Commit
·
36c9461
1
Parent(s):
f0e4e3b
huggingartists
Browse files- README.md +3 -3
- flax_model.msgpack +1 -1
- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- rng_state.pth +2 -2
- scheduler.pt +1 -1
- trainer_state.json +3113 -5
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -45,15 +45,15 @@ from datasets import load_dataset
|
|
| 45 |
dataset = load_dataset("huggingartists/oxxxymiron")
|
| 46 |
```
|
| 47 |
|
| 48 |
-
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/
|
| 49 |
|
| 50 |
## Training procedure
|
| 51 |
|
| 52 |
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Oxxxymiron's lyrics.
|
| 53 |
|
| 54 |
-
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/
|
| 55 |
|
| 56 |
-
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/
|
| 57 |
|
| 58 |
## How to use
|
| 59 |
|
|
|
|
| 45 |
dataset = load_dataset("huggingartists/oxxxymiron")
|
| 46 |
```
|
| 47 |
|
| 48 |
+
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/296e4zy2/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
|
| 49 |
|
| 50 |
## Training procedure
|
| 51 |
|
| 52 |
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Oxxxymiron's lyrics.
|
| 53 |
|
| 54 |
+
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/lyd324n8) for full transparency and reproducibility.
|
| 55 |
|
| 56 |
+
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/lyd324n8/artifacts) is logged and versioned.
|
| 57 |
|
| 58 |
## How to use
|
| 59 |
|
flax_model.msgpack
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 497764120
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bf9a57b36b9276338603832189c301268f56bae13bf00fb4ac5da15b09879d59
|
| 3 |
size 497764120
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 995604017
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6fcbd0efc1ba3f1cf85a1ad153a13999c85e76f9f2b7c9440f87ec591788eaa1
|
| 3 |
size 995604017
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 510403817
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aa19aa16a103bc8d557dbe011a86e4135bafaa6acf98d21f7c8a1f0ef4362155
|
| 3 |
size 510403817
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44cd6576738e9780ee196c4c60bbd639b2c9174b22df17c2b94e485513761999
|
| 3 |
+
size 14439
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3757bed4046f1195d42cf6b407ad0ea93daf46be4a6bcc490f1daeb7bdf87f8c
|
| 3 |
size 623
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -1560,11 +1560,3119 @@
|
|
| 1560 |
"learning_rate": 0.0,
|
| 1561 |
"loss": 1.7624,
|
| 1562 |
"step": 1295
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1563 |
}
|
| 1564 |
],
|
| 1565 |
-
"max_steps":
|
| 1566 |
-
"num_train_epochs":
|
| 1567 |
-
"total_flos":
|
| 1568 |
"trial_name": null,
|
| 1569 |
"trial_params": null
|
| 1570 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 15.0,
|
| 5 |
+
"global_step": 3885,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 1560 |
"learning_rate": 0.0,
|
| 1561 |
"loss": 1.7624,
|
| 1562 |
"step": 1295
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 5.02,
|
| 1566 |
+
"learning_rate": 1.261250123775442e-07,
|
| 1567 |
+
"loss": 1.6889,
|
| 1568 |
+
"step": 1300
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 5.04,
|
| 1572 |
+
"learning_rate": 5.040362734534236e-07,
|
| 1573 |
+
"loss": 1.6715,
|
| 1574 |
+
"step": 1305
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 5.06,
|
| 1578 |
+
"learning_rate": 1.132344160414776e-06,
|
| 1579 |
+
"loss": 1.7252,
|
| 1580 |
+
"step": 1310
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 5.08,
|
| 1584 |
+
"learning_rate": 2.0087383134942512e-06,
|
| 1585 |
+
"loss": 1.7355,
|
| 1586 |
+
"step": 1315
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 5.1,
|
| 1590 |
+
"learning_rate": 3.1299961314264046e-06,
|
| 1591 |
+
"loss": 1.8099,
|
| 1592 |
+
"step": 1320
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 5.12,
|
| 1596 |
+
"learning_rate": 4.491994621320179e-06,
|
| 1597 |
+
"loss": 1.7203,
|
| 1598 |
+
"step": 1325
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 5.14,
|
| 1602 |
+
"learning_rate": 6.089725559373899e-06,
|
| 1603 |
+
"loss": 1.7262,
|
| 1604 |
+
"step": 1330
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 5.15,
|
| 1608 |
+
"learning_rate": 7.917313906685478e-06,
|
| 1609 |
+
"loss": 1.721,
|
| 1610 |
+
"step": 1335
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 5.17,
|
| 1614 |
+
"learning_rate": 9.968039412440962e-06,
|
| 1615 |
+
"loss": 1.6592,
|
| 1616 |
+
"step": 1340
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 5.19,
|
| 1620 |
+
"learning_rate": 1.2234361325042687e-05,
|
| 1621 |
+
"loss": 1.6871,
|
| 1622 |
+
"step": 1345
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 5.21,
|
| 1626 |
+
"learning_rate": 1.4707946120313293e-05,
|
| 1627 |
+
"loss": 1.7319,
|
| 1628 |
+
"step": 1350
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 5.23,
|
| 1632 |
+
"learning_rate": 1.737969814481516e-05,
|
| 1633 |
+
"loss": 1.7102,
|
| 1634 |
+
"step": 1355
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 5.25,
|
| 1638 |
+
"learning_rate": 2.0239793061604692e-05,
|
| 1639 |
+
"loss": 1.6969,
|
| 1640 |
+
"step": 1360
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 5.27,
|
| 1644 |
+
"learning_rate": 2.3277713975440297e-05,
|
| 1645 |
+
"loss": 1.6891,
|
| 1646 |
+
"step": 1365
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 5.29,
|
| 1650 |
+
"learning_rate": 2.648229010460629e-05,
|
| 1651 |
+
"loss": 1.7584,
|
| 1652 |
+
"step": 1370
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 5.31,
|
| 1656 |
+
"learning_rate": 2.9841737857150448e-05,
|
| 1657 |
+
"loss": 1.7074,
|
| 1658 |
+
"step": 1375
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 5.33,
|
| 1662 |
+
"learning_rate": 3.334370416049612e-05,
|
| 1663 |
+
"loss": 1.6984,
|
| 1664 |
+
"step": 1380
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 5.35,
|
| 1668 |
+
"learning_rate": 3.697531188509984e-05,
|
| 1669 |
+
"loss": 1.6734,
|
| 1670 |
+
"step": 1385
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 5.37,
|
| 1674 |
+
"learning_rate": 4.072320719512421e-05,
|
| 1675 |
+
"loss": 1.6728,
|
| 1676 |
+
"step": 1390
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 5.39,
|
| 1680 |
+
"learning_rate": 4.457360865201626e-05,
|
| 1681 |
+
"loss": 1.7748,
|
| 1682 |
+
"step": 1395
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 5.41,
|
| 1686 |
+
"learning_rate": 4.8512357890428555e-05,
|
| 1687 |
+
"loss": 1.6899,
|
| 1688 |
+
"step": 1400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 5.42,
|
| 1692 |
+
"learning_rate": 5.252497168014445e-05,
|
| 1693 |
+
"loss": 1.6741,
|
| 1694 |
+
"step": 1405
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 5.44,
|
| 1698 |
+
"learning_rate": 5.659669518256621e-05,
|
| 1699 |
+
"loss": 1.6965,
|
| 1700 |
+
"step": 1410
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 5.46,
|
| 1704 |
+
"learning_rate": 6.071255620594022e-05,
|
| 1705 |
+
"loss": 1.6641,
|
| 1706 |
+
"step": 1415
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 5.48,
|
| 1710 |
+
"learning_rate": 6.485742025981456e-05,
|
| 1711 |
+
"loss": 1.7494,
|
| 1712 |
+
"step": 1420
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 5.5,
|
| 1716 |
+
"learning_rate": 6.901604620628525e-05,
|
| 1717 |
+
"loss": 1.7469,
|
| 1718 |
+
"step": 1425
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 5.52,
|
| 1722 |
+
"learning_rate": 7.31731423033995e-05,
|
| 1723 |
+
"loss": 1.7503,
|
| 1724 |
+
"step": 1430
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 5.54,
|
| 1728 |
+
"learning_rate": 7.731342243463585e-05,
|
| 1729 |
+
"loss": 1.7359,
|
| 1730 |
+
"step": 1435
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 5.56,
|
| 1734 |
+
"learning_rate": 8.14216623176967e-05,
|
| 1735 |
+
"loss": 1.7196,
|
| 1736 |
+
"step": 1440
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 5.58,
|
| 1740 |
+
"learning_rate": 8.548275548593167e-05,
|
| 1741 |
+
"loss": 1.7352,
|
| 1742 |
+
"step": 1445
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 5.6,
|
| 1746 |
+
"learning_rate": 8.948176883653917e-05,
|
| 1747 |
+
"loss": 1.7017,
|
| 1748 |
+
"step": 1450
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 5.62,
|
| 1752 |
+
"learning_rate": 9.340399754128714e-05,
|
| 1753 |
+
"loss": 1.7402,
|
| 1754 |
+
"step": 1455
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 5.64,
|
| 1758 |
+
"learning_rate": 9.723501911784583e-05,
|
| 1759 |
+
"loss": 1.7463,
|
| 1760 |
+
"step": 1460
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 5.66,
|
| 1764 |
+
"learning_rate": 0.00010096074646289766,
|
| 1765 |
+
"loss": 1.6842,
|
| 1766 |
+
"step": 1465
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 5.68,
|
| 1770 |
+
"learning_rate": 0.00010456747965202592,
|
| 1771 |
+
"loss": 1.7311,
|
| 1772 |
+
"step": 1470
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 5.69,
|
| 1776 |
+
"learning_rate": 0.00010804195631589798,
|
| 1777 |
+
"loss": 1.7528,
|
| 1778 |
+
"step": 1475
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 5.71,
|
| 1782 |
+
"learning_rate": 0.00011137140040750908,
|
| 1783 |
+
"loss": 1.7338,
|
| 1784 |
+
"step": 1480
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 5.73,
|
| 1788 |
+
"learning_rate": 0.00011454356918116697,
|
| 1789 |
+
"loss": 1.8454,
|
| 1790 |
+
"step": 1485
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 5.75,
|
| 1794 |
+
"learning_rate": 0.00011754679821046187,
|
| 1795 |
+
"loss": 1.6556,
|
| 1796 |
+
"step": 1490
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 5.77,
|
| 1800 |
+
"learning_rate": 0.00012037004427969469,
|
| 1801 |
+
"loss": 1.7088,
|
| 1802 |
+
"step": 1495
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 5.79,
|
| 1806 |
+
"learning_rate": 0.00012300292599103937,
|
| 1807 |
+
"loss": 1.7949,
|
| 1808 |
+
"step": 1500
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 5.81,
|
| 1812 |
+
"learning_rate": 0.0001254357619381275,
|
| 1813 |
+
"loss": 1.7168,
|
| 1814 |
+
"step": 1505
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 5.83,
|
| 1818 |
+
"learning_rate": 0.00012765960630568417,
|
| 1819 |
+
"loss": 1.8008,
|
| 1820 |
+
"step": 1510
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 5.85,
|
| 1824 |
+
"learning_rate": 0.00012966628176431028,
|
| 1825 |
+
"loss": 1.7668,
|
| 1826 |
+
"step": 1515
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 5.87,
|
| 1830 |
+
"learning_rate": 0.000131448409539456,
|
| 1831 |
+
"loss": 1.8123,
|
| 1832 |
+
"step": 1520
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 5.89,
|
| 1836 |
+
"learning_rate": 0.00013299943654401658,
|
| 1837 |
+
"loss": 1.6967,
|
| 1838 |
+
"step": 1525
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 5.91,
|
| 1842 |
+
"learning_rate": 0.0001343136594747806,
|
| 1843 |
+
"loss": 1.7767,
|
| 1844 |
+
"step": 1530
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 5.93,
|
| 1848 |
+
"learning_rate": 0.00013538624578412676,
|
| 1849 |
+
"loss": 1.8144,
|
| 1850 |
+
"step": 1535
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 5.95,
|
| 1854 |
+
"learning_rate": 0.0001362132514498528,
|
| 1855 |
+
"loss": 1.7673,
|
| 1856 |
+
"step": 1540
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 5.97,
|
| 1860 |
+
"learning_rate": 0.00013679163547779458,
|
| 1861 |
+
"loss": 1.7239,
|
| 1862 |
+
"step": 1545
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 5.98,
|
| 1866 |
+
"learning_rate": 0.00013711927108390887,
|
| 1867 |
+
"loss": 1.7237,
|
| 1868 |
+
"step": 1550
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 6.0,
|
| 1872 |
+
"learning_rate": 0.00013719495351470075,
|
| 1873 |
+
"loss": 1.8186,
|
| 1874 |
+
"step": 1555
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 6.02,
|
| 1878 |
+
"learning_rate": 0.0001370184044772396,
|
| 1879 |
+
"loss": 1.7571,
|
| 1880 |
+
"step": 1560
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 6.04,
|
| 1884 |
+
"learning_rate": 0.00013659027316247397,
|
| 1885 |
+
"loss": 1.6908,
|
| 1886 |
+
"step": 1565
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 6.06,
|
| 1890 |
+
"learning_rate": 0.00013591213385808238,
|
| 1891 |
+
"loss": 1.6378,
|
| 1892 |
+
"step": 1570
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 6.08,
|
| 1896 |
+
"learning_rate": 0.0001349864801596381,
|
| 1897 |
+
"loss": 1.763,
|
| 1898 |
+
"step": 1575
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 6.1,
|
| 1902 |
+
"learning_rate": 0.00013381671580137334,
|
| 1903 |
+
"loss": 1.8114,
|
| 1904 |
+
"step": 1580
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 6.12,
|
| 1908 |
+
"learning_rate": 0.00013240714214026117,
|
| 1909 |
+
"loss": 1.7691,
|
| 1910 |
+
"step": 1585
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 6.14,
|
| 1914 |
+
"learning_rate": 0.00013076294233943417,
|
| 1915 |
+
"loss": 1.7388,
|
| 1916 |
+
"step": 1590
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 6.16,
|
| 1920 |
+
"learning_rate": 0.0001288901623091032,
|
| 1921 |
+
"loss": 1.6458,
|
| 1922 |
+
"step": 1595
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 6.18,
|
| 1926 |
+
"learning_rate": 0.00012679568847505571,
|
| 1927 |
+
"loss": 1.6852,
|
| 1928 |
+
"step": 1600
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 6.2,
|
| 1932 |
+
"learning_rate": 0.00012448722245648225,
|
| 1933 |
+
"loss": 1.7267,
|
| 1934 |
+
"step": 1605
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 6.22,
|
| 1938 |
+
"learning_rate": 0.00012197325274624507,
|
| 1939 |
+
"loss": 1.7517,
|
| 1940 |
+
"step": 1610
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 6.24,
|
| 1944 |
+
"learning_rate": 0.00011926302349772057,
|
| 1945 |
+
"loss": 1.7343,
|
| 1946 |
+
"step": 1615
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 6.25,
|
| 1950 |
+
"learning_rate": 0.0001163665005329939,
|
| 1951 |
+
"loss": 1.6811,
|
| 1952 |
+
"step": 1620
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 6.27,
|
| 1956 |
+
"learning_rate": 0.00011329433469739406,
|
| 1957 |
+
"loss": 1.7056,
|
| 1958 |
+
"step": 1625
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 6.29,
|
| 1962 |
+
"learning_rate": 0.00011005782269511991,
|
| 1963 |
+
"loss": 1.7447,
|
| 1964 |
+
"step": 1630
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 6.31,
|
| 1968 |
+
"learning_rate": 0.00010666886554997244,
|
| 1969 |
+
"loss": 1.6661,
|
| 1970 |
+
"step": 1635
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 6.33,
|
| 1974 |
+
"learning_rate": 0.00010313992484393024,
|
| 1975 |
+
"loss": 1.723,
|
| 1976 |
+
"step": 1640
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 6.35,
|
| 1980 |
+
"learning_rate": 9.948397689449228e-05,
|
| 1981 |
+
"loss": 1.6887,
|
| 1982 |
+
"step": 1645
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 6.37,
|
| 1986 |
+
"learning_rate": 9.571446503927964e-05,
|
| 1987 |
+
"loss": 1.6767,
|
| 1988 |
+
"step": 1650
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 6.39,
|
| 1992 |
+
"learning_rate": 9.184525020334699e-05,
|
| 1993 |
+
"loss": 1.6593,
|
| 1994 |
+
"step": 1655
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 6.41,
|
| 1998 |
+
"learning_rate": 8.789055993098258e-05,
|
| 1999 |
+
"loss": 1.6807,
|
| 2000 |
+
"step": 1660
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 6.43,
|
| 2004 |
+
"learning_rate": 8.386493606940322e-05,
|
| 2005 |
+
"loss": 1.7043,
|
| 2006 |
+
"step": 1665
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 6.45,
|
| 2010 |
+
"learning_rate": 7.978318129672484e-05,
|
| 2011 |
+
"loss": 1.7188,
|
| 2012 |
+
"step": 1670
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 6.47,
|
| 2016 |
+
"learning_rate": 7.566030469082603e-05,
|
| 2017 |
+
"loss": 1.6494,
|
| 2018 |
+
"step": 1675
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 6.49,
|
| 2022 |
+
"learning_rate": 7.151146653925584e-05,
|
| 2023 |
+
"loss": 1.6752,
|
| 2024 |
+
"step": 1680
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 6.51,
|
| 2028 |
+
"learning_rate": 6.735192259312878e-05,
|
| 2029 |
+
"loss": 1.6569,
|
| 2030 |
+
"step": 1685
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 6.53,
|
| 2034 |
+
"learning_rate": 6.319696796998709e-05,
|
| 2035 |
+
"loss": 1.6728,
|
| 2036 |
+
"step": 1690
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 6.54,
|
| 2040 |
+
"learning_rate": 5.906188091190817e-05,
|
| 2041 |
+
"loss": 1.6875,
|
| 2042 |
+
"step": 1695
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 6.56,
|
| 2046 |
+
"learning_rate": 5.4961866605667284e-05,
|
| 2047 |
+
"loss": 1.6511,
|
| 2048 |
+
"step": 1700
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 6.58,
|
| 2052 |
+
"learning_rate": 5.091200127153063e-05,
|
| 2053 |
+
"loss": 1.6906,
|
| 2054 |
+
"step": 1705
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 6.6,
|
| 2058 |
+
"learning_rate": 4.6927176726273094e-05,
|
| 2059 |
+
"loss": 1.6586,
|
| 2060 |
+
"step": 1710
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 6.62,
|
| 2064 |
+
"learning_rate": 4.302204562427082e-05,
|
| 2065 |
+
"loss": 1.6804,
|
| 2066 |
+
"step": 1715
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 6.64,
|
| 2070 |
+
"learning_rate": 3.921096757801896e-05,
|
| 2071 |
+
"loss": 1.6353,
|
| 2072 |
+
"step": 1720
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 6.66,
|
| 2076 |
+
"learning_rate": 3.550795635619796e-05,
|
| 2077 |
+
"loss": 1.6895,
|
| 2078 |
+
"step": 1725
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 6.68,
|
| 2082 |
+
"learning_rate": 3.192662835344908e-05,
|
| 2083 |
+
"loss": 1.7331,
|
| 2084 |
+
"step": 1730
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 6.7,
|
| 2088 |
+
"learning_rate": 2.8480152521337155e-05,
|
| 2089 |
+
"loss": 1.6892,
|
| 2090 |
+
"step": 1735
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 6.72,
|
| 2094 |
+
"learning_rate": 2.51812019446141e-05,
|
| 2095 |
+
"loss": 1.6841,
|
| 2096 |
+
"step": 1740
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 6.74,
|
| 2100 |
+
"learning_rate": 2.2041907240840133e-05,
|
| 2101 |
+
"loss": 1.7096,
|
| 2102 |
+
"step": 1745
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 6.76,
|
| 2106 |
+
"learning_rate": 1.907381195471957e-05,
|
| 2107 |
+
"loss": 1.7284,
|
| 2108 |
+
"step": 1750
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 6.78,
|
| 2112 |
+
"learning_rate": 1.6287830111171488e-05,
|
| 2113 |
+
"loss": 1.6272,
|
| 2114 |
+
"step": 1755
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 6.8,
|
| 2118 |
+
"learning_rate": 1.3694206083212888e-05,
|
| 2119 |
+
"loss": 1.5783,
|
| 2120 |
+
"step": 1760
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 6.81,
|
| 2124 |
+
"learning_rate": 1.1302476922232546e-05,
|
| 2125 |
+
"loss": 1.5607,
|
| 2126 |
+
"step": 1765
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 6.83,
|
| 2130 |
+
"learning_rate": 9.121437289164463e-06,
|
| 2131 |
+
"loss": 1.6762,
|
| 2132 |
+
"step": 1770
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 6.85,
|
| 2136 |
+
"learning_rate": 7.159107115516178e-06,
|
| 2137 |
+
"loss": 1.6488,
|
| 2138 |
+
"step": 1775
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 6.87,
|
| 2142 |
+
"learning_rate": 5.422702113166627e-06,
|
| 2143 |
+
"loss": 1.6201,
|
| 2144 |
+
"step": 1780
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 6.89,
|
| 2148 |
+
"learning_rate": 3.918607241369662e-06,
|
| 2149 |
+
"loss": 1.7022,
|
| 2150 |
+
"step": 1785
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 6.91,
|
| 2154 |
+
"learning_rate": 2.65235322853129e-06,
|
| 2155 |
+
"loss": 1.632,
|
| 2156 |
+
"step": 1790
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 6.93,
|
| 2160 |
+
"learning_rate": 1.6285962350901147e-06,
|
| 2161 |
+
"loss": 1.6661,
|
| 2162 |
+
"step": 1795
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 6.95,
|
| 2166 |
+
"learning_rate": 8.511007322841488e-07,
|
| 2167 |
+
"loss": 1.6079,
|
| 2168 |
+
"step": 1800
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 6.97,
|
| 2172 |
+
"learning_rate": 3.2272565976125165e-07,
|
| 2173 |
+
"loss": 1.6758,
|
| 2174 |
+
"step": 1805
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 6.99,
|
| 2178 |
+
"learning_rate": 4.541391293127461e-08,
|
| 2179 |
+
"loss": 1.6987,
|
| 2180 |
+
"step": 1810
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 7.01,
|
| 2184 |
+
"learning_rate": 2.018519871846962e-08,
|
| 2185 |
+
"loss": 1.5689,
|
| 2186 |
+
"step": 1815
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 7.03,
|
| 2190 |
+
"learning_rate": 2.471322859826806e-07,
|
| 2191 |
+
"loss": 1.5376,
|
| 2192 |
+
"step": 1820
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 7.05,
|
| 2196 |
+
"learning_rate": 7.254206643976737e-07,
|
| 2197 |
+
"loss": 1.6088,
|
| 2198 |
+
"step": 1825
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 7.07,
|
| 2202 |
+
"learning_rate": 1.4532916130407314e-06,
|
| 2203 |
+
"loss": 1.6132,
|
| 2204 |
+
"step": 1830
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 7.08,
|
| 2208 |
+
"learning_rate": 2.4280686674102744e-06,
|
| 2209 |
+
"loss": 1.6823,
|
| 2210 |
+
"step": 1835
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 7.1,
|
| 2214 |
+
"learning_rate": 3.6461674610908866e-06,
|
| 2215 |
+
"loss": 1.5818,
|
| 2216 |
+
"step": 1840
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 7.12,
|
| 2220 |
+
"learning_rate": 5.103108905877507e-06,
|
| 2221 |
+
"loss": 1.5745,
|
| 2222 |
+
"step": 1845
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 7.14,
|
| 2226 |
+
"learning_rate": 6.793535661894024e-06,
|
| 2227 |
+
"loss": 1.5783,
|
| 2228 |
+
"step": 1850
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 7.16,
|
| 2232 |
+
"learning_rate": 8.7112318371425e-06,
|
| 2233 |
+
"loss": 1.6224,
|
| 2234 |
+
"step": 1855
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 7.18,
|
| 2238 |
+
"learning_rate": 1.0849145844047318e-05,
|
| 2239 |
+
"loss": 1.6016,
|
| 2240 |
+
"step": 1860
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 7.2,
|
| 2244 |
+
"learning_rate": 1.3199416328947412e-05,
|
| 2245 |
+
"loss": 1.5826,
|
| 2246 |
+
"step": 1865
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 7.22,
|
| 2250 |
+
"learning_rate": 1.5753401079189635e-05,
|
| 2251 |
+
"loss": 1.6424,
|
| 2252 |
+
"step": 1870
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 7.24,
|
| 2256 |
+
"learning_rate": 1.8501708801530793e-05,
|
| 2257 |
+
"loss": 1.5944,
|
| 2258 |
+
"step": 1875
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 7.26,
|
| 2262 |
+
"learning_rate": 2.1434233654994585e-05,
|
| 2263 |
+
"loss": 1.5702,
|
| 2264 |
+
"step": 1880
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 7.28,
|
| 2268 |
+
"learning_rate": 2.454019241120068e-05,
|
| 2269 |
+
"loss": 1.5819,
|
| 2270 |
+
"step": 1885
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 7.3,
|
| 2274 |
+
"learning_rate": 2.780816410552581e-05,
|
| 2275 |
+
"loss": 1.5461,
|
| 2276 |
+
"step": 1890
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 7.32,
|
| 2280 |
+
"learning_rate": 3.12261320332941e-05,
|
| 2281 |
+
"loss": 1.6276,
|
| 2282 |
+
"step": 1895
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 7.34,
|
| 2286 |
+
"learning_rate": 3.4781527936569615e-05,
|
| 2287 |
+
"loss": 1.6333,
|
| 2288 |
+
"step": 1900
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 7.36,
|
| 2292 |
+
"learning_rate": 3.8461278219075155e-05,
|
| 2293 |
+
"loss": 1.5744,
|
| 2294 |
+
"step": 1905
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 7.37,
|
| 2298 |
+
"learning_rate": 4.2251852019296586e-05,
|
| 2299 |
+
"loss": 1.601,
|
| 2300 |
+
"step": 1910
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 7.39,
|
| 2304 |
+
"learning_rate": 4.6139310965004655e-05,
|
| 2305 |
+
"loss": 1.4994,
|
| 2306 |
+
"step": 1915
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 7.41,
|
| 2310 |
+
"learning_rate": 5.010936042623934e-05,
|
| 2311 |
+
"loss": 1.5667,
|
| 2312 |
+
"step": 1920
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 7.43,
|
| 2316 |
+
"learning_rate": 5.4147402078293086e-05,
|
| 2317 |
+
"loss": 1.6055,
|
| 2318 |
+
"step": 1925
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 7.45,
|
| 2322 |
+
"learning_rate": 5.823858758141886e-05,
|
| 2323 |
+
"loss": 1.6403,
|
| 2324 |
+
"step": 1930
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 7.47,
|
| 2328 |
+
"learning_rate": 6.236787317986658e-05,
|
| 2329 |
+
"loss": 1.5103,
|
| 2330 |
+
"step": 1935
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 7.49,
|
| 2334 |
+
"learning_rate": 6.65200750194898e-05,
|
| 2335 |
+
"loss": 1.5918,
|
| 2336 |
+
"step": 1940
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 7.51,
|
| 2340 |
+
"learning_rate": 7.067992498051008e-05,
|
| 2341 |
+
"loss": 1.5905,
|
| 2342 |
+
"step": 1945
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 7.53,
|
| 2346 |
+
"learning_rate": 7.48321268201333e-05,
|
| 2347 |
+
"loss": 1.6157,
|
| 2348 |
+
"step": 1950
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 7.55,
|
| 2352 |
+
"learning_rate": 7.896141241858101e-05,
|
| 2353 |
+
"loss": 1.5963,
|
| 2354 |
+
"step": 1955
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 7.57,
|
| 2358 |
+
"learning_rate": 8.305259792170679e-05,
|
| 2359 |
+
"loss": 1.586,
|
| 2360 |
+
"step": 1960
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 7.59,
|
| 2364 |
+
"learning_rate": 8.709063957376054e-05,
|
| 2365 |
+
"loss": 1.6324,
|
| 2366 |
+
"step": 1965
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 7.61,
|
| 2370 |
+
"learning_rate": 9.106068903499522e-05,
|
| 2371 |
+
"loss": 1.6232,
|
| 2372 |
+
"step": 1970
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 7.63,
|
| 2376 |
+
"learning_rate": 9.494814798070329e-05,
|
| 2377 |
+
"loss": 1.6404,
|
| 2378 |
+
"step": 1975
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 7.64,
|
| 2382 |
+
"learning_rate": 9.873872178092473e-05,
|
| 2383 |
+
"loss": 1.6215,
|
| 2384 |
+
"step": 1980
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 7.66,
|
| 2388 |
+
"learning_rate": 0.00010241847206343028,
|
| 2389 |
+
"loss": 1.617,
|
| 2390 |
+
"step": 1985
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 7.68,
|
| 2394 |
+
"learning_rate": 0.0001059738679667058,
|
| 2395 |
+
"loss": 1.6565,
|
| 2396 |
+
"step": 1990
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 7.7,
|
| 2400 |
+
"learning_rate": 0.0001093918358944741,
|
| 2401 |
+
"loss": 1.7342,
|
| 2402 |
+
"step": 1995
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 7.72,
|
| 2406 |
+
"learning_rate": 0.00011265980758879924,
|
| 2407 |
+
"loss": 1.6063,
|
| 2408 |
+
"step": 2000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 7.74,
|
| 2412 |
+
"learning_rate": 0.00011576576634500532,
|
| 2413 |
+
"loss": 1.6993,
|
| 2414 |
+
"step": 2005
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 7.76,
|
| 2418 |
+
"learning_rate": 0.00011869829119846911,
|
| 2419 |
+
"loss": 1.6355,
|
| 2420 |
+
"step": 2010
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 7.78,
|
| 2424 |
+
"learning_rate": 0.00012144659892081027,
|
| 2425 |
+
"loss": 1.6087,
|
| 2426 |
+
"step": 2015
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 7.8,
|
| 2430 |
+
"learning_rate": 0.00012400058367105252,
|
| 2431 |
+
"loss": 1.6668,
|
| 2432 |
+
"step": 2020
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 7.82,
|
| 2436 |
+
"learning_rate": 0.00012635085415595263,
|
| 2437 |
+
"loss": 1.7275,
|
| 2438 |
+
"step": 2025
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 7.84,
|
| 2442 |
+
"learning_rate": 0.00012848876816285744,
|
| 2443 |
+
"loss": 1.6637,
|
| 2444 |
+
"step": 2030
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 7.86,
|
| 2448 |
+
"learning_rate": 0.00013040646433810593,
|
| 2449 |
+
"loss": 1.6713,
|
| 2450 |
+
"step": 2035
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 7.88,
|
| 2454 |
+
"learning_rate": 0.00013209689109412246,
|
| 2455 |
+
"loss": 1.6358,
|
| 2456 |
+
"step": 2040
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 7.9,
|
| 2460 |
+
"learning_rate": 0.00013355383253890908,
|
| 2461 |
+
"loss": 1.6572,
|
| 2462 |
+
"step": 2045
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 7.92,
|
| 2466 |
+
"learning_rate": 0.0001347719313325897,
|
| 2467 |
+
"loss": 1.6781,
|
| 2468 |
+
"step": 2050
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 7.93,
|
| 2472 |
+
"learning_rate": 0.00013574670838695924,
|
| 2473 |
+
"loss": 1.6401,
|
| 2474 |
+
"step": 2055
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 7.95,
|
| 2478 |
+
"learning_rate": 0.0001364745793356023,
|
| 2479 |
+
"loss": 1.673,
|
| 2480 |
+
"step": 2060
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 7.97,
|
| 2484 |
+
"learning_rate": 0.0001369528677140173,
|
| 2485 |
+
"loss": 1.7179,
|
| 2486 |
+
"step": 2065
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 7.99,
|
| 2490 |
+
"learning_rate": 0.00013717981480128154,
|
| 2491 |
+
"loss": 1.7015,
|
| 2492 |
+
"step": 2070
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 8.01,
|
| 2496 |
+
"learning_rate": 0.00013715458608706872,
|
| 2497 |
+
"loss": 1.6596,
|
| 2498 |
+
"step": 2075
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 8.03,
|
| 2502 |
+
"learning_rate": 0.00013687727434023877,
|
| 2503 |
+
"loss": 1.6462,
|
| 2504 |
+
"step": 2080
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 8.05,
|
| 2508 |
+
"learning_rate": 0.00013634889926771588,
|
| 2509 |
+
"loss": 1.6041,
|
| 2510 |
+
"step": 2085
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 8.07,
|
| 2514 |
+
"learning_rate": 0.00013557140376490998,
|
| 2515 |
+
"loss": 1.5571,
|
| 2516 |
+
"step": 2090
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 8.09,
|
| 2520 |
+
"learning_rate": 0.00013454764677146882,
|
| 2521 |
+
"loss": 1.5541,
|
| 2522 |
+
"step": 2095
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 8.11,
|
| 2526 |
+
"learning_rate": 0.00013328139275863037,
|
| 2527 |
+
"loss": 1.6773,
|
| 2528 |
+
"step": 2100
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 8.13,
|
| 2532 |
+
"learning_rate": 0.00013177729788683341,
|
| 2533 |
+
"loss": 1.6003,
|
| 2534 |
+
"step": 2105
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 8.15,
|
| 2538 |
+
"learning_rate": 0.00013004089288448387,
|
| 2539 |
+
"loss": 1.5288,
|
| 2540 |
+
"step": 2110
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 8.17,
|
| 2544 |
+
"learning_rate": 0.0001280785627108356,
|
| 2545 |
+
"loss": 1.6059,
|
| 2546 |
+
"step": 2115
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 8.19,
|
| 2550 |
+
"learning_rate": 0.00012589752307776752,
|
| 2551 |
+
"loss": 1.6265,
|
| 2552 |
+
"step": 2120
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 8.2,
|
| 2556 |
+
"learning_rate": 0.0001235057939167872,
|
| 2557 |
+
"loss": 1.6372,
|
| 2558 |
+
"step": 2125
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 8.22,
|
| 2562 |
+
"learning_rate": 0.00012091216988882845,
|
| 2563 |
+
"loss": 1.7324,
|
| 2564 |
+
"step": 2130
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 8.24,
|
| 2568 |
+
"learning_rate": 0.00011812618804528034,
|
| 2569 |
+
"loss": 1.5938,
|
| 2570 |
+
"step": 2135
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 8.26,
|
| 2574 |
+
"learning_rate": 0.00011515809275915997,
|
| 2575 |
+
"loss": 1.5805,
|
| 2576 |
+
"step": 2140
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 8.28,
|
| 2580 |
+
"learning_rate": 0.00011201879805538599,
|
| 2581 |
+
"loss": 1.62,
|
| 2582 |
+
"step": 2145
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 8.3,
|
| 2586 |
+
"learning_rate": 0.00010871984747866294,
|
| 2587 |
+
"loss": 1.5884,
|
| 2588 |
+
"step": 2150
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 8.32,
|
| 2592 |
+
"learning_rate": 0.00010527337164655102,
|
| 2593 |
+
"loss": 1.576,
|
| 2594 |
+
"step": 2155
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 8.34,
|
| 2598 |
+
"learning_rate": 0.00010169204364380236,
|
| 2599 |
+
"loss": 1.5746,
|
| 2600 |
+
"step": 2160
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 8.36,
|
| 2604 |
+
"learning_rate": 9.798903242198116e-05,
|
| 2605 |
+
"loss": 1.6259,
|
| 2606 |
+
"step": 2165
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 8.38,
|
| 2610 |
+
"learning_rate": 9.417795437572906e-05,
|
| 2611 |
+
"loss": 1.6047,
|
| 2612 |
+
"step": 2170
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 8.4,
|
| 2616 |
+
"learning_rate": 9.027282327372703e-05,
|
| 2617 |
+
"loss": 1.5615,
|
| 2618 |
+
"step": 2175
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 8.42,
|
| 2622 |
+
"learning_rate": 8.628799872846948e-05,
|
| 2623 |
+
"loss": 1.6294,
|
| 2624 |
+
"step": 2180
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 8.44,
|
| 2628 |
+
"learning_rate": 8.223813339433283e-05,
|
| 2629 |
+
"loss": 1.5485,
|
| 2630 |
+
"step": 2185
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 8.46,
|
| 2634 |
+
"learning_rate": 7.813811908809194e-05,
|
| 2635 |
+
"loss": 1.5291,
|
| 2636 |
+
"step": 2190
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 8.47,
|
| 2640 |
+
"learning_rate": 7.400303203001327e-05,
|
| 2641 |
+
"loss": 1.6178,
|
| 2642 |
+
"step": 2195
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 8.49,
|
| 2646 |
+
"learning_rate": 6.98480774068711e-05,
|
| 2647 |
+
"loss": 1.6335,
|
| 2648 |
+
"step": 2200
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 8.51,
|
| 2652 |
+
"learning_rate": 6.568853346074429e-05,
|
| 2653 |
+
"loss": 1.5607,
|
| 2654 |
+
"step": 2205
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 8.53,
|
| 2658 |
+
"learning_rate": 6.15396953091741e-05,
|
| 2659 |
+
"loss": 1.552,
|
| 2660 |
+
"step": 2210
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 8.55,
|
| 2664 |
+
"learning_rate": 5.741681870327528e-05,
|
| 2665 |
+
"loss": 1.6358,
|
| 2666 |
+
"step": 2215
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 8.57,
|
| 2670 |
+
"learning_rate": 5.33350639305969e-05,
|
| 2671 |
+
"loss": 1.6499,
|
| 2672 |
+
"step": 2220
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 8.59,
|
| 2676 |
+
"learning_rate": 4.930944006901777e-05,
|
| 2677 |
+
"loss": 1.5632,
|
| 2678 |
+
"step": 2225
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 8.61,
|
| 2682 |
+
"learning_rate": 4.535474979665314e-05,
|
| 2683 |
+
"loss": 1.5825,
|
| 2684 |
+
"step": 2230
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 8.63,
|
| 2688 |
+
"learning_rate": 4.148553496072023e-05,
|
| 2689 |
+
"loss": 1.6277,
|
| 2690 |
+
"step": 2235
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 8.65,
|
| 2694 |
+
"learning_rate": 3.7716023105507615e-05,
|
| 2695 |
+
"loss": 1.5497,
|
| 2696 |
+
"step": 2240
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 8.67,
|
| 2700 |
+
"learning_rate": 3.406007515606987e-05,
|
| 2701 |
+
"loss": 1.5159,
|
| 2702 |
+
"step": 2245
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 8.69,
|
| 2706 |
+
"learning_rate": 3.0531134450027666e-05,
|
| 2707 |
+
"loss": 1.5683,
|
| 2708 |
+
"step": 2250
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 8.71,
|
| 2712 |
+
"learning_rate": 2.7142177304880198e-05,
|
| 2713 |
+
"loss": 1.5193,
|
| 2714 |
+
"step": 2255
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 8.73,
|
| 2718 |
+
"learning_rate": 2.390566530260624e-05,
|
| 2719 |
+
"loss": 1.6145,
|
| 2720 |
+
"step": 2260
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 8.75,
|
| 2724 |
+
"learning_rate": 2.0833499467006378e-05,
|
| 2725 |
+
"loss": 1.5854,
|
| 2726 |
+
"step": 2265
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 8.76,
|
| 2730 |
+
"learning_rate": 1.7936976502279525e-05,
|
| 2731 |
+
"loss": 1.5426,
|
| 2732 |
+
"step": 2270
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 8.78,
|
| 2736 |
+
"learning_rate": 1.5226747253755011e-05,
|
| 2737 |
+
"loss": 1.5862,
|
| 2738 |
+
"step": 2275
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 8.8,
|
| 2742 |
+
"learning_rate": 1.2712777543517822e-05,
|
| 2743 |
+
"loss": 1.5478,
|
| 2744 |
+
"step": 2280
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 8.82,
|
| 2748 |
+
"learning_rate": 1.0404311524944368e-05,
|
| 2749 |
+
"loss": 1.6329,
|
| 2750 |
+
"step": 2285
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 8.84,
|
| 2754 |
+
"learning_rate": 8.309837690896873e-06,
|
| 2755 |
+
"loss": 1.535,
|
| 2756 |
+
"step": 2290
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 8.86,
|
| 2760 |
+
"learning_rate": 6.43705766056588e-06,
|
| 2761 |
+
"loss": 1.5849,
|
| 2762 |
+
"step": 2295
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 8.88,
|
| 2766 |
+
"learning_rate": 4.792857859738948e-06,
|
| 2767 |
+
"loss": 1.6253,
|
| 2768 |
+
"step": 2300
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 8.9,
|
| 2772 |
+
"learning_rate": 3.3832841986266175e-06,
|
| 2773 |
+
"loss": 1.537,
|
| 2774 |
+
"step": 2305
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 8.92,
|
| 2778 |
+
"learning_rate": 2.213519840361947e-06,
|
| 2779 |
+
"loss": 1.5028,
|
| 2780 |
+
"step": 2310
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 8.94,
|
| 2784 |
+
"learning_rate": 1.2878661419176351e-06,
|
| 2785 |
+
"loss": 1.5904,
|
| 2786 |
+
"step": 2315
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 8.96,
|
| 2790 |
+
"learning_rate": 6.097268375260679e-07,
|
| 2791 |
+
"loss": 1.6009,
|
| 2792 |
+
"step": 2320
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 8.98,
|
| 2796 |
+
"learning_rate": 1.8159552276040752e-07,
|
| 2797 |
+
"loss": 1.5465,
|
| 2798 |
+
"step": 2325
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 9.0,
|
| 2802 |
+
"learning_rate": 5.046485299251069e-09,
|
| 2803 |
+
"loss": 1.5575,
|
| 2804 |
+
"step": 2330
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 9.02,
|
| 2808 |
+
"learning_rate": 8.072891609113784e-08,
|
| 2809 |
+
"loss": 1.5791,
|
| 2810 |
+
"step": 2335
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 9.03,
|
| 2814 |
+
"learning_rate": 4.0836452220544814e-07,
|
| 2815 |
+
"loss": 1.4865,
|
| 2816 |
+
"step": 2340
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 9.05,
|
| 2820 |
+
"learning_rate": 9.867485501471922e-07,
|
| 2821 |
+
"loss": 1.5316,
|
| 2822 |
+
"step": 2345
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 9.07,
|
| 2826 |
+
"learning_rate": 1.813754215873199e-06,
|
| 2827 |
+
"loss": 1.5403,
|
| 2828 |
+
"step": 2350
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 9.09,
|
| 2832 |
+
"learning_rate": 2.8863405252193584e-06,
|
| 2833 |
+
"loss": 1.4183,
|
| 2834 |
+
"step": 2355
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 9.11,
|
| 2838 |
+
"learning_rate": 4.200563455983382e-06,
|
| 2839 |
+
"loss": 1.5547,
|
| 2840 |
+
"step": 2360
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 9.13,
|
| 2844 |
+
"learning_rate": 5.75159046054386e-06,
|
| 2845 |
+
"loss": 1.5414,
|
| 2846 |
+
"step": 2365
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 9.15,
|
| 2850 |
+
"learning_rate": 7.5337182356897725e-06,
|
| 2851 |
+
"loss": 1.5445,
|
| 2852 |
+
"step": 2370
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 9.17,
|
| 2856 |
+
"learning_rate": 9.540393694315775e-06,
|
| 2857 |
+
"loss": 1.4539,
|
| 2858 |
+
"step": 2375
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 9.19,
|
| 2862 |
+
"learning_rate": 1.1764238061872442e-05,
|
| 2863 |
+
"loss": 1.4992,
|
| 2864 |
+
"step": 2380
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 9.21,
|
| 2868 |
+
"learning_rate": 1.4197074008960564e-05,
|
| 2869 |
+
"loss": 1.5203,
|
| 2870 |
+
"step": 2385
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 9.23,
|
| 2874 |
+
"learning_rate": 1.6829955720305234e-05,
|
| 2875 |
+
"loss": 1.4989,
|
| 2876 |
+
"step": 2390
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 9.25,
|
| 2880 |
+
"learning_rate": 1.965320178953787e-05,
|
| 2881 |
+
"loss": 1.5128,
|
| 2882 |
+
"step": 2395
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 9.27,
|
| 2886 |
+
"learning_rate": 2.265643081883295e-05,
|
| 2887 |
+
"loss": 1.5033,
|
| 2888 |
+
"step": 2400
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 9.29,
|
| 2892 |
+
"learning_rate": 2.582859959249101e-05,
|
| 2893 |
+
"loss": 1.4938,
|
| 2894 |
+
"step": 2405
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 9.31,
|
| 2898 |
+
"learning_rate": 2.915804368410211e-05,
|
| 2899 |
+
"loss": 1.5157,
|
| 2900 |
+
"step": 2410
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 9.32,
|
| 2904 |
+
"learning_rate": 3.2632520347973973e-05,
|
| 2905 |
+
"loss": 1.4103,
|
| 2906 |
+
"step": 2415
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 9.34,
|
| 2910 |
+
"learning_rate": 3.623925353710222e-05,
|
| 2911 |
+
"loss": 1.524,
|
| 2912 |
+
"step": 2420
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 9.36,
|
| 2916 |
+
"learning_rate": 3.996498088215406e-05,
|
| 2917 |
+
"loss": 1.5389,
|
| 2918 |
+
"step": 2425
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 9.38,
|
| 2922 |
+
"learning_rate": 4.3796002458712527e-05,
|
| 2923 |
+
"loss": 1.5645,
|
| 2924 |
+
"step": 2430
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 9.4,
|
| 2928 |
+
"learning_rate": 4.7718231163460484e-05,
|
| 2929 |
+
"loss": 1.5511,
|
| 2930 |
+
"step": 2435
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 9.42,
|
| 2934 |
+
"learning_rate": 5.1717244514068206e-05,
|
| 2935 |
+
"loss": 1.5406,
|
| 2936 |
+
"step": 2440
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 9.44,
|
| 2940 |
+
"learning_rate": 5.57783376823034e-05,
|
| 2941 |
+
"loss": 1.567,
|
| 2942 |
+
"step": 2445
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 9.46,
|
| 2946 |
+
"learning_rate": 5.988657756536402e-05,
|
| 2947 |
+
"loss": 1.602,
|
| 2948 |
+
"step": 2450
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 9.48,
|
| 2952 |
+
"learning_rate": 6.402685769660036e-05,
|
| 2953 |
+
"loss": 1.4789,
|
| 2954 |
+
"step": 2455
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 9.5,
|
| 2958 |
+
"learning_rate": 6.818395379371463e-05,
|
| 2959 |
+
"loss": 1.5673,
|
| 2960 |
+
"step": 2460
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 9.52,
|
| 2964 |
+
"learning_rate": 7.234257974018531e-05,
|
| 2965 |
+
"loss": 1.5527,
|
| 2966 |
+
"step": 2465
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 9.54,
|
| 2970 |
+
"learning_rate": 7.64874437940594e-05,
|
| 2971 |
+
"loss": 1.4721,
|
| 2972 |
+
"step": 2470
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 9.56,
|
| 2976 |
+
"learning_rate": 8.060330481743391e-05,
|
| 2977 |
+
"loss": 1.4447,
|
| 2978 |
+
"step": 2475
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 9.58,
|
| 2982 |
+
"learning_rate": 8.467502831985544e-05,
|
| 2983 |
+
"loss": 1.5768,
|
| 2984 |
+
"step": 2480
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 9.59,
|
| 2988 |
+
"learning_rate": 8.868764210957132e-05,
|
| 2989 |
+
"loss": 1.4808,
|
| 2990 |
+
"step": 2485
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 9.61,
|
| 2994 |
+
"learning_rate": 9.262639134798362e-05,
|
| 2995 |
+
"loss": 1.4197,
|
| 2996 |
+
"step": 2490
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 9.63,
|
| 3000 |
+
"learning_rate": 9.647679280487567e-05,
|
| 3001 |
+
"loss": 1.6109,
|
| 3002 |
+
"step": 2495
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 9.65,
|
| 3006 |
+
"learning_rate": 0.00010022468811489983,
|
| 3007 |
+
"loss": 1.5907,
|
| 3008 |
+
"step": 2500
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 9.67,
|
| 3012 |
+
"learning_rate": 0.00010385629583950378,
|
| 3013 |
+
"loss": 1.5902,
|
| 3014 |
+
"step": 2505
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 9.69,
|
| 3018 |
+
"learning_rate": 0.00010735826214284965,
|
| 3019 |
+
"loss": 1.6053,
|
| 3020 |
+
"step": 2510
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 9.71,
|
| 3024 |
+
"learning_rate": 0.00011071770989539361,
|
| 3025 |
+
"loss": 1.552,
|
| 3026 |
+
"step": 2515
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 9.73,
|
| 3030 |
+
"learning_rate": 0.00011392228602455961,
|
| 3031 |
+
"loss": 1.5787,
|
| 3032 |
+
"step": 2520
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 9.75,
|
| 3036 |
+
"learning_rate": 0.00011696020693839523,
|
| 3037 |
+
"loss": 1.4997,
|
| 3038 |
+
"step": 2525
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 9.77,
|
| 3042 |
+
"learning_rate": 0.00011982030185518476,
|
| 3043 |
+
"loss": 1.6354,
|
| 3044 |
+
"step": 2530
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 9.79,
|
| 3048 |
+
"learning_rate": 0.00012249205387968647,
|
| 3049 |
+
"loss": 1.586,
|
| 3050 |
+
"step": 2535
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 9.81,
|
| 3054 |
+
"learning_rate": 0.0001249656386749574,
|
| 3055 |
+
"loss": 1.511,
|
| 3056 |
+
"step": 2540
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 9.83,
|
| 3060 |
+
"learning_rate": 0.000127231960587559,
|
| 3061 |
+
"loss": 1.5002,
|
| 3062 |
+
"step": 2545
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 9.85,
|
| 3066 |
+
"learning_rate": 0.00012928268609331444,
|
| 3067 |
+
"loss": 1.5829,
|
| 3068 |
+
"step": 2550
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 9.86,
|
| 3072 |
+
"learning_rate": 0.00013111027444062605,
|
| 3073 |
+
"loss": 1.6407,
|
| 3074 |
+
"step": 2555
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 9.88,
|
| 3078 |
+
"learning_rate": 0.00013270800537867978,
|
| 3079 |
+
"loss": 1.5058,
|
| 3080 |
+
"step": 2560
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 9.9,
|
| 3084 |
+
"learning_rate": 0.00013407000386857348,
|
| 3085 |
+
"loss": 1.4854,
|
| 3086 |
+
"step": 2565
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 9.92,
|
| 3090 |
+
"learning_rate": 0.0001351912616865057,
|
| 3091 |
+
"loss": 1.4912,
|
| 3092 |
+
"step": 2570
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 9.94,
|
| 3096 |
+
"learning_rate": 0.00013606765583958525,
|
| 3097 |
+
"loss": 1.5218,
|
| 3098 |
+
"step": 2575
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 9.96,
|
| 3102 |
+
"learning_rate": 0.00013669596372654658,
|
| 3103 |
+
"loss": 1.5828,
|
| 3104 |
+
"step": 2580
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 9.98,
|
| 3108 |
+
"learning_rate": 0.00013707387498762246,
|
| 3109 |
+
"loss": 1.5816,
|
| 3110 |
+
"step": 2585
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 10.0,
|
| 3114 |
+
"learning_rate": 0.0001372,
|
| 3115 |
+
"loss": 1.4944,
|
| 3116 |
+
"step": 2590
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 10.02,
|
| 3120 |
+
"learning_rate": 0.00013707387498762246,
|
| 3121 |
+
"loss": 1.5296,
|
| 3122 |
+
"step": 2595
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 10.04,
|
| 3126 |
+
"learning_rate": 0.0001366959637265466,
|
| 3127 |
+
"loss": 1.5509,
|
| 3128 |
+
"step": 2600
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 10.06,
|
| 3132 |
+
"learning_rate": 0.00013606765583958527,
|
| 3133 |
+
"loss": 1.5305,
|
| 3134 |
+
"step": 2605
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 10.08,
|
| 3138 |
+
"learning_rate": 0.00013519126168650574,
|
| 3139 |
+
"loss": 1.4872,
|
| 3140 |
+
"step": 2610
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 10.1,
|
| 3144 |
+
"learning_rate": 0.00013407000386857353,
|
| 3145 |
+
"loss": 1.544,
|
| 3146 |
+
"step": 2615
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 10.12,
|
| 3150 |
+
"learning_rate": 0.00013270800537867983,
|
| 3151 |
+
"loss": 1.5421,
|
| 3152 |
+
"step": 2620
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 10.14,
|
| 3156 |
+
"learning_rate": 0.0001311102744406261,
|
| 3157 |
+
"loss": 1.5468,
|
| 3158 |
+
"step": 2625
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 10.15,
|
| 3162 |
+
"learning_rate": 0.00012928268609331455,
|
| 3163 |
+
"loss": 1.5529,
|
| 3164 |
+
"step": 2630
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 10.17,
|
| 3168 |
+
"learning_rate": 0.00012723196058755907,
|
| 3169 |
+
"loss": 1.5357,
|
| 3170 |
+
"step": 2635
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 10.19,
|
| 3174 |
+
"learning_rate": 0.00012496563867495748,
|
| 3175 |
+
"loss": 1.5077,
|
| 3176 |
+
"step": 2640
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 10.21,
|
| 3180 |
+
"learning_rate": 0.00012249205387968658,
|
| 3181 |
+
"loss": 1.6099,
|
| 3182 |
+
"step": 2645
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 10.23,
|
| 3186 |
+
"learning_rate": 0.00011982030185518488,
|
| 3187 |
+
"loss": 1.5701,
|
| 3188 |
+
"step": 2650
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 10.25,
|
| 3192 |
+
"learning_rate": 0.0001169602069383955,
|
| 3193 |
+
"loss": 1.5479,
|
| 3194 |
+
"step": 2655
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 10.27,
|
| 3198 |
+
"learning_rate": 0.00011392228602455956,
|
| 3199 |
+
"loss": 1.5486,
|
| 3200 |
+
"step": 2660
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 10.29,
|
| 3204 |
+
"learning_rate": 0.00011071770989539373,
|
| 3205 |
+
"loss": 1.5367,
|
| 3206 |
+
"step": 2665
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 10.31,
|
| 3210 |
+
"learning_rate": 0.0001073582621428498,
|
| 3211 |
+
"loss": 1.5361,
|
| 3212 |
+
"step": 2670
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 10.33,
|
| 3216 |
+
"learning_rate": 0.00010385629583950413,
|
| 3217 |
+
"loss": 1.3755,
|
| 3218 |
+
"step": 2675
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 10.35,
|
| 3222 |
+
"learning_rate": 0.00010022468811490019,
|
| 3223 |
+
"loss": 1.464,
|
| 3224 |
+
"step": 2680
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 10.37,
|
| 3228 |
+
"learning_rate": 9.64767928048756e-05,
|
| 3229 |
+
"loss": 1.5142,
|
| 3230 |
+
"step": 2685
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 10.39,
|
| 3234 |
+
"learning_rate": 9.262639134798378e-05,
|
| 3235 |
+
"loss": 1.5207,
|
| 3236 |
+
"step": 2690
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 10.41,
|
| 3240 |
+
"learning_rate": 8.868764210957149e-05,
|
| 3241 |
+
"loss": 1.5357,
|
| 3242 |
+
"step": 2695
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 10.42,
|
| 3246 |
+
"learning_rate": 8.467502831985583e-05,
|
| 3247 |
+
"loss": 1.454,
|
| 3248 |
+
"step": 2700
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 10.44,
|
| 3252 |
+
"learning_rate": 8.06033048174343e-05,
|
| 3253 |
+
"loss": 1.5096,
|
| 3254 |
+
"step": 2705
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 10.46,
|
| 3258 |
+
"learning_rate": 7.648744379405981e-05,
|
| 3259 |
+
"loss": 1.5628,
|
| 3260 |
+
"step": 2710
|
| 3261 |
+
},
|
| 3262 |
+
{
|
| 3263 |
+
"epoch": 10.48,
|
| 3264 |
+
"learning_rate": 7.234257974018524e-05,
|
| 3265 |
+
"loss": 1.468,
|
| 3266 |
+
"step": 2715
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 10.5,
|
| 3270 |
+
"learning_rate": 6.818395379371479e-05,
|
| 3271 |
+
"loss": 1.4858,
|
| 3272 |
+
"step": 2720
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 10.52,
|
| 3276 |
+
"learning_rate": 6.402685769660054e-05,
|
| 3277 |
+
"loss": 1.4885,
|
| 3278 |
+
"step": 2725
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 10.54,
|
| 3282 |
+
"learning_rate": 5.988657756536443e-05,
|
| 3283 |
+
"loss": 1.4577,
|
| 3284 |
+
"step": 2730
|
| 3285 |
+
},
|
| 3286 |
+
{
|
| 3287 |
+
"epoch": 10.56,
|
| 3288 |
+
"learning_rate": 5.577833768230333e-05,
|
| 3289 |
+
"loss": 1.5513,
|
| 3290 |
+
"step": 2735
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 10.58,
|
| 3294 |
+
"learning_rate": 5.171724451406837e-05,
|
| 3295 |
+
"loss": 1.4957,
|
| 3296 |
+
"step": 2740
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 10.6,
|
| 3300 |
+
"learning_rate": 4.7718231163460647e-05,
|
| 3301 |
+
"loss": 1.5075,
|
| 3302 |
+
"step": 2745
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 10.62,
|
| 3306 |
+
"learning_rate": 4.379600245871268e-05,
|
| 3307 |
+
"loss": 1.448,
|
| 3308 |
+
"step": 2750
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 10.64,
|
| 3312 |
+
"learning_rate": 3.996498088215443e-05,
|
| 3313 |
+
"loss": 1.53,
|
| 3314 |
+
"step": 2755
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 10.66,
|
| 3318 |
+
"learning_rate": 3.623925353710258e-05,
|
| 3319 |
+
"loss": 1.5454,
|
| 3320 |
+
"step": 2760
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 10.68,
|
| 3324 |
+
"learning_rate": 3.2632520347973906e-05,
|
| 3325 |
+
"loss": 1.5139,
|
| 3326 |
+
"step": 2765
|
| 3327 |
+
},
|
| 3328 |
+
{
|
| 3329 |
+
"epoch": 10.69,
|
| 3330 |
+
"learning_rate": 2.915804368410225e-05,
|
| 3331 |
+
"loss": 1.5848,
|
| 3332 |
+
"step": 2770
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 10.71,
|
| 3336 |
+
"learning_rate": 2.5828599592491143e-05,
|
| 3337 |
+
"loss": 1.514,
|
| 3338 |
+
"step": 2775
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 10.73,
|
| 3342 |
+
"learning_rate": 2.2656430818833073e-05,
|
| 3343 |
+
"loss": 1.4666,
|
| 3344 |
+
"step": 2780
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 10.75,
|
| 3348 |
+
"learning_rate": 1.965320178953816e-05,
|
| 3349 |
+
"loss": 1.4546,
|
| 3350 |
+
"step": 2785
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 10.77,
|
| 3354 |
+
"learning_rate": 1.682995572030518e-05,
|
| 3355 |
+
"loss": 1.4552,
|
| 3356 |
+
"step": 2790
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 10.79,
|
| 3360 |
+
"learning_rate": 1.4197074008960664e-05,
|
| 3361 |
+
"loss": 1.482,
|
| 3362 |
+
"step": 2795
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 10.81,
|
| 3366 |
+
"learning_rate": 1.1764238061872534e-05,
|
| 3367 |
+
"loss": 1.5422,
|
| 3368 |
+
"step": 2800
|
| 3369 |
+
},
|
| 3370 |
+
{
|
| 3371 |
+
"epoch": 10.83,
|
| 3372 |
+
"learning_rate": 9.54039369431598e-06,
|
| 3373 |
+
"loss": 1.492,
|
| 3374 |
+
"step": 2805
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 10.85,
|
| 3378 |
+
"learning_rate": 7.5337182356897344e-06,
|
| 3379 |
+
"loss": 1.4219,
|
| 3380 |
+
"step": 2810
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 10.87,
|
| 3384 |
+
"learning_rate": 5.75159046054383e-06,
|
| 3385 |
+
"loss": 1.4972,
|
| 3386 |
+
"step": 2815
|
| 3387 |
+
},
|
| 3388 |
+
{
|
| 3389 |
+
"epoch": 10.89,
|
| 3390 |
+
"learning_rate": 4.200563455983359e-06,
|
| 3391 |
+
"loss": 1.4525,
|
| 3392 |
+
"step": 2820
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 10.91,
|
| 3396 |
+
"learning_rate": 2.886340525219404e-06,
|
| 3397 |
+
"loss": 1.4337,
|
| 3398 |
+
"step": 2825
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 10.93,
|
| 3402 |
+
"learning_rate": 1.8137542158732371e-06,
|
| 3403 |
+
"loss": 1.5066,
|
| 3404 |
+
"step": 2830
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 10.95,
|
| 3408 |
+
"learning_rate": 9.867485501472609e-07,
|
| 3409 |
+
"loss": 1.4053,
|
| 3410 |
+
"step": 2835
|
| 3411 |
+
},
|
| 3412 |
+
{
|
| 3413 |
+
"epoch": 10.97,
|
| 3414 |
+
"learning_rate": 4.083645222054405e-07,
|
| 3415 |
+
"loss": 1.4861,
|
| 3416 |
+
"step": 2840
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 10.98,
|
| 3420 |
+
"learning_rate": 8.072891609114545e-08,
|
| 3421 |
+
"loss": 1.4625,
|
| 3422 |
+
"step": 2845
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 11.0,
|
| 3426 |
+
"learning_rate": 5.046485299251069e-09,
|
| 3427 |
+
"loss": 1.4993,
|
| 3428 |
+
"step": 2850
|
| 3429 |
+
},
|
| 3430 |
+
{
|
| 3431 |
+
"epoch": 11.02,
|
| 3432 |
+
"learning_rate": 1.8159552276039227e-07,
|
| 3433 |
+
"loss": 1.4475,
|
| 3434 |
+
"step": 2855
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 11.04,
|
| 3438 |
+
"learning_rate": 6.097268375260069e-07,
|
| 3439 |
+
"loss": 1.4133,
|
| 3440 |
+
"step": 2860
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 11.06,
|
| 3444 |
+
"learning_rate": 1.2878661419176504e-06,
|
| 3445 |
+
"loss": 1.4672,
|
| 3446 |
+
"step": 2865
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 11.08,
|
| 3450 |
+
"learning_rate": 2.21351984036197e-06,
|
| 3451 |
+
"loss": 1.4601,
|
| 3452 |
+
"step": 2870
|
| 3453 |
+
},
|
| 3454 |
+
{
|
| 3455 |
+
"epoch": 11.1,
|
| 3456 |
+
"learning_rate": 3.383284198626564e-06,
|
| 3457 |
+
"loss": 1.4559,
|
| 3458 |
+
"step": 2875
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 11.12,
|
| 3462 |
+
"learning_rate": 4.792857859738887e-06,
|
| 3463 |
+
"loss": 1.3926,
|
| 3464 |
+
"step": 2880
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 11.14,
|
| 3468 |
+
"learning_rate": 6.437057660565811e-06,
|
| 3469 |
+
"loss": 1.3658,
|
| 3470 |
+
"step": 2885
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 11.16,
|
| 3474 |
+
"learning_rate": 8.309837690896675e-06,
|
| 3475 |
+
"loss": 1.3814,
|
| 3476 |
+
"step": 2890
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 11.18,
|
| 3480 |
+
"learning_rate": 1.0404311524944405e-05,
|
| 3481 |
+
"loss": 1.4756,
|
| 3482 |
+
"step": 2895
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 11.2,
|
| 3486 |
+
"learning_rate": 1.271277754351773e-05,
|
| 3487 |
+
"loss": 1.3856,
|
| 3488 |
+
"step": 2900
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 11.22,
|
| 3492 |
+
"learning_rate": 1.5226747253754904e-05,
|
| 3493 |
+
"loss": 1.3775,
|
| 3494 |
+
"step": 2905
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 11.24,
|
| 3498 |
+
"learning_rate": 1.7936976502279244e-05,
|
| 3499 |
+
"loss": 1.4091,
|
| 3500 |
+
"step": 2910
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 11.25,
|
| 3504 |
+
"learning_rate": 2.083349946700608e-05,
|
| 3505 |
+
"loss": 1.4543,
|
| 3506 |
+
"step": 2915
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 11.27,
|
| 3510 |
+
"learning_rate": 2.39056653026063e-05,
|
| 3511 |
+
"loss": 1.4107,
|
| 3512 |
+
"step": 2920
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 11.29,
|
| 3516 |
+
"learning_rate": 2.714217730488006e-05,
|
| 3517 |
+
"loss": 1.4381,
|
| 3518 |
+
"step": 2925
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 11.31,
|
| 3522 |
+
"learning_rate": 3.053113445002753e-05,
|
| 3523 |
+
"loss": 1.4025,
|
| 3524 |
+
"step": 2930
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 11.33,
|
| 3528 |
+
"learning_rate": 3.4060075156069725e-05,
|
| 3529 |
+
"loss": 1.3656,
|
| 3530 |
+
"step": 2935
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 11.35,
|
| 3534 |
+
"learning_rate": 3.771602310550724e-05,
|
| 3535 |
+
"loss": 1.4167,
|
| 3536 |
+
"step": 2940
|
| 3537 |
+
},
|
| 3538 |
+
{
|
| 3539 |
+
"epoch": 11.37,
|
| 3540 |
+
"learning_rate": 4.148553496072031e-05,
|
| 3541 |
+
"loss": 1.4686,
|
| 3542 |
+
"step": 2945
|
| 3543 |
+
},
|
| 3544 |
+
{
|
| 3545 |
+
"epoch": 11.39,
|
| 3546 |
+
"learning_rate": 4.5354749796653205e-05,
|
| 3547 |
+
"loss": 1.3998,
|
| 3548 |
+
"step": 2950
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 11.41,
|
| 3552 |
+
"learning_rate": 4.9309440069017615e-05,
|
| 3553 |
+
"loss": 1.4714,
|
| 3554 |
+
"step": 2955
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 11.43,
|
| 3558 |
+
"learning_rate": 5.333506393059674e-05,
|
| 3559 |
+
"loss": 1.3963,
|
| 3560 |
+
"step": 2960
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 11.45,
|
| 3564 |
+
"learning_rate": 5.7416818703274866e-05,
|
| 3565 |
+
"loss": 1.5068,
|
| 3566 |
+
"step": 2965
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 11.47,
|
| 3570 |
+
"learning_rate": 6.153969530917418e-05,
|
| 3571 |
+
"loss": 1.4338,
|
| 3572 |
+
"step": 2970
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 11.49,
|
| 3576 |
+
"learning_rate": 6.568853346074412e-05,
|
| 3577 |
+
"loss": 1.3606,
|
| 3578 |
+
"step": 2975
|
| 3579 |
+
},
|
| 3580 |
+
{
|
| 3581 |
+
"epoch": 11.51,
|
| 3582 |
+
"learning_rate": 6.984807740687094e-05,
|
| 3583 |
+
"loss": 1.4016,
|
| 3584 |
+
"step": 2980
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 11.53,
|
| 3588 |
+
"learning_rate": 7.400303203001311e-05,
|
| 3589 |
+
"loss": 1.4269,
|
| 3590 |
+
"step": 2985
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 11.54,
|
| 3594 |
+
"learning_rate": 7.813811908809178e-05,
|
| 3595 |
+
"loss": 1.44,
|
| 3596 |
+
"step": 2990
|
| 3597 |
+
},
|
| 3598 |
+
{
|
| 3599 |
+
"epoch": 11.56,
|
| 3600 |
+
"learning_rate": 8.223813339433243e-05,
|
| 3601 |
+
"loss": 1.4977,
|
| 3602 |
+
"step": 2995
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 11.58,
|
| 3606 |
+
"learning_rate": 8.628799872846956e-05,
|
| 3607 |
+
"loss": 1.436,
|
| 3608 |
+
"step": 3000
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 11.6,
|
| 3612 |
+
"learning_rate": 9.027282327372687e-05,
|
| 3613 |
+
"loss": 1.513,
|
| 3614 |
+
"step": 3005
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 11.62,
|
| 3618 |
+
"learning_rate": 9.417795437572891e-05,
|
| 3619 |
+
"loss": 1.4691,
|
| 3620 |
+
"step": 3010
|
| 3621 |
+
},
|
| 3622 |
+
{
|
| 3623 |
+
"epoch": 11.64,
|
| 3624 |
+
"learning_rate": 9.798903242198079e-05,
|
| 3625 |
+
"loss": 1.5133,
|
| 3626 |
+
"step": 3015
|
| 3627 |
+
},
|
| 3628 |
+
{
|
| 3629 |
+
"epoch": 11.66,
|
| 3630 |
+
"learning_rate": 0.000101692043643802,
|
| 3631 |
+
"loss": 1.5047,
|
| 3632 |
+
"step": 3020
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 11.68,
|
| 3636 |
+
"learning_rate": 0.00010527337164655109,
|
| 3637 |
+
"loss": 1.4302,
|
| 3638 |
+
"step": 3025
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 11.7,
|
| 3642 |
+
"learning_rate": 0.00010871984747866282,
|
| 3643 |
+
"loss": 1.5886,
|
| 3644 |
+
"step": 3030
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 11.72,
|
| 3648 |
+
"learning_rate": 0.00011201879805538586,
|
| 3649 |
+
"loss": 1.4925,
|
| 3650 |
+
"step": 3035
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 11.74,
|
| 3654 |
+
"learning_rate": 0.00011515809275915985,
|
| 3655 |
+
"loss": 1.4445,
|
| 3656 |
+
"step": 3040
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 11.76,
|
| 3660 |
+
"learning_rate": 0.00011812618804528006,
|
| 3661 |
+
"loss": 1.4577,
|
| 3662 |
+
"step": 3045
|
| 3663 |
+
},
|
| 3664 |
+
{
|
| 3665 |
+
"epoch": 11.78,
|
| 3666 |
+
"learning_rate": 0.00012091216988882848,
|
| 3667 |
+
"loss": 1.4792,
|
| 3668 |
+
"step": 3050
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 11.8,
|
| 3672 |
+
"learning_rate": 0.00012350579391678723,
|
| 3673 |
+
"loss": 1.4425,
|
| 3674 |
+
"step": 3055
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 11.81,
|
| 3678 |
+
"learning_rate": 0.00012589752307776744,
|
| 3679 |
+
"loss": 1.4294,
|
| 3680 |
+
"step": 3060
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 11.83,
|
| 3684 |
+
"learning_rate": 0.0001280785627108355,
|
| 3685 |
+
"loss": 1.4138,
|
| 3686 |
+
"step": 3065
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 11.85,
|
| 3690 |
+
"learning_rate": 0.0001300408928844837,
|
| 3691 |
+
"loss": 1.5429,
|
| 3692 |
+
"step": 3070
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 11.87,
|
| 3696 |
+
"learning_rate": 0.00013177729788683344,
|
| 3697 |
+
"loss": 1.5001,
|
| 3698 |
+
"step": 3075
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 11.89,
|
| 3702 |
+
"learning_rate": 0.00013328139275863032,
|
| 3703 |
+
"loss": 1.444,
|
| 3704 |
+
"step": 3080
|
| 3705 |
+
},
|
| 3706 |
+
{
|
| 3707 |
+
"epoch": 11.91,
|
| 3708 |
+
"learning_rate": 0.00013454764677146876,
|
| 3709 |
+
"loss": 1.4294,
|
| 3710 |
+
"step": 3085
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 11.93,
|
| 3714 |
+
"learning_rate": 0.00013557140376490993,
|
| 3715 |
+
"loss": 1.5483,
|
| 3716 |
+
"step": 3090
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 11.95,
|
| 3720 |
+
"learning_rate": 0.0001363488992677158,
|
| 3721 |
+
"loss": 1.5026,
|
| 3722 |
+
"step": 3095
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 11.97,
|
| 3726 |
+
"learning_rate": 0.00013687727434023872,
|
| 3727 |
+
"loss": 1.5176,
|
| 3728 |
+
"step": 3100
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 11.99,
|
| 3732 |
+
"learning_rate": 0.00013715458608706872,
|
| 3733 |
+
"loss": 1.4418,
|
| 3734 |
+
"step": 3105
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 12.01,
|
| 3738 |
+
"learning_rate": 0.00013717981480128154,
|
| 3739 |
+
"loss": 1.441,
|
| 3740 |
+
"step": 3110
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 12.03,
|
| 3744 |
+
"learning_rate": 0.00013695286771401734,
|
| 3745 |
+
"loss": 1.3854,
|
| 3746 |
+
"step": 3115
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 12.05,
|
| 3750 |
+
"learning_rate": 0.00013647457933560234,
|
| 3751 |
+
"loss": 1.4397,
|
| 3752 |
+
"step": 3120
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 12.07,
|
| 3756 |
+
"learning_rate": 0.00013574670838695926,
|
| 3757 |
+
"loss": 1.4672,
|
| 3758 |
+
"step": 3125
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 12.08,
|
| 3762 |
+
"learning_rate": 0.0001347719313325897,
|
| 3763 |
+
"loss": 1.4525,
|
| 3764 |
+
"step": 3130
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 12.1,
|
| 3768 |
+
"learning_rate": 0.00013355383253890914,
|
| 3769 |
+
"loss": 1.4068,
|
| 3770 |
+
"step": 3135
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 12.12,
|
| 3774 |
+
"learning_rate": 0.0001320968910941225,
|
| 3775 |
+
"loss": 1.4855,
|
| 3776 |
+
"step": 3140
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 12.14,
|
| 3780 |
+
"learning_rate": 0.0001304064643381061,
|
| 3781 |
+
"loss": 1.4212,
|
| 3782 |
+
"step": 3145
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 12.16,
|
| 3786 |
+
"learning_rate": 0.00012848876816285777,
|
| 3787 |
+
"loss": 1.4849,
|
| 3788 |
+
"step": 3150
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 12.18,
|
| 3792 |
+
"learning_rate": 0.00012635085415595244,
|
| 3793 |
+
"loss": 1.3912,
|
| 3794 |
+
"step": 3155
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 12.2,
|
| 3798 |
+
"learning_rate": 0.00012400058367105247,
|
| 3799 |
+
"loss": 1.483,
|
| 3800 |
+
"step": 3160
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 12.22,
|
| 3804 |
+
"learning_rate": 0.00012144659892081038,
|
| 3805 |
+
"loss": 1.3818,
|
| 3806 |
+
"step": 3165
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 12.24,
|
| 3810 |
+
"learning_rate": 0.00011869829119846924,
|
| 3811 |
+
"loss": 1.4634,
|
| 3812 |
+
"step": 3170
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 12.26,
|
| 3816 |
+
"learning_rate": 0.00011576576634500562,
|
| 3817 |
+
"loss": 1.4034,
|
| 3818 |
+
"step": 3175
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 12.28,
|
| 3822 |
+
"learning_rate": 0.00011265980758879936,
|
| 3823 |
+
"loss": 1.4014,
|
| 3824 |
+
"step": 3180
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 12.3,
|
| 3828 |
+
"learning_rate": 0.00010939183589447423,
|
| 3829 |
+
"loss": 1.4222,
|
| 3830 |
+
"step": 3185
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 12.32,
|
| 3834 |
+
"learning_rate": 0.00010597386796670575,
|
| 3835 |
+
"loss": 1.4854,
|
| 3836 |
+
"step": 3190
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 12.34,
|
| 3840 |
+
"learning_rate": 0.00010241847206343044,
|
| 3841 |
+
"loss": 1.4472,
|
| 3842 |
+
"step": 3195
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 12.36,
|
| 3846 |
+
"learning_rate": 9.87387217809251e-05,
|
| 3847 |
+
"loss": 1.5271,
|
| 3848 |
+
"step": 3200
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 12.37,
|
| 3852 |
+
"learning_rate": 9.494814798070321e-05,
|
| 3853 |
+
"loss": 1.401,
|
| 3854 |
+
"step": 3205
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 12.39,
|
| 3858 |
+
"learning_rate": 9.106068903499514e-05,
|
| 3859 |
+
"loss": 1.5122,
|
| 3860 |
+
"step": 3210
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 12.41,
|
| 3864 |
+
"learning_rate": 8.709063957376094e-05,
|
| 3865 |
+
"loss": 1.4755,
|
| 3866 |
+
"step": 3215
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 12.43,
|
| 3870 |
+
"learning_rate": 8.30525979217072e-05,
|
| 3871 |
+
"loss": 1.4605,
|
| 3872 |
+
"step": 3220
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 12.45,
|
| 3876 |
+
"learning_rate": 7.896141241858118e-05,
|
| 3877 |
+
"loss": 1.3958,
|
| 3878 |
+
"step": 3225
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 12.47,
|
| 3882 |
+
"learning_rate": 7.48321268201337e-05,
|
| 3883 |
+
"loss": 1.4285,
|
| 3884 |
+
"step": 3230
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 12.49,
|
| 3888 |
+
"learning_rate": 7.067992498051e-05,
|
| 3889 |
+
"loss": 1.4276,
|
| 3890 |
+
"step": 3235
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 12.51,
|
| 3894 |
+
"learning_rate": 6.652007501948996e-05,
|
| 3895 |
+
"loss": 1.4174,
|
| 3896 |
+
"step": 3240
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 12.53,
|
| 3900 |
+
"learning_rate": 6.236787317986674e-05,
|
| 3901 |
+
"loss": 1.4845,
|
| 3902 |
+
"step": 3245
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 12.55,
|
| 3906 |
+
"learning_rate": 5.823858758141927e-05,
|
| 3907 |
+
"loss": 1.4357,
|
| 3908 |
+
"step": 3250
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 12.57,
|
| 3912 |
+
"learning_rate": 5.414740207829325e-05,
|
| 3913 |
+
"loss": 1.4382,
|
| 3914 |
+
"step": 3255
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 12.59,
|
| 3918 |
+
"learning_rate": 5.010936042623904e-05,
|
| 3919 |
+
"loss": 1.3395,
|
| 3920 |
+
"step": 3260
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 12.61,
|
| 3924 |
+
"learning_rate": 4.6139310965004824e-05,
|
| 3925 |
+
"loss": 1.3898,
|
| 3926 |
+
"step": 3265
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 12.63,
|
| 3930 |
+
"learning_rate": 4.225185201929675e-05,
|
| 3931 |
+
"loss": 1.4521,
|
| 3932 |
+
"step": 3270
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 12.64,
|
| 3936 |
+
"learning_rate": 3.8461278219075304e-05,
|
| 3937 |
+
"loss": 1.4757,
|
| 3938 |
+
"step": 3275
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 12.66,
|
| 3942 |
+
"learning_rate": 3.478152793656996e-05,
|
| 3943 |
+
"loss": 1.4128,
|
| 3944 |
+
"step": 3280
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 12.68,
|
| 3948 |
+
"learning_rate": 3.122613203329423e-05,
|
| 3949 |
+
"loss": 1.4424,
|
| 3950 |
+
"step": 3285
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 12.7,
|
| 3954 |
+
"learning_rate": 2.780816410552575e-05,
|
| 3955 |
+
"loss": 1.4287,
|
| 3956 |
+
"step": 3290
|
| 3957 |
+
},
|
| 3958 |
+
{
|
| 3959 |
+
"epoch": 12.72,
|
| 3960 |
+
"learning_rate": 2.454019241120062e-05,
|
| 3961 |
+
"loss": 1.3953,
|
| 3962 |
+
"step": 3295
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 12.74,
|
| 3966 |
+
"learning_rate": 2.1434233654994707e-05,
|
| 3967 |
+
"loss": 1.3275,
|
| 3968 |
+
"step": 3300
|
| 3969 |
+
},
|
| 3970 |
+
{
|
| 3971 |
+
"epoch": 12.76,
|
| 3972 |
+
"learning_rate": 1.8501708801531077e-05,
|
| 3973 |
+
"loss": 1.3897,
|
| 3974 |
+
"step": 3305
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 12.78,
|
| 3978 |
+
"learning_rate": 1.575340107918959e-05,
|
| 3979 |
+
"loss": 1.3657,
|
| 3980 |
+
"step": 3310
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 12.8,
|
| 3984 |
+
"learning_rate": 1.319941632894751e-05,
|
| 3985 |
+
"loss": 1.3897,
|
| 3986 |
+
"step": 3315
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 12.82,
|
| 3990 |
+
"learning_rate": 1.0849145844047538e-05,
|
| 3991 |
+
"loss": 1.4783,
|
| 3992 |
+
"step": 3320
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 12.84,
|
| 3996 |
+
"learning_rate": 8.711231837142462e-06,
|
| 3997 |
+
"loss": 1.4102,
|
| 3998 |
+
"step": 3325
|
| 3999 |
+
},
|
| 4000 |
+
{
|
| 4001 |
+
"epoch": 12.86,
|
| 4002 |
+
"learning_rate": 6.793535661894092e-06,
|
| 4003 |
+
"loss": 1.4442,
|
| 4004 |
+
"step": 3330
|
| 4005 |
+
},
|
| 4006 |
+
{
|
| 4007 |
+
"epoch": 12.88,
|
| 4008 |
+
"learning_rate": 5.1031089058776675e-06,
|
| 4009 |
+
"loss": 1.3875,
|
| 4010 |
+
"step": 3335
|
| 4011 |
+
},
|
| 4012 |
+
{
|
| 4013 |
+
"epoch": 12.9,
|
| 4014 |
+
"learning_rate": 3.6461674610908637e-06,
|
| 4015 |
+
"loss": 1.4228,
|
| 4016 |
+
"step": 3340
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 12.92,
|
| 4020 |
+
"learning_rate": 2.42806866741032e-06,
|
| 4021 |
+
"loss": 1.4015,
|
| 4022 |
+
"step": 3345
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 12.93,
|
| 4026 |
+
"learning_rate": 1.453291613040815e-06,
|
| 4027 |
+
"loss": 1.3937,
|
| 4028 |
+
"step": 3350
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 12.95,
|
| 4032 |
+
"learning_rate": 7.254206643977347e-07,
|
| 4033 |
+
"loss": 1.4905,
|
| 4034 |
+
"step": 3355
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 12.97,
|
| 4038 |
+
"learning_rate": 2.4713228598269586e-07,
|
| 4039 |
+
"loss": 1.4419,
|
| 4040 |
+
"step": 3360
|
| 4041 |
+
},
|
| 4042 |
+
{
|
| 4043 |
+
"epoch": 12.99,
|
| 4044 |
+
"learning_rate": 2.0185198718462007e-08,
|
| 4045 |
+
"loss": 1.4331,
|
| 4046 |
+
"step": 3365
|
| 4047 |
+
},
|
| 4048 |
+
{
|
| 4049 |
+
"epoch": 13.01,
|
| 4050 |
+
"learning_rate": 4.5413912931266996e-08,
|
| 4051 |
+
"loss": 1.4014,
|
| 4052 |
+
"step": 3370
|
| 4053 |
+
},
|
| 4054 |
+
{
|
| 4055 |
+
"epoch": 13.03,
|
| 4056 |
+
"learning_rate": 3.227256597612364e-07,
|
| 4057 |
+
"loss": 1.3146,
|
| 4058 |
+
"step": 3375
|
| 4059 |
+
},
|
| 4060 |
+
{
|
| 4061 |
+
"epoch": 13.05,
|
| 4062 |
+
"learning_rate": 8.51100732284126e-07,
|
| 4063 |
+
"loss": 1.3623,
|
| 4064 |
+
"step": 3380
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 13.07,
|
| 4068 |
+
"learning_rate": 1.62859623508997e-06,
|
| 4069 |
+
"loss": 1.3259,
|
| 4070 |
+
"step": 3385
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 13.09,
|
| 4074 |
+
"learning_rate": 2.652353228531244e-06,
|
| 4075 |
+
"loss": 1.3975,
|
| 4076 |
+
"step": 3390
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 13.11,
|
| 4080 |
+
"learning_rate": 3.9186072413696845e-06,
|
| 4081 |
+
"loss": 1.3585,
|
| 4082 |
+
"step": 3395
|
| 4083 |
+
},
|
| 4084 |
+
{
|
| 4085 |
+
"epoch": 13.13,
|
| 4086 |
+
"learning_rate": 5.422702113166566e-06,
|
| 4087 |
+
"loss": 1.3596,
|
| 4088 |
+
"step": 3400
|
| 4089 |
+
},
|
| 4090 |
+
{
|
| 4091 |
+
"epoch": 13.15,
|
| 4092 |
+
"learning_rate": 7.159107115516102e-06,
|
| 4093 |
+
"loss": 1.4021,
|
| 4094 |
+
"step": 3405
|
| 4095 |
+
},
|
| 4096 |
+
{
|
| 4097 |
+
"epoch": 13.17,
|
| 4098 |
+
"learning_rate": 9.121437289164265e-06,
|
| 4099 |
+
"loss": 1.4666,
|
| 4100 |
+
"step": 3410
|
| 4101 |
+
},
|
| 4102 |
+
{
|
| 4103 |
+
"epoch": 13.19,
|
| 4104 |
+
"learning_rate": 1.1302476922232583e-05,
|
| 4105 |
+
"loss": 1.327,
|
| 4106 |
+
"step": 3415
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 13.2,
|
| 4110 |
+
"learning_rate": 1.3694206083212781e-05,
|
| 4111 |
+
"loss": 1.2798,
|
| 4112 |
+
"step": 3420
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 13.22,
|
| 4116 |
+
"learning_rate": 1.628783011117153e-05,
|
| 4117 |
+
"loss": 1.3184,
|
| 4118 |
+
"step": 3425
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 13.24,
|
| 4122 |
+
"learning_rate": 1.9073811954719624e-05,
|
| 4123 |
+
"loss": 1.3236,
|
| 4124 |
+
"step": 3430
|
| 4125 |
+
},
|
| 4126 |
+
{
|
| 4127 |
+
"epoch": 13.26,
|
| 4128 |
+
"learning_rate": 2.2041907240839828e-05,
|
| 4129 |
+
"loss": 1.3766,
|
| 4130 |
+
"step": 3435
|
| 4131 |
+
},
|
| 4132 |
+
{
|
| 4133 |
+
"epoch": 13.28,
|
| 4134 |
+
"learning_rate": 2.518120194461378e-05,
|
| 4135 |
+
"loss": 1.2779,
|
| 4136 |
+
"step": 3440
|
| 4137 |
+
},
|
| 4138 |
+
{
|
| 4139 |
+
"epoch": 13.3,
|
| 4140 |
+
"learning_rate": 2.8480152521337216e-05,
|
| 4141 |
+
"loss": 1.3743,
|
| 4142 |
+
"step": 3445
|
| 4143 |
+
},
|
| 4144 |
+
{
|
| 4145 |
+
"epoch": 13.32,
|
| 4146 |
+
"learning_rate": 3.1926628353448936e-05,
|
| 4147 |
+
"loss": 1.336,
|
| 4148 |
+
"step": 3450
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 13.34,
|
| 4152 |
+
"learning_rate": 3.5507956356197615e-05,
|
| 4153 |
+
"loss": 1.3522,
|
| 4154 |
+
"step": 3455
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 13.36,
|
| 4158 |
+
"learning_rate": 3.9210967578018804e-05,
|
| 4159 |
+
"loss": 1.3693,
|
| 4160 |
+
"step": 3460
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 13.38,
|
| 4164 |
+
"learning_rate": 4.302204562427067e-05,
|
| 4165 |
+
"loss": 1.3374,
|
| 4166 |
+
"step": 3465
|
| 4167 |
+
},
|
| 4168 |
+
{
|
| 4169 |
+
"epoch": 13.4,
|
| 4170 |
+
"learning_rate": 4.692717672627317e-05,
|
| 4171 |
+
"loss": 1.3881,
|
| 4172 |
+
"step": 3470
|
| 4173 |
+
},
|
| 4174 |
+
{
|
| 4175 |
+
"epoch": 13.42,
|
| 4176 |
+
"learning_rate": 5.091200127153047e-05,
|
| 4177 |
+
"loss": 1.2653,
|
| 4178 |
+
"step": 3475
|
| 4179 |
+
},
|
| 4180 |
+
{
|
| 4181 |
+
"epoch": 13.44,
|
| 4182 |
+
"learning_rate": 5.496186660566713e-05,
|
| 4183 |
+
"loss": 1.3907,
|
| 4184 |
+
"step": 3480
|
| 4185 |
+
},
|
| 4186 |
+
{
|
| 4187 |
+
"epoch": 13.46,
|
| 4188 |
+
"learning_rate": 5.906188091190777e-05,
|
| 4189 |
+
"loss": 1.3586,
|
| 4190 |
+
"step": 3485
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 13.47,
|
| 4194 |
+
"learning_rate": 6.319696796998643e-05,
|
| 4195 |
+
"loss": 1.3102,
|
| 4196 |
+
"step": 3490
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 13.49,
|
| 4200 |
+
"learning_rate": 6.735192259312862e-05,
|
| 4201 |
+
"loss": 1.3599,
|
| 4202 |
+
"step": 3495
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 13.51,
|
| 4206 |
+
"learning_rate": 7.151146653925592e-05,
|
| 4207 |
+
"loss": 1.3715,
|
| 4208 |
+
"step": 3500
|
| 4209 |
+
},
|
| 4210 |
+
{
|
| 4211 |
+
"epoch": 13.53,
|
| 4212 |
+
"learning_rate": 7.566030469082585e-05,
|
| 4213 |
+
"loss": 1.4274,
|
| 4214 |
+
"step": 3505
|
| 4215 |
+
},
|
| 4216 |
+
{
|
| 4217 |
+
"epoch": 13.55,
|
| 4218 |
+
"learning_rate": 7.978318129672468e-05,
|
| 4219 |
+
"loss": 1.2634,
|
| 4220 |
+
"step": 3510
|
| 4221 |
+
},
|
| 4222 |
+
{
|
| 4223 |
+
"epoch": 13.57,
|
| 4224 |
+
"learning_rate": 8.386493606940281e-05,
|
| 4225 |
+
"loss": 1.3939,
|
| 4226 |
+
"step": 3515
|
| 4227 |
+
},
|
| 4228 |
+
{
|
| 4229 |
+
"epoch": 13.59,
|
| 4230 |
+
"learning_rate": 8.789055993098241e-05,
|
| 4231 |
+
"loss": 1.4498,
|
| 4232 |
+
"step": 3520
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 13.61,
|
| 4236 |
+
"learning_rate": 9.184525020334682e-05,
|
| 4237 |
+
"loss": 1.4425,
|
| 4238 |
+
"step": 3525
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 13.63,
|
| 4242 |
+
"learning_rate": 9.571446503927972e-05,
|
| 4243 |
+
"loss": 1.3688,
|
| 4244 |
+
"step": 3530
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 13.65,
|
| 4248 |
+
"learning_rate": 9.948397689449235e-05,
|
| 4249 |
+
"loss": 1.3409,
|
| 4250 |
+
"step": 3535
|
| 4251 |
+
},
|
| 4252 |
+
{
|
| 4253 |
+
"epoch": 13.67,
|
| 4254 |
+
"learning_rate": 0.00010313992484392988,
|
| 4255 |
+
"loss": 1.4686,
|
| 4256 |
+
"step": 3540
|
| 4257 |
+
},
|
| 4258 |
+
{
|
| 4259 |
+
"epoch": 13.69,
|
| 4260 |
+
"learning_rate": 0.00010666886554997249,
|
| 4261 |
+
"loss": 1.3646,
|
| 4262 |
+
"step": 3545
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 13.71,
|
| 4266 |
+
"learning_rate": 0.00011005782269511996,
|
| 4267 |
+
"loss": 1.411,
|
| 4268 |
+
"step": 3550
|
| 4269 |
+
},
|
| 4270 |
+
{
|
| 4271 |
+
"epoch": 13.73,
|
| 4272 |
+
"learning_rate": 0.00011329433469739373,
|
| 4273 |
+
"loss": 1.3738,
|
| 4274 |
+
"step": 3555
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 13.75,
|
| 4278 |
+
"learning_rate": 0.0001163665005329936,
|
| 4279 |
+
"loss": 1.3912,
|
| 4280 |
+
"step": 3560
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 13.76,
|
| 4284 |
+
"learning_rate": 0.00011926302349772045,
|
| 4285 |
+
"loss": 1.3728,
|
| 4286 |
+
"step": 3565
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 13.78,
|
| 4290 |
+
"learning_rate": 0.00012197325274624481,
|
| 4291 |
+
"loss": 1.3925,
|
| 4292 |
+
"step": 3570
|
| 4293 |
+
},
|
| 4294 |
+
{
|
| 4295 |
+
"epoch": 13.8,
|
| 4296 |
+
"learning_rate": 0.0001244872224564823,
|
| 4297 |
+
"loss": 1.3735,
|
| 4298 |
+
"step": 3575
|
| 4299 |
+
},
|
| 4300 |
+
{
|
| 4301 |
+
"epoch": 13.82,
|
| 4302 |
+
"learning_rate": 0.0001267956884750556,
|
| 4303 |
+
"loss": 1.4361,
|
| 4304 |
+
"step": 3580
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 13.84,
|
| 4308 |
+
"learning_rate": 0.0001288901623091031,
|
| 4309 |
+
"loss": 1.4661,
|
| 4310 |
+
"step": 3585
|
| 4311 |
+
},
|
| 4312 |
+
{
|
| 4313 |
+
"epoch": 13.86,
|
| 4314 |
+
"learning_rate": 0.000130762942339434,
|
| 4315 |
+
"loss": 1.4177,
|
| 4316 |
+
"step": 3590
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 13.88,
|
| 4320 |
+
"learning_rate": 0.00013240714214026112,
|
| 4321 |
+
"loss": 1.4351,
|
| 4322 |
+
"step": 3595
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 13.9,
|
| 4326 |
+
"learning_rate": 0.00013381671580137345,
|
| 4327 |
+
"loss": 1.4243,
|
| 4328 |
+
"step": 3600
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 13.92,
|
| 4332 |
+
"learning_rate": 0.00013498648015963804,
|
| 4333 |
+
"loss": 1.3258,
|
| 4334 |
+
"step": 3605
|
| 4335 |
+
},
|
| 4336 |
+
{
|
| 4337 |
+
"epoch": 13.94,
|
| 4338 |
+
"learning_rate": 0.00013591213385808236,
|
| 4339 |
+
"loss": 1.3917,
|
| 4340 |
+
"step": 3610
|
| 4341 |
+
},
|
| 4342 |
+
{
|
| 4343 |
+
"epoch": 13.96,
|
| 4344 |
+
"learning_rate": 0.00013659027316247394,
|
| 4345 |
+
"loss": 1.3626,
|
| 4346 |
+
"step": 3615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 13.98,
|
| 4350 |
+
"learning_rate": 0.00013701840447723958,
|
| 4351 |
+
"loss": 1.505,
|
| 4352 |
+
"step": 3620
|
| 4353 |
+
},
|
| 4354 |
+
{
|
| 4355 |
+
"epoch": 14.0,
|
| 4356 |
+
"learning_rate": 0.00013719495351470075,
|
| 4357 |
+
"loss": 1.3238,
|
| 4358 |
+
"step": 3625
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 14.02,
|
| 4362 |
+
"learning_rate": 0.00013711927108390887,
|
| 4363 |
+
"loss": 1.3589,
|
| 4364 |
+
"step": 3630
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 14.03,
|
| 4368 |
+
"learning_rate": 0.00013679163547779456,
|
| 4369 |
+
"loss": 1.4241,
|
| 4370 |
+
"step": 3635
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 14.05,
|
| 4374 |
+
"learning_rate": 0.00013621325144985282,
|
| 4375 |
+
"loss": 1.4179,
|
| 4376 |
+
"step": 3640
|
| 4377 |
+
},
|
| 4378 |
+
{
|
| 4379 |
+
"epoch": 14.07,
|
| 4380 |
+
"learning_rate": 0.00013538624578412686,
|
| 4381 |
+
"loss": 1.3404,
|
| 4382 |
+
"step": 3645
|
| 4383 |
+
},
|
| 4384 |
+
{
|
| 4385 |
+
"epoch": 14.09,
|
| 4386 |
+
"learning_rate": 0.00013431365947478058,
|
| 4387 |
+
"loss": 1.3758,
|
| 4388 |
+
"step": 3650
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 14.11,
|
| 4392 |
+
"learning_rate": 0.00013299943654401664,
|
| 4393 |
+
"loss": 1.4247,
|
| 4394 |
+
"step": 3655
|
| 4395 |
+
},
|
| 4396 |
+
{
|
| 4397 |
+
"epoch": 14.13,
|
| 4398 |
+
"learning_rate": 0.00013144840953945616,
|
| 4399 |
+
"loss": 1.3701,
|
| 4400 |
+
"step": 3660
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 14.15,
|
| 4404 |
+
"learning_rate": 0.00012966628176431025,
|
| 4405 |
+
"loss": 1.3553,
|
| 4406 |
+
"step": 3665
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 14.17,
|
| 4410 |
+
"learning_rate": 0.00012765960630568425,
|
| 4411 |
+
"loss": 1.381,
|
| 4412 |
+
"step": 3670
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 14.19,
|
| 4416 |
+
"learning_rate": 0.00012543576193812774,
|
| 4417 |
+
"loss": 1.442,
|
| 4418 |
+
"step": 3675
|
| 4419 |
+
},
|
| 4420 |
+
{
|
| 4421 |
+
"epoch": 14.21,
|
| 4422 |
+
"learning_rate": 0.0001230029259910393,
|
| 4423 |
+
"loss": 1.3873,
|
| 4424 |
+
"step": 3680
|
| 4425 |
+
},
|
| 4426 |
+
{
|
| 4427 |
+
"epoch": 14.23,
|
| 4428 |
+
"learning_rate": 0.0001203700442796948,
|
| 4429 |
+
"loss": 1.3884,
|
| 4430 |
+
"step": 3685
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 14.25,
|
| 4434 |
+
"learning_rate": 0.00011754679821046217,
|
| 4435 |
+
"loss": 1.3278,
|
| 4436 |
+
"step": 3690
|
| 4437 |
+
},
|
| 4438 |
+
{
|
| 4439 |
+
"epoch": 14.27,
|
| 4440 |
+
"learning_rate": 0.00011454356918116728,
|
| 4441 |
+
"loss": 1.3606,
|
| 4442 |
+
"step": 3695
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 14.29,
|
| 4446 |
+
"learning_rate": 0.00011137140040750922,
|
| 4447 |
+
"loss": 1.2409,
|
| 4448 |
+
"step": 3700
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 14.31,
|
| 4452 |
+
"learning_rate": 0.00010804195631589772,
|
| 4453 |
+
"loss": 1.411,
|
| 4454 |
+
"step": 3705
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 14.32,
|
| 4458 |
+
"learning_rate": 0.00010456747965202607,
|
| 4459 |
+
"loss": 1.38,
|
| 4460 |
+
"step": 3710
|
| 4461 |
+
},
|
| 4462 |
+
{
|
| 4463 |
+
"epoch": 14.34,
|
| 4464 |
+
"learning_rate": 0.00010096074646289782,
|
| 4465 |
+
"loss": 1.3982,
|
| 4466 |
+
"step": 3715
|
| 4467 |
+
},
|
| 4468 |
+
{
|
| 4469 |
+
"epoch": 14.36,
|
| 4470 |
+
"learning_rate": 9.723501911784598e-05,
|
| 4471 |
+
"loss": 1.3883,
|
| 4472 |
+
"step": 3720
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 14.38,
|
| 4476 |
+
"learning_rate": 9.340399754128775e-05,
|
| 4477 |
+
"loss": 1.3611,
|
| 4478 |
+
"step": 3725
|
| 4479 |
+
},
|
| 4480 |
+
{
|
| 4481 |
+
"epoch": 14.4,
|
| 4482 |
+
"learning_rate": 8.948176883653932e-05,
|
| 4483 |
+
"loss": 1.4344,
|
| 4484 |
+
"step": 3730
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 14.42,
|
| 4488 |
+
"learning_rate": 8.548275548593159e-05,
|
| 4489 |
+
"loss": 1.2783,
|
| 4490 |
+
"step": 3735
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 14.44,
|
| 4494 |
+
"learning_rate": 8.142166231769664e-05,
|
| 4495 |
+
"loss": 1.335,
|
| 4496 |
+
"step": 3740
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 14.46,
|
| 4500 |
+
"learning_rate": 7.731342243463601e-05,
|
| 4501 |
+
"loss": 1.3506,
|
| 4502 |
+
"step": 3745
|
| 4503 |
+
},
|
| 4504 |
+
{
|
| 4505 |
+
"epoch": 14.48,
|
| 4506 |
+
"learning_rate": 7.317314230339991e-05,
|
| 4507 |
+
"loss": 1.4243,
|
| 4508 |
+
"step": 3750
|
| 4509 |
+
},
|
| 4510 |
+
{
|
| 4511 |
+
"epoch": 14.5,
|
| 4512 |
+
"learning_rate": 6.901604620628517e-05,
|
| 4513 |
+
"loss": 1.3969,
|
| 4514 |
+
"step": 3755
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 14.52,
|
| 4518 |
+
"learning_rate": 6.485742025981473e-05,
|
| 4519 |
+
"loss": 1.3597,
|
| 4520 |
+
"step": 3760
|
| 4521 |
+
},
|
| 4522 |
+
{
|
| 4523 |
+
"epoch": 14.54,
|
| 4524 |
+
"learning_rate": 6.071255620594063e-05,
|
| 4525 |
+
"loss": 1.4289,
|
| 4526 |
+
"step": 3765
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 14.56,
|
| 4530 |
+
"learning_rate": 5.659669518256613e-05,
|
| 4531 |
+
"loss": 1.3466,
|
| 4532 |
+
"step": 3770
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 14.58,
|
| 4536 |
+
"learning_rate": 5.252497168014461e-05,
|
| 4537 |
+
"loss": 1.279,
|
| 4538 |
+
"step": 3775
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 14.59,
|
| 4542 |
+
"learning_rate": 4.8512357890428955e-05,
|
| 4543 |
+
"loss": 1.3786,
|
| 4544 |
+
"step": 3780
|
| 4545 |
+
},
|
| 4546 |
+
{
|
| 4547 |
+
"epoch": 14.61,
|
| 4548 |
+
"learning_rate": 4.457360865201619e-05,
|
| 4549 |
+
"loss": 1.2442,
|
| 4550 |
+
"step": 3785
|
| 4551 |
+
},
|
| 4552 |
+
{
|
| 4553 |
+
"epoch": 14.63,
|
| 4554 |
+
"learning_rate": 4.072320719512437e-05,
|
| 4555 |
+
"loss": 1.2467,
|
| 4556 |
+
"step": 3790
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 14.65,
|
| 4560 |
+
"learning_rate": 3.697531188510021e-05,
|
| 4561 |
+
"loss": 1.326,
|
| 4562 |
+
"step": 3795
|
| 4563 |
+
},
|
| 4564 |
+
{
|
| 4565 |
+
"epoch": 14.67,
|
| 4566 |
+
"learning_rate": 3.3343704160496265e-05,
|
| 4567 |
+
"loss": 1.3049,
|
| 4568 |
+
"step": 3800
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 14.69,
|
| 4572 |
+
"learning_rate": 2.9841737857150583e-05,
|
| 4573 |
+
"loss": 1.3741,
|
| 4574 |
+
"step": 3805
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 14.71,
|
| 4578 |
+
"learning_rate": 2.648229010460623e-05,
|
| 4579 |
+
"loss": 1.3036,
|
| 4580 |
+
"step": 3810
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 14.73,
|
| 4584 |
+
"learning_rate": 2.3277713975440426e-05,
|
| 4585 |
+
"loss": 1.3118,
|
| 4586 |
+
"step": 3815
|
| 4587 |
+
},
|
| 4588 |
+
{
|
| 4589 |
+
"epoch": 14.75,
|
| 4590 |
+
"learning_rate": 2.0239793061604814e-05,
|
| 4591 |
+
"loss": 1.3541,
|
| 4592 |
+
"step": 3820
|
| 4593 |
+
},
|
| 4594 |
+
{
|
| 4595 |
+
"epoch": 14.77,
|
| 4596 |
+
"learning_rate": 1.7379698144815434e-05,
|
| 4597 |
+
"loss": 1.3646,
|
| 4598 |
+
"step": 3825
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 14.79,
|
| 4602 |
+
"learning_rate": 1.4707946120313696e-05,
|
| 4603 |
+
"loss": 1.3313,
|
| 4604 |
+
"step": 3830
|
| 4605 |
+
},
|
| 4606 |
+
{
|
| 4607 |
+
"epoch": 14.81,
|
| 4608 |
+
"learning_rate": 1.2234361325042786e-05,
|
| 4609 |
+
"loss": 1.3923,
|
| 4610 |
+
"step": 3835
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 14.83,
|
| 4614 |
+
"learning_rate": 9.968039412440925e-06,
|
| 4615 |
+
"loss": 1.2976,
|
| 4616 |
+
"step": 3840
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 14.85,
|
| 4620 |
+
"learning_rate": 7.917313906685554e-06,
|
| 4621 |
+
"loss": 1.3127,
|
| 4622 |
+
"step": 3845
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 14.86,
|
| 4626 |
+
"learning_rate": 6.089725559373968e-06,
|
| 4627 |
+
"loss": 1.3699,
|
| 4628 |
+
"step": 3850
|
| 4629 |
+
},
|
| 4630 |
+
{
|
| 4631 |
+
"epoch": 14.88,
|
| 4632 |
+
"learning_rate": 4.4919946213203235e-06,
|
| 4633 |
+
"loss": 1.2705,
|
| 4634 |
+
"step": 3855
|
| 4635 |
+
},
|
| 4636 |
+
{
|
| 4637 |
+
"epoch": 14.9,
|
| 4638 |
+
"learning_rate": 3.129996131426458e-06,
|
| 4639 |
+
"loss": 1.3474,
|
| 4640 |
+
"step": 3860
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 14.92,
|
| 4644 |
+
"learning_rate": 2.00873831349432e-06,
|
| 4645 |
+
"loss": 1.3704,
|
| 4646 |
+
"step": 3865
|
| 4647 |
+
},
|
| 4648 |
+
{
|
| 4649 |
+
"epoch": 14.94,
|
| 4650 |
+
"learning_rate": 1.1323441604147607e-06,
|
| 4651 |
+
"loss": 1.3555,
|
| 4652 |
+
"step": 3870
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 14.96,
|
| 4656 |
+
"learning_rate": 5.040362734534312e-07,
|
| 4657 |
+
"loss": 1.3937,
|
| 4658 |
+
"step": 3875
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 14.98,
|
| 4662 |
+
"learning_rate": 1.2612501237755945e-07,
|
| 4663 |
+
"loss": 1.425,
|
| 4664 |
+
"step": 3880
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 15.0,
|
| 4668 |
+
"learning_rate": 0.0,
|
| 4669 |
+
"loss": 1.42,
|
| 4670 |
+
"step": 3885
|
| 4671 |
}
|
| 4672 |
],
|
| 4673 |
+
"max_steps": 3885,
|
| 4674 |
+
"num_train_epochs": 15,
|
| 4675 |
+
"total_flos": 4058518487040000.0,
|
| 4676 |
"trial_name": null,
|
| 4677 |
"trial_params": null
|
| 4678 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2671
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c8781197db3a3403466497c5da7316ba5318202ed62c8d7147bd3a02f7bd353
|
| 3 |
size 2671
|