
"learning_rate": 3.4174057649667404e-05,
|
| 4280 |
+
"loss": 1.2638548612594604,
|
| 4281 |
+
"step": 609
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 0.3480245328769077,
|
| 4285 |
+
"grad_norm": 1.0234375,
|
| 4286 |
+
"learning_rate": 3.414634146341464e-05,
|
| 4287 |
+
"loss": 1.2314380407333374,
|
| 4288 |
+
"step": 610
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 0.3485950648980174,
|
| 4292 |
+
"grad_norm": 1.0,
|
| 4293 |
+
"learning_rate": 3.411862527716186e-05,
|
| 4294 |
+
"loss": 1.1847796440124512,
|
| 4295 |
+
"step": 611
|
| 4296 |
+
},
|
| 4297 |
+
{
|
| 4298 |
+
"epoch": 0.3491655969191271,
|
| 4299 |
+
"grad_norm": 1.015625,
|
| 4300 |
+
"learning_rate": 3.409090909090909e-05,
|
| 4301 |
+
"loss": 1.1967138051986694,
|
| 4302 |
+
"step": 612
|
| 4303 |
+
},
|
| 4304 |
+
{
|
| 4305 |
+
"epoch": 0.34973612894023676,
|
| 4306 |
+
"grad_norm": 1.0546875,
|
| 4307 |
+
"learning_rate": 3.406319290465632e-05,
|
| 4308 |
+
"loss": 1.1948060989379883,
|
| 4309 |
+
"step": 613
|
| 4310 |
+
},
|
| 4311 |
+
{
|
| 4312 |
+
"epoch": 0.35030666096134644,
|
| 4313 |
+
"grad_norm": 1.0078125,
|
| 4314 |
+
"learning_rate": 3.4035476718403544e-05,
|
| 4315 |
+
"loss": 1.248701810836792,
|
| 4316 |
+
"step": 614
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 0.3508771929824561,
|
| 4320 |
+
"grad_norm": 0.98046875,
|
| 4321 |
+
"learning_rate": 3.400776053215078e-05,
|
| 4322 |
+
"loss": 1.2076679468154907,
|
| 4323 |
+
"step": 615
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 0.3514477250035658,
|
| 4327 |
+
"grad_norm": 1.0234375,
|
| 4328 |
+
"learning_rate": 3.3980044345898e-05,
|
| 4329 |
+
"loss": 1.20987868309021,
|
| 4330 |
+
"step": 616
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 0.3520182570246755,
|
| 4334 |
+
"grad_norm": 0.9921875,
|
| 4335 |
+
"learning_rate": 3.395232815964523e-05,
|
| 4336 |
+
"loss": 1.1548939943313599,
|
| 4337 |
+
"step": 617
|
| 4338 |
+
},
|
| 4339 |
+
{
|
| 4340 |
+
"epoch": 0.35258878904578517,
|
| 4341 |
+
"grad_norm": 1.03125,
|
| 4342 |
+
"learning_rate": 3.392461197339246e-05,
|
| 4343 |
+
"loss": 1.2160520553588867,
|
| 4344 |
+
"step": 618
|
| 4345 |
+
},
|
| 4346 |
+
{
|
| 4347 |
+
"epoch": 0.3531593210668949,
|
| 4348 |
+
"grad_norm": 1.03125,
|
| 4349 |
+
"learning_rate": 3.389689578713969e-05,
|
| 4350 |
+
"loss": 1.2215287685394287,
|
| 4351 |
+
"step": 619
|
| 4352 |
+
},
|
| 4353 |
+
{
|
| 4354 |
+
"epoch": 0.3537298530880046,
|
| 4355 |
+
"grad_norm": 1.015625,
|
| 4356 |
+
"learning_rate": 3.386917960088692e-05,
|
| 4357 |
+
"loss": 1.2433137893676758,
|
| 4358 |
+
"step": 620
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 0.35430038510911427,
|
| 4362 |
+
"grad_norm": 1.015625,
|
| 4363 |
+
"learning_rate": 3.384146341463415e-05,
|
| 4364 |
+
"loss": 1.2307751178741455,
|
| 4365 |
+
"step": 621
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 0.35487091713022395,
|
| 4369 |
+
"grad_norm": 0.99609375,
|
| 4370 |
+
"learning_rate": 3.381374722838137e-05,
|
| 4371 |
+
"loss": 1.1872355937957764,
|
| 4372 |
+
"step": 622
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 0.35544144915133363,
|
| 4376 |
+
"grad_norm": 1.03125,
|
| 4377 |
+
"learning_rate": 3.378603104212861e-05,
|
| 4378 |
+
"loss": 1.200265645980835,
|
| 4379 |
+
"step": 623
|
| 4380 |
+
},
|
| 4381 |
+
{
|
| 4382 |
+
"epoch": 0.3560119811724433,
|
| 4383 |
+
"grad_norm": 1.0546875,
|
| 4384 |
+
"learning_rate": 3.375831485587583e-05,
|
| 4385 |
+
"loss": 1.3020355701446533,
|
| 4386 |
+
"step": 624
|
| 4387 |
+
},
|
| 4388 |
+
{
|
| 4389 |
+
"epoch": 0.356582513193553,
|
| 4390 |
+
"grad_norm": 1.03125,
|
| 4391 |
+
"learning_rate": 3.373059866962306e-05,
|
| 4392 |
+
"loss": 1.1976819038391113,
|
| 4393 |
+
"step": 625
|
| 4394 |
+
},
|
| 4395 |
+
{
|
| 4396 |
+
"epoch": 0.3571530452146627,
|
| 4397 |
+
"grad_norm": 1.0234375,
|
| 4398 |
+
"learning_rate": 3.370288248337029e-05,
|
| 4399 |
+
"loss": 1.1945629119873047,
|
| 4400 |
+
"step": 626
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 0.35772357723577236,
|
| 4404 |
+
"grad_norm": 0.9921875,
|
| 4405 |
+
"learning_rate": 3.3675166297117514e-05,
|
| 4406 |
+
"loss": 1.2189013957977295,
|
| 4407 |
+
"step": 627
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 0.35829410925688204,
|
| 4411 |
+
"grad_norm": 0.9765625,
|
| 4412 |
+
"learning_rate": 3.364745011086475e-05,
|
| 4413 |
+
"loss": 1.2139533758163452,
|
| 4414 |
+
"step": 628
|
| 4415 |
+
},
|
| 4416 |
+
{
|
| 4417 |
+
"epoch": 0.3588646412779917,
|
| 4418 |
+
"grad_norm": 0.96484375,
|
| 4419 |
+
"learning_rate": 3.361973392461197e-05,
|
| 4420 |
+
"loss": 1.1832334995269775,
|
| 4421 |
+
"step": 629
|
| 4422 |
+
},
|
| 4423 |
+
{
|
| 4424 |
+
"epoch": 0.3594351732991014,
|
| 4425 |
+
"grad_norm": 1.0078125,
|
| 4426 |
+
"learning_rate": 3.35920177383592e-05,
|
| 4427 |
+
"loss": 1.1789777278900146,
|
| 4428 |
+
"step": 630
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 0.3600057053202111,
|
| 4432 |
+
"grad_norm": 0.97265625,
|
| 4433 |
+
"learning_rate": 3.356430155210643e-05,
|
| 4434 |
+
"loss": 1.1401221752166748,
|
| 4435 |
+
"step": 631
|
| 4436 |
+
},
|
| 4437 |
+
{
|
| 4438 |
+
"epoch": 0.36057623734132077,
|
| 4439 |
+
"grad_norm": 1.015625,
|
| 4440 |
+
"learning_rate": 3.353658536585366e-05,
|
| 4441 |
+
"loss": 1.2332661151885986,
|
| 4442 |
+
"step": 632
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 0.36114676936243045,
|
| 4446 |
+
"grad_norm": 0.9609375,
|
| 4447 |
+
"learning_rate": 3.350886917960089e-05,
|
| 4448 |
+
"loss": 1.1867516040802002,
|
| 4449 |
+
"step": 633
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 0.36171730138354014,
|
| 4453 |
+
"grad_norm": 1.0078125,
|
| 4454 |
+
"learning_rate": 3.348115299334812e-05,
|
| 4455 |
+
"loss": 1.2486271858215332,
|
| 4456 |
+
"step": 634
|
| 4457 |
+
},
|
| 4458 |
+
{
|
| 4459 |
+
"epoch": 0.3622878334046498,
|
| 4460 |
+
"grad_norm": 1.0390625,
|
| 4461 |
+
"learning_rate": 3.345343680709534e-05,
|
| 4462 |
+
"loss": 1.1644282341003418,
|
| 4463 |
+
"step": 635
|
| 4464 |
+
},
|
| 4465 |
+
{
|
| 4466 |
+
"epoch": 0.3628583654257595,
|
| 4467 |
+
"grad_norm": 0.98828125,
|
| 4468 |
+
"learning_rate": 3.342572062084257e-05,
|
| 4469 |
+
"loss": 1.1926931142807007,
|
| 4470 |
+
"step": 636
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 0.3634288974468692,
|
| 4474 |
+
"grad_norm": 0.98046875,
|
| 4475 |
+
"learning_rate": 3.33980044345898e-05,
|
| 4476 |
+
"loss": 1.2337167263031006,
|
| 4477 |
+
"step": 637
|
| 4478 |
+
},
|
| 4479 |
+
{
|
| 4480 |
+
"epoch": 0.36399942946797886,
|
| 4481 |
+
"grad_norm": 1.0078125,
|
| 4482 |
+
"learning_rate": 3.337028824833703e-05,
|
| 4483 |
+
"loss": 1.2726258039474487,
|
| 4484 |
+
"step": 638
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 0.3645699614890886,
|
| 4488 |
+
"grad_norm": 1.046875,
|
| 4489 |
+
"learning_rate": 3.334257206208426e-05,
|
| 4490 |
+
"loss": 1.229848861694336,
|
| 4491 |
+
"step": 639
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 0.3651404935101983,
|
| 4495 |
+
"grad_norm": 0.94921875,
|
| 4496 |
+
"learning_rate": 3.3314855875831483e-05,
|
| 4497 |
+
"loss": 1.1424199342727661,
|
| 4498 |
+
"step": 640
|
| 4499 |
+
},
|
| 4500 |
+
{
|
| 4501 |
+
"epoch": 0.36571102553130796,
|
| 4502 |
+
"grad_norm": 1.0234375,
|
| 4503 |
+
"learning_rate": 3.328713968957872e-05,
|
| 4504 |
+
"loss": 1.2158143520355225,
|
| 4505 |
+
"step": 641
|
| 4506 |
+
},
|
| 4507 |
+
{
|
| 4508 |
+
"epoch": 0.36628155755241765,
|
| 4509 |
+
"grad_norm": 1.0234375,
|
| 4510 |
+
"learning_rate": 3.325942350332594e-05,
|
| 4511 |
+
"loss": 1.213433027267456,
|
| 4512 |
+
"step": 642
|
| 4513 |
+
},
|
| 4514 |
+
{
|
| 4515 |
+
"epoch": 0.36685208957352733,
|
| 4516 |
+
"grad_norm": 1.0234375,
|
| 4517 |
+
"learning_rate": 3.323170731707317e-05,
|
| 4518 |
+
"loss": 1.1552369594573975,
|
| 4519 |
+
"step": 643
|
| 4520 |
+
},
|
| 4521 |
+
{
|
| 4522 |
+
"epoch": 0.367422621594637,
|
| 4523 |
+
"grad_norm": 0.9765625,
|
| 4524 |
+
"learning_rate": 3.32039911308204e-05,
|
| 4525 |
+
"loss": 1.1470410823822021,
|
| 4526 |
+
"step": 644
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 0.3679931536157467,
|
| 4530 |
+
"grad_norm": 1.0234375,
|
| 4531 |
+
"learning_rate": 3.317627494456763e-05,
|
| 4532 |
+
"loss": 1.227137804031372,
|
| 4533 |
+
"step": 645
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 0.3685636856368564,
|
| 4537 |
+
"grad_norm": 0.98046875,
|
| 4538 |
+
"learning_rate": 3.314855875831486e-05,
|
| 4539 |
+
"loss": 1.1736478805541992,
|
| 4540 |
+
"step": 646
|
| 4541 |
+
},
|
| 4542 |
+
{
|
| 4543 |
+
"epoch": 0.36913421765796606,
|
| 4544 |
+
"grad_norm": 1.0078125,
|
| 4545 |
+
"learning_rate": 3.312084257206209e-05,
|
| 4546 |
+
"loss": 1.2192144393920898,
|
| 4547 |
+
"step": 647
|
| 4548 |
+
},
|
| 4549 |
+
{
|
| 4550 |
+
"epoch": 0.36970474967907574,
|
| 4551 |
+
"grad_norm": 0.97265625,
|
| 4552 |
+
"learning_rate": 3.309312638580931e-05,
|
| 4553 |
+
"loss": 1.1780518293380737,
|
| 4554 |
+
"step": 648
|
| 4555 |
+
},
|
| 4556 |
+
{
|
| 4557 |
+
"epoch": 0.3702752817001854,
|
| 4558 |
+
"grad_norm": 1.015625,
|
| 4559 |
+
"learning_rate": 3.306541019955654e-05,
|
| 4560 |
+
"loss": 1.2205878496170044,
|
| 4561 |
+
"step": 649
|
| 4562 |
+
},
|
| 4563 |
+
{
|
| 4564 |
+
"epoch": 0.3708458137212951,
|
| 4565 |
+
"grad_norm": 0.99609375,
|
| 4566 |
+
"learning_rate": 3.303769401330377e-05,
|
| 4567 |
+
"loss": 1.2226086854934692,
|
| 4568 |
+
"step": 650
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 0.3714163457424048,
|
| 4572 |
+
"grad_norm": 1.0078125,
|
| 4573 |
+
"learning_rate": 3.3009977827051e-05,
|
| 4574 |
+
"loss": 1.1905219554901123,
|
| 4575 |
+
"step": 651
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 0.37198687776351447,
|
| 4579 |
+
"grad_norm": 1.015625,
|
| 4580 |
+
"learning_rate": 3.298226164079823e-05,
|
| 4581 |
+
"loss": 1.1790423393249512,
|
| 4582 |
+
"step": 652
|
| 4583 |
+
},
|
| 4584 |
+
{
|
| 4585 |
+
"epoch": 0.37255740978462415,
|
| 4586 |
+
"grad_norm": 0.9609375,
|
| 4587 |
+
"learning_rate": 3.295454545454545e-05,
|
| 4588 |
+
"loss": 1.1909444332122803,
|
| 4589 |
+
"step": 653
|
| 4590 |
+
},
|
| 4591 |
+
{
|
| 4592 |
+
"epoch": 0.37312794180573383,
|
| 4593 |
+
"grad_norm": 0.98828125,
|
| 4594 |
+
"learning_rate": 3.292682926829269e-05,
|
| 4595 |
+
"loss": 1.2416154146194458,
|
| 4596 |
+
"step": 654
|
| 4597 |
+
},
|
| 4598 |
+
{
|
| 4599 |
+
"epoch": 0.3736984738268435,
|
| 4600 |
+
"grad_norm": 1.0234375,
|
| 4601 |
+
"learning_rate": 3.289911308203991e-05,
|
| 4602 |
+
"loss": 1.2464513778686523,
|
| 4603 |
+
"step": 655
|
| 4604 |
+
},
|
| 4605 |
+
{
|
| 4606 |
+
"epoch": 0.3742690058479532,
|
| 4607 |
+
"grad_norm": 1.015625,
|
| 4608 |
+
"learning_rate": 3.287139689578714e-05,
|
| 4609 |
+
"loss": 1.239952802658081,
|
| 4610 |
+
"step": 656
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 0.3748395378690629,
|
| 4614 |
+
"grad_norm": 0.99609375,
|
| 4615 |
+
"learning_rate": 3.284368070953437e-05,
|
| 4616 |
+
"loss": 1.2005925178527832,
|
| 4617 |
+
"step": 657
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 0.37541006989017256,
|
| 4621 |
+
"grad_norm": 1.0078125,
|
| 4622 |
+
"learning_rate": 3.28159645232816e-05,
|
| 4623 |
+
"loss": 1.2646636962890625,
|
| 4624 |
+
"step": 658
|
| 4625 |
+
},
|
| 4626 |
+
{
|
| 4627 |
+
"epoch": 0.3759806019112823,
|
| 4628 |
+
"grad_norm": 0.98828125,
|
| 4629 |
+
"learning_rate": 3.278824833702882e-05,
|
| 4630 |
+
"loss": 1.203331470489502,
|
| 4631 |
+
"step": 659
|
| 4632 |
+
},
|
| 4633 |
+
{
|
| 4634 |
+
"epoch": 0.376551133932392,
|
| 4635 |
+
"grad_norm": 0.97265625,
|
| 4636 |
+
"learning_rate": 3.276053215077605e-05,
|
| 4637 |
+
"loss": 1.1849339008331299,
|
| 4638 |
+
"step": 660
|
| 4639 |
+
},
|
| 4640 |
+
{
|
| 4641 |
+
"epoch": 0.37712166595350166,
|
| 4642 |
+
"grad_norm": 1.0078125,
|
| 4643 |
+
"learning_rate": 3.273281596452328e-05,
|
| 4644 |
+
"loss": 1.2010148763656616,
|
| 4645 |
+
"step": 661
|
| 4646 |
+
},
|
| 4647 |
+
{
|
| 4648 |
+
"epoch": 0.37769219797461134,
|
| 4649 |
+
"grad_norm": 0.9921875,
|
| 4650 |
+
"learning_rate": 3.270509977827051e-05,
|
| 4651 |
+
"loss": 1.2384660243988037,
|
| 4652 |
+
"step": 662
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 0.378262729995721,
|
| 4656 |
+
"grad_norm": 0.984375,
|
| 4657 |
+
"learning_rate": 3.267738359201774e-05,
|
| 4658 |
+
"loss": 1.2244110107421875,
|
| 4659 |
+
"step": 663
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 0.3788332620168307,
|
| 4663 |
+
"grad_norm": 1.015625,
|
| 4664 |
+
"learning_rate": 3.2649667405764963e-05,
|
| 4665 |
+
"loss": 1.2706053256988525,
|
| 4666 |
+
"step": 664
|
| 4667 |
+
},
|
| 4668 |
+
{
|
| 4669 |
+
"epoch": 0.3794037940379404,
|
| 4670 |
+
"grad_norm": 0.98828125,
|
| 4671 |
+
"learning_rate": 3.26219512195122e-05,
|
| 4672 |
+
"loss": 1.2451549768447876,
|
| 4673 |
+
"step": 665
|
| 4674 |
+
},
|
| 4675 |
+
{
|
| 4676 |
+
"epoch": 0.37997432605905007,
|
| 4677 |
+
"grad_norm": 1.015625,
|
| 4678 |
+
"learning_rate": 3.259423503325942e-05,
|
| 4679 |
+
"loss": 1.2653909921646118,
|
| 4680 |
+
"step": 666
|
| 4681 |
+
},
|
| 4682 |
+
{
|
| 4683 |
+
"epoch": 0.38054485808015975,
|
| 4684 |
+
"grad_norm": 1.0078125,
|
| 4685 |
+
"learning_rate": 3.256651884700665e-05,
|
| 4686 |
+
"loss": 1.2227097749710083,
|
| 4687 |
+
"step": 667
|
| 4688 |
+
},
|
| 4689 |
+
{
|
| 4690 |
+
"epoch": 0.38111539010126944,
|
| 4691 |
+
"grad_norm": 0.98828125,
|
| 4692 |
+
"learning_rate": 3.253880266075388e-05,
|
| 4693 |
+
"loss": 1.2289211750030518,
|
| 4694 |
+
"step": 668
|
| 4695 |
+
},
|
| 4696 |
+
{
|
| 4697 |
+
"epoch": 0.3816859221223791,
|
| 4698 |
+
"grad_norm": 1.0078125,
|
| 4699 |
+
"learning_rate": 3.251108647450111e-05,
|
| 4700 |
+
"loss": 1.2068843841552734,
|
| 4701 |
+
"step": 669
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 0.3822564541434888,
|
| 4705 |
+
"grad_norm": 0.9453125,
|
| 4706 |
+
"learning_rate": 3.248337028824834e-05,
|
| 4707 |
+
"loss": 1.166361689567566,
|
| 4708 |
+
"step": 670
|
| 4709 |
+
},
|
| 4710 |
+
{
|
| 4711 |
+
"epoch": 0.3828269861645985,
|
| 4712 |
+
"grad_norm": 0.9609375,
|
| 4713 |
+
"learning_rate": 3.245565410199557e-05,
|
| 4714 |
+
"loss": 1.220710277557373,
|
| 4715 |
+
"step": 671
|
| 4716 |
+
},
|
| 4717 |
+
{
|
| 4718 |
+
"epoch": 0.38339751818570816,
|
| 4719 |
+
"grad_norm": 0.984375,
|
| 4720 |
+
"learning_rate": 3.242793791574279e-05,
|
| 4721 |
+
"loss": 1.1663460731506348,
|
| 4722 |
+
"step": 672
|
| 4723 |
+
},
|
| 4724 |
+
{
|
| 4725 |
+
"epoch": 0.38396805020681785,
|
| 4726 |
+
"grad_norm": 1.0078125,
|
| 4727 |
+
"learning_rate": 3.240022172949002e-05,
|
| 4728 |
+
"loss": 1.1803617477416992,
|
| 4729 |
+
"step": 673
|
| 4730 |
+
},
|
| 4731 |
+
{
|
| 4732 |
+
"epoch": 0.38453858222792753,
|
| 4733 |
+
"grad_norm": 0.9921875,
|
| 4734 |
+
"learning_rate": 3.237250554323725e-05,
|
| 4735 |
+
"loss": 1.1342628002166748,
|
| 4736 |
+
"step": 674
|
| 4737 |
+
},
|
| 4738 |
+
{
|
| 4739 |
+
"epoch": 0.3851091142490372,
|
| 4740 |
+
"grad_norm": 0.9921875,
|
| 4741 |
+
"learning_rate": 3.234478935698448e-05,
|
| 4742 |
+
"loss": 1.2325470447540283,
|
| 4743 |
+
"step": 675
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 0.3856796462701469,
|
| 4747 |
+
"grad_norm": 1.0078125,
|
| 4748 |
+
"learning_rate": 3.231707317073171e-05,
|
| 4749 |
+
"loss": 1.1941877603530884,
|
| 4750 |
+
"step": 676
|
| 4751 |
+
},
|
| 4752 |
+
{
|
| 4753 |
+
"epoch": 0.3862501782912566,
|
| 4754 |
+
"grad_norm": 1.0,
|
| 4755 |
+
"learning_rate": 3.228935698447893e-05,
|
| 4756 |
+
"loss": 1.1775301694869995,
|
| 4757 |
+
"step": 677
|
| 4758 |
+
},
|
| 4759 |
+
{
|
| 4760 |
+
"epoch": 0.38682071031236626,
|
| 4761 |
+
"grad_norm": 1.0546875,
|
| 4762 |
+
"learning_rate": 3.226164079822617e-05,
|
| 4763 |
+
"loss": 1.248462438583374,
|
| 4764 |
+
"step": 678
|
| 4765 |
+
},
|
| 4766 |
+
{
|
| 4767 |
+
"epoch": 0.387391242333476,
|
| 4768 |
+
"grad_norm": 1.03125,
|
| 4769 |
+
"learning_rate": 3.223392461197339e-05,
|
| 4770 |
+
"loss": 1.2440953254699707,
|
| 4771 |
+
"step": 679
|
| 4772 |
+
},
|
| 4773 |
+
{
|
| 4774 |
+
"epoch": 0.3879617743545857,
|
| 4775 |
+
"grad_norm": 0.98046875,
|
| 4776 |
+
"learning_rate": 3.220620842572062e-05,
|
| 4777 |
+
"loss": 1.1706881523132324,
|
| 4778 |
+
"step": 680
|
| 4779 |
+
},
|
| 4780 |
+
{
|
| 4781 |
+
"epoch": 0.38853230637569536,
|
| 4782 |
+
"grad_norm": 1.0390625,
|
| 4783 |
+
"learning_rate": 3.217849223946785e-05,
|
| 4784 |
+
"loss": 1.227694034576416,
|
| 4785 |
+
"step": 681
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 0.38910283839680504,
|
| 4789 |
+
"grad_norm": 1.0234375,
|
| 4790 |
+
"learning_rate": 3.215077605321508e-05,
|
| 4791 |
+
"loss": 1.2553303241729736,
|
| 4792 |
+
"step": 682
|
| 4793 |
+
},
|
| 4794 |
+
{
|
| 4795 |
+
"epoch": 0.3896733704179147,
|
| 4796 |
+
"grad_norm": 0.94140625,
|
| 4797 |
+
"learning_rate": 3.212305986696231e-05,
|
| 4798 |
+
"loss": 1.1399942636489868,
|
| 4799 |
+
"step": 683
|
| 4800 |
+
},
|
| 4801 |
+
{
|
| 4802 |
+
"epoch": 0.3902439024390244,
|
| 4803 |
+
"grad_norm": 0.98828125,
|
| 4804 |
+
"learning_rate": 3.209534368070954e-05,
|
| 4805 |
+
"loss": 1.2082273960113525,
|
| 4806 |
+
"step": 684
|
| 4807 |
+
},
|
| 4808 |
+
{
|
| 4809 |
+
"epoch": 0.3908144344601341,
|
| 4810 |
+
"grad_norm": 1.0625,
|
| 4811 |
+
"learning_rate": 3.206762749445676e-05,
|
| 4812 |
+
"loss": 1.2403631210327148,
|
| 4813 |
+
"step": 685
|
| 4814 |
+
},
|
| 4815 |
+
{
|
| 4816 |
+
"epoch": 0.39138496648124377,
|
| 4817 |
+
"grad_norm": 1.03125,
|
| 4818 |
+
"learning_rate": 3.203991130820399e-05,
|
| 4819 |
+
"loss": 1.1668493747711182,
|
| 4820 |
+
"step": 686
|
| 4821 |
+
},
|
| 4822 |
+
{
|
| 4823 |
+
"epoch": 0.39195549850235345,
|
| 4824 |
+
"grad_norm": 1.0,
|
| 4825 |
+
"learning_rate": 3.201219512195122e-05,
|
| 4826 |
+
"loss": 1.1642647981643677,
|
| 4827 |
+
"step": 687
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 0.39252603052346313,
|
| 4831 |
+
"grad_norm": 0.94921875,
|
| 4832 |
+
"learning_rate": 3.198447893569845e-05,
|
| 4833 |
+
"loss": 1.169840693473816,
|
| 4834 |
+
"step": 688
|
| 4835 |
+
},
|
| 4836 |
+
{
|
| 4837 |
+
"epoch": 0.3930965625445728,
|
| 4838 |
+
"grad_norm": 1.0625,
|
| 4839 |
+
"learning_rate": 3.195676274944568e-05,
|
| 4840 |
+
"loss": 1.1918284893035889,
|
| 4841 |
+
"step": 689
|
| 4842 |
+
},
|
| 4843 |
+
{
|
| 4844 |
+
"epoch": 0.3936670945656825,
|
| 4845 |
+
"grad_norm": 1.0546875,
|
| 4846 |
+
"learning_rate": 3.19290465631929e-05,
|
| 4847 |
+
"loss": 1.2486236095428467,
|
| 4848 |
+
"step": 690
|
| 4849 |
+
},
|
| 4850 |
+
{
|
| 4851 |
+
"epoch": 0.3942376265867922,
|
| 4852 |
+
"grad_norm": 1.0,
|
| 4853 |
+
"learning_rate": 3.190133037694014e-05,
|
| 4854 |
+
"loss": 1.212164044380188,
|
| 4855 |
+
"step": 691
|
| 4856 |
+
},
|
| 4857 |
+
{
|
| 4858 |
+
"epoch": 0.39480815860790186,
|
| 4859 |
+
"grad_norm": 1.0,
|
| 4860 |
+
"learning_rate": 3.187361419068736e-05,
|
| 4861 |
+
"loss": 1.2184773683547974,
|
| 4862 |
+
"step": 692
|
| 4863 |
+
},
|
| 4864 |
+
{
|
| 4865 |
+
"epoch": 0.39537869062901154,
|
| 4866 |
+
"grad_norm": 1.0625,
|
| 4867 |
+
"learning_rate": 3.184589800443459e-05,
|
| 4868 |
+
"loss": 1.2665815353393555,
|
| 4869 |
+
"step": 693
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 0.3959492226501212,
|
| 4873 |
+
"grad_norm": 1.0625,
|
| 4874 |
+
"learning_rate": 3.181818181818182e-05,
|
| 4875 |
+
"loss": 1.1956299543380737,
|
| 4876 |
+
"step": 694
|
| 4877 |
+
},
|
| 4878 |
+
{
|
| 4879 |
+
"epoch": 0.3965197546712309,
|
| 4880 |
+
"grad_norm": 0.98828125,
|
| 4881 |
+
"learning_rate": 3.179046563192905e-05,
|
| 4882 |
+
"loss": 1.1868462562561035,
|
| 4883 |
+
"step": 695
|
| 4884 |
+
},
|
| 4885 |
+
{
|
| 4886 |
+
"epoch": 0.3970902866923406,
|
| 4887 |
+
"grad_norm": 1.0234375,
|
| 4888 |
+
"learning_rate": 3.176274944567628e-05,
|
| 4889 |
+
"loss": 1.2558304071426392,
|
| 4890 |
+
"step": 696
|
| 4891 |
+
},
|
| 4892 |
+
{
|
| 4893 |
+
"epoch": 0.39766081871345027,
|
| 4894 |
+
"grad_norm": 1.0078125,
|
| 4895 |
+
"learning_rate": 3.17350332594235e-05,
|
| 4896 |
+
"loss": 1.2197167873382568,
|
| 4897 |
+
"step": 697
|
| 4898 |
+
},
|
| 4899 |
+
{
|
| 4900 |
+
"epoch": 0.39823135073455995,
|
| 4901 |
+
"grad_norm": 1.0390625,
|
| 4902 |
+
"learning_rate": 3.170731707317073e-05,
|
| 4903 |
+
"loss": 1.2546510696411133,
|
| 4904 |
+
"step": 698
|
| 4905 |
+
},
|
| 4906 |
+
{
|
| 4907 |
+
"epoch": 0.39880188275566963,
|
| 4908 |
+
"grad_norm": 1.078125,
|
| 4909 |
+
"learning_rate": 3.167960088691796e-05,
|
| 4910 |
+
"loss": 1.2634811401367188,
|
| 4911 |
+
"step": 699
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 0.39937241477677937,
|
| 4915 |
+
"grad_norm": 0.953125,
|
| 4916 |
+
"learning_rate": 3.165188470066519e-05,
|
| 4917 |
+
"loss": 1.1409438848495483,
|
| 4918 |
+
"step": 700
|
| 4919 |
+
},
|
| 4920 |
+
{
|
| 4921 |
+
"epoch": 0.39994294679788905,
|
| 4922 |
+
"grad_norm": 1.0,
|
| 4923 |
+
"learning_rate": 3.162416851441242e-05,
|
| 4924 |
+
"loss": 1.167540431022644,
|
| 4925 |
+
"step": 701
|
| 4926 |
+
},
|
| 4927 |
+
{
|
| 4928 |
+
"epoch": 0.40051347881899874,
|
| 4929 |
+
"grad_norm": 0.98828125,
|
| 4930 |
+
"learning_rate": 3.159645232815965e-05,
|
| 4931 |
+
"loss": 1.2233819961547852,
|
| 4932 |
+
"step": 702
|
| 4933 |
+
},
|
| 4934 |
+
{
|
| 4935 |
+
"epoch": 0.4010840108401084,
|
| 4936 |
+
"grad_norm": 1.0625,
|
| 4937 |
+
"learning_rate": 3.156873614190687e-05,
|
| 4938 |
+
"loss": 1.2183570861816406,
|
| 4939 |
+
"step": 703
|
| 4940 |
+
},
|
| 4941 |
+
{
|
| 4942 |
+
"epoch": 0.4016545428612181,
|
| 4943 |
+
"grad_norm": 1.0234375,
|
| 4944 |
+
"learning_rate": 3.154101995565411e-05,
|
| 4945 |
+
"loss": 1.2039064168930054,
|
| 4946 |
+
"step": 704
|
| 4947 |
+
},
|
| 4948 |
+
{
|
| 4949 |
+
"epoch": 0.4022250748823278,
|
| 4950 |
+
"grad_norm": 1.0234375,
|
| 4951 |
+
"learning_rate": 3.151330376940133e-05,
|
| 4952 |
+
"loss": 1.2583222389221191,
|
| 4953 |
+
"step": 705
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 0.40279560690343746,
|
| 4957 |
+
"grad_norm": 0.9765625,
|
| 4958 |
+
"learning_rate": 3.148558758314856e-05,
|
| 4959 |
+
"loss": 1.2133885622024536,
|
| 4960 |
+
"step": 706
|
| 4961 |
+
},
|
| 4962 |
+
{
|
| 4963 |
+
"epoch": 0.40336613892454715,
|
| 4964 |
+
"grad_norm": 0.99609375,
|
| 4965 |
+
"learning_rate": 3.145787139689579e-05,
|
| 4966 |
+
"loss": 1.2497689723968506,
|
| 4967 |
+
"step": 707
|
| 4968 |
+
},
|
| 4969 |
+
{
|
| 4970 |
+
"epoch": 0.40393667094565683,
|
| 4971 |
+
"grad_norm": 1.015625,
|
| 4972 |
+
"learning_rate": 3.143015521064302e-05,
|
| 4973 |
+
"loss": 1.1765098571777344,
|
| 4974 |
+
"step": 708
|
| 4975 |
+
},
|
| 4976 |
+
{
|
| 4977 |
+
"epoch": 0.4045072029667665,
|
| 4978 |
+
"grad_norm": 0.9765625,
|
| 4979 |
+
"learning_rate": 3.140243902439025e-05,
|
| 4980 |
+
"loss": 1.1668319702148438,
|
| 4981 |
+
"step": 709
|
| 4982 |
+
},
|
| 4983 |
+
{
|
| 4984 |
+
"epoch": 0.4050777349878762,
|
| 4985 |
+
"grad_norm": 1.0,
|
| 4986 |
+
"learning_rate": 3.137472283813747e-05,
|
| 4987 |
+
"loss": 1.1545255184173584,
|
| 4988 |
+
"step": 710
|
| 4989 |
+
},
|
| 4990 |
+
{
|
| 4991 |
+
"epoch": 0.4056482670089859,
|
| 4992 |
+
"grad_norm": 1.0,
|
| 4993 |
+
"learning_rate": 3.13470066518847e-05,
|
| 4994 |
+
"loss": 1.2044893503189087,
|
| 4995 |
+
"step": 711
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 0.40621879903009556,
|
| 4999 |
+
"grad_norm": 0.99609375,
|
| 5000 |
+
"learning_rate": 3.131929046563193e-05,
|
| 5001 |
+
"loss": 1.2121517658233643,
|
| 5002 |
+
"step": 712
|
| 5003 |
+
},
|
| 5004 |
+
{
|
| 5005 |
+
"epoch": 0.40678933105120524,
|
| 5006 |
+
"grad_norm": 1.0390625,
|
| 5007 |
+
"learning_rate": 3.129157427937916e-05,
|
| 5008 |
+
"loss": 1.276052713394165,
|
| 5009 |
+
"step": 713
|
| 5010 |
+
},
|
| 5011 |
+
{
|
| 5012 |
+
"epoch": 0.4073598630723149,
|
| 5013 |
+
"grad_norm": 1.015625,
|
| 5014 |
+
"learning_rate": 3.126385809312638e-05,
|
| 5015 |
+
"loss": 1.1800833940505981,
|
| 5016 |
+
"step": 714
|
| 5017 |
+
},
|
| 5018 |
+
{
|
| 5019 |
+
"epoch": 0.4079303950934246,
|
| 5020 |
+
"grad_norm": 0.984375,
|
| 5021 |
+
"learning_rate": 3.123614190687362e-05,
|
| 5022 |
+
"loss": 1.1513339281082153,
|
| 5023 |
+
"step": 715
|
| 5024 |
+
},
|
| 5025 |
+
{
|
| 5026 |
+
"epoch": 0.4085009271145343,
|
| 5027 |
+
"grad_norm": 0.98046875,
|
| 5028 |
+
"learning_rate": 3.120842572062084e-05,
|
| 5029 |
+
"loss": 1.2298616170883179,
|
| 5030 |
+
"step": 716
|
| 5031 |
+
},
|
| 5032 |
+
{
|
| 5033 |
+
"epoch": 0.40907145913564397,
|
| 5034 |
+
"grad_norm": 0.9765625,
|
| 5035 |
+
"learning_rate": 3.118070953436808e-05,
|
| 5036 |
+
"loss": 1.1709084510803223,
|
| 5037 |
+
"step": 717
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 0.40964199115675365,
|
| 5041 |
+
"grad_norm": 0.98828125,
|
| 5042 |
+
"learning_rate": 3.11529933481153e-05,
|
| 5043 |
+
"loss": 1.1676058769226074,
|
| 5044 |
+
"step": 718
|
| 5045 |
+
},
|
| 5046 |
+
{
|
| 5047 |
+
"epoch": 0.41021252317786333,
|
| 5048 |
+
"grad_norm": 0.98828125,
|
| 5049 |
+
"learning_rate": 3.112527716186253e-05,
|
| 5050 |
+
"loss": 1.2025721073150635,
|
| 5051 |
+
"step": 719
|
| 5052 |
+
},
|
| 5053 |
+
{
|
| 5054 |
+
"epoch": 0.41078305519897307,
|
| 5055 |
+
"grad_norm": 1.0390625,
|
| 5056 |
+
"learning_rate": 3.109756097560976e-05,
|
| 5057 |
+
"loss": 1.2218658924102783,
|
| 5058 |
+
"step": 720
|
| 5059 |
+
},
|
| 5060 |
+
{
|
| 5061 |
+
"epoch": 0.41135358722008275,
|
| 5062 |
+
"grad_norm": 0.96875,
|
| 5063 |
+
"learning_rate": 3.106984478935698e-05,
|
| 5064 |
+
"loss": 1.1744896173477173,
|
| 5065 |
+
"step": 721
|
| 5066 |
+
},
|
| 5067 |
+
{
|
| 5068 |
+
"epoch": 0.41192411924119243,
|
| 5069 |
+
"grad_norm": 0.94921875,
|
| 5070 |
+
"learning_rate": 3.104212860310421e-05,
|
| 5071 |
+
"loss": 1.1989339590072632,
|
| 5072 |
+
"step": 722
|
| 5073 |
+
},
|
| 5074 |
+
{
|
| 5075 |
+
"epoch": 0.4124946512623021,
|
| 5076 |
+
"grad_norm": 0.9765625,
|
| 5077 |
+
"learning_rate": 3.101441241685144e-05,
|
| 5078 |
+
"loss": 1.2189137935638428,
|
| 5079 |
+
"step": 723
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 0.4130651832834118,
|
| 5083 |
+
"grad_norm": 0.9921875,
|
| 5084 |
+
"learning_rate": 3.098669623059867e-05,
|
| 5085 |
+
"loss": 1.2155076265335083,
|
| 5086 |
+
"step": 724
|
| 5087 |
+
},
|
| 5088 |
+
{
|
| 5089 |
+
"epoch": 0.4136357153045215,
|
| 5090 |
+
"grad_norm": 0.9921875,
|
| 5091 |
+
"learning_rate": 3.09589800443459e-05,
|
| 5092 |
+
"loss": 1.1465799808502197,
|
| 5093 |
+
"step": 725
|
| 5094 |
+
},
|
| 5095 |
+
{
|
| 5096 |
+
"epoch": 0.41420624732563116,
|
| 5097 |
+
"grad_norm": 0.98828125,
|
| 5098 |
+
"learning_rate": 3.093126385809313e-05,
|
| 5099 |
+
"loss": 1.2145007848739624,
|
| 5100 |
+
"step": 726
|
| 5101 |
+
},
|
| 5102 |
+
{
|
| 5103 |
+
"epoch": 0.41477677934674084,
|
| 5104 |
+
"grad_norm": 0.984375,
|
| 5105 |
+
"learning_rate": 3.090354767184035e-05,
|
| 5106 |
+
"loss": 1.2057294845581055,
|
| 5107 |
+
"step": 727
|
| 5108 |
+
},
|
| 5109 |
+
{
|
| 5110 |
+
"epoch": 0.4153473113678505,
|
| 5111 |
+
"grad_norm": 0.9921875,
|
| 5112 |
+
"learning_rate": 3.087583148558759e-05,
|
| 5113 |
+
"loss": 1.2041752338409424,
|
| 5114 |
+
"step": 728
|
| 5115 |
+
},
|
| 5116 |
+
{
|
| 5117 |
+
"epoch": 0.4159178433889602,
|
| 5118 |
+
"grad_norm": 0.9765625,
|
| 5119 |
+
"learning_rate": 3.084811529933481e-05,
|
| 5120 |
+
"loss": 1.1989641189575195,
|
| 5121 |
+
"step": 729
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 0.4164883754100699,
|
| 5125 |
+
"grad_norm": 0.9453125,
|
| 5126 |
+
"learning_rate": 3.082039911308204e-05,
|
| 5127 |
+
"loss": 1.188431739807129,
|
| 5128 |
+
"step": 730
|
| 5129 |
+
},
|
| 5130 |
+
{
|
| 5131 |
+
"epoch": 0.41705890743117957,
|
| 5132 |
+
"grad_norm": 0.96875,
|
| 5133 |
+
"learning_rate": 3.079268292682927e-05,
|
| 5134 |
+
"loss": 1.1488507986068726,
|
| 5135 |
+
"step": 731
|
| 5136 |
+
},
|
| 5137 |
+
{
|
| 5138 |
+
"epoch": 0.41762943945228925,
|
| 5139 |
+
"grad_norm": 1.015625,
|
| 5140 |
+
"learning_rate": 3.07649667405765e-05,
|
| 5141 |
+
"loss": 1.2174850702285767,
|
| 5142 |
+
"step": 732
|
| 5143 |
+
},
|
| 5144 |
+
{
|
| 5145 |
+
"epoch": 0.41819997147339893,
|
| 5146 |
+
"grad_norm": 0.96875,
|
| 5147 |
+
"learning_rate": 3.073725055432373e-05,
|
| 5148 |
+
"loss": 1.2141880989074707,
|
| 5149 |
+
"step": 733
|
| 5150 |
+
},
|
| 5151 |
+
{
|
| 5152 |
+
"epoch": 0.4187705034945086,
|
| 5153 |
+
"grad_norm": 1.03125,
|
| 5154 |
+
"learning_rate": 3.070953436807095e-05,
|
| 5155 |
+
"loss": 1.2875535488128662,
|
| 5156 |
+
"step": 734
|
| 5157 |
+
},
|
| 5158 |
+
{
|
| 5159 |
+
"epoch": 0.4193410355156183,
|
| 5160 |
+
"grad_norm": 0.984375,
|
| 5161 |
+
"learning_rate": 3.068181818181818e-05,
|
| 5162 |
+
"loss": 1.168579339981079,
|
| 5163 |
+
"step": 735
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 0.419911567536728,
|
| 5167 |
+
"grad_norm": 0.96875,
|
| 5168 |
+
"learning_rate": 3.065410199556541e-05,
|
| 5169 |
+
"loss": 1.1168636083602905,
|
| 5170 |
+
"step": 736
|
| 5171 |
+
},
|
| 5172 |
+
{
|
| 5173 |
+
"epoch": 0.42048209955783766,
|
| 5174 |
+
"grad_norm": 0.984375,
|
| 5175 |
+
"learning_rate": 3.062638580931264e-05,
|
| 5176 |
+
"loss": 1.1600708961486816,
|
| 5177 |
+
"step": 737
|
| 5178 |
+
},
|
| 5179 |
+
{
|
| 5180 |
+
"epoch": 0.42105263157894735,
|
| 5181 |
+
"grad_norm": 0.9765625,
|
| 5182 |
+
"learning_rate": 3.059866962305987e-05,
|
| 5183 |
+
"loss": 1.1832588911056519,
|
| 5184 |
+
"step": 738
|
| 5185 |
+
},
|
| 5186 |
+
{
|
| 5187 |
+
"epoch": 0.42105263157894735,
|
| 5188 |
+
"eval_loss": 1.1941628456115723,
|
| 5189 |
+
"eval_runtime": 80.1253,
|
| 5190 |
+
"eval_samples_per_second": 11.931,
|
| 5191 |
+
"eval_steps_per_second": 2.983,
|
| 5192 |
+
"step": 738
|
| 5193 |
}
|
| 5194 |
],
|
| 5195 |
"logging_steps": 1,
|
|
|
|
| 5209 |
"attributes": {}
|
| 5210 |
}
|
| 5211 |
},
|
| 5212 |
+
"total_flos": 6.500297618922824e+17,
|
| 5213 |
"train_batch_size": 8,
|
| 5214 |
"trial_name": null,
|
| 5215 |
"trial_params": null
|