Training in progress, step 1912, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2503003904
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e893d538403d4ac222e2baaf746a33535ee8031c07cf1939cc3355ea15106a0
|
| 3 |
size 2503003904
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5006244836
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbf245997dbc83cd89bcfb5067dfa742724b5f13ff1993cd0ad6d3d60a4c987a
|
| 3 |
size 5006244836
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12f09fa1a152c2febaa1b0be3c98d7abd70a22c5965d994af5b7173cc3e6ff7f
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c767cedc54b733779ba8a20f635d848598fd89e5cfee0706f6c63df8c1e6b2d8
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -11718,6 +11718,1679 @@
|
|
| 11718 |
"learning_rate": 5.6277671951738716e-06,
|
| 11719 |
"loss": 0.7867,
|
| 11720 |
"step": 1673
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11721 |
}
|
| 11722 |
],
|
| 11723 |
"logging_steps": 1,
|
|
@@ -11737,7 +13410,7 @@
|
|
| 11737 |
"attributes": {}
|
| 11738 |
}
|
| 11739 |
},
|
| 11740 |
-
"total_flos":
|
| 11741 |
"train_batch_size": 8,
|
| 11742 |
"trial_name": null,
|
| 11743 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9775051124744376,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1912,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 11718 |
"learning_rate": 5.6277671951738716e-06,
|
| 11719 |
"loss": 0.7867,
|
| 11720 |
"step": 1673
|
| 11721 |
+
},
|
| 11722 |
+
{
|
| 11723 |
+
"epoch": 0.8558282208588958,
|
| 11724 |
+
"grad_norm": 3.041142225265503,
|
| 11725 |
+
"learning_rate": 5.588822025878476e-06,
|
| 11726 |
+
"loss": 0.7922,
|
| 11727 |
+
"step": 1674
|
| 11728 |
+
},
|
| 11729 |
+
{
|
| 11730 |
+
"epoch": 0.8563394683026585,
|
| 11731 |
+
"grad_norm": 3.526838779449463,
|
| 11732 |
+
"learning_rate": 5.550004100195639e-06,
|
| 11733 |
+
"loss": 0.9025,
|
| 11734 |
+
"step": 1675
|
| 11735 |
+
},
|
| 11736 |
+
{
|
| 11737 |
+
"epoch": 0.8568507157464212,
|
| 11738 |
+
"grad_norm": 3.6863765716552734,
|
| 11739 |
+
"learning_rate": 5.5113135293435815e-06,
|
| 11740 |
+
"loss": 0.9744,
|
| 11741 |
+
"step": 1676
|
| 11742 |
+
},
|
| 11743 |
+
{
|
| 11744 |
+
"epoch": 0.8573619631901841,
|
| 11745 |
+
"grad_norm": 3.2295031547546387,
|
| 11746 |
+
"learning_rate": 5.4727504241756874e-06,
|
| 11747 |
+
"loss": 0.8475,
|
| 11748 |
+
"step": 1677
|
| 11749 |
+
},
|
| 11750 |
+
{
|
| 11751 |
+
"epoch": 0.8578732106339468,
|
| 11752 |
+
"grad_norm": 3.341581344604492,
|
| 11753 |
+
"learning_rate": 5.434314895180082e-06,
|
| 11754 |
+
"loss": 0.8515,
|
| 11755 |
+
"step": 1678
|
| 11756 |
+
},
|
| 11757 |
+
{
|
| 11758 |
+
"epoch": 0.8583844580777096,
|
| 11759 |
+
"grad_norm": 3.328876495361328,
|
| 11760 |
+
"learning_rate": 5.396007052479407e-06,
|
| 11761 |
+
"loss": 0.8321,
|
| 11762 |
+
"step": 1679
|
| 11763 |
+
},
|
| 11764 |
+
{
|
| 11765 |
+
"epoch": 0.8588957055214724,
|
| 11766 |
+
"grad_norm": 3.497668743133545,
|
| 11767 |
+
"learning_rate": 5.357827005830435e-06,
|
| 11768 |
+
"loss": 0.8929,
|
| 11769 |
+
"step": 1680
|
| 11770 |
+
},
|
| 11771 |
+
{
|
| 11772 |
+
"epoch": 0.8594069529652352,
|
| 11773 |
+
"grad_norm": 3.690748691558838,
|
| 11774 |
+
"learning_rate": 5.319774864623834e-06,
|
| 11775 |
+
"loss": 0.8603,
|
| 11776 |
+
"step": 1681
|
| 11777 |
+
},
|
| 11778 |
+
{
|
| 11779 |
+
"epoch": 0.8599182004089979,
|
| 11780 |
+
"grad_norm": 3.709012508392334,
|
| 11781 |
+
"learning_rate": 5.281850737883731e-06,
|
| 11782 |
+
"loss": 0.9677,
|
| 11783 |
+
"step": 1682
|
| 11784 |
+
},
|
| 11785 |
+
{
|
| 11786 |
+
"epoch": 0.8604294478527608,
|
| 11787 |
+
"grad_norm": 3.8703484535217285,
|
| 11788 |
+
"learning_rate": 5.2440547342675614e-06,
|
| 11789 |
+
"loss": 0.8876,
|
| 11790 |
+
"step": 1683
|
| 11791 |
+
},
|
| 11792 |
+
{
|
| 11793 |
+
"epoch": 0.8609406952965235,
|
| 11794 |
+
"grad_norm": 3.9959239959716797,
|
| 11795 |
+
"learning_rate": 5.206386962065602e-06,
|
| 11796 |
+
"loss": 0.791,
|
| 11797 |
+
"step": 1684
|
| 11798 |
+
},
|
| 11799 |
+
{
|
| 11800 |
+
"epoch": 0.8614519427402862,
|
| 11801 |
+
"grad_norm": 3.8929603099823,
|
| 11802 |
+
"learning_rate": 5.168847529200782e-06,
|
| 11803 |
+
"loss": 0.8654,
|
| 11804 |
+
"step": 1685
|
| 11805 |
+
},
|
| 11806 |
+
{
|
| 11807 |
+
"epoch": 0.8619631901840491,
|
| 11808 |
+
"grad_norm": 4.002877235412598,
|
| 11809 |
+
"learning_rate": 5.1314365432282904e-06,
|
| 11810 |
+
"loss": 0.8284,
|
| 11811 |
+
"step": 1686
|
| 11812 |
+
},
|
| 11813 |
+
{
|
| 11814 |
+
"epoch": 0.8624744376278118,
|
| 11815 |
+
"grad_norm": 4.066626071929932,
|
| 11816 |
+
"learning_rate": 5.094154111335292e-06,
|
| 11817 |
+
"loss": 0.8603,
|
| 11818 |
+
"step": 1687
|
| 11819 |
+
},
|
| 11820 |
+
{
|
| 11821 |
+
"epoch": 0.8629856850715747,
|
| 11822 |
+
"grad_norm": 4.232320785522461,
|
| 11823 |
+
"learning_rate": 5.057000340340678e-06,
|
| 11824 |
+
"loss": 0.9426,
|
| 11825 |
+
"step": 1688
|
| 11826 |
+
},
|
| 11827 |
+
{
|
| 11828 |
+
"epoch": 0.8634969325153374,
|
| 11829 |
+
"grad_norm": 3.965895175933838,
|
| 11830 |
+
"learning_rate": 5.019975336694649e-06,
|
| 11831 |
+
"loss": 0.7798,
|
| 11832 |
+
"step": 1689
|
| 11833 |
+
},
|
| 11834 |
+
{
|
| 11835 |
+
"epoch": 0.8640081799591002,
|
| 11836 |
+
"grad_norm": 4.249181270599365,
|
| 11837 |
+
"learning_rate": 4.983079206478513e-06,
|
| 11838 |
+
"loss": 0.8211,
|
| 11839 |
+
"step": 1690
|
| 11840 |
+
},
|
| 11841 |
+
{
|
| 11842 |
+
"epoch": 0.864519427402863,
|
| 11843 |
+
"grad_norm": 4.230957508087158,
|
| 11844 |
+
"learning_rate": 4.946312055404328e-06,
|
| 11845 |
+
"loss": 0.9142,
|
| 11846 |
+
"step": 1691
|
| 11847 |
+
},
|
| 11848 |
+
{
|
| 11849 |
+
"epoch": 0.8650306748466258,
|
| 11850 |
+
"grad_norm": 4.479368209838867,
|
| 11851 |
+
"learning_rate": 4.909673988814601e-06,
|
| 11852 |
+
"loss": 0.9803,
|
| 11853 |
+
"step": 1692
|
| 11854 |
+
},
|
| 11855 |
+
{
|
| 11856 |
+
"epoch": 0.8655419222903885,
|
| 11857 |
+
"grad_norm": 4.021700859069824,
|
| 11858 |
+
"learning_rate": 4.873165111681993e-06,
|
| 11859 |
+
"loss": 0.7705,
|
| 11860 |
+
"step": 1693
|
| 11861 |
+
},
|
| 11862 |
+
{
|
| 11863 |
+
"epoch": 0.8660531697341514,
|
| 11864 |
+
"grad_norm": 4.131351470947266,
|
| 11865 |
+
"learning_rate": 4.836785528609051e-06,
|
| 11866 |
+
"loss": 0.7206,
|
| 11867 |
+
"step": 1694
|
| 11868 |
+
},
|
| 11869 |
+
{
|
| 11870 |
+
"epoch": 0.8665644171779141,
|
| 11871 |
+
"grad_norm": 4.708834171295166,
|
| 11872 |
+
"learning_rate": 4.800535343827833e-06,
|
| 11873 |
+
"loss": 0.8698,
|
| 11874 |
+
"step": 1695
|
| 11875 |
+
},
|
| 11876 |
+
{
|
| 11877 |
+
"epoch": 0.8670756646216768,
|
| 11878 |
+
"grad_norm": 4.927844047546387,
|
| 11879 |
+
"learning_rate": 4.764414661199707e-06,
|
| 11880 |
+
"loss": 0.7615,
|
| 11881 |
+
"step": 1696
|
| 11882 |
+
},
|
| 11883 |
+
{
|
| 11884 |
+
"epoch": 0.8675869120654397,
|
| 11885 |
+
"grad_norm": 5.510052680969238,
|
| 11886 |
+
"learning_rate": 4.728423584214947e-06,
|
| 11887 |
+
"loss": 0.9676,
|
| 11888 |
+
"step": 1697
|
| 11889 |
+
},
|
| 11890 |
+
{
|
| 11891 |
+
"epoch": 0.8680981595092024,
|
| 11892 |
+
"grad_norm": 5.557268142700195,
|
| 11893 |
+
"learning_rate": 4.692562215992541e-06,
|
| 11894 |
+
"loss": 0.6631,
|
| 11895 |
+
"step": 1698
|
| 11896 |
+
},
|
| 11897 |
+
{
|
| 11898 |
+
"epoch": 0.8686094069529653,
|
| 11899 |
+
"grad_norm": 6.219789028167725,
|
| 11900 |
+
"learning_rate": 4.656830659279804e-06,
|
| 11901 |
+
"loss": 0.8028,
|
| 11902 |
+
"step": 1699
|
| 11903 |
+
},
|
| 11904 |
+
{
|
| 11905 |
+
"epoch": 0.869120654396728,
|
| 11906 |
+
"grad_norm": 5.629024028778076,
|
| 11907 |
+
"learning_rate": 4.621229016452156e-06,
|
| 11908 |
+
"loss": 0.4033,
|
| 11909 |
+
"step": 1700
|
| 11910 |
+
},
|
| 11911 |
+
{
|
| 11912 |
+
"epoch": 0.8696319018404908,
|
| 11913 |
+
"grad_norm": 1.9563968181610107,
|
| 11914 |
+
"learning_rate": 4.585757389512768e-06,
|
| 11915 |
+
"loss": 0.8627,
|
| 11916 |
+
"step": 1701
|
| 11917 |
+
},
|
| 11918 |
+
{
|
| 11919 |
+
"epoch": 0.8701431492842536,
|
| 11920 |
+
"grad_norm": 2.113006591796875,
|
| 11921 |
+
"learning_rate": 4.550415880092313e-06,
|
| 11922 |
+
"loss": 0.8872,
|
| 11923 |
+
"step": 1702
|
| 11924 |
+
},
|
| 11925 |
+
{
|
| 11926 |
+
"epoch": 0.8706543967280164,
|
| 11927 |
+
"grad_norm": 2.489034652709961,
|
| 11928 |
+
"learning_rate": 4.515204589448674e-06,
|
| 11929 |
+
"loss": 0.9124,
|
| 11930 |
+
"step": 1703
|
| 11931 |
+
},
|
| 11932 |
+
{
|
| 11933 |
+
"epoch": 0.8711656441717791,
|
| 11934 |
+
"grad_norm": 2.6170196533203125,
|
| 11935 |
+
"learning_rate": 4.48012361846662e-06,
|
| 11936 |
+
"loss": 1.0461,
|
| 11937 |
+
"step": 1704
|
| 11938 |
+
},
|
| 11939 |
+
{
|
| 11940 |
+
"epoch": 0.871676891615542,
|
| 11941 |
+
"grad_norm": 2.774226188659668,
|
| 11942 |
+
"learning_rate": 4.445173067657554e-06,
|
| 11943 |
+
"loss": 1.0186,
|
| 11944 |
+
"step": 1705
|
| 11945 |
+
},
|
| 11946 |
+
{
|
| 11947 |
+
"epoch": 0.8721881390593047,
|
| 11948 |
+
"grad_norm": 2.7719779014587402,
|
| 11949 |
+
"learning_rate": 4.410353037159193e-06,
|
| 11950 |
+
"loss": 0.9851,
|
| 11951 |
+
"step": 1706
|
| 11952 |
+
},
|
| 11953 |
+
{
|
| 11954 |
+
"epoch": 0.8726993865030674,
|
| 11955 |
+
"grad_norm": 2.8371808528900146,
|
| 11956 |
+
"learning_rate": 4.3756636267353214e-06,
|
| 11957 |
+
"loss": 1.0079,
|
| 11958 |
+
"step": 1707
|
| 11959 |
+
},
|
| 11960 |
+
{
|
| 11961 |
+
"epoch": 0.8732106339468303,
|
| 11962 |
+
"grad_norm": 3.129444122314453,
|
| 11963 |
+
"learning_rate": 4.341104935775442e-06,
|
| 11964 |
+
"loss": 0.8973,
|
| 11965 |
+
"step": 1708
|
| 11966 |
+
},
|
| 11967 |
+
{
|
| 11968 |
+
"epoch": 0.873721881390593,
|
| 11969 |
+
"grad_norm": 2.850924491882324,
|
| 11970 |
+
"learning_rate": 4.306677063294573e-06,
|
| 11971 |
+
"loss": 0.918,
|
| 11972 |
+
"step": 1709
|
| 11973 |
+
},
|
| 11974 |
+
{
|
| 11975 |
+
"epoch": 0.8742331288343558,
|
| 11976 |
+
"grad_norm": 2.839162826538086,
|
| 11977 |
+
"learning_rate": 4.272380107932888e-06,
|
| 11978 |
+
"loss": 0.9134,
|
| 11979 |
+
"step": 1710
|
| 11980 |
+
},
|
| 11981 |
+
{
|
| 11982 |
+
"epoch": 0.8747443762781186,
|
| 11983 |
+
"grad_norm": 3.149712562561035,
|
| 11984 |
+
"learning_rate": 4.238214167955484e-06,
|
| 11985 |
+
"loss": 0.9679,
|
| 11986 |
+
"step": 1711
|
| 11987 |
+
},
|
| 11988 |
+
{
|
| 11989 |
+
"epoch": 0.8752556237218814,
|
| 11990 |
+
"grad_norm": 2.945416212081909,
|
| 11991 |
+
"learning_rate": 4.2041793412520734e-06,
|
| 11992 |
+
"loss": 0.8602,
|
| 11993 |
+
"step": 1712
|
| 11994 |
+
},
|
| 11995 |
+
{
|
| 11996 |
+
"epoch": 0.8757668711656442,
|
| 11997 |
+
"grad_norm": 3.0397250652313232,
|
| 11998 |
+
"learning_rate": 4.17027572533672e-06,
|
| 11999 |
+
"loss": 0.9156,
|
| 12000 |
+
"step": 1713
|
| 12001 |
+
},
|
| 12002 |
+
{
|
| 12003 |
+
"epoch": 0.876278118609407,
|
| 12004 |
+
"grad_norm": 3.256863832473755,
|
| 12005 |
+
"learning_rate": 4.136503417347554e-06,
|
| 12006 |
+
"loss": 0.7461,
|
| 12007 |
+
"step": 1714
|
| 12008 |
+
},
|
| 12009 |
+
{
|
| 12010 |
+
"epoch": 0.8767893660531697,
|
| 12011 |
+
"grad_norm": 3.1958820819854736,
|
| 12012 |
+
"learning_rate": 4.102862514046474e-06,
|
| 12013 |
+
"loss": 0.8761,
|
| 12014 |
+
"step": 1715
|
| 12015 |
+
},
|
| 12016 |
+
{
|
| 12017 |
+
"epoch": 0.8773006134969326,
|
| 12018 |
+
"grad_norm": 3.4146556854248047,
|
| 12019 |
+
"learning_rate": 4.069353111818913e-06,
|
| 12020 |
+
"loss": 0.9106,
|
| 12021 |
+
"step": 1716
|
| 12022 |
+
},
|
| 12023 |
+
{
|
| 12024 |
+
"epoch": 0.8778118609406953,
|
| 12025 |
+
"grad_norm": 3.057286024093628,
|
| 12026 |
+
"learning_rate": 4.035975306673517e-06,
|
| 12027 |
+
"loss": 0.8755,
|
| 12028 |
+
"step": 1717
|
| 12029 |
+
},
|
| 12030 |
+
{
|
| 12031 |
+
"epoch": 0.878323108384458,
|
| 12032 |
+
"grad_norm": 3.0085763931274414,
|
| 12033 |
+
"learning_rate": 4.0027291942419055e-06,
|
| 12034 |
+
"loss": 0.7805,
|
| 12035 |
+
"step": 1718
|
| 12036 |
+
},
|
| 12037 |
+
{
|
| 12038 |
+
"epoch": 0.8788343558282209,
|
| 12039 |
+
"grad_norm": 3.2898752689361572,
|
| 12040 |
+
"learning_rate": 3.969614869778354e-06,
|
| 12041 |
+
"loss": 0.8877,
|
| 12042 |
+
"step": 1719
|
| 12043 |
+
},
|
| 12044 |
+
{
|
| 12045 |
+
"epoch": 0.8793456032719836,
|
| 12046 |
+
"grad_norm": 3.1025335788726807,
|
| 12047 |
+
"learning_rate": 3.936632428159609e-06,
|
| 12048 |
+
"loss": 0.808,
|
| 12049 |
+
"step": 1720
|
| 12050 |
+
},
|
| 12051 |
+
{
|
| 12052 |
+
"epoch": 0.8798568507157464,
|
| 12053 |
+
"grad_norm": 3.2930991649627686,
|
| 12054 |
+
"learning_rate": 3.903781963884467e-06,
|
| 12055 |
+
"loss": 0.8736,
|
| 12056 |
+
"step": 1721
|
| 12057 |
+
},
|
| 12058 |
+
{
|
| 12059 |
+
"epoch": 0.8803680981595092,
|
| 12060 |
+
"grad_norm": 3.3883445262908936,
|
| 12061 |
+
"learning_rate": 3.871063571073668e-06,
|
| 12062 |
+
"loss": 0.9969,
|
| 12063 |
+
"step": 1722
|
| 12064 |
+
},
|
| 12065 |
+
{
|
| 12066 |
+
"epoch": 0.880879345603272,
|
| 12067 |
+
"grad_norm": 3.6220452785491943,
|
| 12068 |
+
"learning_rate": 3.838477343469516e-06,
|
| 12069 |
+
"loss": 0.88,
|
| 12070 |
+
"step": 1723
|
| 12071 |
+
},
|
| 12072 |
+
{
|
| 12073 |
+
"epoch": 0.8813905930470347,
|
| 12074 |
+
"grad_norm": 3.937267303466797,
|
| 12075 |
+
"learning_rate": 3.8060233744356633e-06,
|
| 12076 |
+
"loss": 0.9832,
|
| 12077 |
+
"step": 1724
|
| 12078 |
+
},
|
| 12079 |
+
{
|
| 12080 |
+
"epoch": 0.8819018404907976,
|
| 12081 |
+
"grad_norm": 3.5126397609710693,
|
| 12082 |
+
"learning_rate": 3.77370175695681e-06,
|
| 12083 |
+
"loss": 0.8513,
|
| 12084 |
+
"step": 1725
|
| 12085 |
+
},
|
| 12086 |
+
{
|
| 12087 |
+
"epoch": 0.8824130879345603,
|
| 12088 |
+
"grad_norm": 3.5317177772521973,
|
| 12089 |
+
"learning_rate": 3.74151258363844e-06,
|
| 12090 |
+
"loss": 0.8387,
|
| 12091 |
+
"step": 1726
|
| 12092 |
+
},
|
| 12093 |
+
{
|
| 12094 |
+
"epoch": 0.8829243353783232,
|
| 12095 |
+
"grad_norm": 3.232076406478882,
|
| 12096 |
+
"learning_rate": 3.7094559467066083e-06,
|
| 12097 |
+
"loss": 0.8725,
|
| 12098 |
+
"step": 1727
|
| 12099 |
+
},
|
| 12100 |
+
{
|
| 12101 |
+
"epoch": 0.8834355828220859,
|
| 12102 |
+
"grad_norm": 3.6095595359802246,
|
| 12103 |
+
"learning_rate": 3.6775319380076e-06,
|
| 12104 |
+
"loss": 0.9068,
|
| 12105 |
+
"step": 1728
|
| 12106 |
+
},
|
| 12107 |
+
{
|
| 12108 |
+
"epoch": 0.8839468302658486,
|
| 12109 |
+
"grad_norm": 3.2999720573425293,
|
| 12110 |
+
"learning_rate": 3.645740649007734e-06,
|
| 12111 |
+
"loss": 0.806,
|
| 12112 |
+
"step": 1729
|
| 12113 |
+
},
|
| 12114 |
+
{
|
| 12115 |
+
"epoch": 0.8844580777096115,
|
| 12116 |
+
"grad_norm": 3.733455181121826,
|
| 12117 |
+
"learning_rate": 3.614082170793021e-06,
|
| 12118 |
+
"loss": 0.8415,
|
| 12119 |
+
"step": 1730
|
| 12120 |
+
},
|
| 12121 |
+
{
|
| 12122 |
+
"epoch": 0.8849693251533742,
|
| 12123 |
+
"grad_norm": 3.5478620529174805,
|
| 12124 |
+
"learning_rate": 3.5825565940690087e-06,
|
| 12125 |
+
"loss": 0.8471,
|
| 12126 |
+
"step": 1731
|
| 12127 |
+
},
|
| 12128 |
+
{
|
| 12129 |
+
"epoch": 0.885480572597137,
|
| 12130 |
+
"grad_norm": 3.889519214630127,
|
| 12131 |
+
"learning_rate": 3.551164009160429e-06,
|
| 12132 |
+
"loss": 0.8128,
|
| 12133 |
+
"step": 1732
|
| 12134 |
+
},
|
| 12135 |
+
{
|
| 12136 |
+
"epoch": 0.8859918200408998,
|
| 12137 |
+
"grad_norm": 3.7842485904693604,
|
| 12138 |
+
"learning_rate": 3.5199045060110013e-06,
|
| 12139 |
+
"loss": 0.9556,
|
| 12140 |
+
"step": 1733
|
| 12141 |
+
},
|
| 12142 |
+
{
|
| 12143 |
+
"epoch": 0.8865030674846626,
|
| 12144 |
+
"grad_norm": 3.5816397666931152,
|
| 12145 |
+
"learning_rate": 3.488778174183116e-06,
|
| 12146 |
+
"loss": 0.8108,
|
| 12147 |
+
"step": 1734
|
| 12148 |
+
},
|
| 12149 |
+
{
|
| 12150 |
+
"epoch": 0.8870143149284253,
|
| 12151 |
+
"grad_norm": 3.812117338180542,
|
| 12152 |
+
"learning_rate": 3.4577851028576523e-06,
|
| 12153 |
+
"loss": 0.7997,
|
| 12154 |
+
"step": 1735
|
| 12155 |
+
},
|
| 12156 |
+
{
|
| 12157 |
+
"epoch": 0.8875255623721882,
|
| 12158 |
+
"grad_norm": 3.7888643741607666,
|
| 12159 |
+
"learning_rate": 3.4269253808336455e-06,
|
| 12160 |
+
"loss": 0.8451,
|
| 12161 |
+
"step": 1736
|
| 12162 |
+
},
|
| 12163 |
+
{
|
| 12164 |
+
"epoch": 0.8880368098159509,
|
| 12165 |
+
"grad_norm": 3.8936595916748047,
|
| 12166 |
+
"learning_rate": 3.3961990965280745e-06,
|
| 12167 |
+
"loss": 0.8154,
|
| 12168 |
+
"step": 1737
|
| 12169 |
+
},
|
| 12170 |
+
{
|
| 12171 |
+
"epoch": 0.8885480572597138,
|
| 12172 |
+
"grad_norm": 4.134406566619873,
|
| 12173 |
+
"learning_rate": 3.36560633797563e-06,
|
| 12174 |
+
"loss": 0.8866,
|
| 12175 |
+
"step": 1738
|
| 12176 |
+
},
|
| 12177 |
+
{
|
| 12178 |
+
"epoch": 0.8890593047034765,
|
| 12179 |
+
"grad_norm": 4.125121116638184,
|
| 12180 |
+
"learning_rate": 3.335147192828403e-06,
|
| 12181 |
+
"loss": 0.7727,
|
| 12182 |
+
"step": 1739
|
| 12183 |
+
},
|
| 12184 |
+
{
|
| 12185 |
+
"epoch": 0.8895705521472392,
|
| 12186 |
+
"grad_norm": 4.411681175231934,
|
| 12187 |
+
"learning_rate": 3.3048217483556744e-06,
|
| 12188 |
+
"loss": 0.8563,
|
| 12189 |
+
"step": 1740
|
| 12190 |
+
},
|
| 12191 |
+
{
|
| 12192 |
+
"epoch": 0.8900817995910021,
|
| 12193 |
+
"grad_norm": 3.7549989223480225,
|
| 12194 |
+
"learning_rate": 3.2746300914436534e-06,
|
| 12195 |
+
"loss": 0.7357,
|
| 12196 |
+
"step": 1741
|
| 12197 |
+
},
|
| 12198 |
+
{
|
| 12199 |
+
"epoch": 0.8905930470347648,
|
| 12200 |
+
"grad_norm": 4.538980960845947,
|
| 12201 |
+
"learning_rate": 3.2445723085952504e-06,
|
| 12202 |
+
"loss": 0.818,
|
| 12203 |
+
"step": 1742
|
| 12204 |
+
},
|
| 12205 |
+
{
|
| 12206 |
+
"epoch": 0.8911042944785276,
|
| 12207 |
+
"grad_norm": 4.389502048492432,
|
| 12208 |
+
"learning_rate": 3.214648485929783e-06,
|
| 12209 |
+
"loss": 0.8799,
|
| 12210 |
+
"step": 1743
|
| 12211 |
+
},
|
| 12212 |
+
{
|
| 12213 |
+
"epoch": 0.8916155419222904,
|
| 12214 |
+
"grad_norm": 4.548320293426514,
|
| 12215 |
+
"learning_rate": 3.184858709182775e-06,
|
| 12216 |
+
"loss": 0.6832,
|
| 12217 |
+
"step": 1744
|
| 12218 |
+
},
|
| 12219 |
+
{
|
| 12220 |
+
"epoch": 0.8921267893660532,
|
| 12221 |
+
"grad_norm": 4.614025115966797,
|
| 12222 |
+
"learning_rate": 3.1552030637056806e-06,
|
| 12223 |
+
"loss": 0.8305,
|
| 12224 |
+
"step": 1745
|
| 12225 |
+
},
|
| 12226 |
+
{
|
| 12227 |
+
"epoch": 0.8926380368098159,
|
| 12228 |
+
"grad_norm": 4.867095470428467,
|
| 12229 |
+
"learning_rate": 3.1256816344656602e-06,
|
| 12230 |
+
"loss": 0.8145,
|
| 12231 |
+
"step": 1746
|
| 12232 |
+
},
|
| 12233 |
+
{
|
| 12234 |
+
"epoch": 0.8931492842535788,
|
| 12235 |
+
"grad_norm": 4.788604259490967,
|
| 12236 |
+
"learning_rate": 3.096294506045311e-06,
|
| 12237 |
+
"loss": 0.8077,
|
| 12238 |
+
"step": 1747
|
| 12239 |
+
},
|
| 12240 |
+
{
|
| 12241 |
+
"epoch": 0.8936605316973415,
|
| 12242 |
+
"grad_norm": 5.478014945983887,
|
| 12243 |
+
"learning_rate": 3.067041762642475e-06,
|
| 12244 |
+
"loss": 0.771,
|
| 12245 |
+
"step": 1748
|
| 12246 |
+
},
|
| 12247 |
+
{
|
| 12248 |
+
"epoch": 0.8941717791411042,
|
| 12249 |
+
"grad_norm": 6.572366237640381,
|
| 12250 |
+
"learning_rate": 3.037923488069927e-06,
|
| 12251 |
+
"loss": 0.7604,
|
| 12252 |
+
"step": 1749
|
| 12253 |
+
},
|
| 12254 |
+
{
|
| 12255 |
+
"epoch": 0.8946830265848671,
|
| 12256 |
+
"grad_norm": 5.753463268280029,
|
| 12257 |
+
"learning_rate": 3.0089397657551865e-06,
|
| 12258 |
+
"loss": 0.4072,
|
| 12259 |
+
"step": 1750
|
| 12260 |
+
},
|
| 12261 |
+
{
|
| 12262 |
+
"epoch": 0.8951942740286298,
|
| 12263 |
+
"grad_norm": 1.9616332054138184,
|
| 12264 |
+
"learning_rate": 2.9800906787402716e-06,
|
| 12265 |
+
"loss": 0.9257,
|
| 12266 |
+
"step": 1751
|
| 12267 |
+
},
|
| 12268 |
+
{
|
| 12269 |
+
"epoch": 0.8957055214723927,
|
| 12270 |
+
"grad_norm": 2.392256259918213,
|
| 12271 |
+
"learning_rate": 2.9513763096814305e-06,
|
| 12272 |
+
"loss": 1.0013,
|
| 12273 |
+
"step": 1752
|
| 12274 |
+
},
|
| 12275 |
+
{
|
| 12276 |
+
"epoch": 0.8962167689161554,
|
| 12277 |
+
"grad_norm": 2.3039324283599854,
|
| 12278 |
+
"learning_rate": 2.9227967408489653e-06,
|
| 12279 |
+
"loss": 0.9961,
|
| 12280 |
+
"step": 1753
|
| 12281 |
+
},
|
| 12282 |
+
{
|
| 12283 |
+
"epoch": 0.8967280163599182,
|
| 12284 |
+
"grad_norm": 2.534642457962036,
|
| 12285 |
+
"learning_rate": 2.89435205412692e-06,
|
| 12286 |
+
"loss": 0.9289,
|
| 12287 |
+
"step": 1754
|
| 12288 |
+
},
|
| 12289 |
+
{
|
| 12290 |
+
"epoch": 0.897239263803681,
|
| 12291 |
+
"grad_norm": 2.941516160964966,
|
| 12292 |
+
"learning_rate": 2.8660423310129135e-06,
|
| 12293 |
+
"loss": 0.957,
|
| 12294 |
+
"step": 1755
|
| 12295 |
+
},
|
| 12296 |
+
{
|
| 12297 |
+
"epoch": 0.8977505112474438,
|
| 12298 |
+
"grad_norm": 2.806290626525879,
|
| 12299 |
+
"learning_rate": 2.8378676526178482e-06,
|
| 12300 |
+
"loss": 1.0622,
|
| 12301 |
+
"step": 1756
|
| 12302 |
+
},
|
| 12303 |
+
{
|
| 12304 |
+
"epoch": 0.8982617586912065,
|
| 12305 |
+
"grad_norm": 2.6641461849212646,
|
| 12306 |
+
"learning_rate": 2.8098280996657456e-06,
|
| 12307 |
+
"loss": 0.9309,
|
| 12308 |
+
"step": 1757
|
| 12309 |
+
},
|
| 12310 |
+
{
|
| 12311 |
+
"epoch": 0.8987730061349694,
|
| 12312 |
+
"grad_norm": 2.6802642345428467,
|
| 12313 |
+
"learning_rate": 2.781923752493437e-06,
|
| 12314 |
+
"loss": 0.8945,
|
| 12315 |
+
"step": 1758
|
| 12316 |
+
},
|
| 12317 |
+
{
|
| 12318 |
+
"epoch": 0.8992842535787321,
|
| 12319 |
+
"grad_norm": 2.851447820663452,
|
| 12320 |
+
"learning_rate": 2.754154691050387e-06,
|
| 12321 |
+
"loss": 0.8722,
|
| 12322 |
+
"step": 1759
|
| 12323 |
+
},
|
| 12324 |
+
{
|
| 12325 |
+
"epoch": 0.8997955010224948,
|
| 12326 |
+
"grad_norm": 2.958563804626465,
|
| 12327 |
+
"learning_rate": 2.7265209948984514e-06,
|
| 12328 |
+
"loss": 0.8529,
|
| 12329 |
+
"step": 1760
|
| 12330 |
+
},
|
| 12331 |
+
{
|
| 12332 |
+
"epoch": 0.9003067484662577,
|
| 12333 |
+
"grad_norm": 2.867089033126831,
|
| 12334 |
+
"learning_rate": 2.6990227432116544e-06,
|
| 12335 |
+
"loss": 0.8957,
|
| 12336 |
+
"step": 1761
|
| 12337 |
+
},
|
| 12338 |
+
{
|
| 12339 |
+
"epoch": 0.9008179959100204,
|
| 12340 |
+
"grad_norm": 3.3198704719543457,
|
| 12341 |
+
"learning_rate": 2.671660014775934e-06,
|
| 12342 |
+
"loss": 0.8905,
|
| 12343 |
+
"step": 1762
|
| 12344 |
+
},
|
| 12345 |
+
{
|
| 12346 |
+
"epoch": 0.9013292433537833,
|
| 12347 |
+
"grad_norm": 3.0108888149261475,
|
| 12348 |
+
"learning_rate": 2.6444328879889622e-06,
|
| 12349 |
+
"loss": 0.8434,
|
| 12350 |
+
"step": 1763
|
| 12351 |
+
},
|
| 12352 |
+
{
|
| 12353 |
+
"epoch": 0.901840490797546,
|
| 12354 |
+
"grad_norm": 2.9155540466308594,
|
| 12355 |
+
"learning_rate": 2.6173414408598827e-06,
|
| 12356 |
+
"loss": 0.8414,
|
| 12357 |
+
"step": 1764
|
| 12358 |
+
},
|
| 12359 |
+
{
|
| 12360 |
+
"epoch": 0.9023517382413088,
|
| 12361 |
+
"grad_norm": 3.178889036178589,
|
| 12362 |
+
"learning_rate": 2.5903857510090835e-06,
|
| 12363 |
+
"loss": 0.9461,
|
| 12364 |
+
"step": 1765
|
| 12365 |
+
},
|
| 12366 |
+
{
|
| 12367 |
+
"epoch": 0.9028629856850716,
|
| 12368 |
+
"grad_norm": 3.119640588760376,
|
| 12369 |
+
"learning_rate": 2.56356589566803e-06,
|
| 12370 |
+
"loss": 0.8996,
|
| 12371 |
+
"step": 1766
|
| 12372 |
+
},
|
| 12373 |
+
{
|
| 12374 |
+
"epoch": 0.9033742331288344,
|
| 12375 |
+
"grad_norm": 3.1831552982330322,
|
| 12376 |
+
"learning_rate": 2.53688195167896e-06,
|
| 12377 |
+
"loss": 0.9139,
|
| 12378 |
+
"step": 1767
|
| 12379 |
+
},
|
| 12380 |
+
{
|
| 12381 |
+
"epoch": 0.9038854805725971,
|
| 12382 |
+
"grad_norm": 3.3082258701324463,
|
| 12383 |
+
"learning_rate": 2.5103339954947626e-06,
|
| 12384 |
+
"loss": 0.8465,
|
| 12385 |
+
"step": 1768
|
| 12386 |
+
},
|
| 12387 |
+
{
|
| 12388 |
+
"epoch": 0.90439672801636,
|
| 12389 |
+
"grad_norm": 3.1402430534362793,
|
| 12390 |
+
"learning_rate": 2.483922103178632e-06,
|
| 12391 |
+
"loss": 0.9071,
|
| 12392 |
+
"step": 1769
|
| 12393 |
+
},
|
| 12394 |
+
{
|
| 12395 |
+
"epoch": 0.9049079754601227,
|
| 12396 |
+
"grad_norm": 3.3881165981292725,
|
| 12397 |
+
"learning_rate": 2.4576463504039913e-06,
|
| 12398 |
+
"loss": 0.9479,
|
| 12399 |
+
"step": 1770
|
| 12400 |
+
},
|
| 12401 |
+
{
|
| 12402 |
+
"epoch": 0.9054192229038854,
|
| 12403 |
+
"grad_norm": 3.3849008083343506,
|
| 12404 |
+
"learning_rate": 2.4315068124541597e-06,
|
| 12405 |
+
"loss": 0.8833,
|
| 12406 |
+
"step": 1771
|
| 12407 |
+
},
|
| 12408 |
+
{
|
| 12409 |
+
"epoch": 0.9059304703476483,
|
| 12410 |
+
"grad_norm": 3.725773334503174,
|
| 12411 |
+
"learning_rate": 2.4055035642222224e-06,
|
| 12412 |
+
"loss": 0.8946,
|
| 12413 |
+
"step": 1772
|
| 12414 |
+
},
|
| 12415 |
+
{
|
| 12416 |
+
"epoch": 0.906441717791411,
|
| 12417 |
+
"grad_norm": 4.276130676269531,
|
| 12418 |
+
"learning_rate": 2.3796366802107394e-06,
|
| 12419 |
+
"loss": 0.9649,
|
| 12420 |
+
"step": 1773
|
| 12421 |
+
},
|
| 12422 |
+
{
|
| 12423 |
+
"epoch": 0.9069529652351738,
|
| 12424 |
+
"grad_norm": 3.543367862701416,
|
| 12425 |
+
"learning_rate": 2.3539062345316e-06,
|
| 12426 |
+
"loss": 0.8461,
|
| 12427 |
+
"step": 1774
|
| 12428 |
+
},
|
| 12429 |
+
{
|
| 12430 |
+
"epoch": 0.9074642126789366,
|
| 12431 |
+
"grad_norm": 3.076871156692505,
|
| 12432 |
+
"learning_rate": 2.3283123009057607e-06,
|
| 12433 |
+
"loss": 0.8243,
|
| 12434 |
+
"step": 1775
|
| 12435 |
+
},
|
| 12436 |
+
{
|
| 12437 |
+
"epoch": 0.9079754601226994,
|
| 12438 |
+
"grad_norm": 3.6242263317108154,
|
| 12439 |
+
"learning_rate": 2.3028549526630583e-06,
|
| 12440 |
+
"loss": 0.8345,
|
| 12441 |
+
"step": 1776
|
| 12442 |
+
},
|
| 12443 |
+
{
|
| 12444 |
+
"epoch": 0.9084867075664622,
|
| 12445 |
+
"grad_norm": 3.5291264057159424,
|
| 12446 |
+
"learning_rate": 2.277534262742015e-06,
|
| 12447 |
+
"loss": 0.892,
|
| 12448 |
+
"step": 1777
|
| 12449 |
+
},
|
| 12450 |
+
{
|
| 12451 |
+
"epoch": 0.908997955010225,
|
| 12452 |
+
"grad_norm": 3.4097137451171875,
|
| 12453 |
+
"learning_rate": 2.2523503036895764e-06,
|
| 12454 |
+
"loss": 0.8539,
|
| 12455 |
+
"step": 1778
|
| 12456 |
+
},
|
| 12457 |
+
{
|
| 12458 |
+
"epoch": 0.9095092024539877,
|
| 12459 |
+
"grad_norm": 3.7175967693328857,
|
| 12460 |
+
"learning_rate": 2.227303147660964e-06,
|
| 12461 |
+
"loss": 0.9083,
|
| 12462 |
+
"step": 1779
|
| 12463 |
+
},
|
| 12464 |
+
{
|
| 12465 |
+
"epoch": 0.9100204498977505,
|
| 12466 |
+
"grad_norm": 3.9345943927764893,
|
| 12467 |
+
"learning_rate": 2.202392866419423e-06,
|
| 12468 |
+
"loss": 0.9569,
|
| 12469 |
+
"step": 1780
|
| 12470 |
+
},
|
| 12471 |
+
{
|
| 12472 |
+
"epoch": 0.9105316973415133,
|
| 12473 |
+
"grad_norm": 3.6897552013397217,
|
| 12474 |
+
"learning_rate": 2.1776195313360505e-06,
|
| 12475 |
+
"loss": 0.8444,
|
| 12476 |
+
"step": 1781
|
| 12477 |
+
},
|
| 12478 |
+
{
|
| 12479 |
+
"epoch": 0.911042944785276,
|
| 12480 |
+
"grad_norm": 3.9733729362487793,
|
| 12481 |
+
"learning_rate": 2.152983213389559e-06,
|
| 12482 |
+
"loss": 0.9426,
|
| 12483 |
+
"step": 1782
|
| 12484 |
+
},
|
| 12485 |
+
{
|
| 12486 |
+
"epoch": 0.9115541922290389,
|
| 12487 |
+
"grad_norm": 4.125808238983154,
|
| 12488 |
+
"learning_rate": 2.1284839831661075e-06,
|
| 12489 |
+
"loss": 0.8886,
|
| 12490 |
+
"step": 1783
|
| 12491 |
+
},
|
| 12492 |
+
{
|
| 12493 |
+
"epoch": 0.9120654396728016,
|
| 12494 |
+
"grad_norm": 3.4890663623809814,
|
| 12495 |
+
"learning_rate": 2.1041219108590692e-06,
|
| 12496 |
+
"loss": 0.8138,
|
| 12497 |
+
"step": 1784
|
| 12498 |
+
},
|
| 12499 |
+
{
|
| 12500 |
+
"epoch": 0.9125766871165644,
|
| 12501 |
+
"grad_norm": 3.9719114303588867,
|
| 12502 |
+
"learning_rate": 2.0798970662688545e-06,
|
| 12503 |
+
"loss": 0.8747,
|
| 12504 |
+
"step": 1785
|
| 12505 |
+
},
|
| 12506 |
+
{
|
| 12507 |
+
"epoch": 0.9130879345603272,
|
| 12508 |
+
"grad_norm": 4.040510654449463,
|
| 12509 |
+
"learning_rate": 2.055809518802676e-06,
|
| 12510 |
+
"loss": 1.0,
|
| 12511 |
+
"step": 1786
|
| 12512 |
+
},
|
| 12513 |
+
{
|
| 12514 |
+
"epoch": 0.91359918200409,
|
| 12515 |
+
"grad_norm": 4.081876754760742,
|
| 12516 |
+
"learning_rate": 2.031859337474407e-06,
|
| 12517 |
+
"loss": 0.7761,
|
| 12518 |
+
"step": 1787
|
| 12519 |
+
},
|
| 12520 |
+
{
|
| 12521 |
+
"epoch": 0.9141104294478528,
|
| 12522 |
+
"grad_norm": 4.227837562561035,
|
| 12523 |
+
"learning_rate": 2.0080465909043113e-06,
|
| 12524 |
+
"loss": 0.8716,
|
| 12525 |
+
"step": 1788
|
| 12526 |
+
},
|
| 12527 |
+
{
|
| 12528 |
+
"epoch": 0.9146216768916156,
|
| 12529 |
+
"grad_norm": 4.269420623779297,
|
| 12530 |
+
"learning_rate": 1.984371347318914e-06,
|
| 12531 |
+
"loss": 0.8066,
|
| 12532 |
+
"step": 1789
|
| 12533 |
+
},
|
| 12534 |
+
{
|
| 12535 |
+
"epoch": 0.9151329243353783,
|
| 12536 |
+
"grad_norm": 4.166466236114502,
|
| 12537 |
+
"learning_rate": 1.9608336745507716e-06,
|
| 12538 |
+
"loss": 0.802,
|
| 12539 |
+
"step": 1790
|
| 12540 |
+
},
|
| 12541 |
+
{
|
| 12542 |
+
"epoch": 0.9156441717791411,
|
| 12543 |
+
"grad_norm": 4.287985801696777,
|
| 12544 |
+
"learning_rate": 1.937433640038261e-06,
|
| 12545 |
+
"loss": 0.7958,
|
| 12546 |
+
"step": 1791
|
| 12547 |
+
},
|
| 12548 |
+
{
|
| 12549 |
+
"epoch": 0.9161554192229039,
|
| 12550 |
+
"grad_norm": 4.09846830368042,
|
| 12551 |
+
"learning_rate": 1.914171310825441e-06,
|
| 12552 |
+
"loss": 0.7164,
|
| 12553 |
+
"step": 1792
|
| 12554 |
+
},
|
| 12555 |
+
{
|
| 12556 |
+
"epoch": 0.9166666666666666,
|
| 12557 |
+
"grad_norm": 4.546944618225098,
|
| 12558 |
+
"learning_rate": 1.8910467535617983e-06,
|
| 12559 |
+
"loss": 0.7325,
|
| 12560 |
+
"step": 1793
|
| 12561 |
+
},
|
| 12562 |
+
{
|
| 12563 |
+
"epoch": 0.9171779141104295,
|
| 12564 |
+
"grad_norm": 4.4474897384643555,
|
| 12565 |
+
"learning_rate": 1.8680600345021171e-06,
|
| 12566 |
+
"loss": 0.7291,
|
| 12567 |
+
"step": 1794
|
| 12568 |
+
},
|
| 12569 |
+
{
|
| 12570 |
+
"epoch": 0.9176891615541922,
|
| 12571 |
+
"grad_norm": 4.722184658050537,
|
| 12572 |
+
"learning_rate": 1.845211219506221e-06,
|
| 12573 |
+
"loss": 0.6715,
|
| 12574 |
+
"step": 1795
|
| 12575 |
+
},
|
| 12576 |
+
{
|
| 12577 |
+
"epoch": 0.918200408997955,
|
| 12578 |
+
"grad_norm": 5.06058931350708,
|
| 12579 |
+
"learning_rate": 1.8225003740388547e-06,
|
| 12580 |
+
"loss": 0.9135,
|
| 12581 |
+
"step": 1796
|
| 12582 |
+
},
|
| 12583 |
+
{
|
| 12584 |
+
"epoch": 0.9187116564417178,
|
| 12585 |
+
"grad_norm": 5.043420791625977,
|
| 12586 |
+
"learning_rate": 1.79992756316944e-06,
|
| 12587 |
+
"loss": 0.7258,
|
| 12588 |
+
"step": 1797
|
| 12589 |
+
},
|
| 12590 |
+
{
|
| 12591 |
+
"epoch": 0.9192229038854806,
|
| 12592 |
+
"grad_norm": 5.176082134246826,
|
| 12593 |
+
"learning_rate": 1.7774928515719157e-06,
|
| 12594 |
+
"loss": 0.7435,
|
| 12595 |
+
"step": 1798
|
| 12596 |
+
},
|
| 12597 |
+
{
|
| 12598 |
+
"epoch": 0.9197341513292433,
|
| 12599 |
+
"grad_norm": 6.064294815063477,
|
| 12600 |
+
"learning_rate": 1.7551963035245588e-06,
|
| 12601 |
+
"loss": 0.7142,
|
| 12602 |
+
"step": 1799
|
| 12603 |
+
},
|
| 12604 |
+
{
|
| 12605 |
+
"epoch": 0.9202453987730062,
|
| 12606 |
+
"grad_norm": 6.77680778503418,
|
| 12607 |
+
"learning_rate": 1.733037982909791e-06,
|
| 12608 |
+
"loss": 0.4713,
|
| 12609 |
+
"step": 1800
|
| 12610 |
+
},
|
| 12611 |
+
{
|
| 12612 |
+
"epoch": 0.9207566462167689,
|
| 12613 |
+
"grad_norm": 1.9696067571640015,
|
| 12614 |
+
"learning_rate": 1.7110179532139781e-06,
|
| 12615 |
+
"loss": 0.8423,
|
| 12616 |
+
"step": 1801
|
| 12617 |
+
},
|
| 12618 |
+
{
|
| 12619 |
+
"epoch": 0.9212678936605317,
|
| 12620 |
+
"grad_norm": 2.4090499877929688,
|
| 12621 |
+
"learning_rate": 1.6891362775272812e-06,
|
| 12622 |
+
"loss": 0.9596,
|
| 12623 |
+
"step": 1802
|
| 12624 |
+
},
|
| 12625 |
+
{
|
| 12626 |
+
"epoch": 0.9217791411042945,
|
| 12627 |
+
"grad_norm": 2.441483974456787,
|
| 12628 |
+
"learning_rate": 1.6673930185434561e-06,
|
| 12629 |
+
"loss": 0.8638,
|
| 12630 |
+
"step": 1803
|
| 12631 |
+
},
|
| 12632 |
+
{
|
| 12633 |
+
"epoch": 0.9222903885480572,
|
| 12634 |
+
"grad_norm": 2.631037712097168,
|
| 12635 |
+
"learning_rate": 1.6457882385596646e-06,
|
| 12636 |
+
"loss": 1.0107,
|
| 12637 |
+
"step": 1804
|
| 12638 |
+
},
|
| 12639 |
+
{
|
| 12640 |
+
"epoch": 0.9228016359918201,
|
| 12641 |
+
"grad_norm": 2.69331431388855,
|
| 12642 |
+
"learning_rate": 1.6243219994763304e-06,
|
| 12643 |
+
"loss": 0.9641,
|
| 12644 |
+
"step": 1805
|
| 12645 |
+
},
|
| 12646 |
+
{
|
| 12647 |
+
"epoch": 0.9233128834355828,
|
| 12648 |
+
"grad_norm": 2.4592180252075195,
|
| 12649 |
+
"learning_rate": 1.6029943627969223e-06,
|
| 12650 |
+
"loss": 0.8723,
|
| 12651 |
+
"step": 1806
|
| 12652 |
+
},
|
| 12653 |
+
{
|
| 12654 |
+
"epoch": 0.9238241308793456,
|
| 12655 |
+
"grad_norm": 2.729130744934082,
|
| 12656 |
+
"learning_rate": 1.5818053896278162e-06,
|
| 12657 |
+
"loss": 0.9765,
|
| 12658 |
+
"step": 1807
|
| 12659 |
+
},
|
| 12660 |
+
{
|
| 12661 |
+
"epoch": 0.9243353783231084,
|
| 12662 |
+
"grad_norm": 2.6787896156311035,
|
| 12663 |
+
"learning_rate": 1.5607551406780717e-06,
|
| 12664 |
+
"loss": 0.8601,
|
| 12665 |
+
"step": 1808
|
| 12666 |
+
},
|
| 12667 |
+
{
|
| 12668 |
+
"epoch": 0.9248466257668712,
|
| 12669 |
+
"grad_norm": 2.9896671772003174,
|
| 12670 |
+
"learning_rate": 1.5398436762593061e-06,
|
| 12671 |
+
"loss": 0.8525,
|
| 12672 |
+
"step": 1809
|
| 12673 |
+
},
|
| 12674 |
+
{
|
| 12675 |
+
"epoch": 0.9253578732106339,
|
| 12676 |
+
"grad_norm": 3.0496749877929688,
|
| 12677 |
+
"learning_rate": 1.519071056285487e-06,
|
| 12678 |
+
"loss": 0.9679,
|
| 12679 |
+
"step": 1810
|
| 12680 |
+
},
|
| 12681 |
+
{
|
| 12682 |
+
"epoch": 0.9258691206543967,
|
| 12683 |
+
"grad_norm": 2.976200819015503,
|
| 12684 |
+
"learning_rate": 1.4984373402728014e-06,
|
| 12685 |
+
"loss": 0.8789,
|
| 12686 |
+
"step": 1811
|
| 12687 |
+
},
|
| 12688 |
+
{
|
| 12689 |
+
"epoch": 0.9263803680981595,
|
| 12690 |
+
"grad_norm": 3.0850539207458496,
|
| 12691 |
+
"learning_rate": 1.4779425873394259e-06,
|
| 12692 |
+
"loss": 0.85,
|
| 12693 |
+
"step": 1812
|
| 12694 |
+
},
|
| 12695 |
+
{
|
| 12696 |
+
"epoch": 0.9268916155419223,
|
| 12697 |
+
"grad_norm": 3.117995023727417,
|
| 12698 |
+
"learning_rate": 1.4575868562054228e-06,
|
| 12699 |
+
"loss": 0.8234,
|
| 12700 |
+
"step": 1813
|
| 12701 |
+
},
|
| 12702 |
+
{
|
| 12703 |
+
"epoch": 0.9274028629856851,
|
| 12704 |
+
"grad_norm": 2.9029812812805176,
|
| 12705 |
+
"learning_rate": 1.4373702051925065e-06,
|
| 12706 |
+
"loss": 0.9211,
|
| 12707 |
+
"step": 1814
|
| 12708 |
+
},
|
| 12709 |
+
{
|
| 12710 |
+
"epoch": 0.9279141104294478,
|
| 12711 |
+
"grad_norm": 3.243557929992676,
|
| 12712 |
+
"learning_rate": 1.4172926922239315e-06,
|
| 12713 |
+
"loss": 0.8936,
|
| 12714 |
+
"step": 1815
|
| 12715 |
+
},
|
| 12716 |
+
{
|
| 12717 |
+
"epoch": 0.9284253578732107,
|
| 12718 |
+
"grad_norm": 2.932993173599243,
|
| 12719 |
+
"learning_rate": 1.3973543748243e-06,
|
| 12720 |
+
"loss": 0.8516,
|
| 12721 |
+
"step": 1816
|
| 12722 |
+
},
|
| 12723 |
+
{
|
| 12724 |
+
"epoch": 0.9289366053169734,
|
| 12725 |
+
"grad_norm": 3.073636054992676,
|
| 12726 |
+
"learning_rate": 1.377555310119405e-06,
|
| 12727 |
+
"loss": 0.8922,
|
| 12728 |
+
"step": 1817
|
| 12729 |
+
},
|
| 12730 |
+
{
|
| 12731 |
+
"epoch": 0.9294478527607362,
|
| 12732 |
+
"grad_norm": 3.3185694217681885,
|
| 12733 |
+
"learning_rate": 1.3578955548360473e-06,
|
| 12734 |
+
"loss": 0.8148,
|
| 12735 |
+
"step": 1818
|
| 12736 |
+
},
|
| 12737 |
+
{
|
| 12738 |
+
"epoch": 0.929959100204499,
|
| 12739 |
+
"grad_norm": 3.1685848236083984,
|
| 12740 |
+
"learning_rate": 1.3383751653019029e-06,
|
| 12741 |
+
"loss": 0.9387,
|
| 12742 |
+
"step": 1819
|
| 12743 |
+
},
|
| 12744 |
+
{
|
| 12745 |
+
"epoch": 0.9304703476482618,
|
| 12746 |
+
"grad_norm": 3.10244083404541,
|
| 12747 |
+
"learning_rate": 1.31899419744535e-06,
|
| 12748 |
+
"loss": 0.7719,
|
| 12749 |
+
"step": 1820
|
| 12750 |
+
},
|
| 12751 |
+
{
|
| 12752 |
+
"epoch": 0.9309815950920245,
|
| 12753 |
+
"grad_norm": 3.1995668411254883,
|
| 12754 |
+
"learning_rate": 1.2997527067952875e-06,
|
| 12755 |
+
"loss": 0.9624,
|
| 12756 |
+
"step": 1821
|
| 12757 |
+
},
|
| 12758 |
+
{
|
| 12759 |
+
"epoch": 0.9314928425357873,
|
| 12760 |
+
"grad_norm": 3.6962995529174805,
|
| 12761 |
+
"learning_rate": 1.2806507484810215e-06,
|
| 12762 |
+
"loss": 0.8737,
|
| 12763 |
+
"step": 1822
|
| 12764 |
+
},
|
| 12765 |
+
{
|
| 12766 |
+
"epoch": 0.9320040899795501,
|
| 12767 |
+
"grad_norm": 3.9050567150115967,
|
| 12768 |
+
"learning_rate": 1.2616883772320508e-06,
|
| 12769 |
+
"loss": 1.0082,
|
| 12770 |
+
"step": 1823
|
| 12771 |
+
},
|
| 12772 |
+
{
|
| 12773 |
+
"epoch": 0.9325153374233128,
|
| 12774 |
+
"grad_norm": 3.417041540145874,
|
| 12775 |
+
"learning_rate": 1.2428656473779721e-06,
|
| 12776 |
+
"loss": 0.8719,
|
| 12777 |
+
"step": 1824
|
| 12778 |
+
},
|
| 12779 |
+
{
|
| 12780 |
+
"epoch": 0.9330265848670757,
|
| 12781 |
+
"grad_norm": 3.355666160583496,
|
| 12782 |
+
"learning_rate": 1.2241826128482625e-06,
|
| 12783 |
+
"loss": 0.9281,
|
| 12784 |
+
"step": 1825
|
| 12785 |
+
},
|
| 12786 |
+
{
|
| 12787 |
+
"epoch": 0.9335378323108384,
|
| 12788 |
+
"grad_norm": 3.6375527381896973,
|
| 12789 |
+
"learning_rate": 1.20563932717217e-06,
|
| 12790 |
+
"loss": 0.8613,
|
| 12791 |
+
"step": 1826
|
| 12792 |
+
},
|
| 12793 |
+
{
|
| 12794 |
+
"epoch": 0.9340490797546013,
|
| 12795 |
+
"grad_norm": 3.6801555156707764,
|
| 12796 |
+
"learning_rate": 1.1872358434785346e-06,
|
| 12797 |
+
"loss": 0.7958,
|
| 12798 |
+
"step": 1827
|
| 12799 |
+
},
|
| 12800 |
+
{
|
| 12801 |
+
"epoch": 0.934560327198364,
|
| 12802 |
+
"grad_norm": 3.775987148284912,
|
| 12803 |
+
"learning_rate": 1.1689722144956671e-06,
|
| 12804 |
+
"loss": 0.8864,
|
| 12805 |
+
"step": 1828
|
| 12806 |
+
},
|
| 12807 |
+
{
|
| 12808 |
+
"epoch": 0.9350715746421268,
|
| 12809 |
+
"grad_norm": 3.3887085914611816,
|
| 12810 |
+
"learning_rate": 1.1508484925511542e-06,
|
| 12811 |
+
"loss": 0.8428,
|
| 12812 |
+
"step": 1829
|
| 12813 |
+
},
|
| 12814 |
+
{
|
| 12815 |
+
"epoch": 0.9355828220858896,
|
| 12816 |
+
"grad_norm": 3.701101064682007,
|
| 12817 |
+
"learning_rate": 1.132864729571731e-06,
|
| 12818 |
+
"loss": 0.8598,
|
| 12819 |
+
"step": 1830
|
| 12820 |
+
},
|
| 12821 |
+
{
|
| 12822 |
+
"epoch": 0.9360940695296524,
|
| 12823 |
+
"grad_norm": 3.671170711517334,
|
| 12824 |
+
"learning_rate": 1.1150209770831588e-06,
|
| 12825 |
+
"loss": 0.8273,
|
| 12826 |
+
"step": 1831
|
| 12827 |
+
},
|
| 12828 |
+
{
|
| 12829 |
+
"epoch": 0.9366053169734151,
|
| 12830 |
+
"grad_norm": 3.679232120513916,
|
| 12831 |
+
"learning_rate": 1.0973172862100145e-06,
|
| 12832 |
+
"loss": 0.899,
|
| 12833 |
+
"step": 1832
|
| 12834 |
+
},
|
| 12835 |
+
{
|
| 12836 |
+
"epoch": 0.9371165644171779,
|
| 12837 |
+
"grad_norm": 3.9870572090148926,
|
| 12838 |
+
"learning_rate": 1.0797537076756127e-06,
|
| 12839 |
+
"loss": 0.9478,
|
| 12840 |
+
"step": 1833
|
| 12841 |
+
},
|
| 12842 |
+
{
|
| 12843 |
+
"epoch": 0.9376278118609407,
|
| 12844 |
+
"grad_norm": 3.767942190170288,
|
| 12845 |
+
"learning_rate": 1.0623302918018108e-06,
|
| 12846 |
+
"loss": 0.8134,
|
| 12847 |
+
"step": 1834
|
| 12848 |
+
},
|
| 12849 |
+
{
|
| 12850 |
+
"epoch": 0.9381390593047034,
|
| 12851 |
+
"grad_norm": 4.18915319442749,
|
| 12852 |
+
"learning_rate": 1.0450470885088937e-06,
|
| 12853 |
+
"loss": 0.9433,
|
| 12854 |
+
"step": 1835
|
| 12855 |
+
},
|
| 12856 |
+
{
|
| 12857 |
+
"epoch": 0.9386503067484663,
|
| 12858 |
+
"grad_norm": 4.106919765472412,
|
| 12859 |
+
"learning_rate": 1.0279041473154116e-06,
|
| 12860 |
+
"loss": 0.8786,
|
| 12861 |
+
"step": 1836
|
| 12862 |
+
},
|
| 12863 |
+
{
|
| 12864 |
+
"epoch": 0.939161554192229,
|
| 12865 |
+
"grad_norm": 4.100769519805908,
|
| 12866 |
+
"learning_rate": 1.010901517338042e-06,
|
| 12867 |
+
"loss": 0.8459,
|
| 12868 |
+
"step": 1837
|
| 12869 |
+
},
|
| 12870 |
+
{
|
| 12871 |
+
"epoch": 0.9396728016359919,
|
| 12872 |
+
"grad_norm": 3.8171629905700684,
|
| 12873 |
+
"learning_rate": 9.94039247291456e-07,
|
| 12874 |
+
"loss": 0.7617,
|
| 12875 |
+
"step": 1838
|
| 12876 |
+
},
|
| 12877 |
+
{
|
| 12878 |
+
"epoch": 0.9401840490797546,
|
| 12879 |
+
"grad_norm": 4.406624794006348,
|
| 12880 |
+
"learning_rate": 9.773173854881913e-07,
|
| 12881 |
+
"loss": 0.7923,
|
| 12882 |
+
"step": 1839
|
| 12883 |
+
},
|
| 12884 |
+
{
|
| 12885 |
+
"epoch": 0.9406952965235174,
|
| 12886 |
+
"grad_norm": 4.20837926864624,
|
| 12887 |
+
"learning_rate": 9.607359798384785e-07,
|
| 12888 |
+
"loss": 0.9773,
|
| 12889 |
+
"step": 1840
|
| 12890 |
+
},
|
| 12891 |
+
{
|
| 12892 |
+
"epoch": 0.9412065439672802,
|
| 12893 |
+
"grad_norm": 4.484842777252197,
|
| 12894 |
+
"learning_rate": 9.442950778501325e-07,
|
| 12895 |
+
"loss": 0.8711,
|
| 12896 |
+
"step": 1841
|
| 12897 |
+
},
|
| 12898 |
+
{
|
| 12899 |
+
"epoch": 0.941717791411043,
|
| 12900 |
+
"grad_norm": 4.719925403594971,
|
| 12901 |
+
"learning_rate": 9.279947266284061e-07,
|
| 12902 |
+
"loss": 0.8392,
|
| 12903 |
+
"step": 1842
|
| 12904 |
+
},
|
| 12905 |
+
{
|
| 12906 |
+
"epoch": 0.9422290388548057,
|
| 12907 |
+
"grad_norm": 4.476531982421875,
|
| 12908 |
+
"learning_rate": 9.118349728758468e-07,
|
| 12909 |
+
"loss": 0.767,
|
| 12910 |
+
"step": 1843
|
| 12911 |
+
},
|
| 12912 |
+
{
|
| 12913 |
+
"epoch": 0.9427402862985685,
|
| 12914 |
+
"grad_norm": 4.48267126083374,
|
| 12915 |
+
"learning_rate": 8.958158628922019e-07,
|
| 12916 |
+
"loss": 0.6777,
|
| 12917 |
+
"step": 1844
|
| 12918 |
+
},
|
| 12919 |
+
{
|
| 12920 |
+
"epoch": 0.9432515337423313,
|
| 12921 |
+
"grad_norm": 4.858860015869141,
|
| 12922 |
+
"learning_rate": 8.799374425742246e-07,
|
| 12923 |
+
"loss": 0.7469,
|
| 12924 |
+
"step": 1845
|
| 12925 |
+
},
|
| 12926 |
+
{
|
| 12927 |
+
"epoch": 0.943762781186094,
|
| 12928 |
+
"grad_norm": 4.577967166900635,
|
| 12929 |
+
"learning_rate": 8.641997574155846e-07,
|
| 12930 |
+
"loss": 0.676,
|
| 12931 |
+
"step": 1846
|
| 12932 |
+
},
|
| 12933 |
+
{
|
| 12934 |
+
"epoch": 0.9442740286298569,
|
| 12935 |
+
"grad_norm": 5.150705337524414,
|
| 12936 |
+
"learning_rate": 8.486028525067358e-07,
|
| 12937 |
+
"loss": 0.7106,
|
| 12938 |
+
"step": 1847
|
| 12939 |
+
},
|
| 12940 |
+
{
|
| 12941 |
+
"epoch": 0.9447852760736196,
|
| 12942 |
+
"grad_norm": 4.760867595672607,
|
| 12943 |
+
"learning_rate": 8.331467725347708e-07,
|
| 12944 |
+
"loss": 0.7528,
|
| 12945 |
+
"step": 1848
|
| 12946 |
+
},
|
| 12947 |
+
{
|
| 12948 |
+
"epoch": 0.9452965235173824,
|
| 12949 |
+
"grad_norm": 6.085768222808838,
|
| 12950 |
+
"learning_rate": 8.178315617832999e-07,
|
| 12951 |
+
"loss": 0.784,
|
| 12952 |
+
"step": 1849
|
| 12953 |
+
},
|
| 12954 |
+
{
|
| 12955 |
+
"epoch": 0.9458077709611452,
|
| 12956 |
+
"grad_norm": 6.886128902435303,
|
| 12957 |
+
"learning_rate": 8.026572641323393e-07,
|
| 12958 |
+
"loss": 0.5051,
|
| 12959 |
+
"step": 1850
|
| 12960 |
+
},
|
| 12961 |
+
{
|
| 12962 |
+
"epoch": 0.946319018404908,
|
| 12963 |
+
"grad_norm": 2.127772808074951,
|
| 12964 |
+
"learning_rate": 7.876239230581506e-07,
|
| 12965 |
+
"loss": 0.9763,
|
| 12966 |
+
"step": 1851
|
| 12967 |
+
},
|
| 12968 |
+
{
|
| 12969 |
+
"epoch": 0.9468302658486708,
|
| 12970 |
+
"grad_norm": 2.3633391857147217,
|
| 12971 |
+
"learning_rate": 7.727315816331515e-07,
|
| 12972 |
+
"loss": 1.012,
|
| 12973 |
+
"step": 1852
|
| 12974 |
+
},
|
| 12975 |
+
{
|
| 12976 |
+
"epoch": 0.9473415132924335,
|
| 12977 |
+
"grad_norm": 2.347792625427246,
|
| 12978 |
+
"learning_rate": 7.579802825257775e-07,
|
| 12979 |
+
"loss": 0.9064,
|
| 12980 |
+
"step": 1853
|
| 12981 |
+
},
|
| 12982 |
+
{
|
| 12983 |
+
"epoch": 0.9478527607361963,
|
| 12984 |
+
"grad_norm": 2.4126203060150146,
|
| 12985 |
+
"learning_rate": 7.43370068000343e-07,
|
| 12986 |
+
"loss": 0.8778,
|
| 12987 |
+
"step": 1854
|
| 12988 |
+
},
|
| 12989 |
+
{
|
| 12990 |
+
"epoch": 0.9483640081799591,
|
| 12991 |
+
"grad_norm": 2.6466588973999023,
|
| 12992 |
+
"learning_rate": 7.289009799169688e-07,
|
| 12993 |
+
"loss": 0.9782,
|
| 12994 |
+
"step": 1855
|
| 12995 |
+
},
|
| 12996 |
+
{
|
| 12997 |
+
"epoch": 0.9488752556237219,
|
| 12998 |
+
"grad_norm": 2.6971752643585205,
|
| 12999 |
+
"learning_rate": 7.145730597314049e-07,
|
| 13000 |
+
"loss": 0.979,
|
| 13001 |
+
"step": 1856
|
| 13002 |
+
},
|
| 13003 |
+
{
|
| 13004 |
+
"epoch": 0.9493865030674846,
|
| 13005 |
+
"grad_norm": 2.6623589992523193,
|
| 13006 |
+
"learning_rate": 7.003863484949413e-07,
|
| 13007 |
+
"loss": 0.9151,
|
| 13008 |
+
"step": 1857
|
| 13009 |
+
},
|
| 13010 |
+
{
|
| 13011 |
+
"epoch": 0.9498977505112475,
|
| 13012 |
+
"grad_norm": 2.865541696548462,
|
| 13013 |
+
"learning_rate": 6.86340886854292e-07,
|
| 13014 |
+
"loss": 0.8949,
|
| 13015 |
+
"step": 1858
|
| 13016 |
+
},
|
| 13017 |
+
{
|
| 13018 |
+
"epoch": 0.9504089979550102,
|
| 13019 |
+
"grad_norm": 3.0056710243225098,
|
| 13020 |
+
"learning_rate": 6.724367150514777e-07,
|
| 13021 |
+
"loss": 0.8936,
|
| 13022 |
+
"step": 1859
|
| 13023 |
+
},
|
| 13024 |
+
{
|
| 13025 |
+
"epoch": 0.950920245398773,
|
| 13026 |
+
"grad_norm": 2.805217742919922,
|
| 13027 |
+
"learning_rate": 6.58673872923693e-07,
|
| 13028 |
+
"loss": 0.831,
|
| 13029 |
+
"step": 1860
|
| 13030 |
+
},
|
| 13031 |
+
{
|
| 13032 |
+
"epoch": 0.9514314928425358,
|
| 13033 |
+
"grad_norm": 3.1351966857910156,
|
| 13034 |
+
"learning_rate": 6.450523999032177e-07,
|
| 13035 |
+
"loss": 1.0224,
|
| 13036 |
+
"step": 1861
|
| 13037 |
+
},
|
| 13038 |
+
{
|
| 13039 |
+
"epoch": 0.9519427402862985,
|
| 13040 |
+
"grad_norm": 2.93005633354187,
|
| 13041 |
+
"learning_rate": 6.315723350172775e-07,
|
| 13042 |
+
"loss": 0.8546,
|
| 13043 |
+
"step": 1862
|
| 13044 |
+
},
|
| 13045 |
+
{
|
| 13046 |
+
"epoch": 0.9524539877300614,
|
| 13047 |
+
"grad_norm": 3.181835651397705,
|
| 13048 |
+
"learning_rate": 6.182337168879671e-07,
|
| 13049 |
+
"loss": 0.903,
|
| 13050 |
+
"step": 1863
|
| 13051 |
+
},
|
| 13052 |
+
{
|
| 13053 |
+
"epoch": 0.9529652351738241,
|
| 13054 |
+
"grad_norm": 3.401155948638916,
|
| 13055 |
+
"learning_rate": 6.050365837320992e-07,
|
| 13056 |
+
"loss": 0.9181,
|
| 13057 |
+
"step": 1864
|
| 13058 |
+
},
|
| 13059 |
+
{
|
| 13060 |
+
"epoch": 0.9534764826175869,
|
| 13061 |
+
"grad_norm": 3.0578250885009766,
|
| 13062 |
+
"learning_rate": 5.919809733611171e-07,
|
| 13063 |
+
"loss": 0.8978,
|
| 13064 |
+
"step": 1865
|
| 13065 |
+
},
|
| 13066 |
+
{
|
| 13067 |
+
"epoch": 0.9539877300613497,
|
| 13068 |
+
"grad_norm": 3.142280101776123,
|
| 13069 |
+
"learning_rate": 5.790669231809875e-07,
|
| 13070 |
+
"loss": 0.9351,
|
| 13071 |
+
"step": 1866
|
| 13072 |
+
},
|
| 13073 |
+
{
|
| 13074 |
+
"epoch": 0.9544989775051125,
|
| 13075 |
+
"grad_norm": 3.1850273609161377,
|
| 13076 |
+
"learning_rate": 5.66294470192097e-07,
|
| 13077 |
+
"loss": 0.9621,
|
| 13078 |
+
"step": 1867
|
| 13079 |
+
},
|
| 13080 |
+
{
|
| 13081 |
+
"epoch": 0.9550102249488752,
|
| 13082 |
+
"grad_norm": 3.274036169052124,
|
| 13083 |
+
"learning_rate": 5.536636509891225e-07,
|
| 13084 |
+
"loss": 0.8428,
|
| 13085 |
+
"step": 1868
|
| 13086 |
+
},
|
| 13087 |
+
{
|
| 13088 |
+
"epoch": 0.9555214723926381,
|
| 13089 |
+
"grad_norm": 3.323293924331665,
|
| 13090 |
+
"learning_rate": 5.411745017609493e-07,
|
| 13091 |
+
"loss": 0.9623,
|
| 13092 |
+
"step": 1869
|
| 13093 |
+
},
|
| 13094 |
+
{
|
| 13095 |
+
"epoch": 0.9560327198364008,
|
| 13096 |
+
"grad_norm": 3.296380043029785,
|
| 13097 |
+
"learning_rate": 5.288270582905708e-07,
|
| 13098 |
+
"loss": 0.86,
|
| 13099 |
+
"step": 1870
|
| 13100 |
+
},
|
| 13101 |
+
{
|
| 13102 |
+
"epoch": 0.9565439672801636,
|
| 13103 |
+
"grad_norm": 3.7464077472686768,
|
| 13104 |
+
"learning_rate": 5.166213559549549e-07,
|
| 13105 |
+
"loss": 0.8665,
|
| 13106 |
+
"step": 1871
|
| 13107 |
+
},
|
| 13108 |
+
{
|
| 13109 |
+
"epoch": 0.9570552147239264,
|
| 13110 |
+
"grad_norm": 3.5171990394592285,
|
| 13111 |
+
"learning_rate": 5.045574297249833e-07,
|
| 13112 |
+
"loss": 0.9063,
|
| 13113 |
+
"step": 1872
|
| 13114 |
+
},
|
| 13115 |
+
{
|
| 13116 |
+
"epoch": 0.9575664621676891,
|
| 13117 |
+
"grad_norm": 3.684037685394287,
|
| 13118 |
+
"learning_rate": 4.926353141653184e-07,
|
| 13119 |
+
"loss": 0.9079,
|
| 13120 |
+
"step": 1873
|
| 13121 |
+
},
|
| 13122 |
+
{
|
| 13123 |
+
"epoch": 0.9580777096114519,
|
| 13124 |
+
"grad_norm": 3.424814462661743,
|
| 13125 |
+
"learning_rate": 4.80855043434325e-07,
|
| 13126 |
+
"loss": 0.8508,
|
| 13127 |
+
"step": 1874
|
| 13128 |
+
},
|
| 13129 |
+
{
|
| 13130 |
+
"epoch": 0.9585889570552147,
|
| 13131 |
+
"grad_norm": 3.3258156776428223,
|
| 13132 |
+
"learning_rate": 4.692166512839491e-07,
|
| 13133 |
+
"loss": 0.8615,
|
| 13134 |
+
"step": 1875
|
| 13135 |
+
},
|
| 13136 |
+
{
|
| 13137 |
+
"epoch": 0.9591002044989775,
|
| 13138 |
+
"grad_norm": 3.7672901153564453,
|
| 13139 |
+
"learning_rate": 4.577201710596612e-07,
|
| 13140 |
+
"loss": 0.9087,
|
| 13141 |
+
"step": 1876
|
| 13142 |
+
},
|
| 13143 |
+
{
|
| 13144 |
+
"epoch": 0.9596114519427403,
|
| 13145 |
+
"grad_norm": 3.638936996459961,
|
| 13146 |
+
"learning_rate": 4.4636563570031873e-07,
|
| 13147 |
+
"loss": 0.921,
|
| 13148 |
+
"step": 1877
|
| 13149 |
+
},
|
| 13150 |
+
{
|
| 13151 |
+
"epoch": 0.9601226993865031,
|
| 13152 |
+
"grad_norm": 3.7481777667999268,
|
| 13153 |
+
"learning_rate": 4.3515307773809855e-07,
|
| 13154 |
+
"loss": 0.8998,
|
| 13155 |
+
"step": 1878
|
| 13156 |
+
},
|
| 13157 |
+
{
|
| 13158 |
+
"epoch": 0.9606339468302658,
|
| 13159 |
+
"grad_norm": 4.0111212730407715,
|
| 13160 |
+
"learning_rate": 4.240825292983808e-07,
|
| 13161 |
+
"loss": 0.8379,
|
| 13162 |
+
"step": 1879
|
| 13163 |
+
},
|
| 13164 |
+
{
|
| 13165 |
+
"epoch": 0.9611451942740287,
|
| 13166 |
+
"grad_norm": 3.8063735961914062,
|
| 13167 |
+
"learning_rate": 4.131540220996877e-07,
|
| 13168 |
+
"loss": 1.0216,
|
| 13169 |
+
"step": 1880
|
| 13170 |
+
},
|
| 13171 |
+
{
|
| 13172 |
+
"epoch": 0.9616564417177914,
|
| 13173 |
+
"grad_norm": 3.818110942840576,
|
| 13174 |
+
"learning_rate": 4.023675874535671e-07,
|
| 13175 |
+
"loss": 0.9189,
|
| 13176 |
+
"step": 1881
|
| 13177 |
+
},
|
| 13178 |
+
{
|
| 13179 |
+
"epoch": 0.9621676891615542,
|
| 13180 |
+
"grad_norm": 3.307727098464966,
|
| 13181 |
+
"learning_rate": 3.917232562645035e-07,
|
| 13182 |
+
"loss": 0.7723,
|
| 13183 |
+
"step": 1882
|
| 13184 |
+
},
|
| 13185 |
+
{
|
| 13186 |
+
"epoch": 0.962678936605317,
|
| 13187 |
+
"grad_norm": 3.7070207595825195,
|
| 13188 |
+
"learning_rate": 3.812210590298515e-07,
|
| 13189 |
+
"loss": 0.8208,
|
| 13190 |
+
"step": 1883
|
| 13191 |
+
},
|
| 13192 |
+
{
|
| 13193 |
+
"epoch": 0.9631901840490797,
|
| 13194 |
+
"grad_norm": 3.999943971633911,
|
| 13195 |
+
"learning_rate": 3.7086102583972494e-07,
|
| 13196 |
+
"loss": 0.9011,
|
| 13197 |
+
"step": 1884
|
| 13198 |
+
},
|
| 13199 |
+
{
|
| 13200 |
+
"epoch": 0.9637014314928425,
|
| 13201 |
+
"grad_norm": 3.9105939865112305,
|
| 13202 |
+
"learning_rate": 3.6064318637693e-07,
|
| 13203 |
+
"loss": 0.9319,
|
| 13204 |
+
"step": 1885
|
| 13205 |
+
},
|
| 13206 |
+
{
|
| 13207 |
+
"epoch": 0.9642126789366053,
|
| 13208 |
+
"grad_norm": 3.4953696727752686,
|
| 13209 |
+
"learning_rate": 3.505675699168487e-07,
|
| 13210 |
+
"loss": 0.836,
|
| 13211 |
+
"step": 1886
|
| 13212 |
+
},
|
| 13213 |
+
{
|
| 13214 |
+
"epoch": 0.9647239263803681,
|
| 13215 |
+
"grad_norm": 4.094799995422363,
|
| 13216 |
+
"learning_rate": 3.406342053274003e-07,
|
| 13217 |
+
"loss": 0.7909,
|
| 13218 |
+
"step": 1887
|
| 13219 |
+
},
|
| 13220 |
+
{
|
| 13221 |
+
"epoch": 0.9652351738241309,
|
| 13222 |
+
"grad_norm": 3.863297700881958,
|
| 13223 |
+
"learning_rate": 3.3084312106892446e-07,
|
| 13224 |
+
"loss": 0.8657,
|
| 13225 |
+
"step": 1888
|
| 13226 |
+
},
|
| 13227 |
+
{
|
| 13228 |
+
"epoch": 0.9657464212678937,
|
| 13229 |
+
"grad_norm": 4.07185173034668,
|
| 13230 |
+
"learning_rate": 3.211943451941035e-07,
|
| 13231 |
+
"loss": 0.8233,
|
| 13232 |
+
"step": 1889
|
| 13233 |
+
},
|
| 13234 |
+
{
|
| 13235 |
+
"epoch": 0.9662576687116564,
|
| 13236 |
+
"grad_norm": 4.35256814956665,
|
| 13237 |
+
"learning_rate": 3.1168790534789605e-07,
|
| 13238 |
+
"loss": 0.9713,
|
| 13239 |
+
"step": 1890
|
| 13240 |
+
},
|
| 13241 |
+
{
|
| 13242 |
+
"epoch": 0.9667689161554193,
|
| 13243 |
+
"grad_norm": 3.9577109813690186,
|
| 13244 |
+
"learning_rate": 3.023238287674479e-07,
|
| 13245 |
+
"loss": 0.8875,
|
| 13246 |
+
"step": 1891
|
| 13247 |
+
},
|
| 13248 |
+
{
|
| 13249 |
+
"epoch": 0.967280163599182,
|
| 13250 |
+
"grad_norm": 4.198614597320557,
|
| 13251 |
+
"learning_rate": 2.9310214228202013e-07,
|
| 13252 |
+
"loss": 0.7918,
|
| 13253 |
+
"step": 1892
|
| 13254 |
+
},
|
| 13255 |
+
{
|
| 13256 |
+
"epoch": 0.9677914110429447,
|
| 13257 |
+
"grad_norm": 4.020992279052734,
|
| 13258 |
+
"learning_rate": 2.840228723129001e-07,
|
| 13259 |
+
"loss": 0.7969,
|
| 13260 |
+
"step": 1893
|
| 13261 |
+
},
|
| 13262 |
+
{
|
| 13263 |
+
"epoch": 0.9683026584867076,
|
| 13264 |
+
"grad_norm": 4.857772350311279,
|
| 13265 |
+
"learning_rate": 2.750860448733461e-07,
|
| 13266 |
+
"loss": 0.7989,
|
| 13267 |
+
"step": 1894
|
| 13268 |
+
},
|
| 13269 |
+
{
|
| 13270 |
+
"epoch": 0.9688139059304703,
|
| 13271 |
+
"grad_norm": 4.296475887298584,
|
| 13272 |
+
"learning_rate": 2.662916855684816e-07,
|
| 13273 |
+
"loss": 0.7896,
|
| 13274 |
+
"step": 1895
|
| 13275 |
+
},
|
| 13276 |
+
{
|
| 13277 |
+
"epoch": 0.9693251533742331,
|
| 13278 |
+
"grad_norm": 4.845669269561768,
|
| 13279 |
+
"learning_rate": 2.5763981959526786e-07,
|
| 13280 |
+
"loss": 0.9473,
|
| 13281 |
+
"step": 1896
|
| 13282 |
+
},
|
| 13283 |
+
{
|
| 13284 |
+
"epoch": 0.9698364008179959,
|
| 13285 |
+
"grad_norm": 5.276214599609375,
|
| 13286 |
+
"learning_rate": 2.4913047174237035e-07,
|
| 13287 |
+
"loss": 0.8389,
|
| 13288 |
+
"step": 1897
|
| 13289 |
+
},
|
| 13290 |
+
{
|
| 13291 |
+
"epoch": 0.9703476482617587,
|
| 13292 |
+
"grad_norm": 5.026844501495361,
|
| 13293 |
+
"learning_rate": 2.407636663901591e-07,
|
| 13294 |
+
"loss": 0.6934,
|
| 13295 |
+
"step": 1898
|
| 13296 |
+
},
|
| 13297 |
+
{
|
| 13298 |
+
"epoch": 0.9708588957055214,
|
| 13299 |
+
"grad_norm": 5.138829231262207,
|
| 13300 |
+
"learning_rate": 2.3253942751056968e-07,
|
| 13301 |
+
"loss": 0.5662,
|
| 13302 |
+
"step": 1899
|
| 13303 |
+
},
|
| 13304 |
+
{
|
| 13305 |
+
"epoch": 0.9713701431492843,
|
| 13306 |
+
"grad_norm": 8.76285457611084,
|
| 13307 |
+
"learning_rate": 2.2445777866709205e-07,
|
| 13308 |
+
"loss": 0.8929,
|
| 13309 |
+
"step": 1900
|
| 13310 |
+
},
|
| 13311 |
+
{
|
| 13312 |
+
"epoch": 0.971881390593047,
|
| 13313 |
+
"grad_norm": 2.1116087436676025,
|
| 13314 |
+
"learning_rate": 2.1651874301465979e-07,
|
| 13315 |
+
"loss": 0.8929,
|
| 13316 |
+
"step": 1901
|
| 13317 |
+
},
|
| 13318 |
+
{
|
| 13319 |
+
"epoch": 0.9723926380368099,
|
| 13320 |
+
"grad_norm": 2.1844208240509033,
|
| 13321 |
+
"learning_rate": 2.087223432996166e-07,
|
| 13322 |
+
"loss": 0.8937,
|
| 13323 |
+
"step": 1902
|
| 13324 |
+
},
|
| 13325 |
+
{
|
| 13326 |
+
"epoch": 0.9729038854805726,
|
| 13327 |
+
"grad_norm": 2.286012887954712,
|
| 13328 |
+
"learning_rate": 2.0106860185962194e-07,
|
| 13329 |
+
"loss": 0.8114,
|
| 13330 |
+
"step": 1903
|
| 13331 |
+
},
|
| 13332 |
+
{
|
| 13333 |
+
"epoch": 0.9734151329243353,
|
| 13334 |
+
"grad_norm": 2.505859613418579,
|
| 13335 |
+
"learning_rate": 1.935575406236123e-07,
|
| 13336 |
+
"loss": 0.9787,
|
| 13337 |
+
"step": 1904
|
| 13338 |
+
},
|
| 13339 |
+
{
|
| 13340 |
+
"epoch": 0.9739263803680982,
|
| 13341 |
+
"grad_norm": 2.708432674407959,
|
| 13342 |
+
"learning_rate": 1.861891811117178e-07,
|
| 13343 |
+
"loss": 0.9853,
|
| 13344 |
+
"step": 1905
|
| 13345 |
+
},
|
| 13346 |
+
{
|
| 13347 |
+
"epoch": 0.9744376278118609,
|
| 13348 |
+
"grad_norm": 2.945042610168457,
|
| 13349 |
+
"learning_rate": 1.7896354443521778e-07,
|
| 13350 |
+
"loss": 0.9986,
|
| 13351 |
+
"step": 1906
|
| 13352 |
+
},
|
| 13353 |
+
{
|
| 13354 |
+
"epoch": 0.9749488752556237,
|
| 13355 |
+
"grad_norm": 2.5554847717285156,
|
| 13356 |
+
"learning_rate": 1.7188065129647435e-07,
|
| 13357 |
+
"loss": 0.9626,
|
| 13358 |
+
"step": 1907
|
| 13359 |
+
},
|
| 13360 |
+
{
|
| 13361 |
+
"epoch": 0.9754601226993865,
|
| 13362 |
+
"grad_norm": 2.838167190551758,
|
| 13363 |
+
"learning_rate": 1.6494052198886555e-07,
|
| 13364 |
+
"loss": 0.9229,
|
| 13365 |
+
"step": 1908
|
| 13366 |
+
},
|
| 13367 |
+
{
|
| 13368 |
+
"epoch": 0.9759713701431493,
|
| 13369 |
+
"grad_norm": 2.6479218006134033,
|
| 13370 |
+
"learning_rate": 1.5814317639673005e-07,
|
| 13371 |
+
"loss": 0.8479,
|
| 13372 |
+
"step": 1909
|
| 13373 |
+
},
|
| 13374 |
+
{
|
| 13375 |
+
"epoch": 0.976482617586912,
|
| 13376 |
+
"grad_norm": 3.0106728076934814,
|
| 13377 |
+
"learning_rate": 1.5148863399532254e-07,
|
| 13378 |
+
"loss": 0.8836,
|
| 13379 |
+
"step": 1910
|
| 13380 |
+
},
|
| 13381 |
+
{
|
| 13382 |
+
"epoch": 0.9769938650306749,
|
| 13383 |
+
"grad_norm": 3.02433705329895,
|
| 13384 |
+
"learning_rate": 1.4497691385074175e-07,
|
| 13385 |
+
"loss": 0.9304,
|
| 13386 |
+
"step": 1911
|
| 13387 |
+
},
|
| 13388 |
+
{
|
| 13389 |
+
"epoch": 0.9775051124744376,
|
| 13390 |
+
"grad_norm": 3.085106134414673,
|
| 13391 |
+
"learning_rate": 1.3860803461989146e-07,
|
| 13392 |
+
"loss": 0.8676,
|
| 13393 |
+
"step": 1912
|
| 13394 |
}
|
| 13395 |
],
|
| 13396 |
"logging_steps": 1,
|
|
|
|
| 13410 |
"attributes": {}
|
| 13411 |
}
|
| 13412 |
},
|
| 13413 |
+
"total_flos": 7.310480002095514e+17,
|
| 13414 |
"train_batch_size": 8,
|
| 13415 |
"trial_name": null,
|
| 13416 |
"trial_params": null
|