Upload folder using huggingface_hub
Browse files
checkpoint-1500/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step2000
|
checkpoint-1500/model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4989973456
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:656da4fa3a8914e9ee5c4eaccae596b2b4ea5c415d7f2d18ebba23a224da7313
|
| 3 |
size 4989973456
|
checkpoint-1500/model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3786358064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:630cb5d37159c1bd6fc51269757ccf7b4719af50345e058774f283dc4497eca4
|
| 3 |
size 3786358064
|
checkpoint-1500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe093ffefd59d129cb1cf11f7e04ccd9d30a8327bb7c240932a1fbb2159d0ed5
|
| 3 |
size 1064
|
checkpoint-1500/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -10507,6 +10507,3506 @@
|
|
| 10507 |
"learning_rate": 1.6687597216781584e-05,
|
| 10508 |
"loss": 0.8145,
|
| 10509 |
"step": 1500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10510 |
}
|
| 10511 |
],
|
| 10512 |
"logging_steps": 1.0,
|
|
@@ -10514,7 +14014,7 @@
|
|
| 10514 |
"num_input_tokens_seen": 0,
|
| 10515 |
"num_train_epochs": 1,
|
| 10516 |
"save_steps": 500,
|
| 10517 |
-
"total_flos":
|
| 10518 |
"train_batch_size": 4,
|
| 10519 |
"trial_name": null,
|
| 10520 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.38513383400731754,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 2000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 10507 |
"learning_rate": 1.6687597216781584e-05,
|
| 10508 |
"loss": 0.8145,
|
| 10509 |
"step": 1500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.29,
|
| 10513 |
+
"grad_norm": 1.6814585945564913,
|
| 10514 |
+
"learning_rate": 1.668295881701969e-05,
|
| 10515 |
+
"loss": 0.9505,
|
| 10516 |
+
"step": 1501
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.29,
|
| 10520 |
+
"grad_norm": 1.8977009968762184,
|
| 10521 |
+
"learning_rate": 1.667831781754954e-05,
|
| 10522 |
+
"loss": 0.9042,
|
| 10523 |
+
"step": 1502
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.29,
|
| 10527 |
+
"grad_norm": 1.8711464591936275,
|
| 10528 |
+
"learning_rate": 1.6673674220176496e-05,
|
| 10529 |
+
"loss": 0.9478,
|
| 10530 |
+
"step": 1503
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.29,
|
| 10534 |
+
"grad_norm": 1.7180539423192296,
|
| 10535 |
+
"learning_rate": 1.666902802670695e-05,
|
| 10536 |
+
"loss": 0.962,
|
| 10537 |
+
"step": 1504
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.29,
|
| 10541 |
+
"grad_norm": 1.6666626104248825,
|
| 10542 |
+
"learning_rate": 1.66643792389483e-05,
|
| 10543 |
+
"loss": 0.9123,
|
| 10544 |
+
"step": 1505
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.29,
|
| 10548 |
+
"grad_norm": 1.1973527998514766,
|
| 10549 |
+
"learning_rate": 1.6659727858708947e-05,
|
| 10550 |
+
"loss": 0.856,
|
| 10551 |
+
"step": 1506
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.29,
|
| 10555 |
+
"grad_norm": 1.5673259814240763,
|
| 10556 |
+
"learning_rate": 1.66550738877983e-05,
|
| 10557 |
+
"loss": 0.9118,
|
| 10558 |
+
"step": 1507
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.29,
|
| 10562 |
+
"grad_norm": 1.5728628446079507,
|
| 10563 |
+
"learning_rate": 1.6650417328026793e-05,
|
| 10564 |
+
"loss": 0.9139,
|
| 10565 |
+
"step": 1508
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.29,
|
| 10569 |
+
"grad_norm": 1.5857396432980118,
|
| 10570 |
+
"learning_rate": 1.6645758181205834e-05,
|
| 10571 |
+
"loss": 0.8822,
|
| 10572 |
+
"step": 1509
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.29,
|
| 10576 |
+
"grad_norm": 1.8890331201978154,
|
| 10577 |
+
"learning_rate": 1.6641096449147875e-05,
|
| 10578 |
+
"loss": 0.9863,
|
| 10579 |
+
"step": 1510
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.29,
|
| 10583 |
+
"grad_norm": 1.6118515343399076,
|
| 10584 |
+
"learning_rate": 1.6636432133666342e-05,
|
| 10585 |
+
"loss": 0.9455,
|
| 10586 |
+
"step": 1511
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.29,
|
| 10590 |
+
"grad_norm": 1.6742412673222384,
|
| 10591 |
+
"learning_rate": 1.663176523657569e-05,
|
| 10592 |
+
"loss": 0.9777,
|
| 10593 |
+
"step": 1512
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.29,
|
| 10597 |
+
"grad_norm": 1.8259098282089554,
|
| 10598 |
+
"learning_rate": 1.6627095759691364e-05,
|
| 10599 |
+
"loss": 1.0421,
|
| 10600 |
+
"step": 1513
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.29,
|
| 10604 |
+
"grad_norm": 1.7330916535087395,
|
| 10605 |
+
"learning_rate": 1.6622423704829812e-05,
|
| 10606 |
+
"loss": 0.8842,
|
| 10607 |
+
"step": 1514
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.29,
|
| 10611 |
+
"grad_norm": 1.6424763863303073,
|
| 10612 |
+
"learning_rate": 1.66177490738085e-05,
|
| 10613 |
+
"loss": 0.8998,
|
| 10614 |
+
"step": 1515
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.29,
|
| 10618 |
+
"grad_norm": 1.7477206896837032,
|
| 10619 |
+
"learning_rate": 1.661307186844588e-05,
|
| 10620 |
+
"loss": 0.9033,
|
| 10621 |
+
"step": 1516
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.29,
|
| 10625 |
+
"grad_norm": 1.557691444398473,
|
| 10626 |
+
"learning_rate": 1.660839209056141e-05,
|
| 10627 |
+
"loss": 0.8975,
|
| 10628 |
+
"step": 1517
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.29,
|
| 10632 |
+
"grad_norm": 1.689261636764078,
|
| 10633 |
+
"learning_rate": 1.6603709741975545e-05,
|
| 10634 |
+
"loss": 0.9635,
|
| 10635 |
+
"step": 1518
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.29,
|
| 10639 |
+
"grad_norm": 1.7840501705145897,
|
| 10640 |
+
"learning_rate": 1.6599024824509752e-05,
|
| 10641 |
+
"loss": 0.9215,
|
| 10642 |
+
"step": 1519
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.29,
|
| 10646 |
+
"grad_norm": 1.9083255044750755,
|
| 10647 |
+
"learning_rate": 1.659433733998649e-05,
|
| 10648 |
+
"loss": 0.9896,
|
| 10649 |
+
"step": 1520
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.29,
|
| 10653 |
+
"grad_norm": 1.6931027780438765,
|
| 10654 |
+
"learning_rate": 1.6589647290229215e-05,
|
| 10655 |
+
"loss": 0.858,
|
| 10656 |
+
"step": 1521
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.29,
|
| 10660 |
+
"grad_norm": 1.7778446951035027,
|
| 10661 |
+
"learning_rate": 1.6584954677062382e-05,
|
| 10662 |
+
"loss": 0.9867,
|
| 10663 |
+
"step": 1522
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.29,
|
| 10667 |
+
"grad_norm": 1.6874904324233353,
|
| 10668 |
+
"learning_rate": 1.6580259502311445e-05,
|
| 10669 |
+
"loss": 1.0151,
|
| 10670 |
+
"step": 1523
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.29,
|
| 10674 |
+
"grad_norm": 1.8272429728687376,
|
| 10675 |
+
"learning_rate": 1.657556176780285e-05,
|
| 10676 |
+
"loss": 0.9711,
|
| 10677 |
+
"step": 1524
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.29,
|
| 10681 |
+
"grad_norm": 1.8411467883969832,
|
| 10682 |
+
"learning_rate": 1.6570861475364046e-05,
|
| 10683 |
+
"loss": 0.9206,
|
| 10684 |
+
"step": 1525
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.29,
|
| 10688 |
+
"grad_norm": 1.8710930737452116,
|
| 10689 |
+
"learning_rate": 1.6566158626823465e-05,
|
| 10690 |
+
"loss": 0.9872,
|
| 10691 |
+
"step": 1526
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.29,
|
| 10695 |
+
"grad_norm": 1.720506195988624,
|
| 10696 |
+
"learning_rate": 1.6561453224010554e-05,
|
| 10697 |
+
"loss": 0.9681,
|
| 10698 |
+
"step": 1527
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.29,
|
| 10702 |
+
"grad_norm": 1.6647183073763208,
|
| 10703 |
+
"learning_rate": 1.655674526875573e-05,
|
| 10704 |
+
"loss": 0.9251,
|
| 10705 |
+
"step": 1528
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.29,
|
| 10709 |
+
"grad_norm": 1.577609447670818,
|
| 10710 |
+
"learning_rate": 1.6552034762890422e-05,
|
| 10711 |
+
"loss": 0.9072,
|
| 10712 |
+
"step": 1529
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.29,
|
| 10716 |
+
"grad_norm": 2.066399167027944,
|
| 10717 |
+
"learning_rate": 1.6547321708247037e-05,
|
| 10718 |
+
"loss": 0.9379,
|
| 10719 |
+
"step": 1530
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.29,
|
| 10723 |
+
"grad_norm": 1.840506676774067,
|
| 10724 |
+
"learning_rate": 1.654260610665898e-05,
|
| 10725 |
+
"loss": 0.9352,
|
| 10726 |
+
"step": 1531
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.3,
|
| 10730 |
+
"grad_norm": 1.7208292126684335,
|
| 10731 |
+
"learning_rate": 1.653788795996065e-05,
|
| 10732 |
+
"loss": 0.899,
|
| 10733 |
+
"step": 1532
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.3,
|
| 10737 |
+
"grad_norm": 1.8414566876249059,
|
| 10738 |
+
"learning_rate": 1.6533167269987435e-05,
|
| 10739 |
+
"loss": 0.8995,
|
| 10740 |
+
"step": 1533
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.3,
|
| 10744 |
+
"grad_norm": 1.8163661936250057,
|
| 10745 |
+
"learning_rate": 1.65284440385757e-05,
|
| 10746 |
+
"loss": 1.0081,
|
| 10747 |
+
"step": 1534
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.3,
|
| 10751 |
+
"grad_norm": 1.705322353027036,
|
| 10752 |
+
"learning_rate": 1.6523718267562815e-05,
|
| 10753 |
+
"loss": 0.9674,
|
| 10754 |
+
"step": 1535
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.3,
|
| 10758 |
+
"grad_norm": 1.7366560844179042,
|
| 10759 |
+
"learning_rate": 1.6518989958787126e-05,
|
| 10760 |
+
"loss": 0.9463,
|
| 10761 |
+
"step": 1536
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.3,
|
| 10765 |
+
"grad_norm": 1.1520789836347844,
|
| 10766 |
+
"learning_rate": 1.651425911408798e-05,
|
| 10767 |
+
"loss": 0.8954,
|
| 10768 |
+
"step": 1537
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.3,
|
| 10772 |
+
"grad_norm": 1.8026482343246513,
|
| 10773 |
+
"learning_rate": 1.6509525735305697e-05,
|
| 10774 |
+
"loss": 0.9674,
|
| 10775 |
+
"step": 1538
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.3,
|
| 10779 |
+
"grad_norm": 1.8502578015677411,
|
| 10780 |
+
"learning_rate": 1.6504789824281586e-05,
|
| 10781 |
+
"loss": 0.9929,
|
| 10782 |
+
"step": 1539
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.3,
|
| 10786 |
+
"grad_norm": 1.7705139842133477,
|
| 10787 |
+
"learning_rate": 1.6500051382857944e-05,
|
| 10788 |
+
"loss": 0.8824,
|
| 10789 |
+
"step": 1540
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.3,
|
| 10793 |
+
"grad_norm": 1.8066747761487192,
|
| 10794 |
+
"learning_rate": 1.6495310412878053e-05,
|
| 10795 |
+
"loss": 0.8852,
|
| 10796 |
+
"step": 1541
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.3,
|
| 10800 |
+
"grad_norm": 1.673141236636469,
|
| 10801 |
+
"learning_rate": 1.649056691618618e-05,
|
| 10802 |
+
"loss": 0.9782,
|
| 10803 |
+
"step": 1542
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.3,
|
| 10807 |
+
"grad_norm": 1.8277718379488843,
|
| 10808 |
+
"learning_rate": 1.648582089462756e-05,
|
| 10809 |
+
"loss": 0.9895,
|
| 10810 |
+
"step": 1543
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.3,
|
| 10814 |
+
"grad_norm": 1.62439404205175,
|
| 10815 |
+
"learning_rate": 1.648107235004843e-05,
|
| 10816 |
+
"loss": 0.9448,
|
| 10817 |
+
"step": 1544
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 0.3,
|
| 10821 |
+
"grad_norm": 1.6323693740805412,
|
| 10822 |
+
"learning_rate": 1.6476321284296005e-05,
|
| 10823 |
+
"loss": 0.9419,
|
| 10824 |
+
"step": 1545
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 0.3,
|
| 10828 |
+
"grad_norm": 1.8284109095149372,
|
| 10829 |
+
"learning_rate": 1.647156769921847e-05,
|
| 10830 |
+
"loss": 0.8998,
|
| 10831 |
+
"step": 1546
|
| 10832 |
+
},
|
| 10833 |
+
{
|
| 10834 |
+
"epoch": 0.3,
|
| 10835 |
+
"grad_norm": 1.6500951414316578,
|
| 10836 |
+
"learning_rate": 1.6466811596664997e-05,
|
| 10837 |
+
"loss": 0.897,
|
| 10838 |
+
"step": 1547
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 0.3,
|
| 10842 |
+
"grad_norm": 1.6636394238778047,
|
| 10843 |
+
"learning_rate": 1.6462052978485733e-05,
|
| 10844 |
+
"loss": 0.991,
|
| 10845 |
+
"step": 1548
|
| 10846 |
+
},
|
| 10847 |
+
{
|
| 10848 |
+
"epoch": 0.3,
|
| 10849 |
+
"grad_norm": 1.6471253017947654,
|
| 10850 |
+
"learning_rate": 1.645729184653181e-05,
|
| 10851 |
+
"loss": 0.9912,
|
| 10852 |
+
"step": 1549
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 0.3,
|
| 10856 |
+
"grad_norm": 1.900480841118098,
|
| 10857 |
+
"learning_rate": 1.645252820265534e-05,
|
| 10858 |
+
"loss": 0.9342,
|
| 10859 |
+
"step": 1550
|
| 10860 |
+
},
|
| 10861 |
+
{
|
| 10862 |
+
"epoch": 0.3,
|
| 10863 |
+
"grad_norm": 1.7274929771579584,
|
| 10864 |
+
"learning_rate": 1.6447762048709397e-05,
|
| 10865 |
+
"loss": 1.0001,
|
| 10866 |
+
"step": 1551
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 0.3,
|
| 10870 |
+
"grad_norm": 1.8024044277346902,
|
| 10871 |
+
"learning_rate": 1.644299338654805e-05,
|
| 10872 |
+
"loss": 0.8785,
|
| 10873 |
+
"step": 1552
|
| 10874 |
+
},
|
| 10875 |
+
{
|
| 10876 |
+
"epoch": 0.3,
|
| 10877 |
+
"grad_norm": 1.650686902566554,
|
| 10878 |
+
"learning_rate": 1.643822221802633e-05,
|
| 10879 |
+
"loss": 0.9584,
|
| 10880 |
+
"step": 1553
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 0.3,
|
| 10884 |
+
"grad_norm": 1.5474371202081496,
|
| 10885 |
+
"learning_rate": 1.6433448545000252e-05,
|
| 10886 |
+
"loss": 0.9775,
|
| 10887 |
+
"step": 1554
|
| 10888 |
+
},
|
| 10889 |
+
{
|
| 10890 |
+
"epoch": 0.3,
|
| 10891 |
+
"grad_norm": 1.5597079451380407,
|
| 10892 |
+
"learning_rate": 1.64286723693268e-05,
|
| 10893 |
+
"loss": 0.9947,
|
| 10894 |
+
"step": 1555
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 0.3,
|
| 10898 |
+
"grad_norm": 1.6385235106949365,
|
| 10899 |
+
"learning_rate": 1.642389369286393e-05,
|
| 10900 |
+
"loss": 0.8831,
|
| 10901 |
+
"step": 1556
|
| 10902 |
+
},
|
| 10903 |
+
{
|
| 10904 |
+
"epoch": 0.3,
|
| 10905 |
+
"grad_norm": 1.6742026888304702,
|
| 10906 |
+
"learning_rate": 1.6419112517470577e-05,
|
| 10907 |
+
"loss": 0.9836,
|
| 10908 |
+
"step": 1557
|
| 10909 |
+
},
|
| 10910 |
+
{
|
| 10911 |
+
"epoch": 0.3,
|
| 10912 |
+
"grad_norm": 1.7075133360117316,
|
| 10913 |
+
"learning_rate": 1.6414328845006644e-05,
|
| 10914 |
+
"loss": 0.9422,
|
| 10915 |
+
"step": 1558
|
| 10916 |
+
},
|
| 10917 |
+
{
|
| 10918 |
+
"epoch": 0.3,
|
| 10919 |
+
"grad_norm": 1.5892520663236174,
|
| 10920 |
+
"learning_rate": 1.6409542677333007e-05,
|
| 10921 |
+
"loss": 0.8709,
|
| 10922 |
+
"step": 1559
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 0.3,
|
| 10926 |
+
"grad_norm": 1.6384876281252254,
|
| 10927 |
+
"learning_rate": 1.640475401631151e-05,
|
| 10928 |
+
"loss": 1.0045,
|
| 10929 |
+
"step": 1560
|
| 10930 |
+
},
|
| 10931 |
+
{
|
| 10932 |
+
"epoch": 0.3,
|
| 10933 |
+
"grad_norm": 1.693979144639973,
|
| 10934 |
+
"learning_rate": 1.6399962863804974e-05,
|
| 10935 |
+
"loss": 0.9034,
|
| 10936 |
+
"step": 1561
|
| 10937 |
+
},
|
| 10938 |
+
{
|
| 10939 |
+
"epoch": 0.3,
|
| 10940 |
+
"grad_norm": 1.8320477072069328,
|
| 10941 |
+
"learning_rate": 1.6395169221677176e-05,
|
| 10942 |
+
"loss": 1.0343,
|
| 10943 |
+
"step": 1562
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 0.3,
|
| 10947 |
+
"grad_norm": 1.7645628809213245,
|
| 10948 |
+
"learning_rate": 1.6390373091792873e-05,
|
| 10949 |
+
"loss": 1.0095,
|
| 10950 |
+
"step": 1563
|
| 10951 |
+
},
|
| 10952 |
+
{
|
| 10953 |
+
"epoch": 0.3,
|
| 10954 |
+
"grad_norm": 1.6613618239883878,
|
| 10955 |
+
"learning_rate": 1.6385574476017795e-05,
|
| 10956 |
+
"loss": 0.9025,
|
| 10957 |
+
"step": 1564
|
| 10958 |
+
},
|
| 10959 |
+
{
|
| 10960 |
+
"epoch": 0.3,
|
| 10961 |
+
"grad_norm": 1.870626964940079,
|
| 10962 |
+
"learning_rate": 1.638077337621861e-05,
|
| 10963 |
+
"loss": 0.9294,
|
| 10964 |
+
"step": 1565
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 0.3,
|
| 10968 |
+
"grad_norm": 1.6781919217989782,
|
| 10969 |
+
"learning_rate": 1.6375969794262997e-05,
|
| 10970 |
+
"loss": 0.9265,
|
| 10971 |
+
"step": 1566
|
| 10972 |
+
},
|
| 10973 |
+
{
|
| 10974 |
+
"epoch": 0.3,
|
| 10975 |
+
"grad_norm": 1.6732809661254076,
|
| 10976 |
+
"learning_rate": 1.637116373201956e-05,
|
| 10977 |
+
"loss": 0.9116,
|
| 10978 |
+
"step": 1567
|
| 10979 |
+
},
|
| 10980 |
+
{
|
| 10981 |
+
"epoch": 0.3,
|
| 10982 |
+
"grad_norm": 2.0888776005788916,
|
| 10983 |
+
"learning_rate": 1.6366355191357885e-05,
|
| 10984 |
+
"loss": 1.0467,
|
| 10985 |
+
"step": 1568
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 0.3,
|
| 10989 |
+
"grad_norm": 1.8650433291396247,
|
| 10990 |
+
"learning_rate": 1.6361544174148527e-05,
|
| 10991 |
+
"loss": 0.9516,
|
| 10992 |
+
"step": 1569
|
| 10993 |
+
},
|
| 10994 |
+
{
|
| 10995 |
+
"epoch": 0.3,
|
| 10996 |
+
"grad_norm": 1.7447193484240076,
|
| 10997 |
+
"learning_rate": 1.6356730682262997e-05,
|
| 10998 |
+
"loss": 0.9328,
|
| 10999 |
+
"step": 1570
|
| 11000 |
+
},
|
| 11001 |
+
{
|
| 11002 |
+
"epoch": 0.3,
|
| 11003 |
+
"grad_norm": 1.544513865520117,
|
| 11004 |
+
"learning_rate": 1.6351914717573767e-05,
|
| 11005 |
+
"loss": 0.8765,
|
| 11006 |
+
"step": 1571
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 0.3,
|
| 11010 |
+
"grad_norm": 1.7101828627501428,
|
| 11011 |
+
"learning_rate": 1.6347096281954277e-05,
|
| 11012 |
+
"loss": 0.9101,
|
| 11013 |
+
"step": 1572
|
| 11014 |
+
},
|
| 11015 |
+
{
|
| 11016 |
+
"epoch": 0.3,
|
| 11017 |
+
"grad_norm": 1.7209199573338974,
|
| 11018 |
+
"learning_rate": 1.6342275377278923e-05,
|
| 11019 |
+
"loss": 0.8928,
|
| 11020 |
+
"step": 1573
|
| 11021 |
+
},
|
| 11022 |
+
{
|
| 11023 |
+
"epoch": 0.3,
|
| 11024 |
+
"grad_norm": 1.730187436877261,
|
| 11025 |
+
"learning_rate": 1.6337452005423067e-05,
|
| 11026 |
+
"loss": 0.9241,
|
| 11027 |
+
"step": 1574
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 0.3,
|
| 11031 |
+
"grad_norm": 1.6976839779004649,
|
| 11032 |
+
"learning_rate": 1.6332626168263025e-05,
|
| 11033 |
+
"loss": 0.8881,
|
| 11034 |
+
"step": 1575
|
| 11035 |
+
},
|
| 11036 |
+
{
|
| 11037 |
+
"epoch": 0.3,
|
| 11038 |
+
"grad_norm": 1.6303874238073626,
|
| 11039 |
+
"learning_rate": 1.6327797867676077e-05,
|
| 11040 |
+
"loss": 0.943,
|
| 11041 |
+
"step": 1576
|
| 11042 |
+
},
|
| 11043 |
+
{
|
| 11044 |
+
"epoch": 0.3,
|
| 11045 |
+
"grad_norm": 1.670673560161254,
|
| 11046 |
+
"learning_rate": 1.6322967105540457e-05,
|
| 11047 |
+
"loss": 0.9151,
|
| 11048 |
+
"step": 1577
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 0.3,
|
| 11052 |
+
"grad_norm": 1.7163060222751576,
|
| 11053 |
+
"learning_rate": 1.6318133883735358e-05,
|
| 11054 |
+
"loss": 0.9639,
|
| 11055 |
+
"step": 1578
|
| 11056 |
+
},
|
| 11057 |
+
{
|
| 11058 |
+
"epoch": 0.3,
|
| 11059 |
+
"grad_norm": 1.687703952131334,
|
| 11060 |
+
"learning_rate": 1.6313298204140934e-05,
|
| 11061 |
+
"loss": 0.9525,
|
| 11062 |
+
"step": 1579
|
| 11063 |
+
},
|
| 11064 |
+
{
|
| 11065 |
+
"epoch": 0.3,
|
| 11066 |
+
"grad_norm": 1.9942262110876094,
|
| 11067 |
+
"learning_rate": 1.6308460068638287e-05,
|
| 11068 |
+
"loss": 0.955,
|
| 11069 |
+
"step": 1580
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 0.3,
|
| 11073 |
+
"grad_norm": 1.8919831342466207,
|
| 11074 |
+
"learning_rate": 1.6303619479109475e-05,
|
| 11075 |
+
"loss": 0.9573,
|
| 11076 |
+
"step": 1581
|
| 11077 |
+
},
|
| 11078 |
+
{
|
| 11079 |
+
"epoch": 0.3,
|
| 11080 |
+
"grad_norm": 1.778747910790833,
|
| 11081 |
+
"learning_rate": 1.6298776437437526e-05,
|
| 11082 |
+
"loss": 0.9421,
|
| 11083 |
+
"step": 1582
|
| 11084 |
+
},
|
| 11085 |
+
{
|
| 11086 |
+
"epoch": 0.3,
|
| 11087 |
+
"grad_norm": 1.745824812138257,
|
| 11088 |
+
"learning_rate": 1.62939309455064e-05,
|
| 11089 |
+
"loss": 0.9841,
|
| 11090 |
+
"step": 1583
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 0.31,
|
| 11094 |
+
"grad_norm": 2.005077387084947,
|
| 11095 |
+
"learning_rate": 1.6289083005201026e-05,
|
| 11096 |
+
"loss": 0.9857,
|
| 11097 |
+
"step": 1584
|
| 11098 |
+
},
|
| 11099 |
+
{
|
| 11100 |
+
"epoch": 0.31,
|
| 11101 |
+
"grad_norm": 1.898514749339766,
|
| 11102 |
+
"learning_rate": 1.6284232618407276e-05,
|
| 11103 |
+
"loss": 0.9036,
|
| 11104 |
+
"step": 1585
|
| 11105 |
+
},
|
| 11106 |
+
{
|
| 11107 |
+
"epoch": 0.31,
|
| 11108 |
+
"grad_norm": 1.5823702179562669,
|
| 11109 |
+
"learning_rate": 1.627937978701198e-05,
|
| 11110 |
+
"loss": 0.9336,
|
| 11111 |
+
"step": 1586
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 0.31,
|
| 11115 |
+
"grad_norm": 1.6580159968617827,
|
| 11116 |
+
"learning_rate": 1.6274524512902912e-05,
|
| 11117 |
+
"loss": 0.9015,
|
| 11118 |
+
"step": 1587
|
| 11119 |
+
},
|
| 11120 |
+
{
|
| 11121 |
+
"epoch": 0.31,
|
| 11122 |
+
"grad_norm": 1.6328945384243476,
|
| 11123 |
+
"learning_rate": 1.626966679796881e-05,
|
| 11124 |
+
"loss": 0.9527,
|
| 11125 |
+
"step": 1588
|
| 11126 |
+
},
|
| 11127 |
+
{
|
| 11128 |
+
"epoch": 0.31,
|
| 11129 |
+
"grad_norm": 1.7353326953684673,
|
| 11130 |
+
"learning_rate": 1.6264806644099347e-05,
|
| 11131 |
+
"loss": 0.9191,
|
| 11132 |
+
"step": 1589
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 0.31,
|
| 11136 |
+
"grad_norm": 1.731152621721883,
|
| 11137 |
+
"learning_rate": 1.6259944053185148e-05,
|
| 11138 |
+
"loss": 0.9689,
|
| 11139 |
+
"step": 1590
|
| 11140 |
+
},
|
| 11141 |
+
{
|
| 11142 |
+
"epoch": 0.31,
|
| 11143 |
+
"grad_norm": 1.8267882683217032,
|
| 11144 |
+
"learning_rate": 1.625507902711779e-05,
|
| 11145 |
+
"loss": 0.9006,
|
| 11146 |
+
"step": 1591
|
| 11147 |
+
},
|
| 11148 |
+
{
|
| 11149 |
+
"epoch": 0.31,
|
| 11150 |
+
"grad_norm": 1.660673153228956,
|
| 11151 |
+
"learning_rate": 1.6250211567789796e-05,
|
| 11152 |
+
"loss": 0.9531,
|
| 11153 |
+
"step": 1592
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 0.31,
|
| 11157 |
+
"grad_norm": 1.7726958038187408,
|
| 11158 |
+
"learning_rate": 1.6245341677094632e-05,
|
| 11159 |
+
"loss": 0.9605,
|
| 11160 |
+
"step": 1593
|
| 11161 |
+
},
|
| 11162 |
+
{
|
| 11163 |
+
"epoch": 0.31,
|
| 11164 |
+
"grad_norm": 1.5638525823490939,
|
| 11165 |
+
"learning_rate": 1.6240469356926714e-05,
|
| 11166 |
+
"loss": 0.9121,
|
| 11167 |
+
"step": 1594
|
| 11168 |
+
},
|
| 11169 |
+
{
|
| 11170 |
+
"epoch": 0.31,
|
| 11171 |
+
"grad_norm": 1.615819492487161,
|
| 11172 |
+
"learning_rate": 1.6235594609181404e-05,
|
| 11173 |
+
"loss": 0.9112,
|
| 11174 |
+
"step": 1595
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 0.31,
|
| 11178 |
+
"grad_norm": 1.1452214964318819,
|
| 11179 |
+
"learning_rate": 1.6230717435755002e-05,
|
| 11180 |
+
"loss": 0.8227,
|
| 11181 |
+
"step": 1596
|
| 11182 |
+
},
|
| 11183 |
+
{
|
| 11184 |
+
"epoch": 0.31,
|
| 11185 |
+
"grad_norm": 1.1150563856276265,
|
| 11186 |
+
"learning_rate": 1.6225837838544763e-05,
|
| 11187 |
+
"loss": 0.8186,
|
| 11188 |
+
"step": 1597
|
| 11189 |
+
},
|
| 11190 |
+
{
|
| 11191 |
+
"epoch": 0.31,
|
| 11192 |
+
"grad_norm": 1.8136187860164037,
|
| 11193 |
+
"learning_rate": 1.622095581944887e-05,
|
| 11194 |
+
"loss": 0.8627,
|
| 11195 |
+
"step": 1598
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 0.31,
|
| 11199 |
+
"grad_norm": 1.9577379459272641,
|
| 11200 |
+
"learning_rate": 1.6216071380366453e-05,
|
| 11201 |
+
"loss": 0.9854,
|
| 11202 |
+
"step": 1599
|
| 11203 |
+
},
|
| 11204 |
+
{
|
| 11205 |
+
"epoch": 0.31,
|
| 11206 |
+
"grad_norm": 1.4458602998328087,
|
| 11207 |
+
"learning_rate": 1.62111845231976e-05,
|
| 11208 |
+
"loss": 0.9476,
|
| 11209 |
+
"step": 1600
|
| 11210 |
+
},
|
| 11211 |
+
{
|
| 11212 |
+
"epoch": 0.31,
|
| 11213 |
+
"grad_norm": 1.6626009143381728,
|
| 11214 |
+
"learning_rate": 1.6206295249843312e-05,
|
| 11215 |
+
"loss": 0.9418,
|
| 11216 |
+
"step": 1601
|
| 11217 |
+
},
|
| 11218 |
+
{
|
| 11219 |
+
"epoch": 0.31,
|
| 11220 |
+
"grad_norm": 1.8417180161902174,
|
| 11221 |
+
"learning_rate": 1.6201403562205547e-05,
|
| 11222 |
+
"loss": 0.928,
|
| 11223 |
+
"step": 1602
|
| 11224 |
+
},
|
| 11225 |
+
{
|
| 11226 |
+
"epoch": 0.31,
|
| 11227 |
+
"grad_norm": 1.6342966301969772,
|
| 11228 |
+
"learning_rate": 1.6196509462187205e-05,
|
| 11229 |
+
"loss": 0.8499,
|
| 11230 |
+
"step": 1603
|
| 11231 |
+
},
|
| 11232 |
+
{
|
| 11233 |
+
"epoch": 0.31,
|
| 11234 |
+
"grad_norm": 1.4967835565015328,
|
| 11235 |
+
"learning_rate": 1.619161295169211e-05,
|
| 11236 |
+
"loss": 0.8553,
|
| 11237 |
+
"step": 1604
|
| 11238 |
+
},
|
| 11239 |
+
{
|
| 11240 |
+
"epoch": 0.31,
|
| 11241 |
+
"grad_norm": 1.7224864536848172,
|
| 11242 |
+
"learning_rate": 1.6186714032625036e-05,
|
| 11243 |
+
"loss": 0.925,
|
| 11244 |
+
"step": 1605
|
| 11245 |
+
},
|
| 11246 |
+
{
|
| 11247 |
+
"epoch": 0.31,
|
| 11248 |
+
"grad_norm": 1.7763197537104958,
|
| 11249 |
+
"learning_rate": 1.618181270689169e-05,
|
| 11250 |
+
"loss": 0.9975,
|
| 11251 |
+
"step": 1606
|
| 11252 |
+
},
|
| 11253 |
+
{
|
| 11254 |
+
"epoch": 0.31,
|
| 11255 |
+
"grad_norm": 1.831926739868879,
|
| 11256 |
+
"learning_rate": 1.6176908976398713e-05,
|
| 11257 |
+
"loss": 1.0196,
|
| 11258 |
+
"step": 1607
|
| 11259 |
+
},
|
| 11260 |
+
{
|
| 11261 |
+
"epoch": 0.31,
|
| 11262 |
+
"grad_norm": 1.4575672638416406,
|
| 11263 |
+
"learning_rate": 1.6172002843053687e-05,
|
| 11264 |
+
"loss": 0.8872,
|
| 11265 |
+
"step": 1608
|
| 11266 |
+
},
|
| 11267 |
+
{
|
| 11268 |
+
"epoch": 0.31,
|
| 11269 |
+
"grad_norm": 1.8265500625497886,
|
| 11270 |
+
"learning_rate": 1.6167094308765122e-05,
|
| 11271 |
+
"loss": 0.8805,
|
| 11272 |
+
"step": 1609
|
| 11273 |
+
},
|
| 11274 |
+
{
|
| 11275 |
+
"epoch": 0.31,
|
| 11276 |
+
"grad_norm": 1.6430601567156466,
|
| 11277 |
+
"learning_rate": 1.6162183375442464e-05,
|
| 11278 |
+
"loss": 0.9035,
|
| 11279 |
+
"step": 1610
|
| 11280 |
+
},
|
| 11281 |
+
{
|
| 11282 |
+
"epoch": 0.31,
|
| 11283 |
+
"grad_norm": 1.4940280767553107,
|
| 11284 |
+
"learning_rate": 1.6157270044996098e-05,
|
| 11285 |
+
"loss": 0.8646,
|
| 11286 |
+
"step": 1611
|
| 11287 |
+
},
|
| 11288 |
+
{
|
| 11289 |
+
"epoch": 0.31,
|
| 11290 |
+
"grad_norm": 1.7627770422295945,
|
| 11291 |
+
"learning_rate": 1.6152354319337336e-05,
|
| 11292 |
+
"loss": 0.8689,
|
| 11293 |
+
"step": 1612
|
| 11294 |
+
},
|
| 11295 |
+
{
|
| 11296 |
+
"epoch": 0.31,
|
| 11297 |
+
"grad_norm": 1.6788474884705797,
|
| 11298 |
+
"learning_rate": 1.614743620037842e-05,
|
| 11299 |
+
"loss": 0.8535,
|
| 11300 |
+
"step": 1613
|
| 11301 |
+
},
|
| 11302 |
+
{
|
| 11303 |
+
"epoch": 0.31,
|
| 11304 |
+
"grad_norm": 1.8638924023543273,
|
| 11305 |
+
"learning_rate": 1.6142515690032525e-05,
|
| 11306 |
+
"loss": 0.9121,
|
| 11307 |
+
"step": 1614
|
| 11308 |
+
},
|
| 11309 |
+
{
|
| 11310 |
+
"epoch": 0.31,
|
| 11311 |
+
"grad_norm": 2.0612905331097275,
|
| 11312 |
+
"learning_rate": 1.6137592790213765e-05,
|
| 11313 |
+
"loss": 0.9246,
|
| 11314 |
+
"step": 1615
|
| 11315 |
+
},
|
| 11316 |
+
{
|
| 11317 |
+
"epoch": 0.31,
|
| 11318 |
+
"grad_norm": 1.7216050549850093,
|
| 11319 |
+
"learning_rate": 1.6132667502837164e-05,
|
| 11320 |
+
"loss": 0.9596,
|
| 11321 |
+
"step": 1616
|
| 11322 |
+
},
|
| 11323 |
+
{
|
| 11324 |
+
"epoch": 0.31,
|
| 11325 |
+
"grad_norm": 1.5634428918355618,
|
| 11326 |
+
"learning_rate": 1.6127739829818695e-05,
|
| 11327 |
+
"loss": 0.8957,
|
| 11328 |
+
"step": 1617
|
| 11329 |
+
},
|
| 11330 |
+
{
|
| 11331 |
+
"epoch": 0.31,
|
| 11332 |
+
"grad_norm": 1.6959549456380303,
|
| 11333 |
+
"learning_rate": 1.6122809773075247e-05,
|
| 11334 |
+
"loss": 0.752,
|
| 11335 |
+
"step": 1618
|
| 11336 |
+
},
|
| 11337 |
+
{
|
| 11338 |
+
"epoch": 0.31,
|
| 11339 |
+
"grad_norm": 1.8052961117745467,
|
| 11340 |
+
"learning_rate": 1.6117877334524635e-05,
|
| 11341 |
+
"loss": 0.9041,
|
| 11342 |
+
"step": 1619
|
| 11343 |
+
},
|
| 11344 |
+
{
|
| 11345 |
+
"epoch": 0.31,
|
| 11346 |
+
"grad_norm": 1.82138238416882,
|
| 11347 |
+
"learning_rate": 1.6112942516085616e-05,
|
| 11348 |
+
"loss": 0.9431,
|
| 11349 |
+
"step": 1620
|
| 11350 |
+
},
|
| 11351 |
+
{
|
| 11352 |
+
"epoch": 0.31,
|
| 11353 |
+
"grad_norm": 1.8821211929280084,
|
| 11354 |
+
"learning_rate": 1.6108005319677853e-05,
|
| 11355 |
+
"loss": 0.9189,
|
| 11356 |
+
"step": 1621
|
| 11357 |
+
},
|
| 11358 |
+
{
|
| 11359 |
+
"epoch": 0.31,
|
| 11360 |
+
"grad_norm": 1.6797384699671483,
|
| 11361 |
+
"learning_rate": 1.6103065747221945e-05,
|
| 11362 |
+
"loss": 0.9559,
|
| 11363 |
+
"step": 1622
|
| 11364 |
+
},
|
| 11365 |
+
{
|
| 11366 |
+
"epoch": 0.31,
|
| 11367 |
+
"grad_norm": 1.6335614363214426,
|
| 11368 |
+
"learning_rate": 1.6098123800639415e-05,
|
| 11369 |
+
"loss": 1.0283,
|
| 11370 |
+
"step": 1623
|
| 11371 |
+
},
|
| 11372 |
+
{
|
| 11373 |
+
"epoch": 0.31,
|
| 11374 |
+
"grad_norm": 1.663077182850145,
|
| 11375 |
+
"learning_rate": 1.6093179481852706e-05,
|
| 11376 |
+
"loss": 0.8679,
|
| 11377 |
+
"step": 1624
|
| 11378 |
+
},
|
| 11379 |
+
{
|
| 11380 |
+
"epoch": 0.31,
|
| 11381 |
+
"grad_norm": 1.736097411032046,
|
| 11382 |
+
"learning_rate": 1.6088232792785187e-05,
|
| 11383 |
+
"loss": 0.9939,
|
| 11384 |
+
"step": 1625
|
| 11385 |
+
},
|
| 11386 |
+
{
|
| 11387 |
+
"epoch": 0.31,
|
| 11388 |
+
"grad_norm": 1.9331864705161093,
|
| 11389 |
+
"learning_rate": 1.608328373536115e-05,
|
| 11390 |
+
"loss": 0.8924,
|
| 11391 |
+
"step": 1626
|
| 11392 |
+
},
|
| 11393 |
+
{
|
| 11394 |
+
"epoch": 0.31,
|
| 11395 |
+
"grad_norm": 1.7154518294143721,
|
| 11396 |
+
"learning_rate": 1.60783323115058e-05,
|
| 11397 |
+
"loss": 0.9131,
|
| 11398 |
+
"step": 1627
|
| 11399 |
+
},
|
| 11400 |
+
{
|
| 11401 |
+
"epoch": 0.31,
|
| 11402 |
+
"grad_norm": 1.6772390452717216,
|
| 11403 |
+
"learning_rate": 1.6073378523145272e-05,
|
| 11404 |
+
"loss": 1.0287,
|
| 11405 |
+
"step": 1628
|
| 11406 |
+
},
|
| 11407 |
+
{
|
| 11408 |
+
"epoch": 0.31,
|
| 11409 |
+
"grad_norm": 1.7439611127964232,
|
| 11410 |
+
"learning_rate": 1.6068422372206613e-05,
|
| 11411 |
+
"loss": 0.9183,
|
| 11412 |
+
"step": 1629
|
| 11413 |
+
},
|
| 11414 |
+
{
|
| 11415 |
+
"epoch": 0.31,
|
| 11416 |
+
"grad_norm": 1.691458239632585,
|
| 11417 |
+
"learning_rate": 1.6063463860617802e-05,
|
| 11418 |
+
"loss": 0.9311,
|
| 11419 |
+
"step": 1630
|
| 11420 |
+
},
|
| 11421 |
+
{
|
| 11422 |
+
"epoch": 0.31,
|
| 11423 |
+
"grad_norm": 1.658304169586574,
|
| 11424 |
+
"learning_rate": 1.605850299030772e-05,
|
| 11425 |
+
"loss": 0.9134,
|
| 11426 |
+
"step": 1631
|
| 11427 |
+
},
|
| 11428 |
+
{
|
| 11429 |
+
"epoch": 0.31,
|
| 11430 |
+
"grad_norm": 1.7083337750889018,
|
| 11431 |
+
"learning_rate": 1.6053539763206177e-05,
|
| 11432 |
+
"loss": 0.9304,
|
| 11433 |
+
"step": 1632
|
| 11434 |
+
},
|
| 11435 |
+
{
|
| 11436 |
+
"epoch": 0.31,
|
| 11437 |
+
"grad_norm": 1.7508484136879259,
|
| 11438 |
+
"learning_rate": 1.6048574181243895e-05,
|
| 11439 |
+
"loss": 0.939,
|
| 11440 |
+
"step": 1633
|
| 11441 |
+
},
|
| 11442 |
+
{
|
| 11443 |
+
"epoch": 0.31,
|
| 11444 |
+
"grad_norm": 1.7852876914422533,
|
| 11445 |
+
"learning_rate": 1.6043606246352516e-05,
|
| 11446 |
+
"loss": 0.9663,
|
| 11447 |
+
"step": 1634
|
| 11448 |
+
},
|
| 11449 |
+
{
|
| 11450 |
+
"epoch": 0.31,
|
| 11451 |
+
"grad_norm": 1.6659108647906151,
|
| 11452 |
+
"learning_rate": 1.6038635960464586e-05,
|
| 11453 |
+
"loss": 0.9219,
|
| 11454 |
+
"step": 1635
|
| 11455 |
+
},
|
| 11456 |
+
{
|
| 11457 |
+
"epoch": 0.32,
|
| 11458 |
+
"grad_norm": 1.8060772275287673,
|
| 11459 |
+
"learning_rate": 1.6033663325513583e-05,
|
| 11460 |
+
"loss": 0.9479,
|
| 11461 |
+
"step": 1636
|
| 11462 |
+
},
|
| 11463 |
+
{
|
| 11464 |
+
"epoch": 0.32,
|
| 11465 |
+
"grad_norm": 1.894086011906853,
|
| 11466 |
+
"learning_rate": 1.602868834343388e-05,
|
| 11467 |
+
"loss": 0.9187,
|
| 11468 |
+
"step": 1637
|
| 11469 |
+
},
|
| 11470 |
+
{
|
| 11471 |
+
"epoch": 0.32,
|
| 11472 |
+
"grad_norm": 1.7600387051701272,
|
| 11473 |
+
"learning_rate": 1.602371101616079e-05,
|
| 11474 |
+
"loss": 0.9938,
|
| 11475 |
+
"step": 1638
|
| 11476 |
+
},
|
| 11477 |
+
{
|
| 11478 |
+
"epoch": 0.32,
|
| 11479 |
+
"grad_norm": 1.6732938933125896,
|
| 11480 |
+
"learning_rate": 1.60187313456305e-05,
|
| 11481 |
+
"loss": 0.941,
|
| 11482 |
+
"step": 1639
|
| 11483 |
+
},
|
| 11484 |
+
{
|
| 11485 |
+
"epoch": 0.32,
|
| 11486 |
+
"grad_norm": 1.543451364693145,
|
| 11487 |
+
"learning_rate": 1.6013749333780143e-05,
|
| 11488 |
+
"loss": 0.9488,
|
| 11489 |
+
"step": 1640
|
| 11490 |
+
},
|
| 11491 |
+
{
|
| 11492 |
+
"epoch": 0.32,
|
| 11493 |
+
"grad_norm": 1.6555084246553577,
|
| 11494 |
+
"learning_rate": 1.600876498254775e-05,
|
| 11495 |
+
"loss": 0.96,
|
| 11496 |
+
"step": 1641
|
| 11497 |
+
},
|
| 11498 |
+
{
|
| 11499 |
+
"epoch": 0.32,
|
| 11500 |
+
"grad_norm": 1.8317018556108886,
|
| 11501 |
+
"learning_rate": 1.600377829387225e-05,
|
| 11502 |
+
"loss": 0.9523,
|
| 11503 |
+
"step": 1642
|
| 11504 |
+
},
|
| 11505 |
+
{
|
| 11506 |
+
"epoch": 0.32,
|
| 11507 |
+
"grad_norm": 1.635513167409773,
|
| 11508 |
+
"learning_rate": 1.5998789269693508e-05,
|
| 11509 |
+
"loss": 0.9135,
|
| 11510 |
+
"step": 1643
|
| 11511 |
+
},
|
| 11512 |
+
{
|
| 11513 |
+
"epoch": 0.32,
|
| 11514 |
+
"grad_norm": 1.6171341568108941,
|
| 11515 |
+
"learning_rate": 1.599379791195227e-05,
|
| 11516 |
+
"loss": 0.8896,
|
| 11517 |
+
"step": 1644
|
| 11518 |
+
},
|
| 11519 |
+
{
|
| 11520 |
+
"epoch": 0.32,
|
| 11521 |
+
"grad_norm": 1.8946636753822537,
|
| 11522 |
+
"learning_rate": 1.598880422259021e-05,
|
| 11523 |
+
"loss": 1.0003,
|
| 11524 |
+
"step": 1645
|
| 11525 |
+
},
|
| 11526 |
+
{
|
| 11527 |
+
"epoch": 0.32,
|
| 11528 |
+
"grad_norm": 1.76766590571786,
|
| 11529 |
+
"learning_rate": 1.598380820354989e-05,
|
| 11530 |
+
"loss": 0.9844,
|
| 11531 |
+
"step": 1646
|
| 11532 |
+
},
|
| 11533 |
+
{
|
| 11534 |
+
"epoch": 0.32,
|
| 11535 |
+
"grad_norm": 1.6685097327074356,
|
| 11536 |
+
"learning_rate": 1.5978809856774807e-05,
|
| 11537 |
+
"loss": 0.9786,
|
| 11538 |
+
"step": 1647
|
| 11539 |
+
},
|
| 11540 |
+
{
|
| 11541 |
+
"epoch": 0.32,
|
| 11542 |
+
"grad_norm": 1.5918070005843183,
|
| 11543 |
+
"learning_rate": 1.5973809184209328e-05,
|
| 11544 |
+
"loss": 0.8771,
|
| 11545 |
+
"step": 1648
|
| 11546 |
+
},
|
| 11547 |
+
{
|
| 11548 |
+
"epoch": 0.32,
|
| 11549 |
+
"grad_norm": 1.7023742031839773,
|
| 11550 |
+
"learning_rate": 1.596880618779875e-05,
|
| 11551 |
+
"loss": 0.8801,
|
| 11552 |
+
"step": 1649
|
| 11553 |
+
},
|
| 11554 |
+
{
|
| 11555 |
+
"epoch": 0.32,
|
| 11556 |
+
"grad_norm": 1.724381937302016,
|
| 11557 |
+
"learning_rate": 1.5963800869489273e-05,
|
| 11558 |
+
"loss": 0.9386,
|
| 11559 |
+
"step": 1650
|
| 11560 |
+
},
|
| 11561 |
+
{
|
| 11562 |
+
"epoch": 0.32,
|
| 11563 |
+
"grad_norm": 1.8494707991079804,
|
| 11564 |
+
"learning_rate": 1.595879323122798e-05,
|
| 11565 |
+
"loss": 1.0036,
|
| 11566 |
+
"step": 1651
|
| 11567 |
+
},
|
| 11568 |
+
{
|
| 11569 |
+
"epoch": 0.32,
|
| 11570 |
+
"grad_norm": 1.5226561189767724,
|
| 11571 |
+
"learning_rate": 1.5953783274962883e-05,
|
| 11572 |
+
"loss": 0.8788,
|
| 11573 |
+
"step": 1652
|
| 11574 |
+
},
|
| 11575 |
+
{
|
| 11576 |
+
"epoch": 0.32,
|
| 11577 |
+
"grad_norm": 1.6318911919928796,
|
| 11578 |
+
"learning_rate": 1.5948771002642874e-05,
|
| 11579 |
+
"loss": 0.9095,
|
| 11580 |
+
"step": 1653
|
| 11581 |
+
},
|
| 11582 |
+
{
|
| 11583 |
+
"epoch": 0.32,
|
| 11584 |
+
"grad_norm": 1.912279971229655,
|
| 11585 |
+
"learning_rate": 1.5943756416217758e-05,
|
| 11586 |
+
"loss": 1.0128,
|
| 11587 |
+
"step": 1654
|
| 11588 |
+
},
|
| 11589 |
+
{
|
| 11590 |
+
"epoch": 0.32,
|
| 11591 |
+
"grad_norm": 1.1650261536863196,
|
| 11592 |
+
"learning_rate": 1.5938739517638236e-05,
|
| 11593 |
+
"loss": 0.8428,
|
| 11594 |
+
"step": 1655
|
| 11595 |
+
},
|
| 11596 |
+
{
|
| 11597 |
+
"epoch": 0.32,
|
| 11598 |
+
"grad_norm": 1.7415643450863434,
|
| 11599 |
+
"learning_rate": 1.5933720308855915e-05,
|
| 11600 |
+
"loss": 0.9211,
|
| 11601 |
+
"step": 1656
|
| 11602 |
+
},
|
| 11603 |
+
{
|
| 11604 |
+
"epoch": 0.32,
|
| 11605 |
+
"grad_norm": 1.8820348107062908,
|
| 11606 |
+
"learning_rate": 1.5928698791823292e-05,
|
| 11607 |
+
"loss": 0.955,
|
| 11608 |
+
"step": 1657
|
| 11609 |
+
},
|
| 11610 |
+
{
|
| 11611 |
+
"epoch": 0.32,
|
| 11612 |
+
"grad_norm": 1.6280225538899704,
|
| 11613 |
+
"learning_rate": 1.5923674968493762e-05,
|
| 11614 |
+
"loss": 1.0581,
|
| 11615 |
+
"step": 1658
|
| 11616 |
+
},
|
| 11617 |
+
{
|
| 11618 |
+
"epoch": 0.32,
|
| 11619 |
+
"grad_norm": 1.6107594870919844,
|
| 11620 |
+
"learning_rate": 1.5918648840821622e-05,
|
| 11621 |
+
"loss": 0.9198,
|
| 11622 |
+
"step": 1659
|
| 11623 |
+
},
|
| 11624 |
+
{
|
| 11625 |
+
"epoch": 0.32,
|
| 11626 |
+
"grad_norm": 1.9027448285073223,
|
| 11627 |
+
"learning_rate": 1.5913620410762066e-05,
|
| 11628 |
+
"loss": 0.9414,
|
| 11629 |
+
"step": 1660
|
| 11630 |
+
},
|
| 11631 |
+
{
|
| 11632 |
+
"epoch": 0.32,
|
| 11633 |
+
"grad_norm": 1.1187156575317108,
|
| 11634 |
+
"learning_rate": 1.590858968027118e-05,
|
| 11635 |
+
"loss": 0.8779,
|
| 11636 |
+
"step": 1661
|
| 11637 |
+
},
|
| 11638 |
+
{
|
| 11639 |
+
"epoch": 0.32,
|
| 11640 |
+
"grad_norm": 1.632013326547582,
|
| 11641 |
+
"learning_rate": 1.5903556651305945e-05,
|
| 11642 |
+
"loss": 0.8975,
|
| 11643 |
+
"step": 1662
|
| 11644 |
+
},
|
| 11645 |
+
{
|
| 11646 |
+
"epoch": 0.32,
|
| 11647 |
+
"grad_norm": 1.6182938839187448,
|
| 11648 |
+
"learning_rate": 1.5898521325824238e-05,
|
| 11649 |
+
"loss": 0.855,
|
| 11650 |
+
"step": 1663
|
| 11651 |
+
},
|
| 11652 |
+
{
|
| 11653 |
+
"epoch": 0.32,
|
| 11654 |
+
"grad_norm": 1.7066599421699202,
|
| 11655 |
+
"learning_rate": 1.589348370578483e-05,
|
| 11656 |
+
"loss": 0.9101,
|
| 11657 |
+
"step": 1664
|
| 11658 |
+
},
|
| 11659 |
+
{
|
| 11660 |
+
"epoch": 0.32,
|
| 11661 |
+
"grad_norm": 1.5337852130580103,
|
| 11662 |
+
"learning_rate": 1.588844379314738e-05,
|
| 11663 |
+
"loss": 0.9258,
|
| 11664 |
+
"step": 1665
|
| 11665 |
+
},
|
| 11666 |
+
{
|
| 11667 |
+
"epoch": 0.32,
|
| 11668 |
+
"grad_norm": 1.5552796689485406,
|
| 11669 |
+
"learning_rate": 1.5883401589872444e-05,
|
| 11670 |
+
"loss": 0.8547,
|
| 11671 |
+
"step": 1666
|
| 11672 |
+
},
|
| 11673 |
+
{
|
| 11674 |
+
"epoch": 0.32,
|
| 11675 |
+
"grad_norm": 1.637891557924487,
|
| 11676 |
+
"learning_rate": 1.587835709792147e-05,
|
| 11677 |
+
"loss": 0.9402,
|
| 11678 |
+
"step": 1667
|
| 11679 |
+
},
|
| 11680 |
+
{
|
| 11681 |
+
"epoch": 0.32,
|
| 11682 |
+
"grad_norm": 1.6360176706001976,
|
| 11683 |
+
"learning_rate": 1.587331031925679e-05,
|
| 11684 |
+
"loss": 0.8344,
|
| 11685 |
+
"step": 1668
|
| 11686 |
+
},
|
| 11687 |
+
{
|
| 11688 |
+
"epoch": 0.32,
|
| 11689 |
+
"grad_norm": 1.7108867430309864,
|
| 11690 |
+
"learning_rate": 1.586826125584163e-05,
|
| 11691 |
+
"loss": 0.8892,
|
| 11692 |
+
"step": 1669
|
| 11693 |
+
},
|
| 11694 |
+
{
|
| 11695 |
+
"epoch": 0.32,
|
| 11696 |
+
"grad_norm": 1.7038609470120825,
|
| 11697 |
+
"learning_rate": 1.5863209909640106e-05,
|
| 11698 |
+
"loss": 0.9929,
|
| 11699 |
+
"step": 1670
|
| 11700 |
+
},
|
| 11701 |
+
{
|
| 11702 |
+
"epoch": 0.32,
|
| 11703 |
+
"grad_norm": 1.5206042599249054,
|
| 11704 |
+
"learning_rate": 1.5858156282617215e-05,
|
| 11705 |
+
"loss": 0.8909,
|
| 11706 |
+
"step": 1671
|
| 11707 |
+
},
|
| 11708 |
+
{
|
| 11709 |
+
"epoch": 0.32,
|
| 11710 |
+
"grad_norm": 1.6711511627920845,
|
| 11711 |
+
"learning_rate": 1.585310037673885e-05,
|
| 11712 |
+
"loss": 0.8614,
|
| 11713 |
+
"step": 1672
|
| 11714 |
+
},
|
| 11715 |
+
{
|
| 11716 |
+
"epoch": 0.32,
|
| 11717 |
+
"grad_norm": 1.8427061122981878,
|
| 11718 |
+
"learning_rate": 1.5848042193971788e-05,
|
| 11719 |
+
"loss": 1.0136,
|
| 11720 |
+
"step": 1673
|
| 11721 |
+
},
|
| 11722 |
+
{
|
| 11723 |
+
"epoch": 0.32,
|
| 11724 |
+
"grad_norm": 1.6340322224497956,
|
| 11725 |
+
"learning_rate": 1.5842981736283686e-05,
|
| 11726 |
+
"loss": 0.977,
|
| 11727 |
+
"step": 1674
|
| 11728 |
+
},
|
| 11729 |
+
{
|
| 11730 |
+
"epoch": 0.32,
|
| 11731 |
+
"grad_norm": 1.6997279400264118,
|
| 11732 |
+
"learning_rate": 1.583791900564309e-05,
|
| 11733 |
+
"loss": 0.9009,
|
| 11734 |
+
"step": 1675
|
| 11735 |
+
},
|
| 11736 |
+
{
|
| 11737 |
+
"epoch": 0.32,
|
| 11738 |
+
"grad_norm": 1.7262098742750869,
|
| 11739 |
+
"learning_rate": 1.583285400401944e-05,
|
| 11740 |
+
"loss": 0.9738,
|
| 11741 |
+
"step": 1676
|
| 11742 |
+
},
|
| 11743 |
+
{
|
| 11744 |
+
"epoch": 0.32,
|
| 11745 |
+
"grad_norm": 1.5265533453619249,
|
| 11746 |
+
"learning_rate": 1.5827786733383033e-05,
|
| 11747 |
+
"loss": 0.855,
|
| 11748 |
+
"step": 1677
|
| 11749 |
+
},
|
| 11750 |
+
{
|
| 11751 |
+
"epoch": 0.32,
|
| 11752 |
+
"grad_norm": 1.5822635288749958,
|
| 11753 |
+
"learning_rate": 1.5822717195705082e-05,
|
| 11754 |
+
"loss": 0.8655,
|
| 11755 |
+
"step": 1678
|
| 11756 |
+
},
|
| 11757 |
+
{
|
| 11758 |
+
"epoch": 0.32,
|
| 11759 |
+
"grad_norm": 1.4231646639501143,
|
| 11760 |
+
"learning_rate": 1.581764539295765e-05,
|
| 11761 |
+
"loss": 0.9177,
|
| 11762 |
+
"step": 1679
|
| 11763 |
+
},
|
| 11764 |
+
{
|
| 11765 |
+
"epoch": 0.32,
|
| 11766 |
+
"grad_norm": 1.6751968342336907,
|
| 11767 |
+
"learning_rate": 1.5812571327113715e-05,
|
| 11768 |
+
"loss": 0.8745,
|
| 11769 |
+
"step": 1680
|
| 11770 |
+
},
|
| 11771 |
+
{
|
| 11772 |
+
"epoch": 0.32,
|
| 11773 |
+
"grad_norm": 1.6747266634292668,
|
| 11774 |
+
"learning_rate": 1.5807495000147103e-05,
|
| 11775 |
+
"loss": 0.867,
|
| 11776 |
+
"step": 1681
|
| 11777 |
+
},
|
| 11778 |
+
{
|
| 11779 |
+
"epoch": 0.32,
|
| 11780 |
+
"grad_norm": 1.8696931576923919,
|
| 11781 |
+
"learning_rate": 1.5802416414032537e-05,
|
| 11782 |
+
"loss": 0.9896,
|
| 11783 |
+
"step": 1682
|
| 11784 |
+
},
|
| 11785 |
+
{
|
| 11786 |
+
"epoch": 0.32,
|
| 11787 |
+
"grad_norm": 1.6926163574834707,
|
| 11788 |
+
"learning_rate": 1.5797335570745614e-05,
|
| 11789 |
+
"loss": 0.868,
|
| 11790 |
+
"step": 1683
|
| 11791 |
+
},
|
| 11792 |
+
{
|
| 11793 |
+
"epoch": 0.32,
|
| 11794 |
+
"grad_norm": 1.7188418247492503,
|
| 11795 |
+
"learning_rate": 1.5792252472262813e-05,
|
| 11796 |
+
"loss": 0.9944,
|
| 11797 |
+
"step": 1684
|
| 11798 |
+
},
|
| 11799 |
+
{
|
| 11800 |
+
"epoch": 0.32,
|
| 11801 |
+
"grad_norm": 1.566218028695576,
|
| 11802 |
+
"learning_rate": 1.578716712056149e-05,
|
| 11803 |
+
"loss": 0.9525,
|
| 11804 |
+
"step": 1685
|
| 11805 |
+
},
|
| 11806 |
+
{
|
| 11807 |
+
"epoch": 0.32,
|
| 11808 |
+
"grad_norm": 1.699566762060431,
|
| 11809 |
+
"learning_rate": 1.578207951761987e-05,
|
| 11810 |
+
"loss": 0.9356,
|
| 11811 |
+
"step": 1686
|
| 11812 |
+
},
|
| 11813 |
+
{
|
| 11814 |
+
"epoch": 0.32,
|
| 11815 |
+
"grad_norm": 1.6877601287806214,
|
| 11816 |
+
"learning_rate": 1.577698966541706e-05,
|
| 11817 |
+
"loss": 1.0246,
|
| 11818 |
+
"step": 1687
|
| 11819 |
+
},
|
| 11820 |
+
{
|
| 11821 |
+
"epoch": 0.33,
|
| 11822 |
+
"grad_norm": 1.514033754520559,
|
| 11823 |
+
"learning_rate": 1.577189756593304e-05,
|
| 11824 |
+
"loss": 0.9412,
|
| 11825 |
+
"step": 1688
|
| 11826 |
+
},
|
| 11827 |
+
{
|
| 11828 |
+
"epoch": 0.33,
|
| 11829 |
+
"grad_norm": 1.8053392098687253,
|
| 11830 |
+
"learning_rate": 1.5766803221148676e-05,
|
| 11831 |
+
"loss": 0.9257,
|
| 11832 |
+
"step": 1689
|
| 11833 |
+
},
|
| 11834 |
+
{
|
| 11835 |
+
"epoch": 0.33,
|
| 11836 |
+
"grad_norm": 1.637881254327796,
|
| 11837 |
+
"learning_rate": 1.5761706633045677e-05,
|
| 11838 |
+
"loss": 0.9698,
|
| 11839 |
+
"step": 1690
|
| 11840 |
+
},
|
| 11841 |
+
{
|
| 11842 |
+
"epoch": 0.33,
|
| 11843 |
+
"grad_norm": 1.4978510106761729,
|
| 11844 |
+
"learning_rate": 1.575660780360666e-05,
|
| 11845 |
+
"loss": 0.8801,
|
| 11846 |
+
"step": 1691
|
| 11847 |
+
},
|
| 11848 |
+
{
|
| 11849 |
+
"epoch": 0.33,
|
| 11850 |
+
"grad_norm": 1.6878496657903825,
|
| 11851 |
+
"learning_rate": 1.575150673481509e-05,
|
| 11852 |
+
"loss": 0.8644,
|
| 11853 |
+
"step": 1692
|
| 11854 |
+
},
|
| 11855 |
+
{
|
| 11856 |
+
"epoch": 0.33,
|
| 11857 |
+
"grad_norm": 1.6593397363940259,
|
| 11858 |
+
"learning_rate": 1.5746403428655318e-05,
|
| 11859 |
+
"loss": 1.023,
|
| 11860 |
+
"step": 1693
|
| 11861 |
+
},
|
| 11862 |
+
{
|
| 11863 |
+
"epoch": 0.33,
|
| 11864 |
+
"grad_norm": 1.6235841003825928,
|
| 11865 |
+
"learning_rate": 1.5741297887112554e-05,
|
| 11866 |
+
"loss": 0.9081,
|
| 11867 |
+
"step": 1694
|
| 11868 |
+
},
|
| 11869 |
+
{
|
| 11870 |
+
"epoch": 0.33,
|
| 11871 |
+
"grad_norm": 1.8109036039046256,
|
| 11872 |
+
"learning_rate": 1.573619011217288e-05,
|
| 11873 |
+
"loss": 1.008,
|
| 11874 |
+
"step": 1695
|
| 11875 |
+
},
|
| 11876 |
+
{
|
| 11877 |
+
"epoch": 0.33,
|
| 11878 |
+
"grad_norm": 1.7366906202563615,
|
| 11879 |
+
"learning_rate": 1.5731080105823253e-05,
|
| 11880 |
+
"loss": 1.0069,
|
| 11881 |
+
"step": 1696
|
| 11882 |
+
},
|
| 11883 |
+
{
|
| 11884 |
+
"epoch": 0.33,
|
| 11885 |
+
"grad_norm": 1.7910543152195322,
|
| 11886 |
+
"learning_rate": 1.572596787005149e-05,
|
| 11887 |
+
"loss": 1.009,
|
| 11888 |
+
"step": 1697
|
| 11889 |
+
},
|
| 11890 |
+
{
|
| 11891 |
+
"epoch": 0.33,
|
| 11892 |
+
"grad_norm": 1.7770513140362456,
|
| 11893 |
+
"learning_rate": 1.5720853406846284e-05,
|
| 11894 |
+
"loss": 0.9124,
|
| 11895 |
+
"step": 1698
|
| 11896 |
+
},
|
| 11897 |
+
{
|
| 11898 |
+
"epoch": 0.33,
|
| 11899 |
+
"grad_norm": 1.663563980279973,
|
| 11900 |
+
"learning_rate": 1.571573671819719e-05,
|
| 11901 |
+
"loss": 0.9501,
|
| 11902 |
+
"step": 1699
|
| 11903 |
+
},
|
| 11904 |
+
{
|
| 11905 |
+
"epoch": 0.33,
|
| 11906 |
+
"grad_norm": 1.798302917814401,
|
| 11907 |
+
"learning_rate": 1.5710617806094623e-05,
|
| 11908 |
+
"loss": 1.0329,
|
| 11909 |
+
"step": 1700
|
| 11910 |
+
},
|
| 11911 |
+
{
|
| 11912 |
+
"epoch": 0.33,
|
| 11913 |
+
"grad_norm": 1.591105191580427,
|
| 11914 |
+
"learning_rate": 1.5705496672529875e-05,
|
| 11915 |
+
"loss": 0.9011,
|
| 11916 |
+
"step": 1701
|
| 11917 |
+
},
|
| 11918 |
+
{
|
| 11919 |
+
"epoch": 0.33,
|
| 11920 |
+
"grad_norm": 1.6919167077809922,
|
| 11921 |
+
"learning_rate": 1.5700373319495087e-05,
|
| 11922 |
+
"loss": 0.954,
|
| 11923 |
+
"step": 1702
|
| 11924 |
+
},
|
| 11925 |
+
{
|
| 11926 |
+
"epoch": 0.33,
|
| 11927 |
+
"grad_norm": 1.6001732503077335,
|
| 11928 |
+
"learning_rate": 1.5695247748983277e-05,
|
| 11929 |
+
"loss": 0.7931,
|
| 11930 |
+
"step": 1703
|
| 11931 |
+
},
|
| 11932 |
+
{
|
| 11933 |
+
"epoch": 0.33,
|
| 11934 |
+
"grad_norm": 1.7339939929632062,
|
| 11935 |
+
"learning_rate": 1.5690119962988325e-05,
|
| 11936 |
+
"loss": 0.915,
|
| 11937 |
+
"step": 1704
|
| 11938 |
+
},
|
| 11939 |
+
{
|
| 11940 |
+
"epoch": 0.33,
|
| 11941 |
+
"grad_norm": 1.548451551637994,
|
| 11942 |
+
"learning_rate": 1.568498996350496e-05,
|
| 11943 |
+
"loss": 0.9463,
|
| 11944 |
+
"step": 1705
|
| 11945 |
+
},
|
| 11946 |
+
{
|
| 11947 |
+
"epoch": 0.33,
|
| 11948 |
+
"grad_norm": 1.748437218428189,
|
| 11949 |
+
"learning_rate": 1.5679857752528788e-05,
|
| 11950 |
+
"loss": 0.9676,
|
| 11951 |
+
"step": 1706
|
| 11952 |
+
},
|
| 11953 |
+
{
|
| 11954 |
+
"epoch": 0.33,
|
| 11955 |
+
"grad_norm": 1.6471556654707846,
|
| 11956 |
+
"learning_rate": 1.567472333205626e-05,
|
| 11957 |
+
"loss": 0.926,
|
| 11958 |
+
"step": 1707
|
| 11959 |
+
},
|
| 11960 |
+
{
|
| 11961 |
+
"epoch": 0.33,
|
| 11962 |
+
"grad_norm": 1.6877746907602753,
|
| 11963 |
+
"learning_rate": 1.5669586704084697e-05,
|
| 11964 |
+
"loss": 1.0535,
|
| 11965 |
+
"step": 1708
|
| 11966 |
+
},
|
| 11967 |
+
{
|
| 11968 |
+
"epoch": 0.33,
|
| 11969 |
+
"grad_norm": 1.7356876446324196,
|
| 11970 |
+
"learning_rate": 1.566444787061228e-05,
|
| 11971 |
+
"loss": 0.9476,
|
| 11972 |
+
"step": 1709
|
| 11973 |
+
},
|
| 11974 |
+
{
|
| 11975 |
+
"epoch": 0.33,
|
| 11976 |
+
"grad_norm": 1.6872419025030534,
|
| 11977 |
+
"learning_rate": 1.565930683363804e-05,
|
| 11978 |
+
"loss": 0.9989,
|
| 11979 |
+
"step": 1710
|
| 11980 |
+
},
|
| 11981 |
+
{
|
| 11982 |
+
"epoch": 0.33,
|
| 11983 |
+
"grad_norm": 1.0437985973600723,
|
| 11984 |
+
"learning_rate": 1.565416359516187e-05,
|
| 11985 |
+
"loss": 0.8111,
|
| 11986 |
+
"step": 1711
|
| 11987 |
+
},
|
| 11988 |
+
{
|
| 11989 |
+
"epoch": 0.33,
|
| 11990 |
+
"grad_norm": 1.0088547902055691,
|
| 11991 |
+
"learning_rate": 1.564901815718452e-05,
|
| 11992 |
+
"loss": 0.8493,
|
| 11993 |
+
"step": 1712
|
| 11994 |
+
},
|
| 11995 |
+
{
|
| 11996 |
+
"epoch": 0.33,
|
| 11997 |
+
"grad_norm": 1.8308929603793904,
|
| 11998 |
+
"learning_rate": 1.5643870521707588e-05,
|
| 11999 |
+
"loss": 0.9714,
|
| 12000 |
+
"step": 1713
|
| 12001 |
+
},
|
| 12002 |
+
{
|
| 12003 |
+
"epoch": 0.33,
|
| 12004 |
+
"grad_norm": 1.6678011598396532,
|
| 12005 |
+
"learning_rate": 1.5638720690733543e-05,
|
| 12006 |
+
"loss": 0.8796,
|
| 12007 |
+
"step": 1714
|
| 12008 |
+
},
|
| 12009 |
+
{
|
| 12010 |
+
"epoch": 0.33,
|
| 12011 |
+
"grad_norm": 1.6676069738231614,
|
| 12012 |
+
"learning_rate": 1.563356866626569e-05,
|
| 12013 |
+
"loss": 0.9325,
|
| 12014 |
+
"step": 1715
|
| 12015 |
+
},
|
| 12016 |
+
{
|
| 12017 |
+
"epoch": 0.33,
|
| 12018 |
+
"grad_norm": 1.687942802391812,
|
| 12019 |
+
"learning_rate": 1.56284144503082e-05,
|
| 12020 |
+
"loss": 0.879,
|
| 12021 |
+
"step": 1716
|
| 12022 |
+
},
|
| 12023 |
+
{
|
| 12024 |
+
"epoch": 0.33,
|
| 12025 |
+
"grad_norm": 1.5751332093640165,
|
| 12026 |
+
"learning_rate": 1.5623258044866087e-05,
|
| 12027 |
+
"loss": 0.9392,
|
| 12028 |
+
"step": 1717
|
| 12029 |
+
},
|
| 12030 |
+
{
|
| 12031 |
+
"epoch": 0.33,
|
| 12032 |
+
"grad_norm": 1.5575620694150407,
|
| 12033 |
+
"learning_rate": 1.561809945194522e-05,
|
| 12034 |
+
"loss": 0.9539,
|
| 12035 |
+
"step": 1718
|
| 12036 |
+
},
|
| 12037 |
+
{
|
| 12038 |
+
"epoch": 0.33,
|
| 12039 |
+
"grad_norm": 1.6785024493599945,
|
| 12040 |
+
"learning_rate": 1.561293867355233e-05,
|
| 12041 |
+
"loss": 0.8994,
|
| 12042 |
+
"step": 1719
|
| 12043 |
+
},
|
| 12044 |
+
{
|
| 12045 |
+
"epoch": 0.33,
|
| 12046 |
+
"grad_norm": 1.5229589448071077,
|
| 12047 |
+
"learning_rate": 1.560777571169498e-05,
|
| 12048 |
+
"loss": 0.9325,
|
| 12049 |
+
"step": 1720
|
| 12050 |
+
},
|
| 12051 |
+
{
|
| 12052 |
+
"epoch": 0.33,
|
| 12053 |
+
"grad_norm": 1.70296765640194,
|
| 12054 |
+
"learning_rate": 1.5602610568381594e-05,
|
| 12055 |
+
"loss": 0.9706,
|
| 12056 |
+
"step": 1721
|
| 12057 |
+
},
|
| 12058 |
+
{
|
| 12059 |
+
"epoch": 0.33,
|
| 12060 |
+
"grad_norm": 1.6492987597197222,
|
| 12061 |
+
"learning_rate": 1.5597443245621444e-05,
|
| 12062 |
+
"loss": 0.9846,
|
| 12063 |
+
"step": 1722
|
| 12064 |
+
},
|
| 12065 |
+
{
|
| 12066 |
+
"epoch": 0.33,
|
| 12067 |
+
"grad_norm": 1.5539259406249721,
|
| 12068 |
+
"learning_rate": 1.5592273745424643e-05,
|
| 12069 |
+
"loss": 0.8847,
|
| 12070 |
+
"step": 1723
|
| 12071 |
+
},
|
| 12072 |
+
{
|
| 12073 |
+
"epoch": 0.33,
|
| 12074 |
+
"grad_norm": 1.9574877365730032,
|
| 12075 |
+
"learning_rate": 1.5587102069802156e-05,
|
| 12076 |
+
"loss": 0.9819,
|
| 12077 |
+
"step": 1724
|
| 12078 |
+
},
|
| 12079 |
+
{
|
| 12080 |
+
"epoch": 0.33,
|
| 12081 |
+
"grad_norm": 1.059334241300379,
|
| 12082 |
+
"learning_rate": 1.5581928220765792e-05,
|
| 12083 |
+
"loss": 0.8633,
|
| 12084 |
+
"step": 1725
|
| 12085 |
+
},
|
| 12086 |
+
{
|
| 12087 |
+
"epoch": 0.33,
|
| 12088 |
+
"grad_norm": 1.8777654599053444,
|
| 12089 |
+
"learning_rate": 1.5576752200328217e-05,
|
| 12090 |
+
"loss": 0.9534,
|
| 12091 |
+
"step": 1726
|
| 12092 |
+
},
|
| 12093 |
+
{
|
| 12094 |
+
"epoch": 0.33,
|
| 12095 |
+
"grad_norm": 1.6369275243616173,
|
| 12096 |
+
"learning_rate": 1.5571574010502925e-05,
|
| 12097 |
+
"loss": 0.9465,
|
| 12098 |
+
"step": 1727
|
| 12099 |
+
},
|
| 12100 |
+
{
|
| 12101 |
+
"epoch": 0.33,
|
| 12102 |
+
"grad_norm": 1.956314276556218,
|
| 12103 |
+
"learning_rate": 1.5566393653304255e-05,
|
| 12104 |
+
"loss": 0.9679,
|
| 12105 |
+
"step": 1728
|
| 12106 |
+
},
|
| 12107 |
+
{
|
| 12108 |
+
"epoch": 0.33,
|
| 12109 |
+
"grad_norm": 1.6159927205594145,
|
| 12110 |
+
"learning_rate": 1.5561211130747405e-05,
|
| 12111 |
+
"loss": 0.892,
|
| 12112 |
+
"step": 1729
|
| 12113 |
+
},
|
| 12114 |
+
{
|
| 12115 |
+
"epoch": 0.33,
|
| 12116 |
+
"grad_norm": 1.7028870865116463,
|
| 12117 |
+
"learning_rate": 1.5556026444848398e-05,
|
| 12118 |
+
"loss": 0.9115,
|
| 12119 |
+
"step": 1730
|
| 12120 |
+
},
|
| 12121 |
+
{
|
| 12122 |
+
"epoch": 0.33,
|
| 12123 |
+
"grad_norm": 1.8208461469027895,
|
| 12124 |
+
"learning_rate": 1.5550839597624106e-05,
|
| 12125 |
+
"loss": 0.8889,
|
| 12126 |
+
"step": 1731
|
| 12127 |
+
},
|
| 12128 |
+
{
|
| 12129 |
+
"epoch": 0.33,
|
| 12130 |
+
"grad_norm": 1.91661243257076,
|
| 12131 |
+
"learning_rate": 1.5545650591092248e-05,
|
| 12132 |
+
"loss": 0.97,
|
| 12133 |
+
"step": 1732
|
| 12134 |
+
},
|
| 12135 |
+
{
|
| 12136 |
+
"epoch": 0.33,
|
| 12137 |
+
"grad_norm": 1.60671095140956,
|
| 12138 |
+
"learning_rate": 1.5540459427271373e-05,
|
| 12139 |
+
"loss": 0.9894,
|
| 12140 |
+
"step": 1733
|
| 12141 |
+
},
|
| 12142 |
+
{
|
| 12143 |
+
"epoch": 0.33,
|
| 12144 |
+
"grad_norm": 1.6111175098454973,
|
| 12145 |
+
"learning_rate": 1.553526610818087e-05,
|
| 12146 |
+
"loss": 0.9202,
|
| 12147 |
+
"step": 1734
|
| 12148 |
+
},
|
| 12149 |
+
{
|
| 12150 |
+
"epoch": 0.33,
|
| 12151 |
+
"grad_norm": 1.7534583643676716,
|
| 12152 |
+
"learning_rate": 1.553007063584097e-05,
|
| 12153 |
+
"loss": 0.8502,
|
| 12154 |
+
"step": 1735
|
| 12155 |
+
},
|
| 12156 |
+
{
|
| 12157 |
+
"epoch": 0.33,
|
| 12158 |
+
"grad_norm": 1.6062441383768804,
|
| 12159 |
+
"learning_rate": 1.5524873012272742e-05,
|
| 12160 |
+
"loss": 1.0042,
|
| 12161 |
+
"step": 1736
|
| 12162 |
+
},
|
| 12163 |
+
{
|
| 12164 |
+
"epoch": 0.33,
|
| 12165 |
+
"grad_norm": 1.5342773548854218,
|
| 12166 |
+
"learning_rate": 1.551967323949809e-05,
|
| 12167 |
+
"loss": 0.867,
|
| 12168 |
+
"step": 1737
|
| 12169 |
+
},
|
| 12170 |
+
{
|
| 12171 |
+
"epoch": 0.33,
|
| 12172 |
+
"grad_norm": 1.6482540145045306,
|
| 12173 |
+
"learning_rate": 1.551447131953976e-05,
|
| 12174 |
+
"loss": 0.9283,
|
| 12175 |
+
"step": 1738
|
| 12176 |
+
},
|
| 12177 |
+
{
|
| 12178 |
+
"epoch": 0.33,
|
| 12179 |
+
"grad_norm": 1.616014640944003,
|
| 12180 |
+
"learning_rate": 1.550926725442132e-05,
|
| 12181 |
+
"loss": 0.8996,
|
| 12182 |
+
"step": 1739
|
| 12183 |
+
},
|
| 12184 |
+
{
|
| 12185 |
+
"epoch": 0.34,
|
| 12186 |
+
"grad_norm": 1.514494552942554,
|
| 12187 |
+
"learning_rate": 1.550406104616718e-05,
|
| 12188 |
+
"loss": 0.9248,
|
| 12189 |
+
"step": 1740
|
| 12190 |
+
},
|
| 12191 |
+
{
|
| 12192 |
+
"epoch": 0.34,
|
| 12193 |
+
"grad_norm": 1.5363529361022905,
|
| 12194 |
+
"learning_rate": 1.5498852696802595e-05,
|
| 12195 |
+
"loss": 0.948,
|
| 12196 |
+
"step": 1741
|
| 12197 |
+
},
|
| 12198 |
+
{
|
| 12199 |
+
"epoch": 0.34,
|
| 12200 |
+
"grad_norm": 1.6602545790223517,
|
| 12201 |
+
"learning_rate": 1.5493642208353627e-05,
|
| 12202 |
+
"loss": 0.9086,
|
| 12203 |
+
"step": 1742
|
| 12204 |
+
},
|
| 12205 |
+
{
|
| 12206 |
+
"epoch": 0.34,
|
| 12207 |
+
"grad_norm": 1.585488148395561,
|
| 12208 |
+
"learning_rate": 1.5488429582847194e-05,
|
| 12209 |
+
"loss": 0.9484,
|
| 12210 |
+
"step": 1743
|
| 12211 |
+
},
|
| 12212 |
+
{
|
| 12213 |
+
"epoch": 0.34,
|
| 12214 |
+
"grad_norm": 1.7637723283346054,
|
| 12215 |
+
"learning_rate": 1.5483214822311036e-05,
|
| 12216 |
+
"loss": 0.9477,
|
| 12217 |
+
"step": 1744
|
| 12218 |
+
},
|
| 12219 |
+
{
|
| 12220 |
+
"epoch": 0.34,
|
| 12221 |
+
"grad_norm": 1.7728810975772227,
|
| 12222 |
+
"learning_rate": 1.5477997928773722e-05,
|
| 12223 |
+
"loss": 0.8948,
|
| 12224 |
+
"step": 1745
|
| 12225 |
+
},
|
| 12226 |
+
{
|
| 12227 |
+
"epoch": 0.34,
|
| 12228 |
+
"grad_norm": 1.627418598739184,
|
| 12229 |
+
"learning_rate": 1.5472778904264655e-05,
|
| 12230 |
+
"loss": 0.9482,
|
| 12231 |
+
"step": 1746
|
| 12232 |
+
},
|
| 12233 |
+
{
|
| 12234 |
+
"epoch": 0.34,
|
| 12235 |
+
"grad_norm": 1.737617924919354,
|
| 12236 |
+
"learning_rate": 1.546755775081406e-05,
|
| 12237 |
+
"loss": 0.9492,
|
| 12238 |
+
"step": 1747
|
| 12239 |
+
},
|
| 12240 |
+
{
|
| 12241 |
+
"epoch": 0.34,
|
| 12242 |
+
"grad_norm": 1.6996510871147033,
|
| 12243 |
+
"learning_rate": 1.5462334470453e-05,
|
| 12244 |
+
"loss": 0.806,
|
| 12245 |
+
"step": 1748
|
| 12246 |
+
},
|
| 12247 |
+
{
|
| 12248 |
+
"epoch": 0.34,
|
| 12249 |
+
"grad_norm": 1.749003030553493,
|
| 12250 |
+
"learning_rate": 1.5457109065213357e-05,
|
| 12251 |
+
"loss": 0.8445,
|
| 12252 |
+
"step": 1749
|
| 12253 |
+
},
|
| 12254 |
+
{
|
| 12255 |
+
"epoch": 0.34,
|
| 12256 |
+
"grad_norm": 1.6613913869959902,
|
| 12257 |
+
"learning_rate": 1.5451881537127846e-05,
|
| 12258 |
+
"loss": 0.8682,
|
| 12259 |
+
"step": 1750
|
| 12260 |
+
},
|
| 12261 |
+
{
|
| 12262 |
+
"epoch": 0.34,
|
| 12263 |
+
"grad_norm": 1.8209928107994644,
|
| 12264 |
+
"learning_rate": 1.5446651888230002e-05,
|
| 12265 |
+
"loss": 1.0161,
|
| 12266 |
+
"step": 1751
|
| 12267 |
+
},
|
| 12268 |
+
{
|
| 12269 |
+
"epoch": 0.34,
|
| 12270 |
+
"grad_norm": 1.7742834895232118,
|
| 12271 |
+
"learning_rate": 1.5441420120554183e-05,
|
| 12272 |
+
"loss": 0.9762,
|
| 12273 |
+
"step": 1752
|
| 12274 |
+
},
|
| 12275 |
+
{
|
| 12276 |
+
"epoch": 0.34,
|
| 12277 |
+
"grad_norm": 1.4903916353227185,
|
| 12278 |
+
"learning_rate": 1.5436186236135586e-05,
|
| 12279 |
+
"loss": 0.8358,
|
| 12280 |
+
"step": 1753
|
| 12281 |
+
},
|
| 12282 |
+
{
|
| 12283 |
+
"epoch": 0.34,
|
| 12284 |
+
"grad_norm": 1.6523896850698925,
|
| 12285 |
+
"learning_rate": 1.5430950237010215e-05,
|
| 12286 |
+
"loss": 0.918,
|
| 12287 |
+
"step": 1754
|
| 12288 |
+
},
|
| 12289 |
+
{
|
| 12290 |
+
"epoch": 0.34,
|
| 12291 |
+
"grad_norm": 1.827097756747899,
|
| 12292 |
+
"learning_rate": 1.5425712125214905e-05,
|
| 12293 |
+
"loss": 0.9376,
|
| 12294 |
+
"step": 1755
|
| 12295 |
+
},
|
| 12296 |
+
{
|
| 12297 |
+
"epoch": 0.34,
|
| 12298 |
+
"grad_norm": 1.6366236729315444,
|
| 12299 |
+
"learning_rate": 1.542047190278731e-05,
|
| 12300 |
+
"loss": 1.0006,
|
| 12301 |
+
"step": 1756
|
| 12302 |
+
},
|
| 12303 |
+
{
|
| 12304 |
+
"epoch": 0.34,
|
| 12305 |
+
"grad_norm": 1.0524696605763608,
|
| 12306 |
+
"learning_rate": 1.54152295717659e-05,
|
| 12307 |
+
"loss": 0.8466,
|
| 12308 |
+
"step": 1757
|
| 12309 |
+
},
|
| 12310 |
+
{
|
| 12311 |
+
"epoch": 0.34,
|
| 12312 |
+
"grad_norm": 1.592256955879349,
|
| 12313 |
+
"learning_rate": 1.5409985134189984e-05,
|
| 12314 |
+
"loss": 0.8885,
|
| 12315 |
+
"step": 1758
|
| 12316 |
+
},
|
| 12317 |
+
{
|
| 12318 |
+
"epoch": 0.34,
|
| 12319 |
+
"grad_norm": 1.5958006272826617,
|
| 12320 |
+
"learning_rate": 1.5404738592099665e-05,
|
| 12321 |
+
"loss": 0.914,
|
| 12322 |
+
"step": 1759
|
| 12323 |
+
},
|
| 12324 |
+
{
|
| 12325 |
+
"epoch": 0.34,
|
| 12326 |
+
"grad_norm": 1.7383735027873934,
|
| 12327 |
+
"learning_rate": 1.5399489947535884e-05,
|
| 12328 |
+
"loss": 0.9507,
|
| 12329 |
+
"step": 1760
|
| 12330 |
+
},
|
| 12331 |
+
{
|
| 12332 |
+
"epoch": 0.34,
|
| 12333 |
+
"grad_norm": 1.7829696954430776,
|
| 12334 |
+
"learning_rate": 1.539423920254039e-05,
|
| 12335 |
+
"loss": 0.9411,
|
| 12336 |
+
"step": 1761
|
| 12337 |
+
},
|
| 12338 |
+
{
|
| 12339 |
+
"epoch": 0.34,
|
| 12340 |
+
"grad_norm": 1.591157229474544,
|
| 12341 |
+
"learning_rate": 1.538898635915576e-05,
|
| 12342 |
+
"loss": 0.9009,
|
| 12343 |
+
"step": 1762
|
| 12344 |
+
},
|
| 12345 |
+
{
|
| 12346 |
+
"epoch": 0.34,
|
| 12347 |
+
"grad_norm": 1.8961964721472366,
|
| 12348 |
+
"learning_rate": 1.5383731419425363e-05,
|
| 12349 |
+
"loss": 0.987,
|
| 12350 |
+
"step": 1763
|
| 12351 |
+
},
|
| 12352 |
+
{
|
| 12353 |
+
"epoch": 0.34,
|
| 12354 |
+
"grad_norm": 1.7139184167666317,
|
| 12355 |
+
"learning_rate": 1.5378474385393412e-05,
|
| 12356 |
+
"loss": 0.9263,
|
| 12357 |
+
"step": 1764
|
| 12358 |
+
},
|
| 12359 |
+
{
|
| 12360 |
+
"epoch": 0.34,
|
| 12361 |
+
"grad_norm": 1.5793588095105051,
|
| 12362 |
+
"learning_rate": 1.537321525910492e-05,
|
| 12363 |
+
"loss": 0.8779,
|
| 12364 |
+
"step": 1765
|
| 12365 |
+
},
|
| 12366 |
+
{
|
| 12367 |
+
"epoch": 0.34,
|
| 12368 |
+
"grad_norm": 1.806016217319507,
|
| 12369 |
+
"learning_rate": 1.536795404260572e-05,
|
| 12370 |
+
"loss": 0.9229,
|
| 12371 |
+
"step": 1766
|
| 12372 |
+
},
|
| 12373 |
+
{
|
| 12374 |
+
"epoch": 0.34,
|
| 12375 |
+
"grad_norm": 1.756089856756367,
|
| 12376 |
+
"learning_rate": 1.5362690737942446e-05,
|
| 12377 |
+
"loss": 0.8605,
|
| 12378 |
+
"step": 1767
|
| 12379 |
+
},
|
| 12380 |
+
{
|
| 12381 |
+
"epoch": 0.34,
|
| 12382 |
+
"grad_norm": 1.8572118885860662,
|
| 12383 |
+
"learning_rate": 1.5357425347162564e-05,
|
| 12384 |
+
"loss": 0.9234,
|
| 12385 |
+
"step": 1768
|
| 12386 |
+
},
|
| 12387 |
+
{
|
| 12388 |
+
"epoch": 0.34,
|
| 12389 |
+
"grad_norm": 1.7856241806716384,
|
| 12390 |
+
"learning_rate": 1.5352157872314327e-05,
|
| 12391 |
+
"loss": 1.0146,
|
| 12392 |
+
"step": 1769
|
| 12393 |
+
},
|
| 12394 |
+
{
|
| 12395 |
+
"epoch": 0.34,
|
| 12396 |
+
"grad_norm": 1.610512365395507,
|
| 12397 |
+
"learning_rate": 1.5346888315446824e-05,
|
| 12398 |
+
"loss": 0.873,
|
| 12399 |
+
"step": 1770
|
| 12400 |
+
},
|
| 12401 |
+
{
|
| 12402 |
+
"epoch": 0.34,
|
| 12403 |
+
"grad_norm": 1.5782211017109056,
|
| 12404 |
+
"learning_rate": 1.5341616678609935e-05,
|
| 12405 |
+
"loss": 0.8567,
|
| 12406 |
+
"step": 1771
|
| 12407 |
+
},
|
| 12408 |
+
{
|
| 12409 |
+
"epoch": 0.34,
|
| 12410 |
+
"grad_norm": 1.654570207794557,
|
| 12411 |
+
"learning_rate": 1.533634296385436e-05,
|
| 12412 |
+
"loss": 0.8263,
|
| 12413 |
+
"step": 1772
|
| 12414 |
+
},
|
| 12415 |
+
{
|
| 12416 |
+
"epoch": 0.34,
|
| 12417 |
+
"grad_norm": 1.393170469294069,
|
| 12418 |
+
"learning_rate": 1.5331067173231603e-05,
|
| 12419 |
+
"loss": 0.882,
|
| 12420 |
+
"step": 1773
|
| 12421 |
+
},
|
| 12422 |
+
{
|
| 12423 |
+
"epoch": 0.34,
|
| 12424 |
+
"grad_norm": 1.5878170235295708,
|
| 12425 |
+
"learning_rate": 1.5325789308793974e-05,
|
| 12426 |
+
"loss": 0.9394,
|
| 12427 |
+
"step": 1774
|
| 12428 |
+
},
|
| 12429 |
+
{
|
| 12430 |
+
"epoch": 0.34,
|
| 12431 |
+
"grad_norm": 1.7372761113657915,
|
| 12432 |
+
"learning_rate": 1.5320509372594596e-05,
|
| 12433 |
+
"loss": 0.9523,
|
| 12434 |
+
"step": 1775
|
| 12435 |
+
},
|
| 12436 |
+
{
|
| 12437 |
+
"epoch": 0.34,
|
| 12438 |
+
"grad_norm": 1.7300471659324526,
|
| 12439 |
+
"learning_rate": 1.5315227366687386e-05,
|
| 12440 |
+
"loss": 1.0319,
|
| 12441 |
+
"step": 1776
|
| 12442 |
+
},
|
| 12443 |
+
{
|
| 12444 |
+
"epoch": 0.34,
|
| 12445 |
+
"grad_norm": 1.9409527939786382,
|
| 12446 |
+
"learning_rate": 1.5309943293127076e-05,
|
| 12447 |
+
"loss": 1.0208,
|
| 12448 |
+
"step": 1777
|
| 12449 |
+
},
|
| 12450 |
+
{
|
| 12451 |
+
"epoch": 0.34,
|
| 12452 |
+
"grad_norm": 1.5078314012936311,
|
| 12453 |
+
"learning_rate": 1.5304657153969206e-05,
|
| 12454 |
+
"loss": 0.9007,
|
| 12455 |
+
"step": 1778
|
| 12456 |
+
},
|
| 12457 |
+
{
|
| 12458 |
+
"epoch": 0.34,
|
| 12459 |
+
"grad_norm": 1.6940177485783197,
|
| 12460 |
+
"learning_rate": 1.529936895127011e-05,
|
| 12461 |
+
"loss": 0.9291,
|
| 12462 |
+
"step": 1779
|
| 12463 |
+
},
|
| 12464 |
+
{
|
| 12465 |
+
"epoch": 0.34,
|
| 12466 |
+
"grad_norm": 1.6538468073075518,
|
| 12467 |
+
"learning_rate": 1.5294078687086926e-05,
|
| 12468 |
+
"loss": 0.8828,
|
| 12469 |
+
"step": 1780
|
| 12470 |
+
},
|
| 12471 |
+
{
|
| 12472 |
+
"epoch": 0.34,
|
| 12473 |
+
"grad_norm": 1.6245630899167927,
|
| 12474 |
+
"learning_rate": 1.5288786363477603e-05,
|
| 12475 |
+
"loss": 0.9154,
|
| 12476 |
+
"step": 1781
|
| 12477 |
+
},
|
| 12478 |
+
{
|
| 12479 |
+
"epoch": 0.34,
|
| 12480 |
+
"grad_norm": 1.644068790077352,
|
| 12481 |
+
"learning_rate": 1.5283491982500874e-05,
|
| 12482 |
+
"loss": 0.9155,
|
| 12483 |
+
"step": 1782
|
| 12484 |
+
},
|
| 12485 |
+
{
|
| 12486 |
+
"epoch": 0.34,
|
| 12487 |
+
"grad_norm": 1.7621859790247505,
|
| 12488 |
+
"learning_rate": 1.5278195546216293e-05,
|
| 12489 |
+
"loss": 0.9127,
|
| 12490 |
+
"step": 1783
|
| 12491 |
+
},
|
| 12492 |
+
{
|
| 12493 |
+
"epoch": 0.34,
|
| 12494 |
+
"grad_norm": 1.6412691064740812,
|
| 12495 |
+
"learning_rate": 1.5272897056684196e-05,
|
| 12496 |
+
"loss": 0.9642,
|
| 12497 |
+
"step": 1784
|
| 12498 |
+
},
|
| 12499 |
+
{
|
| 12500 |
+
"epoch": 0.34,
|
| 12501 |
+
"grad_norm": 1.7284551662658276,
|
| 12502 |
+
"learning_rate": 1.526759651596573e-05,
|
| 12503 |
+
"loss": 0.9682,
|
| 12504 |
+
"step": 1785
|
| 12505 |
+
},
|
| 12506 |
+
{
|
| 12507 |
+
"epoch": 0.34,
|
| 12508 |
+
"grad_norm": 1.7534777399859998,
|
| 12509 |
+
"learning_rate": 1.526229392612283e-05,
|
| 12510 |
+
"loss": 0.9304,
|
| 12511 |
+
"step": 1786
|
| 12512 |
+
},
|
| 12513 |
+
{
|
| 12514 |
+
"epoch": 0.34,
|
| 12515 |
+
"grad_norm": 1.610391858816259,
|
| 12516 |
+
"learning_rate": 1.5256989289218237e-05,
|
| 12517 |
+
"loss": 0.9148,
|
| 12518 |
+
"step": 1787
|
| 12519 |
+
},
|
| 12520 |
+
{
|
| 12521 |
+
"epoch": 0.34,
|
| 12522 |
+
"grad_norm": 1.8330198235703115,
|
| 12523 |
+
"learning_rate": 1.5251682607315485e-05,
|
| 12524 |
+
"loss": 0.9917,
|
| 12525 |
+
"step": 1788
|
| 12526 |
+
},
|
| 12527 |
+
{
|
| 12528 |
+
"epoch": 0.34,
|
| 12529 |
+
"grad_norm": 1.6130949199706817,
|
| 12530 |
+
"learning_rate": 1.5246373882478899e-05,
|
| 12531 |
+
"loss": 0.8627,
|
| 12532 |
+
"step": 1789
|
| 12533 |
+
},
|
| 12534 |
+
{
|
| 12535 |
+
"epoch": 0.34,
|
| 12536 |
+
"grad_norm": 1.6917099405793583,
|
| 12537 |
+
"learning_rate": 1.5241063116773606e-05,
|
| 12538 |
+
"loss": 0.8093,
|
| 12539 |
+
"step": 1790
|
| 12540 |
+
},
|
| 12541 |
+
{
|
| 12542 |
+
"epoch": 0.34,
|
| 12543 |
+
"grad_norm": 1.611514855787027,
|
| 12544 |
+
"learning_rate": 1.5235750312265522e-05,
|
| 12545 |
+
"loss": 0.8823,
|
| 12546 |
+
"step": 1791
|
| 12547 |
+
},
|
| 12548 |
+
{
|
| 12549 |
+
"epoch": 0.35,
|
| 12550 |
+
"grad_norm": 1.508488536223312,
|
| 12551 |
+
"learning_rate": 1.5230435471021356e-05,
|
| 12552 |
+
"loss": 0.8525,
|
| 12553 |
+
"step": 1792
|
| 12554 |
+
},
|
| 12555 |
+
{
|
| 12556 |
+
"epoch": 0.35,
|
| 12557 |
+
"grad_norm": 1.8488504173801166,
|
| 12558 |
+
"learning_rate": 1.5225118595108615e-05,
|
| 12559 |
+
"loss": 0.9723,
|
| 12560 |
+
"step": 1793
|
| 12561 |
+
},
|
| 12562 |
+
{
|
| 12563 |
+
"epoch": 0.35,
|
| 12564 |
+
"grad_norm": 1.81787454238434,
|
| 12565 |
+
"learning_rate": 1.5219799686595588e-05,
|
| 12566 |
+
"loss": 0.9186,
|
| 12567 |
+
"step": 1794
|
| 12568 |
+
},
|
| 12569 |
+
{
|
| 12570 |
+
"epoch": 0.35,
|
| 12571 |
+
"grad_norm": 1.835940653590753,
|
| 12572 |
+
"learning_rate": 1.5214478747551367e-05,
|
| 12573 |
+
"loss": 1.0315,
|
| 12574 |
+
"step": 1795
|
| 12575 |
+
},
|
| 12576 |
+
{
|
| 12577 |
+
"epoch": 0.35,
|
| 12578 |
+
"grad_norm": 1.7183650877031766,
|
| 12579 |
+
"learning_rate": 1.520915578004582e-05,
|
| 12580 |
+
"loss": 0.9409,
|
| 12581 |
+
"step": 1796
|
| 12582 |
+
},
|
| 12583 |
+
{
|
| 12584 |
+
"epoch": 0.35,
|
| 12585 |
+
"grad_norm": 1.6105105683049763,
|
| 12586 |
+
"learning_rate": 1.5203830786149615e-05,
|
| 12587 |
+
"loss": 0.9783,
|
| 12588 |
+
"step": 1797
|
| 12589 |
+
},
|
| 12590 |
+
{
|
| 12591 |
+
"epoch": 0.35,
|
| 12592 |
+
"grad_norm": 1.6828073709950533,
|
| 12593 |
+
"learning_rate": 1.51985037679342e-05,
|
| 12594 |
+
"loss": 0.95,
|
| 12595 |
+
"step": 1798
|
| 12596 |
+
},
|
| 12597 |
+
{
|
| 12598 |
+
"epoch": 0.35,
|
| 12599 |
+
"grad_norm": 1.7659646765890635,
|
| 12600 |
+
"learning_rate": 1.5193174727471822e-05,
|
| 12601 |
+
"loss": 0.9881,
|
| 12602 |
+
"step": 1799
|
| 12603 |
+
},
|
| 12604 |
+
{
|
| 12605 |
+
"epoch": 0.35,
|
| 12606 |
+
"grad_norm": 1.7186747858397031,
|
| 12607 |
+
"learning_rate": 1.5187843666835502e-05,
|
| 12608 |
+
"loss": 0.9142,
|
| 12609 |
+
"step": 1800
|
| 12610 |
+
},
|
| 12611 |
+
{
|
| 12612 |
+
"epoch": 0.35,
|
| 12613 |
+
"grad_norm": 1.5025090660678702,
|
| 12614 |
+
"learning_rate": 1.5182510588099058e-05,
|
| 12615 |
+
"loss": 0.8743,
|
| 12616 |
+
"step": 1801
|
| 12617 |
+
},
|
| 12618 |
+
{
|
| 12619 |
+
"epoch": 0.35,
|
| 12620 |
+
"grad_norm": 1.7934959998223838,
|
| 12621 |
+
"learning_rate": 1.5177175493337077e-05,
|
| 12622 |
+
"loss": 0.9239,
|
| 12623 |
+
"step": 1802
|
| 12624 |
+
},
|
| 12625 |
+
{
|
| 12626 |
+
"epoch": 0.35,
|
| 12627 |
+
"grad_norm": 1.5578282419512874,
|
| 12628 |
+
"learning_rate": 1.5171838384624952e-05,
|
| 12629 |
+
"loss": 0.869,
|
| 12630 |
+
"step": 1803
|
| 12631 |
+
},
|
| 12632 |
+
{
|
| 12633 |
+
"epoch": 0.35,
|
| 12634 |
+
"grad_norm": 1.7377002208663477,
|
| 12635 |
+
"learning_rate": 1.516649926403884e-05,
|
| 12636 |
+
"loss": 0.968,
|
| 12637 |
+
"step": 1804
|
| 12638 |
+
},
|
| 12639 |
+
{
|
| 12640 |
+
"epoch": 0.35,
|
| 12641 |
+
"grad_norm": 1.775042883902276,
|
| 12642 |
+
"learning_rate": 1.5161158133655694e-05,
|
| 12643 |
+
"loss": 0.9978,
|
| 12644 |
+
"step": 1805
|
| 12645 |
+
},
|
| 12646 |
+
{
|
| 12647 |
+
"epoch": 0.35,
|
| 12648 |
+
"grad_norm": 1.5911271575560177,
|
| 12649 |
+
"learning_rate": 1.5155814995553239e-05,
|
| 12650 |
+
"loss": 0.8461,
|
| 12651 |
+
"step": 1806
|
| 12652 |
+
},
|
| 12653 |
+
{
|
| 12654 |
+
"epoch": 0.35,
|
| 12655 |
+
"grad_norm": 1.9003801818751707,
|
| 12656 |
+
"learning_rate": 1.515046985180999e-05,
|
| 12657 |
+
"loss": 0.9565,
|
| 12658 |
+
"step": 1807
|
| 12659 |
+
},
|
| 12660 |
+
{
|
| 12661 |
+
"epoch": 0.35,
|
| 12662 |
+
"grad_norm": 1.7235091223002925,
|
| 12663 |
+
"learning_rate": 1.5145122704505238e-05,
|
| 12664 |
+
"loss": 0.9244,
|
| 12665 |
+
"step": 1808
|
| 12666 |
+
},
|
| 12667 |
+
{
|
| 12668 |
+
"epoch": 0.35,
|
| 12669 |
+
"grad_norm": 1.5836993320305877,
|
| 12670 |
+
"learning_rate": 1.5139773555719046e-05,
|
| 12671 |
+
"loss": 0.9256,
|
| 12672 |
+
"step": 1809
|
| 12673 |
+
},
|
| 12674 |
+
{
|
| 12675 |
+
"epoch": 0.35,
|
| 12676 |
+
"grad_norm": 1.7698468539495165,
|
| 12677 |
+
"learning_rate": 1.5134422407532268e-05,
|
| 12678 |
+
"loss": 0.9067,
|
| 12679 |
+
"step": 1810
|
| 12680 |
+
},
|
| 12681 |
+
{
|
| 12682 |
+
"epoch": 0.35,
|
| 12683 |
+
"grad_norm": 1.6257859529986807,
|
| 12684 |
+
"learning_rate": 1.512906926202653e-05,
|
| 12685 |
+
"loss": 0.879,
|
| 12686 |
+
"step": 1811
|
| 12687 |
+
},
|
| 12688 |
+
{
|
| 12689 |
+
"epoch": 0.35,
|
| 12690 |
+
"grad_norm": 1.6883738442738565,
|
| 12691 |
+
"learning_rate": 1.512371412128424e-05,
|
| 12692 |
+
"loss": 0.9165,
|
| 12693 |
+
"step": 1812
|
| 12694 |
+
},
|
| 12695 |
+
{
|
| 12696 |
+
"epoch": 0.35,
|
| 12697 |
+
"grad_norm": 1.641267657042082,
|
| 12698 |
+
"learning_rate": 1.5118356987388567e-05,
|
| 12699 |
+
"loss": 0.9208,
|
| 12700 |
+
"step": 1813
|
| 12701 |
+
},
|
| 12702 |
+
{
|
| 12703 |
+
"epoch": 0.35,
|
| 12704 |
+
"grad_norm": 1.5234773704787912,
|
| 12705 |
+
"learning_rate": 1.5112997862423472e-05,
|
| 12706 |
+
"loss": 0.939,
|
| 12707 |
+
"step": 1814
|
| 12708 |
+
},
|
| 12709 |
+
{
|
| 12710 |
+
"epoch": 0.35,
|
| 12711 |
+
"grad_norm": 1.699284171054198,
|
| 12712 |
+
"learning_rate": 1.5107636748473687e-05,
|
| 12713 |
+
"loss": 0.927,
|
| 12714 |
+
"step": 1815
|
| 12715 |
+
},
|
| 12716 |
+
{
|
| 12717 |
+
"epoch": 0.35,
|
| 12718 |
+
"grad_norm": 1.7766671571230526,
|
| 12719 |
+
"learning_rate": 1.5102273647624714e-05,
|
| 12720 |
+
"loss": 0.9559,
|
| 12721 |
+
"step": 1816
|
| 12722 |
+
},
|
| 12723 |
+
{
|
| 12724 |
+
"epoch": 0.35,
|
| 12725 |
+
"grad_norm": 1.8071128001622048,
|
| 12726 |
+
"learning_rate": 1.5096908561962824e-05,
|
| 12727 |
+
"loss": 0.9329,
|
| 12728 |
+
"step": 1817
|
| 12729 |
+
},
|
| 12730 |
+
{
|
| 12731 |
+
"epoch": 0.35,
|
| 12732 |
+
"grad_norm": 1.5602195899860873,
|
| 12733 |
+
"learning_rate": 1.509154149357507e-05,
|
| 12734 |
+
"loss": 0.8408,
|
| 12735 |
+
"step": 1818
|
| 12736 |
+
},
|
| 12737 |
+
{
|
| 12738 |
+
"epoch": 0.35,
|
| 12739 |
+
"grad_norm": 1.6681745264243264,
|
| 12740 |
+
"learning_rate": 1.5086172444549274e-05,
|
| 12741 |
+
"loss": 0.9715,
|
| 12742 |
+
"step": 1819
|
| 12743 |
+
},
|
| 12744 |
+
{
|
| 12745 |
+
"epoch": 0.35,
|
| 12746 |
+
"grad_norm": 1.8299212257238782,
|
| 12747 |
+
"learning_rate": 1.508080141697402e-05,
|
| 12748 |
+
"loss": 0.9271,
|
| 12749 |
+
"step": 1820
|
| 12750 |
+
},
|
| 12751 |
+
{
|
| 12752 |
+
"epoch": 0.35,
|
| 12753 |
+
"grad_norm": 1.6199015194976085,
|
| 12754 |
+
"learning_rate": 1.5075428412938672e-05,
|
| 12755 |
+
"loss": 0.8942,
|
| 12756 |
+
"step": 1821
|
| 12757 |
+
},
|
| 12758 |
+
{
|
| 12759 |
+
"epoch": 0.35,
|
| 12760 |
+
"grad_norm": 1.637504821627329,
|
| 12761 |
+
"learning_rate": 1.5070053434533351e-05,
|
| 12762 |
+
"loss": 0.9063,
|
| 12763 |
+
"step": 1822
|
| 12764 |
+
},
|
| 12765 |
+
{
|
| 12766 |
+
"epoch": 0.35,
|
| 12767 |
+
"grad_norm": 1.676609757359368,
|
| 12768 |
+
"learning_rate": 1.5064676483848968e-05,
|
| 12769 |
+
"loss": 0.8435,
|
| 12770 |
+
"step": 1823
|
| 12771 |
+
},
|
| 12772 |
+
{
|
| 12773 |
+
"epoch": 0.35,
|
| 12774 |
+
"grad_norm": 1.759464544373338,
|
| 12775 |
+
"learning_rate": 1.5059297562977174e-05,
|
| 12776 |
+
"loss": 0.9497,
|
| 12777 |
+
"step": 1824
|
| 12778 |
+
},
|
| 12779 |
+
{
|
| 12780 |
+
"epoch": 0.35,
|
| 12781 |
+
"grad_norm": 1.8044880904814558,
|
| 12782 |
+
"learning_rate": 1.5053916674010403e-05,
|
| 12783 |
+
"loss": 0.921,
|
| 12784 |
+
"step": 1825
|
| 12785 |
+
},
|
| 12786 |
+
{
|
| 12787 |
+
"epoch": 0.35,
|
| 12788 |
+
"grad_norm": 1.6629588750577138,
|
| 12789 |
+
"learning_rate": 1.5048533819041853e-05,
|
| 12790 |
+
"loss": 0.9421,
|
| 12791 |
+
"step": 1826
|
| 12792 |
+
},
|
| 12793 |
+
{
|
| 12794 |
+
"epoch": 0.35,
|
| 12795 |
+
"grad_norm": 2.0351132912201115,
|
| 12796 |
+
"learning_rate": 1.5043149000165482e-05,
|
| 12797 |
+
"loss": 0.9275,
|
| 12798 |
+
"step": 1827
|
| 12799 |
+
},
|
| 12800 |
+
{
|
| 12801 |
+
"epoch": 0.35,
|
| 12802 |
+
"grad_norm": 1.7221826608550477,
|
| 12803 |
+
"learning_rate": 1.5037762219476016e-05,
|
| 12804 |
+
"loss": 0.9359,
|
| 12805 |
+
"step": 1828
|
| 12806 |
+
},
|
| 12807 |
+
{
|
| 12808 |
+
"epoch": 0.35,
|
| 12809 |
+
"grad_norm": 1.8863983035535,
|
| 12810 |
+
"learning_rate": 1.503237347906894e-05,
|
| 12811 |
+
"loss": 0.9219,
|
| 12812 |
+
"step": 1829
|
| 12813 |
+
},
|
| 12814 |
+
{
|
| 12815 |
+
"epoch": 0.35,
|
| 12816 |
+
"grad_norm": 1.7877241947739209,
|
| 12817 |
+
"learning_rate": 1.5026982781040511e-05,
|
| 12818 |
+
"loss": 0.9239,
|
| 12819 |
+
"step": 1830
|
| 12820 |
+
},
|
| 12821 |
+
{
|
| 12822 |
+
"epoch": 0.35,
|
| 12823 |
+
"grad_norm": 1.7559006781624251,
|
| 12824 |
+
"learning_rate": 1.5021590127487731e-05,
|
| 12825 |
+
"loss": 0.9135,
|
| 12826 |
+
"step": 1831
|
| 12827 |
+
},
|
| 12828 |
+
{
|
| 12829 |
+
"epoch": 0.35,
|
| 12830 |
+
"grad_norm": 1.7628850458243945,
|
| 12831 |
+
"learning_rate": 1.5016195520508383e-05,
|
| 12832 |
+
"loss": 1.0021,
|
| 12833 |
+
"step": 1832
|
| 12834 |
+
},
|
| 12835 |
+
{
|
| 12836 |
+
"epoch": 0.35,
|
| 12837 |
+
"grad_norm": 1.712262152372984,
|
| 12838 |
+
"learning_rate": 1.501079896220099e-05,
|
| 12839 |
+
"loss": 0.9462,
|
| 12840 |
+
"step": 1833
|
| 12841 |
+
},
|
| 12842 |
+
{
|
| 12843 |
+
"epoch": 0.35,
|
| 12844 |
+
"grad_norm": 1.5312995616872846,
|
| 12845 |
+
"learning_rate": 1.5005400454664847e-05,
|
| 12846 |
+
"loss": 0.8234,
|
| 12847 |
+
"step": 1834
|
| 12848 |
+
},
|
| 12849 |
+
{
|
| 12850 |
+
"epoch": 0.35,
|
| 12851 |
+
"grad_norm": 1.4310758854139343,
|
| 12852 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 12853 |
+
"loss": 0.8858,
|
| 12854 |
+
"step": 1835
|
| 12855 |
+
},
|
| 12856 |
+
{
|
| 12857 |
+
"epoch": 0.35,
|
| 12858 |
+
"grad_norm": 1.546597318526974,
|
| 12859 |
+
"learning_rate": 1.4994597600307263e-05,
|
| 12860 |
+
"loss": 0.9568,
|
| 12861 |
+
"step": 1836
|
| 12862 |
+
},
|
| 12863 |
+
{
|
| 12864 |
+
"epoch": 0.35,
|
| 12865 |
+
"grad_norm": 1.582830465670841,
|
| 12866 |
+
"learning_rate": 1.4989193257688195e-05,
|
| 12867 |
+
"loss": 0.9045,
|
| 12868 |
+
"step": 1837
|
| 12869 |
+
},
|
| 12870 |
+
{
|
| 12871 |
+
"epoch": 0.35,
|
| 12872 |
+
"grad_norm": 1.6151541196243508,
|
| 12873 |
+
"learning_rate": 1.4983786974245118e-05,
|
| 12874 |
+
"loss": 0.9019,
|
| 12875 |
+
"step": 1838
|
| 12876 |
+
},
|
| 12877 |
+
{
|
| 12878 |
+
"epoch": 0.35,
|
| 12879 |
+
"grad_norm": 1.6357268555554538,
|
| 12880 |
+
"learning_rate": 1.4978378752081105e-05,
|
| 12881 |
+
"loss": 0.9621,
|
| 12882 |
+
"step": 1839
|
| 12883 |
+
},
|
| 12884 |
+
{
|
| 12885 |
+
"epoch": 0.35,
|
| 12886 |
+
"grad_norm": 1.7828133609360786,
|
| 12887 |
+
"learning_rate": 1.497296859329998e-05,
|
| 12888 |
+
"loss": 0.8937,
|
| 12889 |
+
"step": 1840
|
| 12890 |
+
},
|
| 12891 |
+
{
|
| 12892 |
+
"epoch": 0.35,
|
| 12893 |
+
"grad_norm": 1.6209326547083096,
|
| 12894 |
+
"learning_rate": 1.4967556500006336e-05,
|
| 12895 |
+
"loss": 0.9472,
|
| 12896 |
+
"step": 1841
|
| 12897 |
+
},
|
| 12898 |
+
{
|
| 12899 |
+
"epoch": 0.35,
|
| 12900 |
+
"grad_norm": 1.6715487795705135,
|
| 12901 |
+
"learning_rate": 1.4962142474305496e-05,
|
| 12902 |
+
"loss": 0.8342,
|
| 12903 |
+
"step": 1842
|
| 12904 |
+
},
|
| 12905 |
+
{
|
| 12906 |
+
"epoch": 0.35,
|
| 12907 |
+
"grad_norm": 1.644054514212963,
|
| 12908 |
+
"learning_rate": 1.4956726518303553e-05,
|
| 12909 |
+
"loss": 0.8793,
|
| 12910 |
+
"step": 1843
|
| 12911 |
+
},
|
| 12912 |
+
{
|
| 12913 |
+
"epoch": 0.36,
|
| 12914 |
+
"grad_norm": 1.6778795735018404,
|
| 12915 |
+
"learning_rate": 1.495130863410734e-05,
|
| 12916 |
+
"loss": 0.9961,
|
| 12917 |
+
"step": 1844
|
| 12918 |
+
},
|
| 12919 |
+
{
|
| 12920 |
+
"epoch": 0.36,
|
| 12921 |
+
"grad_norm": 1.7530107949989808,
|
| 12922 |
+
"learning_rate": 1.4945888823824449e-05,
|
| 12923 |
+
"loss": 0.8523,
|
| 12924 |
+
"step": 1845
|
| 12925 |
+
},
|
| 12926 |
+
{
|
| 12927 |
+
"epoch": 0.36,
|
| 12928 |
+
"grad_norm": 1.600308338620861,
|
| 12929 |
+
"learning_rate": 1.494046708956321e-05,
|
| 12930 |
+
"loss": 0.8763,
|
| 12931 |
+
"step": 1846
|
| 12932 |
+
},
|
| 12933 |
+
{
|
| 12934 |
+
"epoch": 0.36,
|
| 12935 |
+
"grad_norm": 1.7748182051211745,
|
| 12936 |
+
"learning_rate": 1.4935043433432715e-05,
|
| 12937 |
+
"loss": 0.9168,
|
| 12938 |
+
"step": 1847
|
| 12939 |
+
},
|
| 12940 |
+
{
|
| 12941 |
+
"epoch": 0.36,
|
| 12942 |
+
"grad_norm": 1.6428997560126952,
|
| 12943 |
+
"learning_rate": 1.4929617857542793e-05,
|
| 12944 |
+
"loss": 0.8927,
|
| 12945 |
+
"step": 1848
|
| 12946 |
+
},
|
| 12947 |
+
{
|
| 12948 |
+
"epoch": 0.36,
|
| 12949 |
+
"grad_norm": 1.7377574670461893,
|
| 12950 |
+
"learning_rate": 1.4924190364004023e-05,
|
| 12951 |
+
"loss": 0.869,
|
| 12952 |
+
"step": 1849
|
| 12953 |
+
},
|
| 12954 |
+
{
|
| 12955 |
+
"epoch": 0.36,
|
| 12956 |
+
"grad_norm": 1.7021319792833238,
|
| 12957 |
+
"learning_rate": 1.4918760954927729e-05,
|
| 12958 |
+
"loss": 0.9108,
|
| 12959 |
+
"step": 1850
|
| 12960 |
+
},
|
| 12961 |
+
{
|
| 12962 |
+
"epoch": 0.36,
|
| 12963 |
+
"grad_norm": 1.591116905628271,
|
| 12964 |
+
"learning_rate": 1.4913329632425984e-05,
|
| 12965 |
+
"loss": 0.949,
|
| 12966 |
+
"step": 1851
|
| 12967 |
+
},
|
| 12968 |
+
{
|
| 12969 |
+
"epoch": 0.36,
|
| 12970 |
+
"grad_norm": 1.6276173765055855,
|
| 12971 |
+
"learning_rate": 1.4907896398611603e-05,
|
| 12972 |
+
"loss": 0.89,
|
| 12973 |
+
"step": 1852
|
| 12974 |
+
},
|
| 12975 |
+
{
|
| 12976 |
+
"epoch": 0.36,
|
| 12977 |
+
"grad_norm": 1.8225115626463677,
|
| 12978 |
+
"learning_rate": 1.4902461255598141e-05,
|
| 12979 |
+
"loss": 0.9123,
|
| 12980 |
+
"step": 1853
|
| 12981 |
+
},
|
| 12982 |
+
{
|
| 12983 |
+
"epoch": 0.36,
|
| 12984 |
+
"grad_norm": 1.5498562183799023,
|
| 12985 |
+
"learning_rate": 1.4897024205499902e-05,
|
| 12986 |
+
"loss": 0.8801,
|
| 12987 |
+
"step": 1854
|
| 12988 |
+
},
|
| 12989 |
+
{
|
| 12990 |
+
"epoch": 0.36,
|
| 12991 |
+
"grad_norm": 1.7255906804323513,
|
| 12992 |
+
"learning_rate": 1.4891585250431929e-05,
|
| 12993 |
+
"loss": 0.88,
|
| 12994 |
+
"step": 1855
|
| 12995 |
+
},
|
| 12996 |
+
{
|
| 12997 |
+
"epoch": 0.36,
|
| 12998 |
+
"grad_norm": 1.7392433916238563,
|
| 12999 |
+
"learning_rate": 1.4886144392510004e-05,
|
| 13000 |
+
"loss": 0.9908,
|
| 13001 |
+
"step": 1856
|
| 13002 |
+
},
|
| 13003 |
+
{
|
| 13004 |
+
"epoch": 0.36,
|
| 13005 |
+
"grad_norm": 1.819892755574117,
|
| 13006 |
+
"learning_rate": 1.4880701633850652e-05,
|
| 13007 |
+
"loss": 0.9699,
|
| 13008 |
+
"step": 1857
|
| 13009 |
+
},
|
| 13010 |
+
{
|
| 13011 |
+
"epoch": 0.36,
|
| 13012 |
+
"grad_norm": 1.6453268799227307,
|
| 13013 |
+
"learning_rate": 1.4875256976571135e-05,
|
| 13014 |
+
"loss": 0.9985,
|
| 13015 |
+
"step": 1858
|
| 13016 |
+
},
|
| 13017 |
+
{
|
| 13018 |
+
"epoch": 0.36,
|
| 13019 |
+
"grad_norm": 1.7570555784318829,
|
| 13020 |
+
"learning_rate": 1.4869810422789459e-05,
|
| 13021 |
+
"loss": 0.9425,
|
| 13022 |
+
"step": 1859
|
| 13023 |
+
},
|
| 13024 |
+
{
|
| 13025 |
+
"epoch": 0.36,
|
| 13026 |
+
"grad_norm": 1.6103885798432909,
|
| 13027 |
+
"learning_rate": 1.4864361974624357e-05,
|
| 13028 |
+
"loss": 0.9111,
|
| 13029 |
+
"step": 1860
|
| 13030 |
+
},
|
| 13031 |
+
{
|
| 13032 |
+
"epoch": 0.36,
|
| 13033 |
+
"grad_norm": 1.6392043539397223,
|
| 13034 |
+
"learning_rate": 1.4858911634195312e-05,
|
| 13035 |
+
"loss": 0.9116,
|
| 13036 |
+
"step": 1861
|
| 13037 |
+
},
|
| 13038 |
+
{
|
| 13039 |
+
"epoch": 0.36,
|
| 13040 |
+
"grad_norm": 1.6241801480289666,
|
| 13041 |
+
"learning_rate": 1.4853459403622535e-05,
|
| 13042 |
+
"loss": 0.9424,
|
| 13043 |
+
"step": 1862
|
| 13044 |
+
},
|
| 13045 |
+
{
|
| 13046 |
+
"epoch": 0.36,
|
| 13047 |
+
"grad_norm": 1.634986046998513,
|
| 13048 |
+
"learning_rate": 1.4848005285026976e-05,
|
| 13049 |
+
"loss": 0.9494,
|
| 13050 |
+
"step": 1863
|
| 13051 |
+
},
|
| 13052 |
+
{
|
| 13053 |
+
"epoch": 0.36,
|
| 13054 |
+
"grad_norm": 1.5937895894975387,
|
| 13055 |
+
"learning_rate": 1.4842549280530312e-05,
|
| 13056 |
+
"loss": 0.877,
|
| 13057 |
+
"step": 1864
|
| 13058 |
+
},
|
| 13059 |
+
{
|
| 13060 |
+
"epoch": 0.36,
|
| 13061 |
+
"grad_norm": 1.572921397651084,
|
| 13062 |
+
"learning_rate": 1.4837091392254964e-05,
|
| 13063 |
+
"loss": 0.8937,
|
| 13064 |
+
"step": 1865
|
| 13065 |
+
},
|
| 13066 |
+
{
|
| 13067 |
+
"epoch": 0.36,
|
| 13068 |
+
"grad_norm": 1.6988453094864369,
|
| 13069 |
+
"learning_rate": 1.4831631622324078e-05,
|
| 13070 |
+
"loss": 0.8695,
|
| 13071 |
+
"step": 1866
|
| 13072 |
+
},
|
| 13073 |
+
{
|
| 13074 |
+
"epoch": 0.36,
|
| 13075 |
+
"grad_norm": 1.5134644774277317,
|
| 13076 |
+
"learning_rate": 1.4826169972861539e-05,
|
| 13077 |
+
"loss": 0.8927,
|
| 13078 |
+
"step": 1867
|
| 13079 |
+
},
|
| 13080 |
+
{
|
| 13081 |
+
"epoch": 0.36,
|
| 13082 |
+
"grad_norm": 1.627315805696269,
|
| 13083 |
+
"learning_rate": 1.4820706445991954e-05,
|
| 13084 |
+
"loss": 0.9352,
|
| 13085 |
+
"step": 1868
|
| 13086 |
+
},
|
| 13087 |
+
{
|
| 13088 |
+
"epoch": 0.36,
|
| 13089 |
+
"grad_norm": 1.6393371999448805,
|
| 13090 |
+
"learning_rate": 1.4815241043840669e-05,
|
| 13091 |
+
"loss": 0.8528,
|
| 13092 |
+
"step": 1869
|
| 13093 |
+
},
|
| 13094 |
+
{
|
| 13095 |
+
"epoch": 0.36,
|
| 13096 |
+
"grad_norm": 1.8952895153405638,
|
| 13097 |
+
"learning_rate": 1.4809773768533757e-05,
|
| 13098 |
+
"loss": 0.9755,
|
| 13099 |
+
"step": 1870
|
| 13100 |
+
},
|
| 13101 |
+
{
|
| 13102 |
+
"epoch": 0.36,
|
| 13103 |
+
"grad_norm": 1.6091945516084094,
|
| 13104 |
+
"learning_rate": 1.4804304622198013e-05,
|
| 13105 |
+
"loss": 0.8705,
|
| 13106 |
+
"step": 1871
|
| 13107 |
+
},
|
| 13108 |
+
{
|
| 13109 |
+
"epoch": 0.36,
|
| 13110 |
+
"grad_norm": 1.624495437307758,
|
| 13111 |
+
"learning_rate": 1.4798833606960972e-05,
|
| 13112 |
+
"loss": 0.8524,
|
| 13113 |
+
"step": 1872
|
| 13114 |
+
},
|
| 13115 |
+
{
|
| 13116 |
+
"epoch": 0.36,
|
| 13117 |
+
"grad_norm": 1.8744172065380642,
|
| 13118 |
+
"learning_rate": 1.4793360724950887e-05,
|
| 13119 |
+
"loss": 1.0719,
|
| 13120 |
+
"step": 1873
|
| 13121 |
+
},
|
| 13122 |
+
{
|
| 13123 |
+
"epoch": 0.36,
|
| 13124 |
+
"grad_norm": 1.5575380698894332,
|
| 13125 |
+
"learning_rate": 1.4787885978296737e-05,
|
| 13126 |
+
"loss": 0.9028,
|
| 13127 |
+
"step": 1874
|
| 13128 |
+
},
|
| 13129 |
+
{
|
| 13130 |
+
"epoch": 0.36,
|
| 13131 |
+
"grad_norm": 1.7892688635911647,
|
| 13132 |
+
"learning_rate": 1.478240936912823e-05,
|
| 13133 |
+
"loss": 0.9931,
|
| 13134 |
+
"step": 1875
|
| 13135 |
+
},
|
| 13136 |
+
{
|
| 13137 |
+
"epoch": 0.36,
|
| 13138 |
+
"grad_norm": 1.6951433638470188,
|
| 13139 |
+
"learning_rate": 1.4776930899575801e-05,
|
| 13140 |
+
"loss": 0.9601,
|
| 13141 |
+
"step": 1876
|
| 13142 |
+
},
|
| 13143 |
+
{
|
| 13144 |
+
"epoch": 0.36,
|
| 13145 |
+
"grad_norm": 1.8511925621033531,
|
| 13146 |
+
"learning_rate": 1.4771450571770603e-05,
|
| 13147 |
+
"loss": 1.0311,
|
| 13148 |
+
"step": 1877
|
| 13149 |
+
},
|
| 13150 |
+
{
|
| 13151 |
+
"epoch": 0.36,
|
| 13152 |
+
"grad_norm": 1.1442810135716395,
|
| 13153 |
+
"learning_rate": 1.4765968387844516e-05,
|
| 13154 |
+
"loss": 0.884,
|
| 13155 |
+
"step": 1878
|
| 13156 |
+
},
|
| 13157 |
+
{
|
| 13158 |
+
"epoch": 0.36,
|
| 13159 |
+
"grad_norm": 1.9225596331812123,
|
| 13160 |
+
"learning_rate": 1.4760484349930134e-05,
|
| 13161 |
+
"loss": 0.8523,
|
| 13162 |
+
"step": 1879
|
| 13163 |
+
},
|
| 13164 |
+
{
|
| 13165 |
+
"epoch": 0.36,
|
| 13166 |
+
"grad_norm": 1.808853808526827,
|
| 13167 |
+
"learning_rate": 1.4754998460160784e-05,
|
| 13168 |
+
"loss": 0.9429,
|
| 13169 |
+
"step": 1880
|
| 13170 |
+
},
|
| 13171 |
+
{
|
| 13172 |
+
"epoch": 0.36,
|
| 13173 |
+
"grad_norm": 1.5600398733677634,
|
| 13174 |
+
"learning_rate": 1.4749510720670506e-05,
|
| 13175 |
+
"loss": 0.9736,
|
| 13176 |
+
"step": 1881
|
| 13177 |
+
},
|
| 13178 |
+
{
|
| 13179 |
+
"epoch": 0.36,
|
| 13180 |
+
"grad_norm": 1.5320412874387632,
|
| 13181 |
+
"learning_rate": 1.4744021133594059e-05,
|
| 13182 |
+
"loss": 0.8701,
|
| 13183 |
+
"step": 1882
|
| 13184 |
+
},
|
| 13185 |
+
{
|
| 13186 |
+
"epoch": 0.36,
|
| 13187 |
+
"grad_norm": 1.523014076301379,
|
| 13188 |
+
"learning_rate": 1.4738529701066928e-05,
|
| 13189 |
+
"loss": 0.8883,
|
| 13190 |
+
"step": 1883
|
| 13191 |
+
},
|
| 13192 |
+
{
|
| 13193 |
+
"epoch": 0.36,
|
| 13194 |
+
"grad_norm": 1.7505097977570019,
|
| 13195 |
+
"learning_rate": 1.4733036425225306e-05,
|
| 13196 |
+
"loss": 0.9172,
|
| 13197 |
+
"step": 1884
|
| 13198 |
+
},
|
| 13199 |
+
{
|
| 13200 |
+
"epoch": 0.36,
|
| 13201 |
+
"grad_norm": 1.5554959069631156,
|
| 13202 |
+
"learning_rate": 1.4727541308206114e-05,
|
| 13203 |
+
"loss": 0.8644,
|
| 13204 |
+
"step": 1885
|
| 13205 |
+
},
|
| 13206 |
+
{
|
| 13207 |
+
"epoch": 0.36,
|
| 13208 |
+
"grad_norm": 1.7958310637324932,
|
| 13209 |
+
"learning_rate": 1.4722044352146978e-05,
|
| 13210 |
+
"loss": 0.866,
|
| 13211 |
+
"step": 1886
|
| 13212 |
+
},
|
| 13213 |
+
{
|
| 13214 |
+
"epoch": 0.36,
|
| 13215 |
+
"grad_norm": 1.855213005785535,
|
| 13216 |
+
"learning_rate": 1.4716545559186244e-05,
|
| 13217 |
+
"loss": 0.8861,
|
| 13218 |
+
"step": 1887
|
| 13219 |
+
},
|
| 13220 |
+
{
|
| 13221 |
+
"epoch": 0.36,
|
| 13222 |
+
"grad_norm": 1.660788717810155,
|
| 13223 |
+
"learning_rate": 1.4711044931462976e-05,
|
| 13224 |
+
"loss": 0.888,
|
| 13225 |
+
"step": 1888
|
| 13226 |
+
},
|
| 13227 |
+
{
|
| 13228 |
+
"epoch": 0.36,
|
| 13229 |
+
"grad_norm": 1.5748794402819912,
|
| 13230 |
+
"learning_rate": 1.4705542471116949e-05,
|
| 13231 |
+
"loss": 0.9492,
|
| 13232 |
+
"step": 1889
|
| 13233 |
+
},
|
| 13234 |
+
{
|
| 13235 |
+
"epoch": 0.36,
|
| 13236 |
+
"grad_norm": 1.671940289814532,
|
| 13237 |
+
"learning_rate": 1.470003818028865e-05,
|
| 13238 |
+
"loss": 0.9392,
|
| 13239 |
+
"step": 1890
|
| 13240 |
+
},
|
| 13241 |
+
{
|
| 13242 |
+
"epoch": 0.36,
|
| 13243 |
+
"grad_norm": 1.6682541425119886,
|
| 13244 |
+
"learning_rate": 1.4694532061119277e-05,
|
| 13245 |
+
"loss": 0.9954,
|
| 13246 |
+
"step": 1891
|
| 13247 |
+
},
|
| 13248 |
+
{
|
| 13249 |
+
"epoch": 0.36,
|
| 13250 |
+
"grad_norm": 1.6270272723817845,
|
| 13251 |
+
"learning_rate": 1.4689024115750745e-05,
|
| 13252 |
+
"loss": 0.9257,
|
| 13253 |
+
"step": 1892
|
| 13254 |
+
},
|
| 13255 |
+
{
|
| 13256 |
+
"epoch": 0.36,
|
| 13257 |
+
"grad_norm": 1.6944012589743433,
|
| 13258 |
+
"learning_rate": 1.4683514346325671e-05,
|
| 13259 |
+
"loss": 0.8054,
|
| 13260 |
+
"step": 1893
|
| 13261 |
+
},
|
| 13262 |
+
{
|
| 13263 |
+
"epoch": 0.36,
|
| 13264 |
+
"grad_norm": 1.7534054491921223,
|
| 13265 |
+
"learning_rate": 1.4678002754987388e-05,
|
| 13266 |
+
"loss": 0.9034,
|
| 13267 |
+
"step": 1894
|
| 13268 |
+
},
|
| 13269 |
+
{
|
| 13270 |
+
"epoch": 0.36,
|
| 13271 |
+
"grad_norm": 1.658585836806205,
|
| 13272 |
+
"learning_rate": 1.4672489343879939e-05,
|
| 13273 |
+
"loss": 0.9457,
|
| 13274 |
+
"step": 1895
|
| 13275 |
+
},
|
| 13276 |
+
{
|
| 13277 |
+
"epoch": 0.37,
|
| 13278 |
+
"grad_norm": 1.640884368651283,
|
| 13279 |
+
"learning_rate": 1.4666974115148065e-05,
|
| 13280 |
+
"loss": 0.8148,
|
| 13281 |
+
"step": 1896
|
| 13282 |
+
},
|
| 13283 |
+
{
|
| 13284 |
+
"epoch": 0.37,
|
| 13285 |
+
"grad_norm": 1.5939722795602012,
|
| 13286 |
+
"learning_rate": 1.4661457070937226e-05,
|
| 13287 |
+
"loss": 0.8452,
|
| 13288 |
+
"step": 1897
|
| 13289 |
+
},
|
| 13290 |
+
{
|
| 13291 |
+
"epoch": 0.37,
|
| 13292 |
+
"grad_norm": 1.6167631945289578,
|
| 13293 |
+
"learning_rate": 1.4655938213393582e-05,
|
| 13294 |
+
"loss": 0.9033,
|
| 13295 |
+
"step": 1898
|
| 13296 |
+
},
|
| 13297 |
+
{
|
| 13298 |
+
"epoch": 0.37,
|
| 13299 |
+
"grad_norm": 1.9454139532441512,
|
| 13300 |
+
"learning_rate": 1.4650417544663997e-05,
|
| 13301 |
+
"loss": 0.9275,
|
| 13302 |
+
"step": 1899
|
| 13303 |
+
},
|
| 13304 |
+
{
|
| 13305 |
+
"epoch": 0.37,
|
| 13306 |
+
"grad_norm": 1.5858648706624496,
|
| 13307 |
+
"learning_rate": 1.4644895066896046e-05,
|
| 13308 |
+
"loss": 0.9054,
|
| 13309 |
+
"step": 1900
|
| 13310 |
+
},
|
| 13311 |
+
{
|
| 13312 |
+
"epoch": 0.37,
|
| 13313 |
+
"grad_norm": 1.6936034257279007,
|
| 13314 |
+
"learning_rate": 1.4639370782238e-05,
|
| 13315 |
+
"loss": 0.9825,
|
| 13316 |
+
"step": 1901
|
| 13317 |
+
},
|
| 13318 |
+
{
|
| 13319 |
+
"epoch": 0.37,
|
| 13320 |
+
"grad_norm": 1.6703890968967032,
|
| 13321 |
+
"learning_rate": 1.4633844692838837e-05,
|
| 13322 |
+
"loss": 0.9524,
|
| 13323 |
+
"step": 1902
|
| 13324 |
+
},
|
| 13325 |
+
{
|
| 13326 |
+
"epoch": 0.37,
|
| 13327 |
+
"grad_norm": 1.5551617347389055,
|
| 13328 |
+
"learning_rate": 1.4628316800848241e-05,
|
| 13329 |
+
"loss": 0.9375,
|
| 13330 |
+
"step": 1903
|
| 13331 |
+
},
|
| 13332 |
+
{
|
| 13333 |
+
"epoch": 0.37,
|
| 13334 |
+
"grad_norm": 1.8783960822922252,
|
| 13335 |
+
"learning_rate": 1.4622787108416585e-05,
|
| 13336 |
+
"loss": 1.0336,
|
| 13337 |
+
"step": 1904
|
| 13338 |
+
},
|
| 13339 |
+
{
|
| 13340 |
+
"epoch": 0.37,
|
| 13341 |
+
"grad_norm": 1.7771071816180781,
|
| 13342 |
+
"learning_rate": 1.4617255617694957e-05,
|
| 13343 |
+
"loss": 0.8875,
|
| 13344 |
+
"step": 1905
|
| 13345 |
+
},
|
| 13346 |
+
{
|
| 13347 |
+
"epoch": 0.37,
|
| 13348 |
+
"grad_norm": 1.6400552487600897,
|
| 13349 |
+
"learning_rate": 1.4611722330835133e-05,
|
| 13350 |
+
"loss": 0.9945,
|
| 13351 |
+
"step": 1906
|
| 13352 |
+
},
|
| 13353 |
+
{
|
| 13354 |
+
"epoch": 0.37,
|
| 13355 |
+
"grad_norm": 1.7424809111938788,
|
| 13356 |
+
"learning_rate": 1.4606187249989598e-05,
|
| 13357 |
+
"loss": 0.947,
|
| 13358 |
+
"step": 1907
|
| 13359 |
+
},
|
| 13360 |
+
{
|
| 13361 |
+
"epoch": 0.37,
|
| 13362 |
+
"grad_norm": 1.6042784586765482,
|
| 13363 |
+
"learning_rate": 1.4600650377311523e-05,
|
| 13364 |
+
"loss": 0.9005,
|
| 13365 |
+
"step": 1908
|
| 13366 |
+
},
|
| 13367 |
+
{
|
| 13368 |
+
"epoch": 0.37,
|
| 13369 |
+
"grad_norm": 1.6758776492466088,
|
| 13370 |
+
"learning_rate": 1.4595111714954787e-05,
|
| 13371 |
+
"loss": 0.9794,
|
| 13372 |
+
"step": 1909
|
| 13373 |
+
},
|
| 13374 |
+
{
|
| 13375 |
+
"epoch": 0.37,
|
| 13376 |
+
"grad_norm": 1.5913304752712094,
|
| 13377 |
+
"learning_rate": 1.4589571265073959e-05,
|
| 13378 |
+
"loss": 0.7905,
|
| 13379 |
+
"step": 1910
|
| 13380 |
+
},
|
| 13381 |
+
{
|
| 13382 |
+
"epoch": 0.37,
|
| 13383 |
+
"grad_norm": 1.7798933897483598,
|
| 13384 |
+
"learning_rate": 1.4584029029824305e-05,
|
| 13385 |
+
"loss": 0.9255,
|
| 13386 |
+
"step": 1911
|
| 13387 |
+
},
|
| 13388 |
+
{
|
| 13389 |
+
"epoch": 0.37,
|
| 13390 |
+
"grad_norm": 1.6437473812141559,
|
| 13391 |
+
"learning_rate": 1.4578485011361783e-05,
|
| 13392 |
+
"loss": 0.8905,
|
| 13393 |
+
"step": 1912
|
| 13394 |
+
},
|
| 13395 |
+
{
|
| 13396 |
+
"epoch": 0.37,
|
| 13397 |
+
"grad_norm": 1.701664671213589,
|
| 13398 |
+
"learning_rate": 1.4572939211843051e-05,
|
| 13399 |
+
"loss": 0.9686,
|
| 13400 |
+
"step": 1913
|
| 13401 |
+
},
|
| 13402 |
+
{
|
| 13403 |
+
"epoch": 0.37,
|
| 13404 |
+
"grad_norm": 1.611101839884725,
|
| 13405 |
+
"learning_rate": 1.4567391633425455e-05,
|
| 13406 |
+
"loss": 0.9226,
|
| 13407 |
+
"step": 1914
|
| 13408 |
+
},
|
| 13409 |
+
{
|
| 13410 |
+
"epoch": 0.37,
|
| 13411 |
+
"grad_norm": 1.4477358561753013,
|
| 13412 |
+
"learning_rate": 1.4561842278267033e-05,
|
| 13413 |
+
"loss": 0.956,
|
| 13414 |
+
"step": 1915
|
| 13415 |
+
},
|
| 13416 |
+
{
|
| 13417 |
+
"epoch": 0.37,
|
| 13418 |
+
"grad_norm": 1.5313991293553062,
|
| 13419 |
+
"learning_rate": 1.4556291148526516e-05,
|
| 13420 |
+
"loss": 0.8799,
|
| 13421 |
+
"step": 1916
|
| 13422 |
+
},
|
| 13423 |
+
{
|
| 13424 |
+
"epoch": 0.37,
|
| 13425 |
+
"grad_norm": 1.570437900284019,
|
| 13426 |
+
"learning_rate": 1.4550738246363322e-05,
|
| 13427 |
+
"loss": 0.8546,
|
| 13428 |
+
"step": 1917
|
| 13429 |
+
},
|
| 13430 |
+
{
|
| 13431 |
+
"epoch": 0.37,
|
| 13432 |
+
"grad_norm": 1.6393386291966954,
|
| 13433 |
+
"learning_rate": 1.4545183573937566e-05,
|
| 13434 |
+
"loss": 0.8923,
|
| 13435 |
+
"step": 1918
|
| 13436 |
+
},
|
| 13437 |
+
{
|
| 13438 |
+
"epoch": 0.37,
|
| 13439 |
+
"grad_norm": 1.7235853278804611,
|
| 13440 |
+
"learning_rate": 1.4539627133410042e-05,
|
| 13441 |
+
"loss": 0.8886,
|
| 13442 |
+
"step": 1919
|
| 13443 |
+
},
|
| 13444 |
+
{
|
| 13445 |
+
"epoch": 0.37,
|
| 13446 |
+
"grad_norm": 1.6947056773766112,
|
| 13447 |
+
"learning_rate": 1.4534068926942235e-05,
|
| 13448 |
+
"loss": 0.8453,
|
| 13449 |
+
"step": 1920
|
| 13450 |
+
},
|
| 13451 |
+
{
|
| 13452 |
+
"epoch": 0.37,
|
| 13453 |
+
"grad_norm": 1.5835750935087545,
|
| 13454 |
+
"learning_rate": 1.4528508956696326e-05,
|
| 13455 |
+
"loss": 0.9339,
|
| 13456 |
+
"step": 1921
|
| 13457 |
+
},
|
| 13458 |
+
{
|
| 13459 |
+
"epoch": 0.37,
|
| 13460 |
+
"grad_norm": 1.742088241485645,
|
| 13461 |
+
"learning_rate": 1.4522947224835165e-05,
|
| 13462 |
+
"loss": 0.9946,
|
| 13463 |
+
"step": 1922
|
| 13464 |
+
},
|
| 13465 |
+
{
|
| 13466 |
+
"epoch": 0.37,
|
| 13467 |
+
"grad_norm": 1.4949963929651728,
|
| 13468 |
+
"learning_rate": 1.4517383733522304e-05,
|
| 13469 |
+
"loss": 0.927,
|
| 13470 |
+
"step": 1923
|
| 13471 |
+
},
|
| 13472 |
+
{
|
| 13473 |
+
"epoch": 0.37,
|
| 13474 |
+
"grad_norm": 1.5253164901288647,
|
| 13475 |
+
"learning_rate": 1.451181848492197e-05,
|
| 13476 |
+
"loss": 0.8769,
|
| 13477 |
+
"step": 1924
|
| 13478 |
+
},
|
| 13479 |
+
{
|
| 13480 |
+
"epoch": 0.37,
|
| 13481 |
+
"grad_norm": 1.4889989022162844,
|
| 13482 |
+
"learning_rate": 1.4506251481199071e-05,
|
| 13483 |
+
"loss": 0.872,
|
| 13484 |
+
"step": 1925
|
| 13485 |
+
},
|
| 13486 |
+
{
|
| 13487 |
+
"epoch": 0.37,
|
| 13488 |
+
"grad_norm": 1.8096496107868196,
|
| 13489 |
+
"learning_rate": 1.450068272451921e-05,
|
| 13490 |
+
"loss": 0.9849,
|
| 13491 |
+
"step": 1926
|
| 13492 |
+
},
|
| 13493 |
+
{
|
| 13494 |
+
"epoch": 0.37,
|
| 13495 |
+
"grad_norm": 1.6165599569378286,
|
| 13496 |
+
"learning_rate": 1.449511221704866e-05,
|
| 13497 |
+
"loss": 0.8824,
|
| 13498 |
+
"step": 1927
|
| 13499 |
+
},
|
| 13500 |
+
{
|
| 13501 |
+
"epoch": 0.37,
|
| 13502 |
+
"grad_norm": 1.696888487069437,
|
| 13503 |
+
"learning_rate": 1.4489539960954382e-05,
|
| 13504 |
+
"loss": 0.9493,
|
| 13505 |
+
"step": 1928
|
| 13506 |
+
},
|
| 13507 |
+
{
|
| 13508 |
+
"epoch": 0.37,
|
| 13509 |
+
"grad_norm": 1.5617807092050424,
|
| 13510 |
+
"learning_rate": 1.4483965958404012e-05,
|
| 13511 |
+
"loss": 0.9603,
|
| 13512 |
+
"step": 1929
|
| 13513 |
+
},
|
| 13514 |
+
{
|
| 13515 |
+
"epoch": 0.37,
|
| 13516 |
+
"grad_norm": 1.7772163500967342,
|
| 13517 |
+
"learning_rate": 1.447839021156587e-05,
|
| 13518 |
+
"loss": 0.8918,
|
| 13519 |
+
"step": 1930
|
| 13520 |
+
},
|
| 13521 |
+
{
|
| 13522 |
+
"epoch": 0.37,
|
| 13523 |
+
"grad_norm": 1.060365842099134,
|
| 13524 |
+
"learning_rate": 1.4472812722608948e-05,
|
| 13525 |
+
"loss": 0.8678,
|
| 13526 |
+
"step": 1931
|
| 13527 |
+
},
|
| 13528 |
+
{
|
| 13529 |
+
"epoch": 0.37,
|
| 13530 |
+
"grad_norm": 1.6120849117340765,
|
| 13531 |
+
"learning_rate": 1.446723349370293e-05,
|
| 13532 |
+
"loss": 0.9337,
|
| 13533 |
+
"step": 1932
|
| 13534 |
+
},
|
| 13535 |
+
{
|
| 13536 |
+
"epoch": 0.37,
|
| 13537 |
+
"grad_norm": 1.0949366371155416,
|
| 13538 |
+
"learning_rate": 1.4461652527018158e-05,
|
| 13539 |
+
"loss": 0.8726,
|
| 13540 |
+
"step": 1933
|
| 13541 |
+
},
|
| 13542 |
+
{
|
| 13543 |
+
"epoch": 0.37,
|
| 13544 |
+
"grad_norm": 1.6705665592860641,
|
| 13545 |
+
"learning_rate": 1.4456069824725663e-05,
|
| 13546 |
+
"loss": 0.9106,
|
| 13547 |
+
"step": 1934
|
| 13548 |
+
},
|
| 13549 |
+
{
|
| 13550 |
+
"epoch": 0.37,
|
| 13551 |
+
"grad_norm": 1.842606010652722,
|
| 13552 |
+
"learning_rate": 1.4450485388997149e-05,
|
| 13553 |
+
"loss": 0.9954,
|
| 13554 |
+
"step": 1935
|
| 13555 |
+
},
|
| 13556 |
+
{
|
| 13557 |
+
"epoch": 0.37,
|
| 13558 |
+
"grad_norm": 1.057321791400095,
|
| 13559 |
+
"learning_rate": 1.444489922200499e-05,
|
| 13560 |
+
"loss": 0.8959,
|
| 13561 |
+
"step": 1936
|
| 13562 |
+
},
|
| 13563 |
+
{
|
| 13564 |
+
"epoch": 0.37,
|
| 13565 |
+
"grad_norm": 1.8140608348248906,
|
| 13566 |
+
"learning_rate": 1.4439311325922235e-05,
|
| 13567 |
+
"loss": 0.9943,
|
| 13568 |
+
"step": 1937
|
| 13569 |
+
},
|
| 13570 |
+
{
|
| 13571 |
+
"epoch": 0.37,
|
| 13572 |
+
"grad_norm": 1.7098410462471028,
|
| 13573 |
+
"learning_rate": 1.4433721702922607e-05,
|
| 13574 |
+
"loss": 0.8261,
|
| 13575 |
+
"step": 1938
|
| 13576 |
+
},
|
| 13577 |
+
{
|
| 13578 |
+
"epoch": 0.37,
|
| 13579 |
+
"grad_norm": 1.0270872514099714,
|
| 13580 |
+
"learning_rate": 1.4428130355180505e-05,
|
| 13581 |
+
"loss": 0.8597,
|
| 13582 |
+
"step": 1939
|
| 13583 |
+
},
|
| 13584 |
+
{
|
| 13585 |
+
"epoch": 0.37,
|
| 13586 |
+
"grad_norm": 1.596577887827628,
|
| 13587 |
+
"learning_rate": 1.4422537284870993e-05,
|
| 13588 |
+
"loss": 0.8603,
|
| 13589 |
+
"step": 1940
|
| 13590 |
+
},
|
| 13591 |
+
{
|
| 13592 |
+
"epoch": 0.37,
|
| 13593 |
+
"grad_norm": 1.0437894211858543,
|
| 13594 |
+
"learning_rate": 1.4416942494169801e-05,
|
| 13595 |
+
"loss": 0.8863,
|
| 13596 |
+
"step": 1941
|
| 13597 |
+
},
|
| 13598 |
+
{
|
| 13599 |
+
"epoch": 0.37,
|
| 13600 |
+
"grad_norm": 1.022257618024313,
|
| 13601 |
+
"learning_rate": 1.441134598525334e-05,
|
| 13602 |
+
"loss": 0.8304,
|
| 13603 |
+
"step": 1942
|
| 13604 |
+
},
|
| 13605 |
+
{
|
| 13606 |
+
"epoch": 0.37,
|
| 13607 |
+
"grad_norm": 1.630469813224106,
|
| 13608 |
+
"learning_rate": 1.4405747760298681e-05,
|
| 13609 |
+
"loss": 0.8952,
|
| 13610 |
+
"step": 1943
|
| 13611 |
+
},
|
| 13612 |
+
{
|
| 13613 |
+
"epoch": 0.37,
|
| 13614 |
+
"grad_norm": 1.7663508160729156,
|
| 13615 |
+
"learning_rate": 1.4400147821483564e-05,
|
| 13616 |
+
"loss": 0.9825,
|
| 13617 |
+
"step": 1944
|
| 13618 |
+
},
|
| 13619 |
+
{
|
| 13620 |
+
"epoch": 0.37,
|
| 13621 |
+
"grad_norm": 1.6935303282233554,
|
| 13622 |
+
"learning_rate": 1.4394546170986394e-05,
|
| 13623 |
+
"loss": 0.9867,
|
| 13624 |
+
"step": 1945
|
| 13625 |
+
},
|
| 13626 |
+
{
|
| 13627 |
+
"epoch": 0.37,
|
| 13628 |
+
"grad_norm": 1.811332894249063,
|
| 13629 |
+
"learning_rate": 1.4388942810986253e-05,
|
| 13630 |
+
"loss": 0.9215,
|
| 13631 |
+
"step": 1946
|
| 13632 |
+
},
|
| 13633 |
+
{
|
| 13634 |
+
"epoch": 0.37,
|
| 13635 |
+
"grad_norm": 1.5936262053219343,
|
| 13636 |
+
"learning_rate": 1.4383337743662872e-05,
|
| 13637 |
+
"loss": 1.0064,
|
| 13638 |
+
"step": 1947
|
| 13639 |
+
},
|
| 13640 |
+
{
|
| 13641 |
+
"epoch": 0.38,
|
| 13642 |
+
"grad_norm": 1.2221389486212109,
|
| 13643 |
+
"learning_rate": 1.4377730971196658e-05,
|
| 13644 |
+
"loss": 0.8219,
|
| 13645 |
+
"step": 1948
|
| 13646 |
+
},
|
| 13647 |
+
{
|
| 13648 |
+
"epoch": 0.38,
|
| 13649 |
+
"grad_norm": 1.692758782000757,
|
| 13650 |
+
"learning_rate": 1.437212249576867e-05,
|
| 13651 |
+
"loss": 0.8998,
|
| 13652 |
+
"step": 1949
|
| 13653 |
+
},
|
| 13654 |
+
{
|
| 13655 |
+
"epoch": 0.38,
|
| 13656 |
+
"grad_norm": 1.6050066128005005,
|
| 13657 |
+
"learning_rate": 1.4366512319560642e-05,
|
| 13658 |
+
"loss": 0.9318,
|
| 13659 |
+
"step": 1950
|
| 13660 |
+
},
|
| 13661 |
+
{
|
| 13662 |
+
"epoch": 0.38,
|
| 13663 |
+
"grad_norm": 1.820927381354136,
|
| 13664 |
+
"learning_rate": 1.4360900444754959e-05,
|
| 13665 |
+
"loss": 0.9633,
|
| 13666 |
+
"step": 1951
|
| 13667 |
+
},
|
| 13668 |
+
{
|
| 13669 |
+
"epoch": 0.38,
|
| 13670 |
+
"grad_norm": 1.759938786543528,
|
| 13671 |
+
"learning_rate": 1.4355286873534677e-05,
|
| 13672 |
+
"loss": 0.9468,
|
| 13673 |
+
"step": 1952
|
| 13674 |
+
},
|
| 13675 |
+
{
|
| 13676 |
+
"epoch": 0.38,
|
| 13677 |
+
"grad_norm": 1.6752343970748882,
|
| 13678 |
+
"learning_rate": 1.4349671608083504e-05,
|
| 13679 |
+
"loss": 0.8669,
|
| 13680 |
+
"step": 1953
|
| 13681 |
+
},
|
| 13682 |
+
{
|
| 13683 |
+
"epoch": 0.38,
|
| 13684 |
+
"grad_norm": 1.7184980534359486,
|
| 13685 |
+
"learning_rate": 1.4344054650585813e-05,
|
| 13686 |
+
"loss": 0.9065,
|
| 13687 |
+
"step": 1954
|
| 13688 |
+
},
|
| 13689 |
+
{
|
| 13690 |
+
"epoch": 0.38,
|
| 13691 |
+
"grad_norm": 1.672450031095198,
|
| 13692 |
+
"learning_rate": 1.4338436003226623e-05,
|
| 13693 |
+
"loss": 1.0158,
|
| 13694 |
+
"step": 1955
|
| 13695 |
+
},
|
| 13696 |
+
{
|
| 13697 |
+
"epoch": 0.38,
|
| 13698 |
+
"grad_norm": 1.7822544858188532,
|
| 13699 |
+
"learning_rate": 1.433281566819163e-05,
|
| 13700 |
+
"loss": 0.8884,
|
| 13701 |
+
"step": 1956
|
| 13702 |
+
},
|
| 13703 |
+
{
|
| 13704 |
+
"epoch": 0.38,
|
| 13705 |
+
"grad_norm": 1.2323855892676456,
|
| 13706 |
+
"learning_rate": 1.4327193647667167e-05,
|
| 13707 |
+
"loss": 0.857,
|
| 13708 |
+
"step": 1957
|
| 13709 |
+
},
|
| 13710 |
+
{
|
| 13711 |
+
"epoch": 0.38,
|
| 13712 |
+
"grad_norm": 1.6749892119624912,
|
| 13713 |
+
"learning_rate": 1.4321569943840235e-05,
|
| 13714 |
+
"loss": 0.9273,
|
| 13715 |
+
"step": 1958
|
| 13716 |
+
},
|
| 13717 |
+
{
|
| 13718 |
+
"epoch": 0.38,
|
| 13719 |
+
"grad_norm": 1.562532316981109,
|
| 13720 |
+
"learning_rate": 1.4315944558898486e-05,
|
| 13721 |
+
"loss": 0.9543,
|
| 13722 |
+
"step": 1959
|
| 13723 |
+
},
|
| 13724 |
+
{
|
| 13725 |
+
"epoch": 0.38,
|
| 13726 |
+
"grad_norm": 1.733048227533296,
|
| 13727 |
+
"learning_rate": 1.4310317495030226e-05,
|
| 13728 |
+
"loss": 0.9416,
|
| 13729 |
+
"step": 1960
|
| 13730 |
+
},
|
| 13731 |
+
{
|
| 13732 |
+
"epoch": 0.38,
|
| 13733 |
+
"grad_norm": 1.6187249948968137,
|
| 13734 |
+
"learning_rate": 1.4304688754424417e-05,
|
| 13735 |
+
"loss": 0.9877,
|
| 13736 |
+
"step": 1961
|
| 13737 |
+
},
|
| 13738 |
+
{
|
| 13739 |
+
"epoch": 0.38,
|
| 13740 |
+
"grad_norm": 1.544184212908568,
|
| 13741 |
+
"learning_rate": 1.4299058339270663e-05,
|
| 13742 |
+
"loss": 0.8356,
|
| 13743 |
+
"step": 1962
|
| 13744 |
+
},
|
| 13745 |
+
{
|
| 13746 |
+
"epoch": 0.38,
|
| 13747 |
+
"grad_norm": 1.5597864305215874,
|
| 13748 |
+
"learning_rate": 1.4293426251759234e-05,
|
| 13749 |
+
"loss": 0.8891,
|
| 13750 |
+
"step": 1963
|
| 13751 |
+
},
|
| 13752 |
+
{
|
| 13753 |
+
"epoch": 0.38,
|
| 13754 |
+
"grad_norm": 1.6764875535211585,
|
| 13755 |
+
"learning_rate": 1.4287792494081042e-05,
|
| 13756 |
+
"loss": 0.9978,
|
| 13757 |
+
"step": 1964
|
| 13758 |
+
},
|
| 13759 |
+
{
|
| 13760 |
+
"epoch": 0.38,
|
| 13761 |
+
"grad_norm": 1.628972353736311,
|
| 13762 |
+
"learning_rate": 1.428215706842765e-05,
|
| 13763 |
+
"loss": 0.8752,
|
| 13764 |
+
"step": 1965
|
| 13765 |
+
},
|
| 13766 |
+
{
|
| 13767 |
+
"epoch": 0.38,
|
| 13768 |
+
"grad_norm": 1.6175343059783494,
|
| 13769 |
+
"learning_rate": 1.4276519976991267e-05,
|
| 13770 |
+
"loss": 0.9888,
|
| 13771 |
+
"step": 1966
|
| 13772 |
+
},
|
| 13773 |
+
{
|
| 13774 |
+
"epoch": 0.38,
|
| 13775 |
+
"grad_norm": 1.6569498876168287,
|
| 13776 |
+
"learning_rate": 1.4270881221964756e-05,
|
| 13777 |
+
"loss": 0.9243,
|
| 13778 |
+
"step": 1967
|
| 13779 |
+
},
|
| 13780 |
+
{
|
| 13781 |
+
"epoch": 0.38,
|
| 13782 |
+
"grad_norm": 1.5486333423207015,
|
| 13783 |
+
"learning_rate": 1.4265240805541628e-05,
|
| 13784 |
+
"loss": 0.875,
|
| 13785 |
+
"step": 1968
|
| 13786 |
+
},
|
| 13787 |
+
{
|
| 13788 |
+
"epoch": 0.38,
|
| 13789 |
+
"grad_norm": 1.6050782117840383,
|
| 13790 |
+
"learning_rate": 1.4259598729916027e-05,
|
| 13791 |
+
"loss": 0.8632,
|
| 13792 |
+
"step": 1969
|
| 13793 |
+
},
|
| 13794 |
+
{
|
| 13795 |
+
"epoch": 0.38,
|
| 13796 |
+
"grad_norm": 1.613167697354811,
|
| 13797 |
+
"learning_rate": 1.425395499728276e-05,
|
| 13798 |
+
"loss": 0.8139,
|
| 13799 |
+
"step": 1970
|
| 13800 |
+
},
|
| 13801 |
+
{
|
| 13802 |
+
"epoch": 0.38,
|
| 13803 |
+
"grad_norm": 1.0995807826218689,
|
| 13804 |
+
"learning_rate": 1.4248309609837262e-05,
|
| 13805 |
+
"loss": 0.8318,
|
| 13806 |
+
"step": 1971
|
| 13807 |
+
},
|
| 13808 |
+
{
|
| 13809 |
+
"epoch": 0.38,
|
| 13810 |
+
"grad_norm": 1.7339742844852268,
|
| 13811 |
+
"learning_rate": 1.4242662569775632e-05,
|
| 13812 |
+
"loss": 1.0169,
|
| 13813 |
+
"step": 1972
|
| 13814 |
+
},
|
| 13815 |
+
{
|
| 13816 |
+
"epoch": 0.38,
|
| 13817 |
+
"grad_norm": 1.775570046531735,
|
| 13818 |
+
"learning_rate": 1.423701387929459e-05,
|
| 13819 |
+
"loss": 0.8689,
|
| 13820 |
+
"step": 1973
|
| 13821 |
+
},
|
| 13822 |
+
{
|
| 13823 |
+
"epoch": 0.38,
|
| 13824 |
+
"grad_norm": 1.712780324624818,
|
| 13825 |
+
"learning_rate": 1.4231363540591512e-05,
|
| 13826 |
+
"loss": 0.8534,
|
| 13827 |
+
"step": 1974
|
| 13828 |
+
},
|
| 13829 |
+
{
|
| 13830 |
+
"epoch": 0.38,
|
| 13831 |
+
"grad_norm": 1.9214141386062493,
|
| 13832 |
+
"learning_rate": 1.4225711555864413e-05,
|
| 13833 |
+
"loss": 0.9083,
|
| 13834 |
+
"step": 1975
|
| 13835 |
+
},
|
| 13836 |
+
{
|
| 13837 |
+
"epoch": 0.38,
|
| 13838 |
+
"grad_norm": 1.0979553405135083,
|
| 13839 |
+
"learning_rate": 1.4220057927311944e-05,
|
| 13840 |
+
"loss": 0.8881,
|
| 13841 |
+
"step": 1976
|
| 13842 |
+
},
|
| 13843 |
+
{
|
| 13844 |
+
"epoch": 0.38,
|
| 13845 |
+
"grad_norm": 1.6731728263110943,
|
| 13846 |
+
"learning_rate": 1.4214402657133396e-05,
|
| 13847 |
+
"loss": 0.968,
|
| 13848 |
+
"step": 1977
|
| 13849 |
+
},
|
| 13850 |
+
{
|
| 13851 |
+
"epoch": 0.38,
|
| 13852 |
+
"grad_norm": 1.5442188820300313,
|
| 13853 |
+
"learning_rate": 1.4208745747528705e-05,
|
| 13854 |
+
"loss": 0.8429,
|
| 13855 |
+
"step": 1978
|
| 13856 |
+
},
|
| 13857 |
+
{
|
| 13858 |
+
"epoch": 0.38,
|
| 13859 |
+
"grad_norm": 1.560545156571736,
|
| 13860 |
+
"learning_rate": 1.420308720069844e-05,
|
| 13861 |
+
"loss": 0.8704,
|
| 13862 |
+
"step": 1979
|
| 13863 |
+
},
|
| 13864 |
+
{
|
| 13865 |
+
"epoch": 0.38,
|
| 13866 |
+
"grad_norm": 1.5465513905600146,
|
| 13867 |
+
"learning_rate": 1.4197427018843804e-05,
|
| 13868 |
+
"loss": 0.8807,
|
| 13869 |
+
"step": 1980
|
| 13870 |
+
},
|
| 13871 |
+
{
|
| 13872 |
+
"epoch": 0.38,
|
| 13873 |
+
"grad_norm": 1.9032763143445839,
|
| 13874 |
+
"learning_rate": 1.4191765204166643e-05,
|
| 13875 |
+
"loss": 0.8731,
|
| 13876 |
+
"step": 1981
|
| 13877 |
+
},
|
| 13878 |
+
{
|
| 13879 |
+
"epoch": 0.38,
|
| 13880 |
+
"grad_norm": 1.5074974422961198,
|
| 13881 |
+
"learning_rate": 1.418610175886943e-05,
|
| 13882 |
+
"loss": 0.8785,
|
| 13883 |
+
"step": 1982
|
| 13884 |
+
},
|
| 13885 |
+
{
|
| 13886 |
+
"epoch": 0.38,
|
| 13887 |
+
"grad_norm": 1.7085486460822743,
|
| 13888 |
+
"learning_rate": 1.4180436685155284e-05,
|
| 13889 |
+
"loss": 0.9437,
|
| 13890 |
+
"step": 1983
|
| 13891 |
+
},
|
| 13892 |
+
{
|
| 13893 |
+
"epoch": 0.38,
|
| 13894 |
+
"grad_norm": 1.0614399683461917,
|
| 13895 |
+
"learning_rate": 1.4174769985227943e-05,
|
| 13896 |
+
"loss": 0.869,
|
| 13897 |
+
"step": 1984
|
| 13898 |
+
},
|
| 13899 |
+
{
|
| 13900 |
+
"epoch": 0.38,
|
| 13901 |
+
"grad_norm": 1.8102639056663725,
|
| 13902 |
+
"learning_rate": 1.4169101661291789e-05,
|
| 13903 |
+
"loss": 0.9407,
|
| 13904 |
+
"step": 1985
|
| 13905 |
+
},
|
| 13906 |
+
{
|
| 13907 |
+
"epoch": 0.38,
|
| 13908 |
+
"grad_norm": 1.677696426013255,
|
| 13909 |
+
"learning_rate": 1.416343171555183e-05,
|
| 13910 |
+
"loss": 0.9164,
|
| 13911 |
+
"step": 1986
|
| 13912 |
+
},
|
| 13913 |
+
{
|
| 13914 |
+
"epoch": 0.38,
|
| 13915 |
+
"grad_norm": 1.5983097724232058,
|
| 13916 |
+
"learning_rate": 1.4157760150213706e-05,
|
| 13917 |
+
"loss": 0.8976,
|
| 13918 |
+
"step": 1987
|
| 13919 |
+
},
|
| 13920 |
+
{
|
| 13921 |
+
"epoch": 0.38,
|
| 13922 |
+
"grad_norm": 1.5633357942983659,
|
| 13923 |
+
"learning_rate": 1.415208696748369e-05,
|
| 13924 |
+
"loss": 0.9583,
|
| 13925 |
+
"step": 1988
|
| 13926 |
+
},
|
| 13927 |
+
{
|
| 13928 |
+
"epoch": 0.38,
|
| 13929 |
+
"grad_norm": 1.8757667991615432,
|
| 13930 |
+
"learning_rate": 1.414641216956868e-05,
|
| 13931 |
+
"loss": 0.8841,
|
| 13932 |
+
"step": 1989
|
| 13933 |
+
},
|
| 13934 |
+
{
|
| 13935 |
+
"epoch": 0.38,
|
| 13936 |
+
"grad_norm": 1.7104011632366787,
|
| 13937 |
+
"learning_rate": 1.4140735758676205e-05,
|
| 13938 |
+
"loss": 0.9663,
|
| 13939 |
+
"step": 1990
|
| 13940 |
+
},
|
| 13941 |
+
{
|
| 13942 |
+
"epoch": 0.38,
|
| 13943 |
+
"grad_norm": 1.6278015170837083,
|
| 13944 |
+
"learning_rate": 1.4135057737014416e-05,
|
| 13945 |
+
"loss": 0.9017,
|
| 13946 |
+
"step": 1991
|
| 13947 |
+
},
|
| 13948 |
+
{
|
| 13949 |
+
"epoch": 0.38,
|
| 13950 |
+
"grad_norm": 1.6085948855074665,
|
| 13951 |
+
"learning_rate": 1.4129378106792103e-05,
|
| 13952 |
+
"loss": 0.9748,
|
| 13953 |
+
"step": 1992
|
| 13954 |
+
},
|
| 13955 |
+
{
|
| 13956 |
+
"epoch": 0.38,
|
| 13957 |
+
"grad_norm": 1.7212729021378976,
|
| 13958 |
+
"learning_rate": 1.412369687021867e-05,
|
| 13959 |
+
"loss": 0.8887,
|
| 13960 |
+
"step": 1993
|
| 13961 |
+
},
|
| 13962 |
+
{
|
| 13963 |
+
"epoch": 0.38,
|
| 13964 |
+
"grad_norm": 1.4847775674762957,
|
| 13965 |
+
"learning_rate": 1.411801402950415e-05,
|
| 13966 |
+
"loss": 0.8589,
|
| 13967 |
+
"step": 1994
|
| 13968 |
+
},
|
| 13969 |
+
{
|
| 13970 |
+
"epoch": 0.38,
|
| 13971 |
+
"grad_norm": 1.7254985997235885,
|
| 13972 |
+
"learning_rate": 1.4112329586859196e-05,
|
| 13973 |
+
"loss": 0.9851,
|
| 13974 |
+
"step": 1995
|
| 13975 |
+
},
|
| 13976 |
+
{
|
| 13977 |
+
"epoch": 0.38,
|
| 13978 |
+
"grad_norm": 1.6497964331392245,
|
| 13979 |
+
"learning_rate": 1.4106643544495092e-05,
|
| 13980 |
+
"loss": 0.9671,
|
| 13981 |
+
"step": 1996
|
| 13982 |
+
},
|
| 13983 |
+
{
|
| 13984 |
+
"epoch": 0.38,
|
| 13985 |
+
"grad_norm": 1.7500475617926596,
|
| 13986 |
+
"learning_rate": 1.4100955904623742e-05,
|
| 13987 |
+
"loss": 0.9816,
|
| 13988 |
+
"step": 1997
|
| 13989 |
+
},
|
| 13990 |
+
{
|
| 13991 |
+
"epoch": 0.38,
|
| 13992 |
+
"grad_norm": 1.6055781534173013,
|
| 13993 |
+
"learning_rate": 1.4095266669457664e-05,
|
| 13994 |
+
"loss": 0.7968,
|
| 13995 |
+
"step": 1998
|
| 13996 |
+
},
|
| 13997 |
+
{
|
| 13998 |
+
"epoch": 0.38,
|
| 13999 |
+
"grad_norm": 1.7201656182922893,
|
| 14000 |
+
"learning_rate": 1.4089575841210004e-05,
|
| 14001 |
+
"loss": 0.9469,
|
| 14002 |
+
"step": 1999
|
| 14003 |
+
},
|
| 14004 |
+
{
|
| 14005 |
+
"epoch": 0.39,
|
| 14006 |
+
"grad_norm": 1.7198154929517466,
|
| 14007 |
+
"learning_rate": 1.4083883422094528e-05,
|
| 14008 |
+
"loss": 0.9505,
|
| 14009 |
+
"step": 2000
|
| 14010 |
}
|
| 14011 |
],
|
| 14012 |
"logging_steps": 1.0,
|
|
|
|
| 14014 |
"num_input_tokens_seen": 0,
|
| 14015 |
"num_train_epochs": 1,
|
| 14016 |
"save_steps": 500,
|
| 14017 |
+
"total_flos": 2671624441298944.0,
|
| 14018 |
"train_batch_size": 4,
|
| 14019 |
"trial_name": null,
|
| 14020 |
"trial_params": null
|