Training in progress, step 1371, 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 137714904
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83da40a9154613a638d1fdae5980fa35c0d809fd92c3a4927c462d08f897d627
|
| 3 |
size 137714904
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 70674644
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d93aa50a0e61430fcc810f6fefe5a083c6f8c03dfc57b8c49e577afe7b421a7
|
| 3 |
size 70674644
|
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 sha256:8e6390c92d64390f8337877d3c9863e3759ab2f814e191b929f8e60f0f1a489f
|
| 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:b70c66781e1a6941a34654d4b8193be447c36adcde318297ffbdeb2c57bc7975
|
| 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,
|
|
@@ -7210,6 +7210,2400 @@
|
|
| 7210 |
"learning_rate": 1.4685340263950558e-05,
|
| 7211 |
"loss": 0.6155,
|
| 7212 |
"step": 1029
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7213 |
}
|
| 7214 |
],
|
| 7215 |
"logging_steps": 1,
|
|
@@ -7224,12 +9618,12 @@
|
|
| 7224 |
"should_evaluate": false,
|
| 7225 |
"should_log": false,
|
| 7226 |
"should_save": true,
|
| 7227 |
-
"should_training_stop":
|
| 7228 |
},
|
| 7229 |
"attributes": {}
|
| 7230 |
}
|
| 7231 |
},
|
| 7232 |
-
"total_flos": 1.
|
| 7233 |
"train_batch_size": 4,
|
| 7234 |
"trial_name": null,
|
| 7235 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.2313631185925832,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1371,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7210 |
"learning_rate": 1.4685340263950558e-05,
|
| 7211 |
"loss": 0.6155,
|
| 7212 |
"step": 1029
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 0.17381766021178754,
|
| 7216 |
+
"grad_norm": 1.0319809913635254,
|
| 7217 |
+
"learning_rate": 1.4604028368654037e-05,
|
| 7218 |
+
"loss": 0.5403,
|
| 7219 |
+
"step": 1030
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 0.17398641522170188,
|
| 7223 |
+
"grad_norm": 1.1701350212097168,
|
| 7224 |
+
"learning_rate": 1.4522903693250905e-05,
|
| 7225 |
+
"loss": 0.5764,
|
| 7226 |
+
"step": 1031
|
| 7227 |
+
},
|
| 7228 |
+
{
|
| 7229 |
+
"epoch": 0.17415517023161625,
|
| 7230 |
+
"grad_norm": 1.1631454229354858,
|
| 7231 |
+
"learning_rate": 1.444196666683389e-05,
|
| 7232 |
+
"loss": 0.5583,
|
| 7233 |
+
"step": 1032
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 0.17432392524153062,
|
| 7237 |
+
"grad_norm": 1.08467435836792,
|
| 7238 |
+
"learning_rate": 1.4361217717503145e-05,
|
| 7239 |
+
"loss": 0.5135,
|
| 7240 |
+
"step": 1033
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 0.17449268025144496,
|
| 7244 |
+
"grad_norm": 1.1960421800613403,
|
| 7245 |
+
"learning_rate": 1.4280657272364035e-05,
|
| 7246 |
+
"loss": 0.5745,
|
| 7247 |
+
"step": 1034
|
| 7248 |
+
},
|
| 7249 |
+
{
|
| 7250 |
+
"epoch": 0.17466143526135933,
|
| 7251 |
+
"grad_norm": 1.4493260383605957,
|
| 7252 |
+
"learning_rate": 1.4200285757524895e-05,
|
| 7253 |
+
"loss": 0.7502,
|
| 7254 |
+
"step": 1035
|
| 7255 |
+
},
|
| 7256 |
+
{
|
| 7257 |
+
"epoch": 0.17483019027127367,
|
| 7258 |
+
"grad_norm": 1.250547170639038,
|
| 7259 |
+
"learning_rate": 1.412010359809472e-05,
|
| 7260 |
+
"loss": 0.637,
|
| 7261 |
+
"step": 1036
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 0.17499894528118803,
|
| 7265 |
+
"grad_norm": 1.2481725215911865,
|
| 7266 |
+
"learning_rate": 1.4040111218180967e-05,
|
| 7267 |
+
"loss": 0.6221,
|
| 7268 |
+
"step": 1037
|
| 7269 |
+
},
|
| 7270 |
+
{
|
| 7271 |
+
"epoch": 0.1751677002911024,
|
| 7272 |
+
"grad_norm": 1.3377397060394287,
|
| 7273 |
+
"learning_rate": 1.3960309040887309e-05,
|
| 7274 |
+
"loss": 0.7129,
|
| 7275 |
+
"step": 1038
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 0.17533645530101674,
|
| 7279 |
+
"grad_norm": 1.292580008506775,
|
| 7280 |
+
"learning_rate": 1.3880697488311329e-05,
|
| 7281 |
+
"loss": 0.6901,
|
| 7282 |
+
"step": 1039
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 0.1755052103109311,
|
| 7286 |
+
"grad_norm": 1.2922450304031372,
|
| 7287 |
+
"learning_rate": 1.3801276981542388e-05,
|
| 7288 |
+
"loss": 0.6255,
|
| 7289 |
+
"step": 1040
|
| 7290 |
+
},
|
| 7291 |
+
{
|
| 7292 |
+
"epoch": 0.17567396532084545,
|
| 7293 |
+
"grad_norm": 1.4294100999832153,
|
| 7294 |
+
"learning_rate": 1.3722047940659328e-05,
|
| 7295 |
+
"loss": 0.708,
|
| 7296 |
+
"step": 1041
|
| 7297 |
+
},
|
| 7298 |
+
{
|
| 7299 |
+
"epoch": 0.17584272033075982,
|
| 7300 |
+
"grad_norm": 1.4303593635559082,
|
| 7301 |
+
"learning_rate": 1.3643010784728245e-05,
|
| 7302 |
+
"loss": 0.6473,
|
| 7303 |
+
"step": 1042
|
| 7304 |
+
},
|
| 7305 |
+
{
|
| 7306 |
+
"epoch": 0.17601147534067418,
|
| 7307 |
+
"grad_norm": 1.499804139137268,
|
| 7308 |
+
"learning_rate": 1.356416593180036e-05,
|
| 7309 |
+
"loss": 0.7227,
|
| 7310 |
+
"step": 1043
|
| 7311 |
+
},
|
| 7312 |
+
{
|
| 7313 |
+
"epoch": 0.17618023035058852,
|
| 7314 |
+
"grad_norm": 1.4877763986587524,
|
| 7315 |
+
"learning_rate": 1.34855137989097e-05,
|
| 7316 |
+
"loss": 0.7049,
|
| 7317 |
+
"step": 1044
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 0.1763489853605029,
|
| 7321 |
+
"grad_norm": 1.396185040473938,
|
| 7322 |
+
"learning_rate": 1.3407054802070924e-05,
|
| 7323 |
+
"loss": 0.6869,
|
| 7324 |
+
"step": 1045
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 0.17651774037041726,
|
| 7328 |
+
"grad_norm": 1.8223285675048828,
|
| 7329 |
+
"learning_rate": 1.3328789356277194e-05,
|
| 7330 |
+
"loss": 0.7735,
|
| 7331 |
+
"step": 1046
|
| 7332 |
+
},
|
| 7333 |
+
{
|
| 7334 |
+
"epoch": 0.1766864953803316,
|
| 7335 |
+
"grad_norm": 1.9497090578079224,
|
| 7336 |
+
"learning_rate": 1.3250717875497864e-05,
|
| 7337 |
+
"loss": 0.808,
|
| 7338 |
+
"step": 1047
|
| 7339 |
+
},
|
| 7340 |
+
{
|
| 7341 |
+
"epoch": 0.17685525039024597,
|
| 7342 |
+
"grad_norm": 1.8972411155700684,
|
| 7343 |
+
"learning_rate": 1.3172840772676387e-05,
|
| 7344 |
+
"loss": 0.8421,
|
| 7345 |
+
"step": 1048
|
| 7346 |
+
},
|
| 7347 |
+
{
|
| 7348 |
+
"epoch": 0.1770240054001603,
|
| 7349 |
+
"grad_norm": 1.9117318391799927,
|
| 7350 |
+
"learning_rate": 1.309515845972809e-05,
|
| 7351 |
+
"loss": 0.8546,
|
| 7352 |
+
"step": 1049
|
| 7353 |
+
},
|
| 7354 |
+
{
|
| 7355 |
+
"epoch": 0.17719276041007467,
|
| 7356 |
+
"grad_norm": 2.5941343307495117,
|
| 7357 |
+
"learning_rate": 1.3017671347537969e-05,
|
| 7358 |
+
"loss": 1.081,
|
| 7359 |
+
"step": 1050
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 0.17736151541998904,
|
| 7363 |
+
"grad_norm": 0.4715406000614166,
|
| 7364 |
+
"learning_rate": 1.294037984595859e-05,
|
| 7365 |
+
"loss": 0.5136,
|
| 7366 |
+
"step": 1051
|
| 7367 |
+
},
|
| 7368 |
+
{
|
| 7369 |
+
"epoch": 0.17753027042990338,
|
| 7370 |
+
"grad_norm": 0.5191971659660339,
|
| 7371 |
+
"learning_rate": 1.2863284363807831e-05,
|
| 7372 |
+
"loss": 0.5824,
|
| 7373 |
+
"step": 1052
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 0.17769902543981775,
|
| 7377 |
+
"grad_norm": 0.5693216919898987,
|
| 7378 |
+
"learning_rate": 1.2786385308866772e-05,
|
| 7379 |
+
"loss": 0.5587,
|
| 7380 |
+
"step": 1053
|
| 7381 |
+
},
|
| 7382 |
+
{
|
| 7383 |
+
"epoch": 0.1778677804497321,
|
| 7384 |
+
"grad_norm": 0.6372774243354797,
|
| 7385 |
+
"learning_rate": 1.2709683087877573e-05,
|
| 7386 |
+
"loss": 0.6168,
|
| 7387 |
+
"step": 1054
|
| 7388 |
+
},
|
| 7389 |
+
{
|
| 7390 |
+
"epoch": 0.17803653545964646,
|
| 7391 |
+
"grad_norm": 0.6607767343521118,
|
| 7392 |
+
"learning_rate": 1.2633178106541216e-05,
|
| 7393 |
+
"loss": 0.5187,
|
| 7394 |
+
"step": 1055
|
| 7395 |
+
},
|
| 7396 |
+
{
|
| 7397 |
+
"epoch": 0.17820529046956082,
|
| 7398 |
+
"grad_norm": 0.6850890517234802,
|
| 7399 |
+
"learning_rate": 1.2556870769515493e-05,
|
| 7400 |
+
"loss": 0.5701,
|
| 7401 |
+
"step": 1056
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 0.17837404547947516,
|
| 7405 |
+
"grad_norm": 0.7265092730522156,
|
| 7406 |
+
"learning_rate": 1.248076148041274e-05,
|
| 7407 |
+
"loss": 0.6247,
|
| 7408 |
+
"step": 1057
|
| 7409 |
+
},
|
| 7410 |
+
{
|
| 7411 |
+
"epoch": 0.17854280048938953,
|
| 7412 |
+
"grad_norm": 0.6663525700569153,
|
| 7413 |
+
"learning_rate": 1.240485064179776e-05,
|
| 7414 |
+
"loss": 0.5468,
|
| 7415 |
+
"step": 1058
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 0.1787115554993039,
|
| 7419 |
+
"grad_norm": 0.7386102676391602,
|
| 7420 |
+
"learning_rate": 1.2329138655185735e-05,
|
| 7421 |
+
"loss": 0.5882,
|
| 7422 |
+
"step": 1059
|
| 7423 |
+
},
|
| 7424 |
+
{
|
| 7425 |
+
"epoch": 0.17888031050921824,
|
| 7426 |
+
"grad_norm": 0.7752735614776611,
|
| 7427 |
+
"learning_rate": 1.2253625921040046e-05,
|
| 7428 |
+
"loss": 0.6626,
|
| 7429 |
+
"step": 1060
|
| 7430 |
+
},
|
| 7431 |
+
{
|
| 7432 |
+
"epoch": 0.1790490655191326,
|
| 7433 |
+
"grad_norm": 0.8515707850456238,
|
| 7434 |
+
"learning_rate": 1.2178312838770117e-05,
|
| 7435 |
+
"loss": 0.6572,
|
| 7436 |
+
"step": 1061
|
| 7437 |
+
},
|
| 7438 |
+
{
|
| 7439 |
+
"epoch": 0.17921782052904695,
|
| 7440 |
+
"grad_norm": 0.7872768044471741,
|
| 7441 |
+
"learning_rate": 1.2103199806729438e-05,
|
| 7442 |
+
"loss": 0.5535,
|
| 7443 |
+
"step": 1062
|
| 7444 |
+
},
|
| 7445 |
+
{
|
| 7446 |
+
"epoch": 0.1793865755389613,
|
| 7447 |
+
"grad_norm": 0.8094439506530762,
|
| 7448 |
+
"learning_rate": 1.2028287222213286e-05,
|
| 7449 |
+
"loss": 0.4774,
|
| 7450 |
+
"step": 1063
|
| 7451 |
+
},
|
| 7452 |
+
{
|
| 7453 |
+
"epoch": 0.17955533054887568,
|
| 7454 |
+
"grad_norm": 0.8031929731369019,
|
| 7455 |
+
"learning_rate": 1.1953575481456792e-05,
|
| 7456 |
+
"loss": 0.4854,
|
| 7457 |
+
"step": 1064
|
| 7458 |
+
},
|
| 7459 |
+
{
|
| 7460 |
+
"epoch": 0.17972408555879002,
|
| 7461 |
+
"grad_norm": 0.8686875700950623,
|
| 7462 |
+
"learning_rate": 1.1879064979632704e-05,
|
| 7463 |
+
"loss": 0.5507,
|
| 7464 |
+
"step": 1065
|
| 7465 |
+
},
|
| 7466 |
+
{
|
| 7467 |
+
"epoch": 0.1798928405687044,
|
| 7468 |
+
"grad_norm": 0.9079012274742126,
|
| 7469 |
+
"learning_rate": 1.1804756110849363e-05,
|
| 7470 |
+
"loss": 0.6092,
|
| 7471 |
+
"step": 1066
|
| 7472 |
+
},
|
| 7473 |
+
{
|
| 7474 |
+
"epoch": 0.18006159557861873,
|
| 7475 |
+
"grad_norm": 0.9290295243263245,
|
| 7476 |
+
"learning_rate": 1.1730649268148664e-05,
|
| 7477 |
+
"loss": 0.5932,
|
| 7478 |
+
"step": 1067
|
| 7479 |
+
},
|
| 7480 |
+
{
|
| 7481 |
+
"epoch": 0.1802303505885331,
|
| 7482 |
+
"grad_norm": 0.9496090412139893,
|
| 7483 |
+
"learning_rate": 1.1656744843503874e-05,
|
| 7484 |
+
"loss": 0.5311,
|
| 7485 |
+
"step": 1068
|
| 7486 |
+
},
|
| 7487 |
+
{
|
| 7488 |
+
"epoch": 0.18039910559844746,
|
| 7489 |
+
"grad_norm": 0.9013460278511047,
|
| 7490 |
+
"learning_rate": 1.1583043227817609e-05,
|
| 7491 |
+
"loss": 0.5556,
|
| 7492 |
+
"step": 1069
|
| 7493 |
+
},
|
| 7494 |
+
{
|
| 7495 |
+
"epoch": 0.1805678606083618,
|
| 7496 |
+
"grad_norm": 0.9510790109634399,
|
| 7497 |
+
"learning_rate": 1.1509544810919798e-05,
|
| 7498 |
+
"loss": 0.5625,
|
| 7499 |
+
"step": 1070
|
| 7500 |
+
},
|
| 7501 |
+
{
|
| 7502 |
+
"epoch": 0.18073661561827617,
|
| 7503 |
+
"grad_norm": 0.9159001111984253,
|
| 7504 |
+
"learning_rate": 1.1436249981565577e-05,
|
| 7505 |
+
"loss": 0.5722,
|
| 7506 |
+
"step": 1071
|
| 7507 |
+
},
|
| 7508 |
+
{
|
| 7509 |
+
"epoch": 0.18090537062819054,
|
| 7510 |
+
"grad_norm": 0.9898829460144043,
|
| 7511 |
+
"learning_rate": 1.1363159127433265e-05,
|
| 7512 |
+
"loss": 0.6594,
|
| 7513 |
+
"step": 1072
|
| 7514 |
+
},
|
| 7515 |
+
{
|
| 7516 |
+
"epoch": 0.18107412563810488,
|
| 7517 |
+
"grad_norm": 0.8602355718612671,
|
| 7518 |
+
"learning_rate": 1.1290272635122256e-05,
|
| 7519 |
+
"loss": 0.412,
|
| 7520 |
+
"step": 1073
|
| 7521 |
+
},
|
| 7522 |
+
{
|
| 7523 |
+
"epoch": 0.18124288064801924,
|
| 7524 |
+
"grad_norm": 0.9539443254470825,
|
| 7525 |
+
"learning_rate": 1.121759089015103e-05,
|
| 7526 |
+
"loss": 0.5553,
|
| 7527 |
+
"step": 1074
|
| 7528 |
+
},
|
| 7529 |
+
{
|
| 7530 |
+
"epoch": 0.18141163565793358,
|
| 7531 |
+
"grad_norm": 0.9541051387786865,
|
| 7532 |
+
"learning_rate": 1.1145114276955121e-05,
|
| 7533 |
+
"loss": 0.5062,
|
| 7534 |
+
"step": 1075
|
| 7535 |
+
},
|
| 7536 |
+
{
|
| 7537 |
+
"epoch": 0.18158039066784795,
|
| 7538 |
+
"grad_norm": 0.982779324054718,
|
| 7539 |
+
"learning_rate": 1.1072843178885023e-05,
|
| 7540 |
+
"loss": 0.5422,
|
| 7541 |
+
"step": 1076
|
| 7542 |
+
},
|
| 7543 |
+
{
|
| 7544 |
+
"epoch": 0.18174914567776232,
|
| 7545 |
+
"grad_norm": 1.096030831336975,
|
| 7546 |
+
"learning_rate": 1.1000777978204214e-05,
|
| 7547 |
+
"loss": 0.501,
|
| 7548 |
+
"step": 1077
|
| 7549 |
+
},
|
| 7550 |
+
{
|
| 7551 |
+
"epoch": 0.18191790068767666,
|
| 7552 |
+
"grad_norm": 1.1027871370315552,
|
| 7553 |
+
"learning_rate": 1.0928919056087134e-05,
|
| 7554 |
+
"loss": 0.5381,
|
| 7555 |
+
"step": 1078
|
| 7556 |
+
},
|
| 7557 |
+
{
|
| 7558 |
+
"epoch": 0.18208665569759103,
|
| 7559 |
+
"grad_norm": 1.1163361072540283,
|
| 7560 |
+
"learning_rate": 1.0857266792617122e-05,
|
| 7561 |
+
"loss": 0.6133,
|
| 7562 |
+
"step": 1079
|
| 7563 |
+
},
|
| 7564 |
+
{
|
| 7565 |
+
"epoch": 0.18225541070750537,
|
| 7566 |
+
"grad_norm": 1.085625410079956,
|
| 7567 |
+
"learning_rate": 1.0785821566784482e-05,
|
| 7568 |
+
"loss": 0.5697,
|
| 7569 |
+
"step": 1080
|
| 7570 |
+
},
|
| 7571 |
+
{
|
| 7572 |
+
"epoch": 0.18242416571741973,
|
| 7573 |
+
"grad_norm": 1.2379124164581299,
|
| 7574 |
+
"learning_rate": 1.071458375648438e-05,
|
| 7575 |
+
"loss": 0.6964,
|
| 7576 |
+
"step": 1081
|
| 7577 |
+
},
|
| 7578 |
+
{
|
| 7579 |
+
"epoch": 0.1825929207273341,
|
| 7580 |
+
"grad_norm": 1.1840894222259521,
|
| 7581 |
+
"learning_rate": 1.0643553738514939e-05,
|
| 7582 |
+
"loss": 0.6278,
|
| 7583 |
+
"step": 1082
|
| 7584 |
+
},
|
| 7585 |
+
{
|
| 7586 |
+
"epoch": 0.18276167573724844,
|
| 7587 |
+
"grad_norm": 1.2528572082519531,
|
| 7588 |
+
"learning_rate": 1.057273188857521e-05,
|
| 7589 |
+
"loss": 0.5366,
|
| 7590 |
+
"step": 1083
|
| 7591 |
+
},
|
| 7592 |
+
{
|
| 7593 |
+
"epoch": 0.1829304307471628,
|
| 7594 |
+
"grad_norm": 1.1213513612747192,
|
| 7595 |
+
"learning_rate": 1.0502118581263148e-05,
|
| 7596 |
+
"loss": 0.4806,
|
| 7597 |
+
"step": 1084
|
| 7598 |
+
},
|
| 7599 |
+
{
|
| 7600 |
+
"epoch": 0.18309918575707718,
|
| 7601 |
+
"grad_norm": 1.1215614080429077,
|
| 7602 |
+
"learning_rate": 1.0431714190073672e-05,
|
| 7603 |
+
"loss": 0.4659,
|
| 7604 |
+
"step": 1085
|
| 7605 |
+
},
|
| 7606 |
+
{
|
| 7607 |
+
"epoch": 0.18326794076699152,
|
| 7608 |
+
"grad_norm": 1.3397492170333862,
|
| 7609 |
+
"learning_rate": 1.0361519087396715e-05,
|
| 7610 |
+
"loss": 0.6759,
|
| 7611 |
+
"step": 1086
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 0.18343669577690588,
|
| 7615 |
+
"grad_norm": 1.2844408750534058,
|
| 7616 |
+
"learning_rate": 1.0291533644515167e-05,
|
| 7617 |
+
"loss": 0.668,
|
| 7618 |
+
"step": 1087
|
| 7619 |
+
},
|
| 7620 |
+
{
|
| 7621 |
+
"epoch": 0.18360545078682022,
|
| 7622 |
+
"grad_norm": 1.2681591510772705,
|
| 7623 |
+
"learning_rate": 1.0221758231603024e-05,
|
| 7624 |
+
"loss": 0.6039,
|
| 7625 |
+
"step": 1088
|
| 7626 |
+
},
|
| 7627 |
+
{
|
| 7628 |
+
"epoch": 0.1837742057967346,
|
| 7629 |
+
"grad_norm": 1.201204776763916,
|
| 7630 |
+
"learning_rate": 1.0152193217723316e-05,
|
| 7631 |
+
"loss": 0.5369,
|
| 7632 |
+
"step": 1089
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 0.18394296080664896,
|
| 7636 |
+
"grad_norm": 1.266699194908142,
|
| 7637 |
+
"learning_rate": 1.008283897082623e-05,
|
| 7638 |
+
"loss": 0.5924,
|
| 7639 |
+
"step": 1090
|
| 7640 |
+
},
|
| 7641 |
+
{
|
| 7642 |
+
"epoch": 0.1841117158165633,
|
| 7643 |
+
"grad_norm": 1.433534860610962,
|
| 7644 |
+
"learning_rate": 1.0013695857747174e-05,
|
| 7645 |
+
"loss": 0.6841,
|
| 7646 |
+
"step": 1091
|
| 7647 |
+
},
|
| 7648 |
+
{
|
| 7649 |
+
"epoch": 0.18428047082647767,
|
| 7650 |
+
"grad_norm": 1.545581579208374,
|
| 7651 |
+
"learning_rate": 9.94476424420474e-06,
|
| 7652 |
+
"loss": 0.6661,
|
| 7653 |
+
"step": 1092
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 0.184449225836392,
|
| 7657 |
+
"grad_norm": 1.4983086585998535,
|
| 7658 |
+
"learning_rate": 9.876044494798898e-06,
|
| 7659 |
+
"loss": 0.7183,
|
| 7660 |
+
"step": 1093
|
| 7661 |
+
},
|
| 7662 |
+
{
|
| 7663 |
+
"epoch": 0.18461798084630637,
|
| 7664 |
+
"grad_norm": 1.6680443286895752,
|
| 7665 |
+
"learning_rate": 9.807536973008985e-06,
|
| 7666 |
+
"loss": 0.6989,
|
| 7667 |
+
"step": 1094
|
| 7668 |
+
},
|
| 7669 |
+
{
|
| 7670 |
+
"epoch": 0.18478673585622074,
|
| 7671 |
+
"grad_norm": 1.5908889770507812,
|
| 7672 |
+
"learning_rate": 9.739242041191781e-06,
|
| 7673 |
+
"loss": 0.8823,
|
| 7674 |
+
"step": 1095
|
| 7675 |
+
},
|
| 7676 |
+
{
|
| 7677 |
+
"epoch": 0.18495549086613508,
|
| 7678 |
+
"grad_norm": 1.6729180812835693,
|
| 7679 |
+
"learning_rate": 9.671160060579648e-06,
|
| 7680 |
+
"loss": 0.6335,
|
| 7681 |
+
"step": 1096
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 0.18512424587604945,
|
| 7685 |
+
"grad_norm": 1.7792624235153198,
|
| 7686 |
+
"learning_rate": 9.603291391278568e-06,
|
| 7687 |
+
"loss": 0.7025,
|
| 7688 |
+
"step": 1097
|
| 7689 |
+
},
|
| 7690 |
+
{
|
| 7691 |
+
"epoch": 0.18529300088596382,
|
| 7692 |
+
"grad_norm": 1.8781849145889282,
|
| 7693 |
+
"learning_rate": 9.535636392266234e-06,
|
| 7694 |
+
"loss": 0.8383,
|
| 7695 |
+
"step": 1098
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 0.18546175589587816,
|
| 7699 |
+
"grad_norm": 1.8984248638153076,
|
| 7700 |
+
"learning_rate": 9.46819542139023e-06,
|
| 7701 |
+
"loss": 0.6684,
|
| 7702 |
+
"step": 1099
|
| 7703 |
+
},
|
| 7704 |
+
{
|
| 7705 |
+
"epoch": 0.18563051090579252,
|
| 7706 |
+
"grad_norm": 2.2846457958221436,
|
| 7707 |
+
"learning_rate": 9.40096883536602e-06,
|
| 7708 |
+
"loss": 0.7748,
|
| 7709 |
+
"step": 1100
|
| 7710 |
+
},
|
| 7711 |
+
{
|
| 7712 |
+
"epoch": 0.18579926591570686,
|
| 7713 |
+
"grad_norm": 0.4258973002433777,
|
| 7714 |
+
"learning_rate": 9.33395698977515e-06,
|
| 7715 |
+
"loss": 0.5428,
|
| 7716 |
+
"step": 1101
|
| 7717 |
+
},
|
| 7718 |
+
{
|
| 7719 |
+
"epoch": 0.18596802092562123,
|
| 7720 |
+
"grad_norm": 0.5763824582099915,
|
| 7721 |
+
"learning_rate": 9.267160239063356e-06,
|
| 7722 |
+
"loss": 0.6497,
|
| 7723 |
+
"step": 1102
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 0.1861367759355356,
|
| 7727 |
+
"grad_norm": 0.5722477436065674,
|
| 7728 |
+
"learning_rate": 9.200578936538629e-06,
|
| 7729 |
+
"loss": 0.5187,
|
| 7730 |
+
"step": 1103
|
| 7731 |
+
},
|
| 7732 |
+
{
|
| 7733 |
+
"epoch": 0.18630553094544994,
|
| 7734 |
+
"grad_norm": 0.6180588006973267,
|
| 7735 |
+
"learning_rate": 9.134213434369426e-06,
|
| 7736 |
+
"loss": 0.583,
|
| 7737 |
+
"step": 1104
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 0.1864742859553643,
|
| 7741 |
+
"grad_norm": 0.6868796348571777,
|
| 7742 |
+
"learning_rate": 9.068064083582789e-06,
|
| 7743 |
+
"loss": 0.5695,
|
| 7744 |
+
"step": 1105
|
| 7745 |
+
},
|
| 7746 |
+
{
|
| 7747 |
+
"epoch": 0.18664304096527864,
|
| 7748 |
+
"grad_norm": 0.7807045578956604,
|
| 7749 |
+
"learning_rate": 9.0021312340624e-06,
|
| 7750 |
+
"loss": 0.6661,
|
| 7751 |
+
"step": 1106
|
| 7752 |
+
},
|
| 7753 |
+
{
|
| 7754 |
+
"epoch": 0.186811795975193,
|
| 7755 |
+
"grad_norm": 0.7304069995880127,
|
| 7756 |
+
"learning_rate": 8.93641523454688e-06,
|
| 7757 |
+
"loss": 0.5707,
|
| 7758 |
+
"step": 1107
|
| 7759 |
+
},
|
| 7760 |
+
{
|
| 7761 |
+
"epoch": 0.18698055098510738,
|
| 7762 |
+
"grad_norm": 0.774566650390625,
|
| 7763 |
+
"learning_rate": 8.870916432627813e-06,
|
| 7764 |
+
"loss": 0.6282,
|
| 7765 |
+
"step": 1108
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 0.18714930599502172,
|
| 7769 |
+
"grad_norm": 0.7647138833999634,
|
| 7770 |
+
"learning_rate": 8.805635174747962e-06,
|
| 7771 |
+
"loss": 0.5705,
|
| 7772 |
+
"step": 1109
|
| 7773 |
+
},
|
| 7774 |
+
{
|
| 7775 |
+
"epoch": 0.1873180610049361,
|
| 7776 |
+
"grad_norm": 0.8842149972915649,
|
| 7777 |
+
"learning_rate": 8.740571806199477e-06,
|
| 7778 |
+
"loss": 0.6233,
|
| 7779 |
+
"step": 1110
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 0.18748681601485043,
|
| 7783 |
+
"grad_norm": 0.8789076805114746,
|
| 7784 |
+
"learning_rate": 8.67572667112197e-06,
|
| 7785 |
+
"loss": 0.6155,
|
| 7786 |
+
"step": 1111
|
| 7787 |
+
},
|
| 7788 |
+
{
|
| 7789 |
+
"epoch": 0.1876555710247648,
|
| 7790 |
+
"grad_norm": 0.8563818335533142,
|
| 7791 |
+
"learning_rate": 8.611100112500808e-06,
|
| 7792 |
+
"loss": 0.583,
|
| 7793 |
+
"step": 1112
|
| 7794 |
+
},
|
| 7795 |
+
{
|
| 7796 |
+
"epoch": 0.18782432603467916,
|
| 7797 |
+
"grad_norm": 0.7744738459587097,
|
| 7798 |
+
"learning_rate": 8.546692472165196e-06,
|
| 7799 |
+
"loss": 0.5526,
|
| 7800 |
+
"step": 1113
|
| 7801 |
+
},
|
| 7802 |
+
{
|
| 7803 |
+
"epoch": 0.1879930810445935,
|
| 7804 |
+
"grad_norm": 1.013342022895813,
|
| 7805 |
+
"learning_rate": 8.48250409078642e-06,
|
| 7806 |
+
"loss": 0.6404,
|
| 7807 |
+
"step": 1114
|
| 7808 |
+
},
|
| 7809 |
+
{
|
| 7810 |
+
"epoch": 0.18816183605450787,
|
| 7811 |
+
"grad_norm": 0.8554475903511047,
|
| 7812 |
+
"learning_rate": 8.418535307876058e-06,
|
| 7813 |
+
"loss": 0.5184,
|
| 7814 |
+
"step": 1115
|
| 7815 |
+
},
|
| 7816 |
+
{
|
| 7817 |
+
"epoch": 0.18833059106442224,
|
| 7818 |
+
"grad_norm": 0.8989919424057007,
|
| 7819 |
+
"learning_rate": 8.354786461784175e-06,
|
| 7820 |
+
"loss": 0.6057,
|
| 7821 |
+
"step": 1116
|
| 7822 |
+
},
|
| 7823 |
+
{
|
| 7824 |
+
"epoch": 0.18849934607433658,
|
| 7825 |
+
"grad_norm": 0.9024717807769775,
|
| 7826 |
+
"learning_rate": 8.291257889697484e-06,
|
| 7827 |
+
"loss": 0.4968,
|
| 7828 |
+
"step": 1117
|
| 7829 |
+
},
|
| 7830 |
+
{
|
| 7831 |
+
"epoch": 0.18866810108425094,
|
| 7832 |
+
"grad_norm": 0.9358378648757935,
|
| 7833 |
+
"learning_rate": 8.227949927637651e-06,
|
| 7834 |
+
"loss": 0.4916,
|
| 7835 |
+
"step": 1118
|
| 7836 |
+
},
|
| 7837 |
+
{
|
| 7838 |
+
"epoch": 0.18883685609416528,
|
| 7839 |
+
"grad_norm": 0.953552782535553,
|
| 7840 |
+
"learning_rate": 8.16486291045942e-06,
|
| 7841 |
+
"loss": 0.4996,
|
| 7842 |
+
"step": 1119
|
| 7843 |
+
},
|
| 7844 |
+
{
|
| 7845 |
+
"epoch": 0.18900561110407965,
|
| 7846 |
+
"grad_norm": 1.07173752784729,
|
| 7847 |
+
"learning_rate": 8.101997171848941e-06,
|
| 7848 |
+
"loss": 0.5665,
|
| 7849 |
+
"step": 1120
|
| 7850 |
+
},
|
| 7851 |
+
{
|
| 7852 |
+
"epoch": 0.18917436611399402,
|
| 7853 |
+
"grad_norm": 0.9952334761619568,
|
| 7854 |
+
"learning_rate": 8.039353044321918e-06,
|
| 7855 |
+
"loss": 0.4867,
|
| 7856 |
+
"step": 1121
|
| 7857 |
+
},
|
| 7858 |
+
{
|
| 7859 |
+
"epoch": 0.18934312112390836,
|
| 7860 |
+
"grad_norm": 1.0533599853515625,
|
| 7861 |
+
"learning_rate": 7.976930859221887e-06,
|
| 7862 |
+
"loss": 0.5156,
|
| 7863 |
+
"step": 1122
|
| 7864 |
+
},
|
| 7865 |
+
{
|
| 7866 |
+
"epoch": 0.18951187613382273,
|
| 7867 |
+
"grad_norm": 1.1125342845916748,
|
| 7868 |
+
"learning_rate": 7.914730946718508e-06,
|
| 7869 |
+
"loss": 0.6671,
|
| 7870 |
+
"step": 1123
|
| 7871 |
+
},
|
| 7872 |
+
{
|
| 7873 |
+
"epoch": 0.18968063114373707,
|
| 7874 |
+
"grad_norm": 1.0971851348876953,
|
| 7875 |
+
"learning_rate": 7.85275363580572e-06,
|
| 7876 |
+
"loss": 0.585,
|
| 7877 |
+
"step": 1124
|
| 7878 |
+
},
|
| 7879 |
+
{
|
| 7880 |
+
"epoch": 0.18984938615365143,
|
| 7881 |
+
"grad_norm": 1.0686278343200684,
|
| 7882 |
+
"learning_rate": 7.79099925430008e-06,
|
| 7883 |
+
"loss": 0.56,
|
| 7884 |
+
"step": 1125
|
| 7885 |
+
},
|
| 7886 |
+
{
|
| 7887 |
+
"epoch": 0.1900181411635658,
|
| 7888 |
+
"grad_norm": 1.0261794328689575,
|
| 7889 |
+
"learning_rate": 7.729468128839007e-06,
|
| 7890 |
+
"loss": 0.5615,
|
| 7891 |
+
"step": 1126
|
| 7892 |
+
},
|
| 7893 |
+
{
|
| 7894 |
+
"epoch": 0.19018689617348014,
|
| 7895 |
+
"grad_norm": 1.0555497407913208,
|
| 7896 |
+
"learning_rate": 7.66816058487907e-06,
|
| 7897 |
+
"loss": 0.5144,
|
| 7898 |
+
"step": 1127
|
| 7899 |
+
},
|
| 7900 |
+
{
|
| 7901 |
+
"epoch": 0.1903556511833945,
|
| 7902 |
+
"grad_norm": 1.1610735654830933,
|
| 7903 |
+
"learning_rate": 7.607076946694202e-06,
|
| 7904 |
+
"loss": 0.5332,
|
| 7905 |
+
"step": 1128
|
| 7906 |
+
},
|
| 7907 |
+
{
|
| 7908 |
+
"epoch": 0.19052440619330888,
|
| 7909 |
+
"grad_norm": 1.1955081224441528,
|
| 7910 |
+
"learning_rate": 7.546217537374073e-06,
|
| 7911 |
+
"loss": 0.6426,
|
| 7912 |
+
"step": 1129
|
| 7913 |
+
},
|
| 7914 |
+
{
|
| 7915 |
+
"epoch": 0.19069316120322322,
|
| 7916 |
+
"grad_norm": 1.1631394624710083,
|
| 7917 |
+
"learning_rate": 7.485582678822306e-06,
|
| 7918 |
+
"loss": 0.5139,
|
| 7919 |
+
"step": 1130
|
| 7920 |
+
},
|
| 7921 |
+
{
|
| 7922 |
+
"epoch": 0.19086191621313758,
|
| 7923 |
+
"grad_norm": 1.2080230712890625,
|
| 7924 |
+
"learning_rate": 7.42517269175485e-06,
|
| 7925 |
+
"loss": 0.5853,
|
| 7926 |
+
"step": 1131
|
| 7927 |
+
},
|
| 7928 |
+
{
|
| 7929 |
+
"epoch": 0.19103067122305192,
|
| 7930 |
+
"grad_norm": 1.1186884641647339,
|
| 7931 |
+
"learning_rate": 7.3649878956982055e-06,
|
| 7932 |
+
"loss": 0.5308,
|
| 7933 |
+
"step": 1132
|
| 7934 |
+
},
|
| 7935 |
+
{
|
| 7936 |
+
"epoch": 0.1911994262329663,
|
| 7937 |
+
"grad_norm": 1.2984037399291992,
|
| 7938 |
+
"learning_rate": 7.305028608987763e-06,
|
| 7939 |
+
"loss": 0.687,
|
| 7940 |
+
"step": 1133
|
| 7941 |
+
},
|
| 7942 |
+
{
|
| 7943 |
+
"epoch": 0.19136818124288066,
|
| 7944 |
+
"grad_norm": 1.2159451246261597,
|
| 7945 |
+
"learning_rate": 7.245295148766173e-06,
|
| 7946 |
+
"loss": 0.5976,
|
| 7947 |
+
"step": 1134
|
| 7948 |
+
},
|
| 7949 |
+
{
|
| 7950 |
+
"epoch": 0.191536936252795,
|
| 7951 |
+
"grad_norm": 1.2357873916625977,
|
| 7952 |
+
"learning_rate": 7.185787830981571e-06,
|
| 7953 |
+
"loss": 0.5822,
|
| 7954 |
+
"step": 1135
|
| 7955 |
+
},
|
| 7956 |
+
{
|
| 7957 |
+
"epoch": 0.19170569126270937,
|
| 7958 |
+
"grad_norm": 1.2717851400375366,
|
| 7959 |
+
"learning_rate": 7.126506970386005e-06,
|
| 7960 |
+
"loss": 0.5941,
|
| 7961 |
+
"step": 1136
|
| 7962 |
+
},
|
| 7963 |
+
{
|
| 7964 |
+
"epoch": 0.1918744462726237,
|
| 7965 |
+
"grad_norm": 1.5165656805038452,
|
| 7966 |
+
"learning_rate": 7.0674528805336645e-06,
|
| 7967 |
+
"loss": 0.6845,
|
| 7968 |
+
"step": 1137
|
| 7969 |
+
},
|
| 7970 |
+
{
|
| 7971 |
+
"epoch": 0.19204320128253807,
|
| 7972 |
+
"grad_norm": 1.2401753664016724,
|
| 7973 |
+
"learning_rate": 7.008625873779356e-06,
|
| 7974 |
+
"loss": 0.5117,
|
| 7975 |
+
"step": 1138
|
| 7976 |
+
},
|
| 7977 |
+
{
|
| 7978 |
+
"epoch": 0.19221195629245244,
|
| 7979 |
+
"grad_norm": 1.247753381729126,
|
| 7980 |
+
"learning_rate": 6.950026261276698e-06,
|
| 7981 |
+
"loss": 0.5991,
|
| 7982 |
+
"step": 1139
|
| 7983 |
+
},
|
| 7984 |
+
{
|
| 7985 |
+
"epoch": 0.19238071130236678,
|
| 7986 |
+
"grad_norm": 1.3807613849639893,
|
| 7987 |
+
"learning_rate": 6.891654352976612e-06,
|
| 7988 |
+
"loss": 0.6656,
|
| 7989 |
+
"step": 1140
|
| 7990 |
+
},
|
| 7991 |
+
{
|
| 7992 |
+
"epoch": 0.19254946631228115,
|
| 7993 |
+
"grad_norm": 1.4891915321350098,
|
| 7994 |
+
"learning_rate": 6.833510457625586e-06,
|
| 7995 |
+
"loss": 0.6281,
|
| 7996 |
+
"step": 1141
|
| 7997 |
+
},
|
| 7998 |
+
{
|
| 7999 |
+
"epoch": 0.19271822132219552,
|
| 8000 |
+
"grad_norm": 1.439781665802002,
|
| 8001 |
+
"learning_rate": 6.775594882764108e-06,
|
| 8002 |
+
"loss": 0.5462,
|
| 8003 |
+
"step": 1142
|
| 8004 |
+
},
|
| 8005 |
+
{
|
| 8006 |
+
"epoch": 0.19288697633210986,
|
| 8007 |
+
"grad_norm": 1.380350112915039,
|
| 8008 |
+
"learning_rate": 6.717907934724982e-06,
|
| 8009 |
+
"loss": 0.5989,
|
| 8010 |
+
"step": 1143
|
| 8011 |
+
},
|
| 8012 |
+
{
|
| 8013 |
+
"epoch": 0.19305573134202422,
|
| 8014 |
+
"grad_norm": 1.393906593322754,
|
| 8015 |
+
"learning_rate": 6.660449918631762e-06,
|
| 8016 |
+
"loss": 0.5525,
|
| 8017 |
+
"step": 1144
|
| 8018 |
+
},
|
| 8019 |
+
{
|
| 8020 |
+
"epoch": 0.19322448635193856,
|
| 8021 |
+
"grad_norm": 1.5263879299163818,
|
| 8022 |
+
"learning_rate": 6.603221138397103e-06,
|
| 8023 |
+
"loss": 0.6114,
|
| 8024 |
+
"step": 1145
|
| 8025 |
+
},
|
| 8026 |
+
{
|
| 8027 |
+
"epoch": 0.19339324136185293,
|
| 8028 |
+
"grad_norm": 1.6503808498382568,
|
| 8029 |
+
"learning_rate": 6.546221896721138e-06,
|
| 8030 |
+
"loss": 0.6674,
|
| 8031 |
+
"step": 1146
|
| 8032 |
+
},
|
| 8033 |
+
{
|
| 8034 |
+
"epoch": 0.1935619963717673,
|
| 8035 |
+
"grad_norm": 1.6666897535324097,
|
| 8036 |
+
"learning_rate": 6.48945249508996e-06,
|
| 8037 |
+
"loss": 0.6877,
|
| 8038 |
+
"step": 1147
|
| 8039 |
+
},
|
| 8040 |
+
{
|
| 8041 |
+
"epoch": 0.19373075138168164,
|
| 8042 |
+
"grad_norm": 1.6416525840759277,
|
| 8043 |
+
"learning_rate": 6.432913233773913e-06,
|
| 8044 |
+
"loss": 0.5663,
|
| 8045 |
+
"step": 1148
|
| 8046 |
+
},
|
| 8047 |
+
{
|
| 8048 |
+
"epoch": 0.193899506391596,
|
| 8049 |
+
"grad_norm": 1.962034821510315,
|
| 8050 |
+
"learning_rate": 6.376604411826071e-06,
|
| 8051 |
+
"loss": 0.6805,
|
| 8052 |
+
"step": 1149
|
| 8053 |
+
},
|
| 8054 |
+
{
|
| 8055 |
+
"epoch": 0.19406826140151034,
|
| 8056 |
+
"grad_norm": 2.3873307704925537,
|
| 8057 |
+
"learning_rate": 6.3205263270806665e-06,
|
| 8058 |
+
"loss": 0.8197,
|
| 8059 |
+
"step": 1150
|
| 8060 |
+
},
|
| 8061 |
+
{
|
| 8062 |
+
"epoch": 0.1942370164114247,
|
| 8063 |
+
"grad_norm": 0.4620276093482971,
|
| 8064 |
+
"learning_rate": 6.264679276151486e-06,
|
| 8065 |
+
"loss": 0.6271,
|
| 8066 |
+
"step": 1151
|
| 8067 |
+
},
|
| 8068 |
+
{
|
| 8069 |
+
"epoch": 0.19440577142133908,
|
| 8070 |
+
"grad_norm": 0.5824121236801147,
|
| 8071 |
+
"learning_rate": 6.2090635544303115e-06,
|
| 8072 |
+
"loss": 0.5112,
|
| 8073 |
+
"step": 1152
|
| 8074 |
+
},
|
| 8075 |
+
{
|
| 8076 |
+
"epoch": 0.19457452643125342,
|
| 8077 |
+
"grad_norm": 0.5734692811965942,
|
| 8078 |
+
"learning_rate": 6.153679456085343e-06,
|
| 8079 |
+
"loss": 0.5177,
|
| 8080 |
+
"step": 1153
|
| 8081 |
+
},
|
| 8082 |
+
{
|
| 8083 |
+
"epoch": 0.1947432814411678,
|
| 8084 |
+
"grad_norm": 0.6562824249267578,
|
| 8085 |
+
"learning_rate": 6.098527274059651e-06,
|
| 8086 |
+
"loss": 0.5824,
|
| 8087 |
+
"step": 1154
|
| 8088 |
+
},
|
| 8089 |
+
{
|
| 8090 |
+
"epoch": 0.19491203645108215,
|
| 8091 |
+
"grad_norm": 0.6037938594818115,
|
| 8092 |
+
"learning_rate": 6.043607300069654e-06,
|
| 8093 |
+
"loss": 0.4827,
|
| 8094 |
+
"step": 1155
|
| 8095 |
+
},
|
| 8096 |
+
{
|
| 8097 |
+
"epoch": 0.1950807914609965,
|
| 8098 |
+
"grad_norm": 0.7912740111351013,
|
| 8099 |
+
"learning_rate": 5.988919824603545e-06,
|
| 8100 |
+
"loss": 0.6638,
|
| 8101 |
+
"step": 1156
|
| 8102 |
+
},
|
| 8103 |
+
{
|
| 8104 |
+
"epoch": 0.19524954647091086,
|
| 8105 |
+
"grad_norm": 0.7995325326919556,
|
| 8106 |
+
"learning_rate": 5.934465136919737e-06,
|
| 8107 |
+
"loss": 0.6254,
|
| 8108 |
+
"step": 1157
|
| 8109 |
+
},
|
| 8110 |
+
{
|
| 8111 |
+
"epoch": 0.1954183014808252,
|
| 8112 |
+
"grad_norm": 0.7708985805511475,
|
| 8113 |
+
"learning_rate": 5.880243525045398e-06,
|
| 8114 |
+
"loss": 0.6051,
|
| 8115 |
+
"step": 1158
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 0.19558705649073957,
|
| 8119 |
+
"grad_norm": 0.8647124171257019,
|
| 8120 |
+
"learning_rate": 5.826255275774862e-06,
|
| 8121 |
+
"loss": 0.6107,
|
| 8122 |
+
"step": 1159
|
| 8123 |
+
},
|
| 8124 |
+
{
|
| 8125 |
+
"epoch": 0.19575581150065394,
|
| 8126 |
+
"grad_norm": 0.8052390217781067,
|
| 8127 |
+
"learning_rate": 5.772500674668157e-06,
|
| 8128 |
+
"loss": 0.5808,
|
| 8129 |
+
"step": 1160
|
| 8130 |
+
},
|
| 8131 |
+
{
|
| 8132 |
+
"epoch": 0.19592456651056828,
|
| 8133 |
+
"grad_norm": 0.8364976644515991,
|
| 8134 |
+
"learning_rate": 5.718980006049446e-06,
|
| 8135 |
+
"loss": 0.5398,
|
| 8136 |
+
"step": 1161
|
| 8137 |
+
},
|
| 8138 |
+
{
|
| 8139 |
+
"epoch": 0.19609332152048264,
|
| 8140 |
+
"grad_norm": 0.8715554475784302,
|
| 8141 |
+
"learning_rate": 5.665693553005586e-06,
|
| 8142 |
+
"loss": 0.6081,
|
| 8143 |
+
"step": 1162
|
| 8144 |
+
},
|
| 8145 |
+
{
|
| 8146 |
+
"epoch": 0.19626207653039698,
|
| 8147 |
+
"grad_norm": 0.908320963382721,
|
| 8148 |
+
"learning_rate": 5.612641597384588e-06,
|
| 8149 |
+
"loss": 0.5801,
|
| 8150 |
+
"step": 1163
|
| 8151 |
+
},
|
| 8152 |
+
{
|
| 8153 |
+
"epoch": 0.19643083154031135,
|
| 8154 |
+
"grad_norm": 0.9266404509544373,
|
| 8155 |
+
"learning_rate": 5.559824419794129e-06,
|
| 8156 |
+
"loss": 0.6676,
|
| 8157 |
+
"step": 1164
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 0.19659958655022572,
|
| 8161 |
+
"grad_norm": 0.7900256514549255,
|
| 8162 |
+
"learning_rate": 5.5072422996000625e-06,
|
| 8163 |
+
"loss": 0.4707,
|
| 8164 |
+
"step": 1165
|
| 8165 |
+
},
|
| 8166 |
+
{
|
| 8167 |
+
"epoch": 0.19676834156014006,
|
| 8168 |
+
"grad_norm": 0.8521780967712402,
|
| 8169 |
+
"learning_rate": 5.45489551492499e-06,
|
| 8170 |
+
"loss": 0.5061,
|
| 8171 |
+
"step": 1166
|
| 8172 |
+
},
|
| 8173 |
+
{
|
| 8174 |
+
"epoch": 0.19693709657005443,
|
| 8175 |
+
"grad_norm": 1.0226095914840698,
|
| 8176 |
+
"learning_rate": 5.402784342646711e-06,
|
| 8177 |
+
"loss": 0.6136,
|
| 8178 |
+
"step": 1167
|
| 8179 |
+
},
|
| 8180 |
+
{
|
| 8181 |
+
"epoch": 0.1971058515799688,
|
| 8182 |
+
"grad_norm": 0.9329948425292969,
|
| 8183 |
+
"learning_rate": 5.350909058396836e-06,
|
| 8184 |
+
"loss": 0.5079,
|
| 8185 |
+
"step": 1168
|
| 8186 |
+
},
|
| 8187 |
+
{
|
| 8188 |
+
"epoch": 0.19727460658988313,
|
| 8189 |
+
"grad_norm": 0.9524698257446289,
|
| 8190 |
+
"learning_rate": 5.299269936559276e-06,
|
| 8191 |
+
"loss": 0.504,
|
| 8192 |
+
"step": 1169
|
| 8193 |
+
},
|
| 8194 |
+
{
|
| 8195 |
+
"epoch": 0.1974433615997975,
|
| 8196 |
+
"grad_norm": 0.9691932797431946,
|
| 8197 |
+
"learning_rate": 5.2478672502687955e-06,
|
| 8198 |
+
"loss": 0.5377,
|
| 8199 |
+
"step": 1170
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 0.19761211660971184,
|
| 8203 |
+
"grad_norm": 0.9309558272361755,
|
| 8204 |
+
"learning_rate": 5.196701271409616e-06,
|
| 8205 |
+
"loss": 0.4994,
|
| 8206 |
+
"step": 1171
|
| 8207 |
+
},
|
| 8208 |
+
{
|
| 8209 |
+
"epoch": 0.1977808716196262,
|
| 8210 |
+
"grad_norm": 1.0124459266662598,
|
| 8211 |
+
"learning_rate": 5.145772270613902e-06,
|
| 8212 |
+
"loss": 0.5391,
|
| 8213 |
+
"step": 1172
|
| 8214 |
+
},
|
| 8215 |
+
{
|
| 8216 |
+
"epoch": 0.19794962662954058,
|
| 8217 |
+
"grad_norm": 1.3132346868515015,
|
| 8218 |
+
"learning_rate": 5.0950805172603975e-06,
|
| 8219 |
+
"loss": 0.515,
|
| 8220 |
+
"step": 1173
|
| 8221 |
+
},
|
| 8222 |
+
{
|
| 8223 |
+
"epoch": 0.19811838163945492,
|
| 8224 |
+
"grad_norm": 0.8934838175773621,
|
| 8225 |
+
"learning_rate": 5.044626279472975e-06,
|
| 8226 |
+
"loss": 0.4449,
|
| 8227 |
+
"step": 1174
|
| 8228 |
+
},
|
| 8229 |
+
{
|
| 8230 |
+
"epoch": 0.19828713664936928,
|
| 8231 |
+
"grad_norm": 1.0268865823745728,
|
| 8232 |
+
"learning_rate": 4.9944098241191894e-06,
|
| 8233 |
+
"loss": 0.5443,
|
| 8234 |
+
"step": 1175
|
| 8235 |
+
},
|
| 8236 |
+
{
|
| 8237 |
+
"epoch": 0.19845589165928362,
|
| 8238 |
+
"grad_norm": 0.9692090153694153,
|
| 8239 |
+
"learning_rate": 4.944431416808931e-06,
|
| 8240 |
+
"loss": 0.4432,
|
| 8241 |
+
"step": 1176
|
| 8242 |
+
},
|
| 8243 |
+
{
|
| 8244 |
+
"epoch": 0.198624646669198,
|
| 8245 |
+
"grad_norm": 1.1712415218353271,
|
| 8246 |
+
"learning_rate": 4.894691321892947e-06,
|
| 8247 |
+
"loss": 0.5674,
|
| 8248 |
+
"step": 1177
|
| 8249 |
+
},
|
| 8250 |
+
{
|
| 8251 |
+
"epoch": 0.19879340167911236,
|
| 8252 |
+
"grad_norm": 1.1192066669464111,
|
| 8253 |
+
"learning_rate": 4.845189802461491e-06,
|
| 8254 |
+
"loss": 0.5008,
|
| 8255 |
+
"step": 1178
|
| 8256 |
+
},
|
| 8257 |
+
{
|
| 8258 |
+
"epoch": 0.1989621566890267,
|
| 8259 |
+
"grad_norm": 1.1353431940078735,
|
| 8260 |
+
"learning_rate": 4.795927120342941e-06,
|
| 8261 |
+
"loss": 0.541,
|
| 8262 |
+
"step": 1179
|
| 8263 |
+
},
|
| 8264 |
+
{
|
| 8265 |
+
"epoch": 0.19913091169894107,
|
| 8266 |
+
"grad_norm": 1.1462347507476807,
|
| 8267 |
+
"learning_rate": 4.746903536102359e-06,
|
| 8268 |
+
"loss": 0.5037,
|
| 8269 |
+
"step": 1180
|
| 8270 |
+
},
|
| 8271 |
+
{
|
| 8272 |
+
"epoch": 0.19929966670885543,
|
| 8273 |
+
"grad_norm": 1.2885733842849731,
|
| 8274 |
+
"learning_rate": 4.698119309040161e-06,
|
| 8275 |
+
"loss": 0.5725,
|
| 8276 |
+
"step": 1181
|
| 8277 |
+
},
|
| 8278 |
+
{
|
| 8279 |
+
"epoch": 0.19946842171876977,
|
| 8280 |
+
"grad_norm": 1.2179657220840454,
|
| 8281 |
+
"learning_rate": 4.6495746971907505e-06,
|
| 8282 |
+
"loss": 0.5973,
|
| 8283 |
+
"step": 1182
|
| 8284 |
+
},
|
| 8285 |
+
{
|
| 8286 |
+
"epoch": 0.19963717672868414,
|
| 8287 |
+
"grad_norm": 1.1196248531341553,
|
| 8288 |
+
"learning_rate": 4.601269957321091e-06,
|
| 8289 |
+
"loss": 0.5392,
|
| 8290 |
+
"step": 1183
|
| 8291 |
+
},
|
| 8292 |
+
{
|
| 8293 |
+
"epoch": 0.19980593173859848,
|
| 8294 |
+
"grad_norm": 1.3646570444107056,
|
| 8295 |
+
"learning_rate": 4.553205344929429e-06,
|
| 8296 |
+
"loss": 0.715,
|
| 8297 |
+
"step": 1184
|
| 8298 |
+
},
|
| 8299 |
+
{
|
| 8300 |
+
"epoch": 0.19997468674851285,
|
| 8301 |
+
"grad_norm": 1.1928213834762573,
|
| 8302 |
+
"learning_rate": 4.5053811142439065e-06,
|
| 8303 |
+
"loss": 0.602,
|
| 8304 |
+
"step": 1185
|
| 8305 |
+
},
|
| 8306 |
+
{
|
| 8307 |
+
"epoch": 0.20014344175842722,
|
| 8308 |
+
"grad_norm": 1.301818609237671,
|
| 8309 |
+
"learning_rate": 4.457797518221174e-06,
|
| 8310 |
+
"loss": 0.6285,
|
| 8311 |
+
"step": 1186
|
| 8312 |
+
},
|
| 8313 |
+
{
|
| 8314 |
+
"epoch": 0.20031219676834155,
|
| 8315 |
+
"grad_norm": 1.324830174446106,
|
| 8316 |
+
"learning_rate": 4.410454808545144e-06,
|
| 8317 |
+
"loss": 0.6589,
|
| 8318 |
+
"step": 1187
|
| 8319 |
+
},
|
| 8320 |
+
{
|
| 8321 |
+
"epoch": 0.20048095177825592,
|
| 8322 |
+
"grad_norm": 1.338803768157959,
|
| 8323 |
+
"learning_rate": 4.363353235625578e-06,
|
| 8324 |
+
"loss": 0.6009,
|
| 8325 |
+
"step": 1188
|
| 8326 |
+
},
|
| 8327 |
+
{
|
| 8328 |
+
"epoch": 0.20064970678817026,
|
| 8329 |
+
"grad_norm": 1.3291209936141968,
|
| 8330 |
+
"learning_rate": 4.316493048596787e-06,
|
| 8331 |
+
"loss": 0.6428,
|
| 8332 |
+
"step": 1189
|
| 8333 |
+
},
|
| 8334 |
+
{
|
| 8335 |
+
"epoch": 0.20081846179808463,
|
| 8336 |
+
"grad_norm": 1.343998908996582,
|
| 8337 |
+
"learning_rate": 4.269874495316351e-06,
|
| 8338 |
+
"loss": 0.6447,
|
| 8339 |
+
"step": 1190
|
| 8340 |
+
},
|
| 8341 |
+
{
|
| 8342 |
+
"epoch": 0.200987216807999,
|
| 8343 |
+
"grad_norm": 1.454965591430664,
|
| 8344 |
+
"learning_rate": 4.2234978223637365e-06,
|
| 8345 |
+
"loss": 0.7031,
|
| 8346 |
+
"step": 1191
|
| 8347 |
+
},
|
| 8348 |
+
{
|
| 8349 |
+
"epoch": 0.20115597181791334,
|
| 8350 |
+
"grad_norm": 1.5052690505981445,
|
| 8351 |
+
"learning_rate": 4.177363275039076e-06,
|
| 8352 |
+
"loss": 0.6319,
|
| 8353 |
+
"step": 1192
|
| 8354 |
+
},
|
| 8355 |
+
{
|
| 8356 |
+
"epoch": 0.2013247268278277,
|
| 8357 |
+
"grad_norm": 1.415196418762207,
|
| 8358 |
+
"learning_rate": 4.1314710973618e-06,
|
| 8359 |
+
"loss": 0.6923,
|
| 8360 |
+
"step": 1193
|
| 8361 |
+
},
|
| 8362 |
+
{
|
| 8363 |
+
"epoch": 0.20149348183774204,
|
| 8364 |
+
"grad_norm": 1.6829357147216797,
|
| 8365 |
+
"learning_rate": 4.0858215320693674e-06,
|
| 8366 |
+
"loss": 0.6939,
|
| 8367 |
+
"step": 1194
|
| 8368 |
+
},
|
| 8369 |
+
{
|
| 8370 |
+
"epoch": 0.2016622368476564,
|
| 8371 |
+
"grad_norm": 1.6496363878250122,
|
| 8372 |
+
"learning_rate": 4.040414820616007e-06,
|
| 8373 |
+
"loss": 0.7792,
|
| 8374 |
+
"step": 1195
|
| 8375 |
+
},
|
| 8376 |
+
{
|
| 8377 |
+
"epoch": 0.20183099185757078,
|
| 8378 |
+
"grad_norm": 1.9069215059280396,
|
| 8379 |
+
"learning_rate": 3.995251203171418e-06,
|
| 8380 |
+
"loss": 0.8317,
|
| 8381 |
+
"step": 1196
|
| 8382 |
+
},
|
| 8383 |
+
{
|
| 8384 |
+
"epoch": 0.20199974686748512,
|
| 8385 |
+
"grad_norm": 1.902542233467102,
|
| 8386 |
+
"learning_rate": 3.950330918619488e-06,
|
| 8387 |
+
"loss": 0.7696,
|
| 8388 |
+
"step": 1197
|
| 8389 |
+
},
|
| 8390 |
+
{
|
| 8391 |
+
"epoch": 0.2021685018773995,
|
| 8392 |
+
"grad_norm": 1.7510250806808472,
|
| 8393 |
+
"learning_rate": 3.905654204557069e-06,
|
| 8394 |
+
"loss": 0.6077,
|
| 8395 |
+
"step": 1198
|
| 8396 |
+
},
|
| 8397 |
+
{
|
| 8398 |
+
"epoch": 0.20233725688731385,
|
| 8399 |
+
"grad_norm": 2.122150182723999,
|
| 8400 |
+
"learning_rate": 3.861221297292655e-06,
|
| 8401 |
+
"loss": 0.7429,
|
| 8402 |
+
"step": 1199
|
| 8403 |
+
},
|
| 8404 |
+
{
|
| 8405 |
+
"epoch": 0.2025060118972282,
|
| 8406 |
+
"grad_norm": 3.036611795425415,
|
| 8407 |
+
"learning_rate": 3.817032431845224e-06,
|
| 8408 |
+
"loss": 1.1305,
|
| 8409 |
+
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 0.20267476690714256,
|
| 8413 |
+
"grad_norm": 0.46971753239631653,
|
| 8414 |
+
"learning_rate": 3.773087841942907e-06,
|
| 8415 |
+
"loss": 0.5726,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 0.2028435219170569,
|
| 8420 |
+
"grad_norm": 0.5580636262893677,
|
| 8421 |
+
"learning_rate": 3.729387760021785e-06,
|
| 8422 |
+
"loss": 0.6074,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 0.20301227692697127,
|
| 8427 |
+
"grad_norm": 0.6104133725166321,
|
| 8428 |
+
"learning_rate": 3.6859324172247025e-06,
|
| 8429 |
+
"loss": 0.5953,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 0.20318103193688564,
|
| 8434 |
+
"grad_norm": 0.6379888653755188,
|
| 8435 |
+
"learning_rate": 3.642722043399971e-06,
|
| 8436 |
+
"loss": 0.6084,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 0.20334978694679998,
|
| 8441 |
+
"grad_norm": 0.6631214618682861,
|
| 8442 |
+
"learning_rate": 3.5997568671001847e-06,
|
| 8443 |
+
"loss": 0.5401,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 0.20351854195671434,
|
| 8448 |
+
"grad_norm": 0.7642641067504883,
|
| 8449 |
+
"learning_rate": 3.557037115581041e-06,
|
| 8450 |
+
"loss": 0.6306,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 0.20368729696662868,
|
| 8455 |
+
"grad_norm": 0.7262935042381287,
|
| 8456 |
+
"learning_rate": 3.5145630148000985e-06,
|
| 8457 |
+
"loss": 0.5239,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 0.20385605197654305,
|
| 8462 |
+
"grad_norm": 0.7599579691886902,
|
| 8463 |
+
"learning_rate": 3.472334789415588e-06,
|
| 8464 |
+
"loss": 0.5652,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 0.20402480698645742,
|
| 8469 |
+
"grad_norm": 0.7735958695411682,
|
| 8470 |
+
"learning_rate": 3.430352662785247e-06,
|
| 8471 |
+
"loss": 0.5854,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 0.20419356199637176,
|
| 8476 |
+
"grad_norm": 0.7749413251876831,
|
| 8477 |
+
"learning_rate": 3.388616856965099e-06,
|
| 8478 |
+
"loss": 0.631,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 0.20436231700628613,
|
| 8483 |
+
"grad_norm": 0.830774188041687,
|
| 8484 |
+
"learning_rate": 3.3471275927083434e-06,
|
| 8485 |
+
"loss": 0.5967,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 0.2045310720162005,
|
| 8490 |
+
"grad_norm": 0.8539721369743347,
|
| 8491 |
+
"learning_rate": 3.3058850894641114e-06,
|
| 8492 |
+
"loss": 0.5401,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 0.20469982702611483,
|
| 8497 |
+
"grad_norm": 0.8092423677444458,
|
| 8498 |
+
"learning_rate": 3.264889565376339e-06,
|
| 8499 |
+
"loss": 0.5026,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 0.2048685820360292,
|
| 8504 |
+
"grad_norm": 0.9337773323059082,
|
| 8505 |
+
"learning_rate": 3.2241412372826517e-06,
|
| 8506 |
+
"loss": 0.6087,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 0.20503733704594354,
|
| 8511 |
+
"grad_norm": 0.9452828168869019,
|
| 8512 |
+
"learning_rate": 3.1836403207131383e-06,
|
| 8513 |
+
"loss": 0.5723,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 0.2052060920558579,
|
| 8518 |
+
"grad_norm": 0.9098586440086365,
|
| 8519 |
+
"learning_rate": 3.1433870298892996e-06,
|
| 8520 |
+
"loss": 0.5383,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 0.20537484706577228,
|
| 8525 |
+
"grad_norm": 0.9220548272132874,
|
| 8526 |
+
"learning_rate": 3.1033815777228136e-06,
|
| 8527 |
+
"loss": 0.5559,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 0.20554360207568662,
|
| 8532 |
+
"grad_norm": 0.9416666030883789,
|
| 8533 |
+
"learning_rate": 3.063624175814528e-06,
|
| 8534 |
+
"loss": 0.5704,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 0.20571235708560098,
|
| 8539 |
+
"grad_norm": 0.9549572467803955,
|
| 8540 |
+
"learning_rate": 3.0241150344532233e-06,
|
| 8541 |
+
"loss": 0.5763,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 0.20588111209551532,
|
| 8546 |
+
"grad_norm": 0.9730275273323059,
|
| 8547 |
+
"learning_rate": 2.9848543626145976e-06,
|
| 8548 |
+
"loss": 0.555,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 0.2060498671054297,
|
| 8553 |
+
"grad_norm": 0.992160439491272,
|
| 8554 |
+
"learning_rate": 2.9458423679600833e-06,
|
| 8555 |
+
"loss": 0.5707,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 0.20621862211534406,
|
| 8560 |
+
"grad_norm": 0.9168452620506287,
|
| 8561 |
+
"learning_rate": 2.907079256835815e-06,
|
| 8562 |
+
"loss": 0.4435,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 0.2063873771252584,
|
| 8567 |
+
"grad_norm": 1.0583746433258057,
|
| 8568 |
+
"learning_rate": 2.868565234271486e-06,
|
| 8569 |
+
"loss": 0.5124,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 0.20655613213517277,
|
| 8574 |
+
"grad_norm": 1.0423890352249146,
|
| 8575 |
+
"learning_rate": 2.8303005039793108e-06,
|
| 8576 |
+
"loss": 0.5583,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 0.20672488714508713,
|
| 8581 |
+
"grad_norm": 1.1211144924163818,
|
| 8582 |
+
"learning_rate": 2.79228526835289e-06,
|
| 8583 |
+
"loss": 0.5367,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 0.20689364215500147,
|
| 8588 |
+
"grad_norm": 1.0966883897781372,
|
| 8589 |
+
"learning_rate": 2.7545197284661926e-06,
|
| 8590 |
+
"loss": 0.5624,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 0.20706239716491584,
|
| 8595 |
+
"grad_norm": 1.232539415359497,
|
| 8596 |
+
"learning_rate": 2.717004084072472e-06,
|
| 8597 |
+
"loss": 0.6477,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 0.20723115217483018,
|
| 8602 |
+
"grad_norm": 1.1842130422592163,
|
| 8603 |
+
"learning_rate": 2.6797385336031987e-06,
|
| 8604 |
+
"loss": 0.5834,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 0.20739990718474455,
|
| 8609 |
+
"grad_norm": 1.1409043073654175,
|
| 8610 |
+
"learning_rate": 2.6427232741670362e-06,
|
| 8611 |
+
"loss": 0.5485,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 0.20756866219465891,
|
| 8616 |
+
"grad_norm": 1.2359825372695923,
|
| 8617 |
+
"learning_rate": 2.605958501548772e-06,
|
| 8618 |
+
"loss": 0.5593,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 0.20773741720457325,
|
| 8623 |
+
"grad_norm": 1.173651099205017,
|
| 8624 |
+
"learning_rate": 2.5694444102082937e-06,
|
| 8625 |
+
"loss": 0.6044,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 0.20790617221448762,
|
| 8630 |
+
"grad_norm": 1.1993813514709473,
|
| 8631 |
+
"learning_rate": 2.533181193279577e-06,
|
| 8632 |
+
"loss": 0.5074,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 0.20807492722440196,
|
| 8637 |
+
"grad_norm": 1.0826400518417358,
|
| 8638 |
+
"learning_rate": 2.4971690425696302e-06,
|
| 8639 |
+
"loss": 0.4711,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 0.20824368223431633,
|
| 8644 |
+
"grad_norm": 1.3286949396133423,
|
| 8645 |
+
"learning_rate": 2.4614081485574923e-06,
|
| 8646 |
+
"loss": 0.6885,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 0.2084124372442307,
|
| 8651 |
+
"grad_norm": 1.2414344549179077,
|
| 8652 |
+
"learning_rate": 2.425898700393253e-06,
|
| 8653 |
+
"loss": 0.6246,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 0.20858119225414504,
|
| 8658 |
+
"grad_norm": 1.2540428638458252,
|
| 8659 |
+
"learning_rate": 2.3906408858970164e-06,
|
| 8660 |
+
"loss": 0.682,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 0.2087499472640594,
|
| 8665 |
+
"grad_norm": 1.4568560123443604,
|
| 8666 |
+
"learning_rate": 2.3556348915579063e-06,
|
| 8667 |
+
"loss": 0.6347,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 0.20891870227397377,
|
| 8672 |
+
"grad_norm": 1.3099277019500732,
|
| 8673 |
+
"learning_rate": 2.320880902533129e-06,
|
| 8674 |
+
"loss": 0.6438,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 0.2090874572838881,
|
| 8679 |
+
"grad_norm": 1.4133312702178955,
|
| 8680 |
+
"learning_rate": 2.286379102646924e-06,
|
| 8681 |
+
"loss": 0.6298,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 0.20925621229380248,
|
| 8686 |
+
"grad_norm": 1.3855383396148682,
|
| 8687 |
+
"learning_rate": 2.252129674389658e-06,
|
| 8688 |
+
"loss": 0.5608,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 0.20942496730371682,
|
| 8693 |
+
"grad_norm": 1.6507779359817505,
|
| 8694 |
+
"learning_rate": 2.2181327989168e-06,
|
| 8695 |
+
"loss": 0.7996,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 0.2095937223136312,
|
| 8700 |
+
"grad_norm": 1.5349948406219482,
|
| 8701 |
+
"learning_rate": 2.184388656048003e-06,
|
| 8702 |
+
"loss": 0.6817,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 0.20976247732354555,
|
| 8707 |
+
"grad_norm": 1.7056605815887451,
|
| 8708 |
+
"learning_rate": 2.1508974242661627e-06,
|
| 8709 |
+
"loss": 0.8419,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 0.2099312323334599,
|
| 8714 |
+
"grad_norm": 1.7766491174697876,
|
| 8715 |
+
"learning_rate": 2.117659280716422e-06,
|
| 8716 |
+
"loss": 0.8449,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 0.21009998734337426,
|
| 8721 |
+
"grad_norm": 1.524853229522705,
|
| 8722 |
+
"learning_rate": 2.084674401205261e-06,
|
| 8723 |
+
"loss": 0.6829,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 0.2102687423532886,
|
| 8728 |
+
"grad_norm": 1.7471916675567627,
|
| 8729 |
+
"learning_rate": 2.05194296019961e-06,
|
| 8730 |
+
"loss": 0.8148,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 0.21043749736320297,
|
| 8735 |
+
"grad_norm": 1.6696993112564087,
|
| 8736 |
+
"learning_rate": 2.019465130825837e-06,
|
| 8737 |
+
"loss": 0.6402,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 0.21060625237311734,
|
| 8742 |
+
"grad_norm": 1.6932615041732788,
|
| 8743 |
+
"learning_rate": 1.9872410848689293e-06,
|
| 8744 |
+
"loss": 0.7545,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 0.21077500738303168,
|
| 8749 |
+
"grad_norm": 1.9034663438796997,
|
| 8750 |
+
"learning_rate": 1.955270992771507e-06,
|
| 8751 |
+
"loss": 0.6396,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 0.21094376239294604,
|
| 8756 |
+
"grad_norm": 2.3862850666046143,
|
| 8757 |
+
"learning_rate": 1.9235550236329556e-06,
|
| 8758 |
+
"loss": 0.9427,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 0.2111125174028604,
|
| 8763 |
+
"grad_norm": 0.5152022242546082,
|
| 8764 |
+
"learning_rate": 1.8920933452085399e-06,
|
| 8765 |
+
"loss": 0.5653,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 0.21128127241277475,
|
| 8770 |
+
"grad_norm": 0.5416737198829651,
|
| 8771 |
+
"learning_rate": 1.8608861239084951e-06,
|
| 8772 |
+
"loss": 0.5184,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 0.21145002742268912,
|
| 8777 |
+
"grad_norm": 0.5705904960632324,
|
| 8778 |
+
"learning_rate": 1.8299335247971562e-06,
|
| 8779 |
+
"loss": 0.5535,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 0.21161878243260346,
|
| 8784 |
+
"grad_norm": 0.6200594305992126,
|
| 8785 |
+
"learning_rate": 1.799235711592101e-06,
|
| 8786 |
+
"loss": 0.4932,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 0.21178753744251783,
|
| 8791 |
+
"grad_norm": 0.6781433820724487,
|
| 8792 |
+
"learning_rate": 1.768792846663242e-06,
|
| 8793 |
+
"loss": 0.5442,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 0.2119562924524322,
|
| 8798 |
+
"grad_norm": 0.6853250861167908,
|
| 8799 |
+
"learning_rate": 1.7386050910320194e-06,
|
| 8800 |
+
"loss": 0.544,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 0.21212504746234653,
|
| 8805 |
+
"grad_norm": 0.7194390892982483,
|
| 8806 |
+
"learning_rate": 1.7086726043705093e-06,
|
| 8807 |
+
"loss": 0.6269,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 0.2122938024722609,
|
| 8812 |
+
"grad_norm": 0.7299390435218811,
|
| 8813 |
+
"learning_rate": 1.6789955450005946e-06,
|
| 8814 |
+
"loss": 0.4949,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 0.21246255748217524,
|
| 8819 |
+
"grad_norm": 0.7063441872596741,
|
| 8820 |
+
"learning_rate": 1.6495740698931284e-06,
|
| 8821 |
+
"loss": 0.4427,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 0.2126313124920896,
|
| 8826 |
+
"grad_norm": 0.803770124912262,
|
| 8827 |
+
"learning_rate": 1.6204083346671061e-06,
|
| 8828 |
+
"loss": 0.5222,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 0.21280006750200398,
|
| 8833 |
+
"grad_norm": 0.8003373742103577,
|
| 8834 |
+
"learning_rate": 1.5914984935888277e-06,
|
| 8835 |
+
"loss": 0.5384,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 0.21296882251191832,
|
| 8840 |
+
"grad_norm": 0.8615963459014893,
|
| 8841 |
+
"learning_rate": 1.5628446995711144e-06,
|
| 8842 |
+
"loss": 0.5575,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 0.21313757752183268,
|
| 8847 |
+
"grad_norm": 1.0034806728363037,
|
| 8848 |
+
"learning_rate": 1.5344471041724483e-06,
|
| 8849 |
+
"loss": 0.594,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 0.21330633253174702,
|
| 8854 |
+
"grad_norm": 0.7860152721405029,
|
| 8855 |
+
"learning_rate": 1.5063058575962185e-06,
|
| 8856 |
+
"loss": 0.4435,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 0.2134750875416614,
|
| 8861 |
+
"grad_norm": 0.8901642560958862,
|
| 8862 |
+
"learning_rate": 1.4784211086899147e-06,
|
| 8863 |
+
"loss": 0.5453,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 0.21364384255157576,
|
| 8868 |
+
"grad_norm": 0.9191757440567017,
|
| 8869 |
+
"learning_rate": 1.450793004944312e-06,
|
| 8870 |
+
"loss": 0.6428,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 0.2138125975614901,
|
| 8875 |
+
"grad_norm": 1.0055803060531616,
|
| 8876 |
+
"learning_rate": 1.423421692492738e-06,
|
| 8877 |
+
"loss": 0.5978,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 0.21398135257140447,
|
| 8882 |
+
"grad_norm": 0.9184823036193848,
|
| 8883 |
+
"learning_rate": 1.3963073161102457e-06,
|
| 8884 |
+
"loss": 0.5046,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 0.21415010758131883,
|
| 8889 |
+
"grad_norm": 0.9808287024497986,
|
| 8890 |
+
"learning_rate": 1.3694500192128977e-06,
|
| 8891 |
+
"loss": 0.6761,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 0.21431886259123317,
|
| 8896 |
+
"grad_norm": 0.9972100257873535,
|
| 8897 |
+
"learning_rate": 1.3428499438569886e-06,
|
| 8898 |
+
"loss": 0.6078,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 0.21448761760114754,
|
| 8903 |
+
"grad_norm": 0.9385042190551758,
|
| 8904 |
+
"learning_rate": 1.316507230738262e-06,
|
| 8905 |
+
"loss": 0.5393,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 0.21465637261106188,
|
| 8910 |
+
"grad_norm": 0.9871394038200378,
|
| 8911 |
+
"learning_rate": 1.2904220191912398e-06,
|
| 8912 |
+
"loss": 0.6001,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 0.21482512762097625,
|
| 8917 |
+
"grad_norm": 0.9962787628173828,
|
| 8918 |
+
"learning_rate": 1.2645944471883996e-06,
|
| 8919 |
+
"loss": 0.5253,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 0.21499388263089061,
|
| 8924 |
+
"grad_norm": 1.0044738054275513,
|
| 8925 |
+
"learning_rate": 1.23902465133951e-06,
|
| 8926 |
+
"loss": 0.5064,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 0.21516263764080495,
|
| 8931 |
+
"grad_norm": 1.224851369857788,
|
| 8932 |
+
"learning_rate": 1.213712766890873e-06,
|
| 8933 |
+
"loss": 0.6911,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 0.21533139265071932,
|
| 8938 |
+
"grad_norm": 1.0862836837768555,
|
| 8939 |
+
"learning_rate": 1.188658927724634e-06,
|
| 8940 |
+
"loss": 0.6702,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 0.21550014766063366,
|
| 8945 |
+
"grad_norm": 1.0600032806396484,
|
| 8946 |
+
"learning_rate": 1.1638632663580452e-06,
|
| 8947 |
+
"loss": 0.5098,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 0.21566890267054803,
|
| 8952 |
+
"grad_norm": 1.1393179893493652,
|
| 8953 |
+
"learning_rate": 1.1393259139427904e-06,
|
| 8954 |
+
"loss": 0.5113,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 0.2158376576804624,
|
| 8959 |
+
"grad_norm": 1.2485078573226929,
|
| 8960 |
+
"learning_rate": 1.1150470002642687e-06,
|
| 8961 |
+
"loss": 0.5597,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 0.21600641269037674,
|
| 8966 |
+
"grad_norm": 1.165150761604309,
|
| 8967 |
+
"learning_rate": 1.0910266537409285e-06,
|
| 8968 |
+
"loss": 0.572,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 0.2161751677002911,
|
| 8973 |
+
"grad_norm": 1.0968618392944336,
|
| 8974 |
+
"learning_rate": 1.067265001423573e-06,
|
| 8975 |
+
"loss": 0.5741,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 0.21634392271020547,
|
| 8980 |
+
"grad_norm": 1.0635594129562378,
|
| 8981 |
+
"learning_rate": 1.0437621689946941e-06,
|
| 8982 |
+
"loss": 0.4003,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 0.2165126777201198,
|
| 8987 |
+
"grad_norm": 1.2414653301239014,
|
| 8988 |
+
"learning_rate": 1.0205182807678183e-06,
|
| 8989 |
+
"loss": 0.6257,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 0.21668143273003418,
|
| 8994 |
+
"grad_norm": 1.1371922492980957,
|
| 8995 |
+
"learning_rate": 9.975334596868225e-07,
|
| 8996 |
+
"loss": 0.4359,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 0.21685018773994852,
|
| 9001 |
+
"grad_norm": 1.3592020273208618,
|
| 9002 |
+
"learning_rate": 9.748078273253137e-07,
|
| 9003 |
+
"loss": 0.6194,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 0.2170189427498629,
|
| 9008 |
+
"grad_norm": 1.3216999769210815,
|
| 9009 |
+
"learning_rate": 9.523415038859618e-07,
|
| 9010 |
+
"loss": 0.6652,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 0.21718769775977725,
|
| 9015 |
+
"grad_norm": 1.2840315103530884,
|
| 9016 |
+
"learning_rate": 9.301346081998841e-07,
|
| 9017 |
+
"loss": 0.558,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 0.2173564527696916,
|
| 9022 |
+
"grad_norm": 1.1984096765518188,
|
| 9023 |
+
"learning_rate": 9.081872577260064e-07,
|
| 9024 |
+
"loss": 0.481,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 0.21752520777960596,
|
| 9029 |
+
"grad_norm": 1.289543628692627,
|
| 9030 |
+
"learning_rate": 8.864995685504251e-07,
|
| 9031 |
+
"loss": 0.5848,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 0.2176939627895203,
|
| 9036 |
+
"grad_norm": 1.4564785957336426,
|
| 9037 |
+
"learning_rate": 8.650716553858296e-07,
|
| 9038 |
+
"loss": 0.785,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 0.21786271779943467,
|
| 9043 |
+
"grad_norm": 1.3245304822921753,
|
| 9044 |
+
"learning_rate": 8.439036315708693e-07,
|
| 9045 |
+
"loss": 0.5533,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 0.21803147280934904,
|
| 9050 |
+
"grad_norm": 1.427384853363037,
|
| 9051 |
+
"learning_rate": 8.229956090695656e-07,
|
| 9052 |
+
"loss": 0.6575,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 0.21820022781926338,
|
| 9057 |
+
"grad_norm": 1.597827672958374,
|
| 9058 |
+
"learning_rate": 8.023476984706957e-07,
|
| 9059 |
+
"loss": 0.7157,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 0.21836898282917774,
|
| 9064 |
+
"grad_norm": 1.5298324823379517,
|
| 9065 |
+
"learning_rate": 7.819600089872592e-07,
|
| 9066 |
+
"loss": 0.5826,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 0.2185377378390921,
|
| 9071 |
+
"grad_norm": 1.614680290222168,
|
| 9072 |
+
"learning_rate": 7.618326484558402e-07,
|
| 9073 |
+
"loss": 0.6945,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 0.21870649284900645,
|
| 9078 |
+
"grad_norm": 1.6210813522338867,
|
| 9079 |
+
"learning_rate": 7.419657233360799e-07,
|
| 9080 |
+
"loss": 0.6229,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 0.21887524785892082,
|
| 9085 |
+
"grad_norm": 1.5343276262283325,
|
| 9086 |
+
"learning_rate": 7.22359338710088e-07,
|
| 9087 |
+
"loss": 0.6332,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 0.21904400286883516,
|
| 9092 |
+
"grad_norm": 1.890868067741394,
|
| 9093 |
+
"learning_rate": 7.030135982819097e-07,
|
| 9094 |
+
"loss": 0.7546,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 0.21921275787874953,
|
| 9099 |
+
"grad_norm": 1.822385311126709,
|
| 9100 |
+
"learning_rate": 6.839286043769655e-07,
|
| 9101 |
+
"loss": 0.6091,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 0.2193815128886639,
|
| 9106 |
+
"grad_norm": 2.7738308906555176,
|
| 9107 |
+
"learning_rate": 6.651044579414956e-07,
|
| 9108 |
+
"loss": 1.0128,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 0.21955026789857823,
|
| 9113 |
+
"grad_norm": 0.4590780735015869,
|
| 9114 |
+
"learning_rate": 6.465412585420438e-07,
|
| 9115 |
+
"loss": 0.5627,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 0.2197190229084926,
|
| 9120 |
+
"grad_norm": 0.5761750340461731,
|
| 9121 |
+
"learning_rate": 6.282391043649415e-07,
|
| 9122 |
+
"loss": 0.5758,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 0.21988777791840694,
|
| 9127 |
+
"grad_norm": 0.5773003697395325,
|
| 9128 |
+
"learning_rate": 6.101980922157524e-07,
|
| 9129 |
+
"loss": 0.606,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 0.2200565329283213,
|
| 9134 |
+
"grad_norm": 0.6117558479309082,
|
| 9135 |
+
"learning_rate": 5.924183175187948e-07,
|
| 9136 |
+
"loss": 0.4897,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 0.22022528793823568,
|
| 9141 |
+
"grad_norm": 0.6885523796081543,
|
| 9142 |
+
"learning_rate": 5.748998743166256e-07,
|
| 9143 |
+
"loss": 0.5727,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 0.22039404294815002,
|
| 9148 |
+
"grad_norm": 0.7196971774101257,
|
| 9149 |
+
"learning_rate": 5.576428552695301e-07,
|
| 9150 |
+
"loss": 0.5318,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 0.22056279795806438,
|
| 9155 |
+
"grad_norm": 0.687440812587738,
|
| 9156 |
+
"learning_rate": 5.406473516550603e-07,
|
| 9157 |
+
"loss": 0.5473,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 0.22073155296797875,
|
| 9162 |
+
"grad_norm": 0.7595966458320618,
|
| 9163 |
+
"learning_rate": 5.239134533675083e-07,
|
| 9164 |
+
"loss": 0.6343,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 0.2209003079778931,
|
| 9169 |
+
"grad_norm": 0.7847376465797424,
|
| 9170 |
+
"learning_rate": 5.074412489174896e-07,
|
| 9171 |
+
"loss": 0.4844,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 0.22106906298780746,
|
| 9176 |
+
"grad_norm": 0.7519535422325134,
|
| 9177 |
+
"learning_rate": 4.912308254314158e-07,
|
| 9178 |
+
"loss": 0.4785,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 0.2212378179977218,
|
| 9183 |
+
"grad_norm": 0.8657416701316833,
|
| 9184 |
+
"learning_rate": 4.7528226865107275e-07,
|
| 9185 |
+
"loss": 0.514,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 0.22140657300763616,
|
| 9190 |
+
"grad_norm": 0.7888314127922058,
|
| 9191 |
+
"learning_rate": 4.5959566293315995e-07,
|
| 9192 |
+
"loss": 0.4713,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 0.22157532801755053,
|
| 9197 |
+
"grad_norm": 0.8790714144706726,
|
| 9198 |
+
"learning_rate": 4.4417109124882395e-07,
|
| 9199 |
+
"loss": 0.5936,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 0.22174408302746487,
|
| 9204 |
+
"grad_norm": 0.8749496340751648,
|
| 9205 |
+
"learning_rate": 4.290086351832423e-07,
|
| 9206 |
+
"loss": 0.5844,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 0.22191283803737924,
|
| 9211 |
+
"grad_norm": 0.9149898290634155,
|
| 9212 |
+
"learning_rate": 4.141083749351959e-07,
|
| 9213 |
+
"loss": 0.5873,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 0.22208159304729358,
|
| 9218 |
+
"grad_norm": 0.9576611518859863,
|
| 9219 |
+
"learning_rate": 3.9947038931661964e-07,
|
| 9220 |
+
"loss": 0.5523,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 0.22225034805720795,
|
| 9225 |
+
"grad_norm": 1.0207158327102661,
|
| 9226 |
+
"learning_rate": 3.850947557521911e-07,
|
| 9227 |
+
"loss": 0.5631,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 0.22241910306712231,
|
| 9232 |
+
"grad_norm": 1.0233733654022217,
|
| 9233 |
+
"learning_rate": 3.7098155027895375e-07,
|
| 9234 |
+
"loss": 0.6176,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 0.22258785807703665,
|
| 9239 |
+
"grad_norm": 0.9032954573631287,
|
| 9240 |
+
"learning_rate": 3.571308475458723e-07,
|
| 9241 |
+
"loss": 0.4697,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 0.22275661308695102,
|
| 9246 |
+
"grad_norm": 1.0094698667526245,
|
| 9247 |
+
"learning_rate": 3.435427208134667e-07,
|
| 9248 |
+
"loss": 0.6105,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 0.2229253680968654,
|
| 9253 |
+
"grad_norm": 0.967095136642456,
|
| 9254 |
+
"learning_rate": 3.3021724195340107e-07,
|
| 9255 |
+
"loss": 0.4839,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 0.22309412310677973,
|
| 9260 |
+
"grad_norm": 0.9922171831130981,
|
| 9261 |
+
"learning_rate": 3.171544814481231e-07,
|
| 9262 |
+
"loss": 0.4597,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 0.2232628781166941,
|
| 9267 |
+
"grad_norm": 1.0829986333847046,
|
| 9268 |
+
"learning_rate": 3.0435450839049197e-07,
|
| 9269 |
+
"loss": 0.5619,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 0.22343163312660844,
|
| 9274 |
+
"grad_norm": 1.030965805053711,
|
| 9275 |
+
"learning_rate": 2.9181739048340095e-07,
|
| 9276 |
+
"loss": 0.579,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 0.2236003881365228,
|
| 9281 |
+
"grad_norm": 0.9884082078933716,
|
| 9282 |
+
"learning_rate": 2.7954319403940553e-07,
|
| 9283 |
+
"loss": 0.4673,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 0.22376914314643717,
|
| 9288 |
+
"grad_norm": 1.1055041551589966,
|
| 9289 |
+
"learning_rate": 2.67531983980418e-07,
|
| 9290 |
+
"loss": 0.5504,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 0.2239378981563515,
|
| 9295 |
+
"grad_norm": 1.2018651962280273,
|
| 9296 |
+
"learning_rate": 2.5578382383732444e-07,
|
| 9297 |
+
"loss": 0.5453,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 0.22410665316626588,
|
| 9302 |
+
"grad_norm": 1.1222630739212036,
|
| 9303 |
+
"learning_rate": 2.4429877574965734e-07,
|
| 9304 |
+
"loss": 0.6074,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 0.22427540817618022,
|
| 9309 |
+
"grad_norm": 0.9751061797142029,
|
| 9310 |
+
"learning_rate": 2.3307690046527885e-07,
|
| 9311 |
+
"loss": 0.4812,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 0.2244441631860946,
|
| 9316 |
+
"grad_norm": 1.1505897045135498,
|
| 9317 |
+
"learning_rate": 2.2211825734004266e-07,
|
| 9318 |
+
"loss": 0.6279,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 0.22461291819600895,
|
| 9323 |
+
"grad_norm": 1.1970534324645996,
|
| 9324 |
+
"learning_rate": 2.1142290433750488e-07,
|
| 9325 |
+
"loss": 0.6253,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 0.2247816732059233,
|
| 9330 |
+
"grad_norm": 1.2056388854980469,
|
| 9331 |
+
"learning_rate": 2.0099089802858017e-07,
|
| 9332 |
+
"loss": 0.567,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 0.22495042821583766,
|
| 9337 |
+
"grad_norm": 1.0611746311187744,
|
| 9338 |
+
"learning_rate": 1.9082229359127512e-07,
|
| 9339 |
+
"loss": 0.523,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.225119183225752,
|
| 9344 |
+
"grad_norm": 1.3161405324935913,
|
| 9345 |
+
"learning_rate": 1.8091714481038858e-07,
|
| 9346 |
+
"loss": 0.6177,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.22528793823566637,
|
| 9351 |
+
"grad_norm": 1.363237977027893,
|
| 9352 |
+
"learning_rate": 1.7127550407721182e-07,
|
| 9353 |
+
"loss": 0.6685,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.22545669324558074,
|
| 9358 |
+
"grad_norm": 1.3239537477493286,
|
| 9359 |
+
"learning_rate": 1.618974223892733e-07,
|
| 9360 |
+
"loss": 0.6109,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.22562544825549508,
|
| 9365 |
+
"grad_norm": 1.0830307006835938,
|
| 9366 |
+
"learning_rate": 1.5278294935006098e-07,
|
| 9367 |
+
"loss": 0.4791,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.22579420326540944,
|
| 9372 |
+
"grad_norm": 1.2744098901748657,
|
| 9373 |
+
"learning_rate": 1.4393213316873931e-07,
|
| 9374 |
+
"loss": 0.6279,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.2259629582753238,
|
| 9379 |
+
"grad_norm": 1.4248663187026978,
|
| 9380 |
+
"learning_rate": 1.3534502065993826e-07,
|
| 9381 |
+
"loss": 0.6801,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.22613171328523815,
|
| 9386 |
+
"grad_norm": 1.3996230363845825,
|
| 9387 |
+
"learning_rate": 1.270216572434646e-07,
|
| 9388 |
+
"loss": 0.5948,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.22630046829515252,
|
| 9393 |
+
"grad_norm": 1.4179376363754272,
|
| 9394 |
+
"learning_rate": 1.1896208694406885e-07,
|
| 9395 |
+
"loss": 0.6589,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.22646922330506686,
|
| 9400 |
+
"grad_norm": 1.4643542766571045,
|
| 9401 |
+
"learning_rate": 1.111663523912454e-07,
|
| 9402 |
+
"loss": 0.6187,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.22663797831498123,
|
| 9407 |
+
"grad_norm": 1.5181905031204224,
|
| 9408 |
+
"learning_rate": 1.0363449481896603e-07,
|
| 9409 |
+
"loss": 0.6958,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.2268067333248956,
|
| 9414 |
+
"grad_norm": 1.714362382888794,
|
| 9415 |
+
"learning_rate": 9.636655406546902e-08,
|
| 9416 |
+
"loss": 0.745,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.22697548833480993,
|
| 9421 |
+
"grad_norm": 1.5242582559585571,
|
| 9422 |
+
"learning_rate": 8.9362568573087e-08,
|
| 9423 |
+
"loss": 0.7357,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.2271442433447243,
|
| 9428 |
+
"grad_norm": 1.657204270362854,
|
| 9429 |
+
"learning_rate": 8.262257538798612e-08,
|
| 9430 |
+
"loss": 0.632,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.22731299835463864,
|
| 9435 |
+
"grad_norm": 1.9706053733825684,
|
| 9436 |
+
"learning_rate": 7.614661016001057e-08,
|
| 9437 |
+
"loss": 0.8604,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.227481753364553,
|
| 9442 |
+
"grad_norm": 1.9471381902694702,
|
| 9443 |
+
"learning_rate": 6.993470714248274e-08,
|
| 9444 |
+
"loss": 0.7653,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.22765050837446738,
|
| 9449 |
+
"grad_norm": 1.659595012664795,
|
| 9450 |
+
"learning_rate": 6.39868991920145e-08,
|
| 9451 |
+
"loss": 0.5757,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.22781926338438171,
|
| 9456 |
+
"grad_norm": 2.402801513671875,
|
| 9457 |
+
"learning_rate": 5.830321776834069e-08,
|
| 9458 |
+
"loss": 0.8009,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.22798801839429608,
|
| 9463 |
+
"grad_norm": 0.47180652618408203,
|
| 9464 |
+
"learning_rate": 5.288369293415807e-08,
|
| 9465 |
+
"loss": 0.5805,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.22815677340421045,
|
| 9470 |
+
"grad_norm": 0.5464998483657837,
|
| 9471 |
+
"learning_rate": 4.7728353354958843e-08,
|
| 9472 |
+
"loss": 0.5027,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.2283255284141248,
|
| 9477 |
+
"grad_norm": 0.6459171772003174,
|
| 9478 |
+
"learning_rate": 4.283722629887521e-08,
|
| 9479 |
+
"loss": 0.6744,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.22849428342403916,
|
| 9484 |
+
"grad_norm": 0.6989730596542358,
|
| 9485 |
+
"learning_rate": 3.821033763654058e-08,
|
| 9486 |
+
"loss": 0.6458,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.2286630384339535,
|
| 9491 |
+
"grad_norm": 0.6470995545387268,
|
| 9492 |
+
"learning_rate": 3.384771184095081e-08,
|
| 9493 |
+
"loss": 0.5209,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.22883179344386786,
|
| 9498 |
+
"grad_norm": 0.7597262859344482,
|
| 9499 |
+
"learning_rate": 2.974937198735317e-08,
|
| 9500 |
+
"loss": 0.5776,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.22900054845378223,
|
| 9505 |
+
"grad_norm": 0.716862142086029,
|
| 9506 |
+
"learning_rate": 2.5915339753085356e-08,
|
| 9507 |
+
"loss": 0.5438,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.22916930346369657,
|
| 9512 |
+
"grad_norm": 0.6981287598609924,
|
| 9513 |
+
"learning_rate": 2.234563541749224e-08,
|
| 9514 |
+
"loss": 0.5289,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.22933805847361094,
|
| 9519 |
+
"grad_norm": 0.7905397415161133,
|
| 9520 |
+
"learning_rate": 1.9040277861814836e-08,
|
| 9521 |
+
"loss": 0.5844,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.22950681348352528,
|
| 9526 |
+
"grad_norm": 0.7727534174919128,
|
| 9527 |
+
"learning_rate": 1.5999284569068184e-08,
|
| 9528 |
+
"loss": 0.5605,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.22967556849343965,
|
| 9533 |
+
"grad_norm": 0.7689732313156128,
|
| 9534 |
+
"learning_rate": 1.3222671623991378e-08,
|
| 9535 |
+
"loss": 0.5119,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.22984432350335401,
|
| 9540 |
+
"grad_norm": 0.7805505394935608,
|
| 9541 |
+
"learning_rate": 1.0710453712903245e-08,
|
| 9542 |
+
"loss": 0.4539,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.23001307851326835,
|
| 9547 |
+
"grad_norm": 0.7606160640716553,
|
| 9548 |
+
"learning_rate": 8.462644123696794e-09,
|
| 9549 |
+
"loss": 0.4911,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.23018183352318272,
|
| 9554 |
+
"grad_norm": 0.8344932794570923,
|
| 9555 |
+
"learning_rate": 6.4792547456948845e-09,
|
| 9556 |
+
"loss": 0.5379,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.2303505885330971,
|
| 9561 |
+
"grad_norm": 0.8788904547691345,
|
| 9562 |
+
"learning_rate": 4.760296069639125e-09,
|
| 9563 |
+
"loss": 0.5596,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.23051934354301143,
|
| 9568 |
+
"grad_norm": 0.9263433814048767,
|
| 9569 |
+
"learning_rate": 3.30577718762326e-09,
|
| 9570 |
+
"loss": 0.5956,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.2306880985529258,
|
| 9575 |
+
"grad_norm": 0.8482834696769714,
|
| 9576 |
+
"learning_rate": 2.115705793032108e-09,
|
| 9577 |
+
"loss": 0.4577,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.23085685356284014,
|
| 9582 |
+
"grad_norm": 0.8887785077095032,
|
| 9583 |
+
"learning_rate": 1.1900881805082532e-09,
|
| 9584 |
+
"loss": 0.4727,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.2310256085727545,
|
| 9589 |
+
"grad_norm": 0.9328110218048096,
|
| 9590 |
+
"learning_rate": 5.289292459187411e-10,
|
| 9591 |
+
"loss": 0.5215,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.23119436358266887,
|
| 9596 |
+
"grad_norm": 0.9682706594467163,
|
| 9597 |
+
"learning_rate": 1.3223248633287277e-10,
|
| 9598 |
+
"loss": 0.5327,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.2313631185925832,
|
| 9603 |
+
"grad_norm": 0.8991033434867859,
|
| 9604 |
+
"learning_rate": 0.0,
|
| 9605 |
+
"loss": 0.5499,
|
| 9606 |
+
"step": 1371
|
| 9607 |
}
|
| 9608 |
],
|
| 9609 |
"logging_steps": 1,
|
|
|
|
| 9618 |
"should_evaluate": false,
|
| 9619 |
"should_log": false,
|
| 9620 |
"should_save": true,
|
| 9621 |
+
"should_training_stop": true
|
| 9622 |
},
|
| 9623 |
"attributes": {}
|
| 9624 |
}
|
| 9625 |
},
|
| 9626 |
+
"total_flos": 1.8906786208048742e+18,
|
| 9627 |
"train_batch_size": 4,
|
| 9628 |
"trial_name": null,
|
| 9629 |
"trial_params": null
|