Commit
•
2922f4d
1
Parent(s):
ef79ee9
huggingartists
Browse files- README.md +3 -3
- config.json +1 -1
- flax_model.msgpack +1 -1
- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- rng_state.pth +2 -2
- scheduler.pt +1 -1
- tokenizer_config.json +1 -1
- trainer_state.json +3365 -5
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -45,15 +45,15 @@ from datasets import load_dataset
|
|
| 45 |
dataset = load_dataset("huggingartists/ghostemane")
|
| 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 Ghostemane'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/ghostemane")
|
| 46 |
```
|
| 47 |
|
| 48 |
+
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/1ou29taa/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 Ghostemane's lyrics.
|
| 53 |
|
| 54 |
+
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/futdflju) for full transparency and reproducibility.
|
| 55 |
|
| 56 |
+
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/futdflju/artifacts) is logged and versioned.
|
| 57 |
|
| 58 |
## How to use
|
| 59 |
|
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPT2LMHeadModel"
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "huggingartists/ghostemane",
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPT2LMHeadModel"
|
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 sha256:0619e683bf55b59a27572f81d7d90c5b852ca70ec315b91727e9182fab0ebd1c
|
| 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 sha256:5ef5d476ae960840e9088b80d802bab5d691b7cda03c4a1fe7b038922aae4e6e
|
| 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 sha256:0a676ce8b88b5df21334331fe4d6429cbd36e74bb55f07a72d4b58bbd153aae3
|
| 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 sha256:0ade46d75af9421f8aa3fe1ab5cd95c44e6a4b80c62953c17cc8a140f9ed79bf
|
| 3 |
+
size 14503
|
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:c651c7b45d256dfba9a5dac3539dfc89264e9f5f466b8dc998f7e482845d6518
|
| 3 |
size 623
|
tokenizer_config.json
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "
|
|
|
|
| 1 |
+
{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "huggingartists/ghostemane", "tokenizer_class": "GPT2Tokenizer"}
|
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,
|
|
@@ -678,11 +678,3371 @@
|
|
| 678 |
"learning_rate": 0.0001372,
|
| 679 |
"loss": 2.1707,
|
| 680 |
"step": 560
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 681 |
}
|
| 682 |
],
|
| 683 |
-
"max_steps":
|
| 684 |
-
"num_train_epochs":
|
| 685 |
-
"total_flos":
|
| 686 |
"trial_name": null,
|
| 687 |
"trial_params": null
|
| 688 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"global_step": 3360,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 678 |
"learning_rate": 0.0001372,
|
| 679 |
"loss": 2.1707,
|
| 680 |
"step": 560
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 10.09,
|
| 684 |
+
"learning_rate": 0.00013451892828543393,
|
| 685 |
+
"loss": 1.7839,
|
| 686 |
+
"step": 565
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 10.18,
|
| 690 |
+
"learning_rate": 0.0001266852800670604,
|
| 691 |
+
"loss": 1.931,
|
| 692 |
+
"step": 570
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 10.27,
|
| 696 |
+
"learning_rate": 0.00011431137524750779,
|
| 697 |
+
"loss": 1.981,
|
| 698 |
+
"step": 575
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 10.36,
|
| 702 |
+
"learning_rate": 9.836442450346467e-05,
|
| 703 |
+
"loss": 1.9186,
|
| 704 |
+
"step": 580
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 10.45,
|
| 708 |
+
"learning_rate": 8.009092691870492e-05,
|
| 709 |
+
"loss": 1.9439,
|
| 710 |
+
"step": 585
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 10.54,
|
| 714 |
+
"learning_rate": 6.0919236939312867e-05,
|
| 715 |
+
"loss": 1.7981,
|
| 716 |
+
"step": 590
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 10.62,
|
| 720 |
+
"learning_rate": 4.23479165397549e-05,
|
| 721 |
+
"loss": 1.8503,
|
| 722 |
+
"step": 595
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 10.71,
|
| 726 |
+
"learning_rate": 2.5828599592491143e-05,
|
| 727 |
+
"loss": 2.1472,
|
| 728 |
+
"step": 600
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 10.8,
|
| 732 |
+
"learning_rate": 1.265252438939482e-05,
|
| 733 |
+
"loss": 1.8879,
|
| 734 |
+
"step": 605
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 10.89,
|
| 738 |
+
"learning_rate": 3.849603540846114e-06,
|
| 739 |
+
"loss": 2.0268,
|
| 740 |
+
"step": 610
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 10.98,
|
| 744 |
+
"learning_rate": 1.0792048977779616e-07,
|
| 745 |
+
"loss": 2.0845,
|
| 746 |
+
"step": 615
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 11.07,
|
| 750 |
+
"learning_rate": 1.7199452243269073e-06,
|
| 751 |
+
"loss": 2.1026,
|
| 752 |
+
"step": 620
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 11.16,
|
| 756 |
+
"learning_rate": 8.559673257059612e-06,
|
| 757 |
+
"loss": 1.8359,
|
| 758 |
+
"step": 625
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 11.25,
|
| 762 |
+
"learning_rate": 2.0092474810602812e-05,
|
| 763 |
+
"loss": 1.6721,
|
| 764 |
+
"step": 630
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 11.34,
|
| 768 |
+
"learning_rate": 3.541688434458027e-05,
|
| 769 |
+
"loss": 1.6882,
|
| 770 |
+
"step": 635
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 11.43,
|
| 774 |
+
"learning_rate": 5.333506393059674e-05,
|
| 775 |
+
"loss": 1.7527,
|
| 776 |
+
"step": 640
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 11.52,
|
| 780 |
+
"learning_rate": 7.244643268047099e-05,
|
| 781 |
+
"loss": 1.7121,
|
| 782 |
+
"step": 645
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 11.61,
|
| 786 |
+
"learning_rate": 9.125714365012432e-05,
|
| 787 |
+
"loss": 1.8074,
|
| 788 |
+
"step": 650
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 11.7,
|
| 792 |
+
"learning_rate": 0.00010829685091793471,
|
| 793 |
+
"loss": 1.8594,
|
| 794 |
+
"step": 655
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 11.79,
|
| 798 |
+
"learning_rate": 0.000122233639697307,
|
| 799 |
+
"loss": 1.8073,
|
| 800 |
+
"step": 660
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 11.88,
|
| 804 |
+
"learning_rate": 0.00013197813593027427,
|
| 805 |
+
"loss": 1.8472,
|
| 806 |
+
"step": 665
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 11.96,
|
| 810 |
+
"learning_rate": 0.00013676865759867642,
|
| 811 |
+
"loss": 1.8812,
|
| 812 |
+
"step": 670
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 12.05,
|
| 816 |
+
"learning_rate": 0.000136230751870351,
|
| 817 |
+
"loss": 1.5812,
|
| 818 |
+
"step": 675
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 12.14,
|
| 822 |
+
"learning_rate": 0.0001304064643381061,
|
| 823 |
+
"loss": 1.6567,
|
| 824 |
+
"step": 680
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 12.23,
|
| 828 |
+
"learning_rate": 0.00011975105251098525,
|
| 829 |
+
"loss": 1.6487,
|
| 830 |
+
"step": 685
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 12.32,
|
| 834 |
+
"learning_rate": 0.000105097400448952,
|
| 835 |
+
"loss": 1.698,
|
| 836 |
+
"step": 690
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 12.41,
|
| 840 |
+
"learning_rate": 8.759091608374439e-05,
|
| 841 |
+
"loss": 1.7092,
|
| 842 |
+
"step": 695
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 12.5,
|
| 846 |
+
"learning_rate": 6.859999999999999e-05,
|
| 847 |
+
"loss": 1.5692,
|
| 848 |
+
"step": 700
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 12.59,
|
| 852 |
+
"learning_rate": 4.960908391625558e-05,
|
| 853 |
+
"loss": 1.7959,
|
| 854 |
+
"step": 705
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 12.68,
|
| 858 |
+
"learning_rate": 3.210259955104795e-05,
|
| 859 |
+
"loss": 1.769,
|
| 860 |
+
"step": 710
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 12.77,
|
| 864 |
+
"learning_rate": 1.7448947489015055e-05,
|
| 865 |
+
"loss": 1.672,
|
| 866 |
+
"step": 715
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 12.86,
|
| 870 |
+
"learning_rate": 6.793535661894092e-06,
|
| 871 |
+
"loss": 1.6729,
|
| 872 |
+
"step": 720
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 12.95,
|
| 876 |
+
"learning_rate": 9.692481296490868e-07,
|
| 877 |
+
"loss": 1.6378,
|
| 878 |
+
"step": 725
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 13.04,
|
| 882 |
+
"learning_rate": 4.3134240132354215e-07,
|
| 883 |
+
"loss": 1.5157,
|
| 884 |
+
"step": 730
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 13.12,
|
| 888 |
+
"learning_rate": 5.221864069725745e-06,
|
| 889 |
+
"loss": 1.5695,
|
| 890 |
+
"step": 735
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 13.21,
|
| 894 |
+
"learning_rate": 1.496636030269301e-05,
|
| 895 |
+
"loss": 1.4221,
|
| 896 |
+
"step": 740
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 13.3,
|
| 900 |
+
"learning_rate": 2.890314908206531e-05,
|
| 901 |
+
"loss": 1.5325,
|
| 902 |
+
"step": 745
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 13.39,
|
| 906 |
+
"learning_rate": 4.5942856349875256e-05,
|
| 907 |
+
"loss": 1.5675,
|
| 908 |
+
"step": 750
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 13.48,
|
| 912 |
+
"learning_rate": 6.475356731952856e-05,
|
| 913 |
+
"loss": 1.4491,
|
| 914 |
+
"step": 755
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 13.57,
|
| 918 |
+
"learning_rate": 8.386493606940281e-05,
|
| 919 |
+
"loss": 1.4567,
|
| 920 |
+
"step": 760
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 13.66,
|
| 924 |
+
"learning_rate": 0.00010178311565541931,
|
| 925 |
+
"loss": 1.4152,
|
| 926 |
+
"step": 765
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 13.75,
|
| 930 |
+
"learning_rate": 0.00011710752518939722,
|
| 931 |
+
"loss": 1.6904,
|
| 932 |
+
"step": 770
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 13.84,
|
| 936 |
+
"learning_rate": 0.00012864032674294042,
|
| 937 |
+
"loss": 1.7226,
|
| 938 |
+
"step": 775
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 13.93,
|
| 942 |
+
"learning_rate": 0.0001354800547756731,
|
| 943 |
+
"loss": 1.6131,
|
| 944 |
+
"step": 780
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 14.02,
|
| 948 |
+
"learning_rate": 0.00013709207951022223,
|
| 949 |
+
"loss": 1.4874,
|
| 950 |
+
"step": 785
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 14.11,
|
| 954 |
+
"learning_rate": 0.00013335039645915404,
|
| 955 |
+
"loss": 1.321,
|
| 956 |
+
"step": 790
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 14.2,
|
| 960 |
+
"learning_rate": 0.00012454747561060542,
|
| 961 |
+
"loss": 1.5378,
|
| 962 |
+
"step": 795
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 14.29,
|
| 966 |
+
"learning_rate": 0.00011137140040750922,
|
| 967 |
+
"loss": 1.5117,
|
| 968 |
+
"step": 800
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 14.38,
|
| 972 |
+
"learning_rate": 9.485208346024507e-05,
|
| 973 |
+
"loss": 1.3334,
|
| 974 |
+
"step": 805
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 14.46,
|
| 978 |
+
"learning_rate": 7.62807630606871e-05,
|
| 979 |
+
"loss": 1.418,
|
| 980 |
+
"step": 810
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 14.55,
|
| 984 |
+
"learning_rate": 5.710907308129505e-05,
|
| 985 |
+
"loss": 1.4658,
|
| 986 |
+
"step": 815
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 14.64,
|
| 990 |
+
"learning_rate": 3.883557549653573e-05,
|
| 991 |
+
"loss": 1.6222,
|
| 992 |
+
"step": 820
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 14.73,
|
| 996 |
+
"learning_rate": 2.2888624752492553e-05,
|
| 997 |
+
"loss": 1.4192,
|
| 998 |
+
"step": 825
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 14.82,
|
| 1002 |
+
"learning_rate": 1.0514719932939869e-05,
|
| 1003 |
+
"loss": 1.346,
|
| 1004 |
+
"step": 830
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 14.91,
|
| 1008 |
+
"learning_rate": 2.681071714566198e-06,
|
| 1009 |
+
"loss": 1.3806,
|
| 1010 |
+
"step": 835
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 15.0,
|
| 1014 |
+
"learning_rate": 0.0,
|
| 1015 |
+
"loss": 1.384,
|
| 1016 |
+
"step": 840
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 15.09,
|
| 1020 |
+
"learning_rate": 2.6810717145661294e-06,
|
| 1021 |
+
"loss": 1.2256,
|
| 1022 |
+
"step": 845
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 15.18,
|
| 1026 |
+
"learning_rate": 1.0514719932939732e-05,
|
| 1027 |
+
"loss": 1.3474,
|
| 1028 |
+
"step": 850
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 15.27,
|
| 1032 |
+
"learning_rate": 2.2888624752492363e-05,
|
| 1033 |
+
"loss": 1.3215,
|
| 1034 |
+
"step": 855
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 15.36,
|
| 1038 |
+
"learning_rate": 3.88355754965355e-05,
|
| 1039 |
+
"loss": 1.2651,
|
| 1040 |
+
"step": 860
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 15.45,
|
| 1044 |
+
"learning_rate": 5.710907308129481e-05,
|
| 1045 |
+
"loss": 1.3224,
|
| 1046 |
+
"step": 865
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 15.54,
|
| 1050 |
+
"learning_rate": 7.628076306068686e-05,
|
| 1051 |
+
"loss": 1.1728,
|
| 1052 |
+
"step": 870
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 15.62,
|
| 1056 |
+
"learning_rate": 9.485208346024484e-05,
|
| 1057 |
+
"loss": 1.3578,
|
| 1058 |
+
"step": 875
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 15.71,
|
| 1062 |
+
"learning_rate": 0.00011137140040750902,
|
| 1063 |
+
"loss": 1.1825,
|
| 1064 |
+
"step": 880
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 15.8,
|
| 1068 |
+
"learning_rate": 0.0001245474756106053,
|
| 1069 |
+
"loss": 1.3246,
|
| 1070 |
+
"step": 885
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 15.89,
|
| 1074 |
+
"learning_rate": 0.00013335039645915393,
|
| 1075 |
+
"loss": 1.332,
|
| 1076 |
+
"step": 890
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 15.98,
|
| 1080 |
+
"learning_rate": 0.00013709207951022223,
|
| 1081 |
+
"loss": 1.388,
|
| 1082 |
+
"step": 895
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 16.07,
|
| 1086 |
+
"learning_rate": 0.00013548005477567304,
|
| 1087 |
+
"loss": 1.2832,
|
| 1088 |
+
"step": 900
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 16.16,
|
| 1092 |
+
"learning_rate": 0.00012864032674294074,
|
| 1093 |
+
"loss": 1.2149,
|
| 1094 |
+
"step": 905
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 16.25,
|
| 1098 |
+
"learning_rate": 0.00011710752518939739,
|
| 1099 |
+
"loss": 1.0402,
|
| 1100 |
+
"step": 910
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 16.34,
|
| 1104 |
+
"learning_rate": 0.00010178311565541954,
|
| 1105 |
+
"loss": 1.2199,
|
| 1106 |
+
"step": 915
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 16.43,
|
| 1110 |
+
"learning_rate": 8.386493606940354e-05,
|
| 1111 |
+
"loss": 1.2836,
|
| 1112 |
+
"step": 920
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 16.52,
|
| 1116 |
+
"learning_rate": 6.47535673195288e-05,
|
| 1117 |
+
"loss": 1.2179,
|
| 1118 |
+
"step": 925
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 16.61,
|
| 1122 |
+
"learning_rate": 4.59428563498755e-05,
|
| 1123 |
+
"loss": 1.2299,
|
| 1124 |
+
"step": 930
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 16.7,
|
| 1128 |
+
"learning_rate": 2.8903149082065114e-05,
|
| 1129 |
+
"loss": 1.072,
|
| 1130 |
+
"step": 935
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 16.79,
|
| 1134 |
+
"learning_rate": 1.4966360302693468e-05,
|
| 1135 |
+
"loss": 1.2192,
|
| 1136 |
+
"step": 940
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 16.88,
|
| 1140 |
+
"learning_rate": 5.221864069725844e-06,
|
| 1141 |
+
"loss": 1.2081,
|
| 1142 |
+
"step": 945
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 16.96,
|
| 1146 |
+
"learning_rate": 4.3134240132356497e-07,
|
| 1147 |
+
"loss": 1.3213,
|
| 1148 |
+
"step": 950
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 17.05,
|
| 1152 |
+
"learning_rate": 9.692481296489572e-07,
|
| 1153 |
+
"loss": 1.0468,
|
| 1154 |
+
"step": 955
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 17.14,
|
| 1158 |
+
"learning_rate": 6.793535661893986e-06,
|
| 1159 |
+
"loss": 1.1071,
|
| 1160 |
+
"step": 960
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 17.23,
|
| 1164 |
+
"learning_rate": 1.7448947489014885e-05,
|
| 1165 |
+
"loss": 1.0509,
|
| 1166 |
+
"step": 965
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 17.32,
|
| 1170 |
+
"learning_rate": 3.210259955104815e-05,
|
| 1171 |
+
"loss": 0.96,
|
| 1172 |
+
"step": 970
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 17.41,
|
| 1176 |
+
"learning_rate": 4.9609083916254864e-05,
|
| 1177 |
+
"loss": 1.1523,
|
| 1178 |
+
"step": 975
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 17.5,
|
| 1182 |
+
"learning_rate": 6.859999999999973e-05,
|
| 1183 |
+
"loss": 0.9922,
|
| 1184 |
+
"step": 980
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 17.59,
|
| 1188 |
+
"learning_rate": 8.759091608374462e-05,
|
| 1189 |
+
"loss": 1.1601,
|
| 1190 |
+
"step": 985
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 17.68,
|
| 1194 |
+
"learning_rate": 0.0001050974004489518,
|
| 1195 |
+
"loss": 0.9979,
|
| 1196 |
+
"step": 990
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 17.77,
|
| 1200 |
+
"learning_rate": 0.00011975105251098509,
|
| 1201 |
+
"loss": 1.1048,
|
| 1202 |
+
"step": 995
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 17.86,
|
| 1206 |
+
"learning_rate": 0.00013040646433810598,
|
| 1207 |
+
"loss": 1.0999,
|
| 1208 |
+
"step": 1000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 17.95,
|
| 1212 |
+
"learning_rate": 0.00013623075187035104,
|
| 1213 |
+
"loss": 1.2753,
|
| 1214 |
+
"step": 1005
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 18.04,
|
| 1218 |
+
"learning_rate": 0.0001367686575986765,
|
| 1219 |
+
"loss": 1.0046,
|
| 1220 |
+
"step": 1010
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 18.12,
|
| 1224 |
+
"learning_rate": 0.00013197813593027435,
|
| 1225 |
+
"loss": 0.9683,
|
| 1226 |
+
"step": 1015
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 18.21,
|
| 1230 |
+
"learning_rate": 0.00012223363969730686,
|
| 1231 |
+
"loss": 1.0153,
|
| 1232 |
+
"step": 1020
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 18.3,
|
| 1236 |
+
"learning_rate": 0.00010829685091793493,
|
| 1237 |
+
"loss": 0.9562,
|
| 1238 |
+
"step": 1025
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 18.39,
|
| 1242 |
+
"learning_rate": 9.125714365012455e-05,
|
| 1243 |
+
"loss": 1.0145,
|
| 1244 |
+
"step": 1030
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 18.48,
|
| 1248 |
+
"learning_rate": 7.244643268047124e-05,
|
| 1249 |
+
"loss": 1.0414,
|
| 1250 |
+
"step": 1035
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 18.57,
|
| 1254 |
+
"learning_rate": 5.333506393059651e-05,
|
| 1255 |
+
"loss": 1.0547,
|
| 1256 |
+
"step": 1040
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 18.66,
|
| 1260 |
+
"learning_rate": 3.541688434458093e-05,
|
| 1261 |
+
"loss": 1.038,
|
| 1262 |
+
"step": 1045
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 18.75,
|
| 1266 |
+
"learning_rate": 2.0092474810602995e-05,
|
| 1267 |
+
"loss": 0.9887,
|
| 1268 |
+
"step": 1050
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 18.84,
|
| 1272 |
+
"learning_rate": 8.559673257059497e-06,
|
| 1273 |
+
"loss": 1.0939,
|
| 1274 |
+
"step": 1055
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 18.93,
|
| 1278 |
+
"learning_rate": 1.7199452243269606e-06,
|
| 1279 |
+
"loss": 1.0271,
|
| 1280 |
+
"step": 1060
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 19.02,
|
| 1284 |
+
"learning_rate": 1.0792048977778093e-07,
|
| 1285 |
+
"loss": 0.8933,
|
| 1286 |
+
"step": 1065
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 19.11,
|
| 1290 |
+
"learning_rate": 3.84960354084603e-06,
|
| 1291 |
+
"loss": 0.9411,
|
| 1292 |
+
"step": 1070
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 19.2,
|
| 1296 |
+
"learning_rate": 1.2652524389394958e-05,
|
| 1297 |
+
"loss": 0.8984,
|
| 1298 |
+
"step": 1075
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 19.29,
|
| 1302 |
+
"learning_rate": 2.5828599592490564e-05,
|
| 1303 |
+
"loss": 0.8203,
|
| 1304 |
+
"step": 1080
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 19.38,
|
| 1308 |
+
"learning_rate": 4.234791653975466e-05,
|
| 1309 |
+
"loss": 0.8312,
|
| 1310 |
+
"step": 1085
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 19.46,
|
| 1314 |
+
"learning_rate": 6.09192369393131e-05,
|
| 1315 |
+
"loss": 0.7923,
|
| 1316 |
+
"step": 1090
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 19.55,
|
| 1320 |
+
"learning_rate": 8.009092691870466e-05,
|
| 1321 |
+
"loss": 0.8649,
|
| 1322 |
+
"step": 1095
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 19.64,
|
| 1326 |
+
"learning_rate": 9.836442450346445e-05,
|
| 1327 |
+
"loss": 0.939,
|
| 1328 |
+
"step": 1100
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 19.73,
|
| 1332 |
+
"learning_rate": 0.0001143113752475076,
|
| 1333 |
+
"loss": 0.8501,
|
| 1334 |
+
"step": 1105
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 19.82,
|
| 1338 |
+
"learning_rate": 0.0001266852800670605,
|
| 1339 |
+
"loss": 0.9123,
|
| 1340 |
+
"step": 1110
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 19.91,
|
| 1344 |
+
"learning_rate": 0.0001345189282854337,
|
| 1345 |
+
"loss": 0.9598,
|
| 1346 |
+
"step": 1115
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 20.0,
|
| 1350 |
+
"learning_rate": 0.0001372,
|
| 1351 |
+
"loss": 0.9292,
|
| 1352 |
+
"step": 1120
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 20.09,
|
| 1356 |
+
"learning_rate": 0.00013451892828543382,
|
| 1357 |
+
"loss": 0.8035,
|
| 1358 |
+
"step": 1125
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 20.18,
|
| 1362 |
+
"learning_rate": 0.00012668528006706069,
|
| 1363 |
+
"loss": 0.792,
|
| 1364 |
+
"step": 1130
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 20.27,
|
| 1368 |
+
"learning_rate": 0.00011431137524750785,
|
| 1369 |
+
"loss": 0.8557,
|
| 1370 |
+
"step": 1135
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 20.36,
|
| 1374 |
+
"learning_rate": 9.836442450346476e-05,
|
| 1375 |
+
"loss": 0.8628,
|
| 1376 |
+
"step": 1140
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 20.45,
|
| 1380 |
+
"learning_rate": 8.0090926918705e-05,
|
| 1381 |
+
"loss": 0.8219,
|
| 1382 |
+
"step": 1145
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 20.54,
|
| 1386 |
+
"learning_rate": 6.091923693931392e-05,
|
| 1387 |
+
"loss": 0.8509,
|
| 1388 |
+
"step": 1150
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 20.62,
|
| 1392 |
+
"learning_rate": 4.234791653975543e-05,
|
| 1393 |
+
"loss": 0.8817,
|
| 1394 |
+
"step": 1155
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 20.71,
|
| 1398 |
+
"learning_rate": 2.5828599592491204e-05,
|
| 1399 |
+
"loss": 0.8186,
|
| 1400 |
+
"step": 1160
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 20.8,
|
| 1404 |
+
"learning_rate": 1.2652524389394875e-05,
|
| 1405 |
+
"loss": 0.8184,
|
| 1406 |
+
"step": 1165
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 20.89,
|
| 1410 |
+
"learning_rate": 3.849603540845984e-06,
|
| 1411 |
+
"loss": 0.8315,
|
| 1412 |
+
"step": 1170
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 20.98,
|
| 1416 |
+
"learning_rate": 1.0792048977777332e-07,
|
| 1417 |
+
"loss": 0.8831,
|
| 1418 |
+
"step": 1175
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 21.07,
|
| 1422 |
+
"learning_rate": 1.7199452243269987e-06,
|
| 1423 |
+
"loss": 0.737,
|
| 1424 |
+
"step": 1180
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 21.16,
|
| 1428 |
+
"learning_rate": 8.559673257059337e-06,
|
| 1429 |
+
"loss": 0.776,
|
| 1430 |
+
"step": 1185
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 21.25,
|
| 1434 |
+
"learning_rate": 2.009247481060276e-05,
|
| 1435 |
+
"loss": 0.6944,
|
| 1436 |
+
"step": 1190
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 21.34,
|
| 1440 |
+
"learning_rate": 3.541688434458063e-05,
|
| 1441 |
+
"loss": 0.6937,
|
| 1442 |
+
"step": 1195
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 21.43,
|
| 1446 |
+
"learning_rate": 5.333506393059618e-05,
|
| 1447 |
+
"loss": 0.7423,
|
| 1448 |
+
"step": 1200
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 21.52,
|
| 1452 |
+
"learning_rate": 7.24464326804709e-05,
|
| 1453 |
+
"loss": 0.7534,
|
| 1454 |
+
"step": 1205
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 21.61,
|
| 1458 |
+
"learning_rate": 9.125714365012422e-05,
|
| 1459 |
+
"loss": 0.6734,
|
| 1460 |
+
"step": 1210
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 21.7,
|
| 1464 |
+
"learning_rate": 0.00010829685091793466,
|
| 1465 |
+
"loss": 0.7151,
|
| 1466 |
+
"step": 1215
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 21.79,
|
| 1470 |
+
"learning_rate": 0.00012223363969730635,
|
| 1471 |
+
"loss": 0.8225,
|
| 1472 |
+
"step": 1220
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 21.88,
|
| 1476 |
+
"learning_rate": 0.00013197813593027405,
|
| 1477 |
+
"loss": 0.6728,
|
| 1478 |
+
"step": 1225
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 21.96,
|
| 1482 |
+
"learning_rate": 0.0001367686575986764,
|
| 1483 |
+
"loss": 0.7781,
|
| 1484 |
+
"step": 1230
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 22.05,
|
| 1488 |
+
"learning_rate": 0.00013623075187035101,
|
| 1489 |
+
"loss": 0.724,
|
| 1490 |
+
"step": 1235
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 22.14,
|
| 1494 |
+
"learning_rate": 0.00013040646433810593,
|
| 1495 |
+
"loss": 0.5813,
|
| 1496 |
+
"step": 1240
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 22.23,
|
| 1500 |
+
"learning_rate": 0.00011975105251098498,
|
| 1501 |
+
"loss": 0.7065,
|
| 1502 |
+
"step": 1245
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 22.32,
|
| 1506 |
+
"learning_rate": 0.00010509740044895168,
|
| 1507 |
+
"loss": 0.6825,
|
| 1508 |
+
"step": 1250
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 22.41,
|
| 1512 |
+
"learning_rate": 8.759091608374493e-05,
|
| 1513 |
+
"loss": 0.7155,
|
| 1514 |
+
"step": 1255
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 22.5,
|
| 1518 |
+
"learning_rate": 6.860000000000005e-05,
|
| 1519 |
+
"loss": 0.7597,
|
| 1520 |
+
"step": 1260
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 22.59,
|
| 1524 |
+
"learning_rate": 4.9609083916255196e-05,
|
| 1525 |
+
"loss": 0.749,
|
| 1526 |
+
"step": 1265
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 22.68,
|
| 1530 |
+
"learning_rate": 3.2102599551048435e-05,
|
| 1531 |
+
"loss": 0.7066,
|
| 1532 |
+
"step": 1270
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 22.77,
|
| 1536 |
+
"learning_rate": 1.7448947489015106e-05,
|
| 1537 |
+
"loss": 0.6666,
|
| 1538 |
+
"step": 1275
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 22.86,
|
| 1542 |
+
"learning_rate": 6.7935356618941304e-06,
|
| 1543 |
+
"loss": 0.6573,
|
| 1544 |
+
"step": 1280
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 22.95,
|
| 1548 |
+
"learning_rate": 9.692481296490182e-07,
|
| 1549 |
+
"loss": 0.7112,
|
| 1550 |
+
"step": 1285
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 23.04,
|
| 1554 |
+
"learning_rate": 4.313424013234736e-07,
|
| 1555 |
+
"loss": 0.5373,
|
| 1556 |
+
"step": 1290
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 23.12,
|
| 1560 |
+
"learning_rate": 5.221864069725524e-06,
|
| 1561 |
+
"loss": 0.5479,
|
| 1562 |
+
"step": 1295
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 23.21,
|
| 1566 |
+
"learning_rate": 1.4966360302692958e-05,
|
| 1567 |
+
"loss": 0.5882,
|
| 1568 |
+
"step": 1300
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 23.3,
|
| 1572 |
+
"learning_rate": 2.8903149082065243e-05,
|
| 1573 |
+
"loss": 0.5555,
|
| 1574 |
+
"step": 1305
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 23.39,
|
| 1578 |
+
"learning_rate": 4.5942856349875636e-05,
|
| 1579 |
+
"loss": 0.5804,
|
| 1580 |
+
"step": 1310
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 23.48,
|
| 1584 |
+
"learning_rate": 6.475356731952897e-05,
|
| 1585 |
+
"loss": 0.5679,
|
| 1586 |
+
"step": 1315
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 23.57,
|
| 1590 |
+
"learning_rate": 8.386493606940368e-05,
|
| 1591 |
+
"loss": 0.6024,
|
| 1592 |
+
"step": 1320
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 23.66,
|
| 1596 |
+
"learning_rate": 0.00010178311565541925,
|
| 1597 |
+
"loss": 0.552,
|
| 1598 |
+
"step": 1325
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 23.75,
|
| 1602 |
+
"learning_rate": 0.00011710752518939715,
|
| 1603 |
+
"loss": 0.6404,
|
| 1604 |
+
"step": 1330
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 23.84,
|
| 1608 |
+
"learning_rate": 0.0001286403267429406,
|
| 1609 |
+
"loss": 0.581,
|
| 1610 |
+
"step": 1335
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 23.93,
|
| 1614 |
+
"learning_rate": 0.00013548005477567298,
|
| 1615 |
+
"loss": 0.6956,
|
| 1616 |
+
"step": 1340
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 24.02,
|
| 1620 |
+
"learning_rate": 0.00013709207951022223,
|
| 1621 |
+
"loss": 0.6156,
|
| 1622 |
+
"step": 1345
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 24.11,
|
| 1626 |
+
"learning_rate": 0.00013335039645915407,
|
| 1627 |
+
"loss": 0.5472,
|
| 1628 |
+
"step": 1350
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 24.2,
|
| 1632 |
+
"learning_rate": 0.0001245474756106052,
|
| 1633 |
+
"loss": 0.5691,
|
| 1634 |
+
"step": 1355
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 24.29,
|
| 1638 |
+
"learning_rate": 0.00011137140040750965,
|
| 1639 |
+
"loss": 0.5632,
|
| 1640 |
+
"step": 1360
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 24.38,
|
| 1644 |
+
"learning_rate": 9.485208346024561e-05,
|
| 1645 |
+
"loss": 0.6406,
|
| 1646 |
+
"step": 1365
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 24.46,
|
| 1650 |
+
"learning_rate": 7.628076306068718e-05,
|
| 1651 |
+
"loss": 0.5618,
|
| 1652 |
+
"step": 1370
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 24.55,
|
| 1656 |
+
"learning_rate": 5.710907308129514e-05,
|
| 1657 |
+
"loss": 0.5619,
|
| 1658 |
+
"step": 1375
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 24.64,
|
| 1662 |
+
"learning_rate": 3.8835575496535365e-05,
|
| 1663 |
+
"loss": 0.5453,
|
| 1664 |
+
"step": 1380
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 24.73,
|
| 1668 |
+
"learning_rate": 2.288862475249225e-05,
|
| 1669 |
+
"loss": 0.6172,
|
| 1670 |
+
"step": 1385
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 24.82,
|
| 1674 |
+
"learning_rate": 1.0514719932939396e-05,
|
| 1675 |
+
"loss": 0.5714,
|
| 1676 |
+
"step": 1390
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 24.91,
|
| 1680 |
+
"learning_rate": 2.681071714566221e-06,
|
| 1681 |
+
"loss": 0.5045,
|
| 1682 |
+
"step": 1395
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 25.0,
|
| 1686 |
+
"learning_rate": 0.0,
|
| 1687 |
+
"loss": 0.5848,
|
| 1688 |
+
"step": 1400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 25.09,
|
| 1692 |
+
"learning_rate": 2.681071714566236e-06,
|
| 1693 |
+
"loss": 0.4364,
|
| 1694 |
+
"step": 1405
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 25.18,
|
| 1698 |
+
"learning_rate": 1.0514719932939435e-05,
|
| 1699 |
+
"loss": 0.4791,
|
| 1700 |
+
"step": 1410
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 25.27,
|
| 1704 |
+
"learning_rate": 2.2888624752492302e-05,
|
| 1705 |
+
"loss": 0.4854,
|
| 1706 |
+
"step": 1415
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 25.36,
|
| 1710 |
+
"learning_rate": 3.8835575496535426e-05,
|
| 1711 |
+
"loss": 0.3908,
|
| 1712 |
+
"step": 1420
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 25.45,
|
| 1716 |
+
"learning_rate": 5.7109073081295205e-05,
|
| 1717 |
+
"loss": 0.4823,
|
| 1718 |
+
"step": 1425
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 25.54,
|
| 1722 |
+
"learning_rate": 7.628076306068627e-05,
|
| 1723 |
+
"loss": 0.478,
|
| 1724 |
+
"step": 1430
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 25.62,
|
| 1728 |
+
"learning_rate": 9.485208346024477e-05,
|
| 1729 |
+
"loss": 0.4768,
|
| 1730 |
+
"step": 1435
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 25.71,
|
| 1734 |
+
"learning_rate": 0.00011137140040750896,
|
| 1735 |
+
"loss": 0.4795,
|
| 1736 |
+
"step": 1440
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 25.8,
|
| 1740 |
+
"learning_rate": 0.0001245474756106047,
|
| 1741 |
+
"loss": 0.4712,
|
| 1742 |
+
"step": 1445
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 25.89,
|
| 1746 |
+
"learning_rate": 0.00013335039645915377,
|
| 1747 |
+
"loss": 0.5162,
|
| 1748 |
+
"step": 1450
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 25.98,
|
| 1752 |
+
"learning_rate": 0.0001370920795102222,
|
| 1753 |
+
"loss": 0.6066,
|
| 1754 |
+
"step": 1455
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 26.07,
|
| 1758 |
+
"learning_rate": 0.0001354800547756732,
|
| 1759 |
+
"loss": 0.4381,
|
| 1760 |
+
"step": 1460
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 26.16,
|
| 1764 |
+
"learning_rate": 0.00012864032674294058,
|
| 1765 |
+
"loss": 0.4356,
|
| 1766 |
+
"step": 1465
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 26.25,
|
| 1770 |
+
"learning_rate": 0.0001171075251893971,
|
| 1771 |
+
"loss": 0.4588,
|
| 1772 |
+
"step": 1470
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 26.34,
|
| 1776 |
+
"learning_rate": 0.00010178311565541919,
|
| 1777 |
+
"loss": 0.4378,
|
| 1778 |
+
"step": 1475
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 26.43,
|
| 1782 |
+
"learning_rate": 8.386493606940363e-05,
|
| 1783 |
+
"loss": 0.4972,
|
| 1784 |
+
"step": 1480
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 26.52,
|
| 1788 |
+
"learning_rate": 6.47535673195289e-05,
|
| 1789 |
+
"loss": 0.5317,
|
| 1790 |
+
"step": 1485
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 26.61,
|
| 1794 |
+
"learning_rate": 4.5942856349875575e-05,
|
| 1795 |
+
"loss": 0.4449,
|
| 1796 |
+
"step": 1490
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 26.7,
|
| 1800 |
+
"learning_rate": 2.8903149082065182e-05,
|
| 1801 |
+
"loss": 0.5294,
|
| 1802 |
+
"step": 1495
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 26.79,
|
| 1806 |
+
"learning_rate": 1.496636030269352e-05,
|
| 1807 |
+
"loss": 0.4735,
|
| 1808 |
+
"step": 1500
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 26.88,
|
| 1812 |
+
"learning_rate": 5.221864069725874e-06,
|
| 1813 |
+
"loss": 0.4522,
|
| 1814 |
+
"step": 1505
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 26.96,
|
| 1818 |
+
"learning_rate": 4.313424013235802e-07,
|
| 1819 |
+
"loss": 0.4696,
|
| 1820 |
+
"step": 1510
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 27.05,
|
| 1824 |
+
"learning_rate": 9.69248129648866e-07,
|
| 1825 |
+
"loss": 0.3814,
|
| 1826 |
+
"step": 1515
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 27.14,
|
| 1830 |
+
"learning_rate": 6.793535661893734e-06,
|
| 1831 |
+
"loss": 0.3927,
|
| 1832 |
+
"step": 1520
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 27.23,
|
| 1836 |
+
"learning_rate": 1.7448947489014506e-05,
|
| 1837 |
+
"loss": 0.4072,
|
| 1838 |
+
"step": 1525
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 27.32,
|
| 1842 |
+
"learning_rate": 3.210259955104767e-05,
|
| 1843 |
+
"loss": 0.4177,
|
| 1844 |
+
"step": 1530
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 27.41,
|
| 1848 |
+
"learning_rate": 4.9609083916255264e-05,
|
| 1849 |
+
"loss": 0.3721,
|
| 1850 |
+
"step": 1535
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 27.5,
|
| 1854 |
+
"learning_rate": 6.860000000000014e-05,
|
| 1855 |
+
"loss": 0.3918,
|
| 1856 |
+
"step": 1540
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 27.59,
|
| 1860 |
+
"learning_rate": 8.7590916083745e-05,
|
| 1861 |
+
"loss": 0.391,
|
| 1862 |
+
"step": 1545
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 27.68,
|
| 1866 |
+
"learning_rate": 0.00010509740044895174,
|
| 1867 |
+
"loss": 0.368,
|
| 1868 |
+
"step": 1550
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 27.77,
|
| 1872 |
+
"learning_rate": 0.00011975105251098503,
|
| 1873 |
+
"loss": 0.4225,
|
| 1874 |
+
"step": 1555
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 27.86,
|
| 1878 |
+
"learning_rate": 0.00013040646433810595,
|
| 1879 |
+
"loss": 0.3877,
|
| 1880 |
+
"step": 1560
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 27.95,
|
| 1884 |
+
"learning_rate": 0.00013623075187035101,
|
| 1885 |
+
"loss": 0.4493,
|
| 1886 |
+
"step": 1565
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 28.04,
|
| 1890 |
+
"learning_rate": 0.00013676865759867652,
|
| 1891 |
+
"loss": 0.4004,
|
| 1892 |
+
"step": 1570
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 28.12,
|
| 1896 |
+
"learning_rate": 0.0001319781359302744,
|
| 1897 |
+
"loss": 0.3739,
|
| 1898 |
+
"step": 1575
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 28.21,
|
| 1902 |
+
"learning_rate": 0.00012223363969730692,
|
| 1903 |
+
"loss": 0.3801,
|
| 1904 |
+
"step": 1580
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 28.3,
|
| 1908 |
+
"learning_rate": 0.00010829685091793539,
|
| 1909 |
+
"loss": 0.4067,
|
| 1910 |
+
"step": 1585
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 28.39,
|
| 1914 |
+
"learning_rate": 9.125714365012509e-05,
|
| 1915 |
+
"loss": 0.3734,
|
| 1916 |
+
"step": 1590
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 28.48,
|
| 1920 |
+
"learning_rate": 7.244643268047182e-05,
|
| 1921 |
+
"loss": 0.3794,
|
| 1922 |
+
"step": 1595
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 28.57,
|
| 1926 |
+
"learning_rate": 5.3335063930597066e-05,
|
| 1927 |
+
"loss": 0.4073,
|
| 1928 |
+
"step": 1600
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 28.66,
|
| 1932 |
+
"learning_rate": 3.541688434458058e-05,
|
| 1933 |
+
"loss": 0.4003,
|
| 1934 |
+
"step": 1605
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 28.75,
|
| 1938 |
+
"learning_rate": 2.0092474810602707e-05,
|
| 1939 |
+
"loss": 0.3867,
|
| 1940 |
+
"step": 1610
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 28.84,
|
| 1944 |
+
"learning_rate": 8.559673257059307e-06,
|
| 1945 |
+
"loss": 0.368,
|
| 1946 |
+
"step": 1615
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 28.93,
|
| 1950 |
+
"learning_rate": 1.7199452243269835e-06,
|
| 1951 |
+
"loss": 0.3861,
|
| 1952 |
+
"step": 1620
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 29.02,
|
| 1956 |
+
"learning_rate": 1.0792048977777332e-07,
|
| 1957 |
+
"loss": 0.3113,
|
| 1958 |
+
"step": 1625
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 29.11,
|
| 1962 |
+
"learning_rate": 3.849603540846007e-06,
|
| 1963 |
+
"loss": 0.2884,
|
| 1964 |
+
"step": 1630
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 29.2,
|
| 1968 |
+
"learning_rate": 1.2652524389394912e-05,
|
| 1969 |
+
"loss": 0.341,
|
| 1970 |
+
"step": 1635
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 29.29,
|
| 1974 |
+
"learning_rate": 2.5828599592490496e-05,
|
| 1975 |
+
"loss": 0.2955,
|
| 1976 |
+
"step": 1640
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 29.38,
|
| 1980 |
+
"learning_rate": 4.234791653975459e-05,
|
| 1981 |
+
"loss": 0.282,
|
| 1982 |
+
"step": 1645
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 29.46,
|
| 1986 |
+
"learning_rate": 6.0919236939313016e-05,
|
| 1987 |
+
"loss": 0.2748,
|
| 1988 |
+
"step": 1650
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 29.55,
|
| 1992 |
+
"learning_rate": 8.009092691870409e-05,
|
| 1993 |
+
"loss": 0.3101,
|
| 1994 |
+
"step": 1655
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 29.64,
|
| 1998 |
+
"learning_rate": 9.836442450346394e-05,
|
| 1999 |
+
"loss": 0.334,
|
| 2000 |
+
"step": 1660
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 29.73,
|
| 2004 |
+
"learning_rate": 0.00011431137524750716,
|
| 2005 |
+
"loss": 0.3008,
|
| 2006 |
+
"step": 1665
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 29.82,
|
| 2010 |
+
"learning_rate": 0.0001266852800670602,
|
| 2011 |
+
"loss": 0.3448,
|
| 2012 |
+
"step": 1670
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 29.91,
|
| 2016 |
+
"learning_rate": 0.00013451892828543358,
|
| 2017 |
+
"loss": 0.3626,
|
| 2018 |
+
"step": 1675
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 30.0,
|
| 2022 |
+
"learning_rate": 0.0001372,
|
| 2023 |
+
"loss": 0.3804,
|
| 2024 |
+
"step": 1680
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 30.09,
|
| 2028 |
+
"learning_rate": 0.0001345189282854337,
|
| 2029 |
+
"loss": 0.2925,
|
| 2030 |
+
"step": 1685
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 30.18,
|
| 2034 |
+
"learning_rate": 0.00012668528006706047,
|
| 2035 |
+
"loss": 0.3217,
|
| 2036 |
+
"step": 1690
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 30.27,
|
| 2040 |
+
"learning_rate": 0.00011431137524750754,
|
| 2041 |
+
"loss": 0.3431,
|
| 2042 |
+
"step": 1695
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 30.36,
|
| 2046 |
+
"learning_rate": 9.83644245034644e-05,
|
| 2047 |
+
"loss": 0.2936,
|
| 2048 |
+
"step": 1700
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 30.45,
|
| 2052 |
+
"learning_rate": 8.009092691870459e-05,
|
| 2053 |
+
"loss": 0.3465,
|
| 2054 |
+
"step": 1705
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 30.54,
|
| 2058 |
+
"learning_rate": 6.091923693931352e-05,
|
| 2059 |
+
"loss": 0.3303,
|
| 2060 |
+
"step": 1710
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 30.62,
|
| 2064 |
+
"learning_rate": 4.234791653975505e-05,
|
| 2065 |
+
"loss": 0.3262,
|
| 2066 |
+
"step": 1715
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 30.71,
|
| 2070 |
+
"learning_rate": 2.582859959249089e-05,
|
| 2071 |
+
"loss": 0.3351,
|
| 2072 |
+
"step": 1720
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 30.8,
|
| 2076 |
+
"learning_rate": 1.2652524389395202e-05,
|
| 2077 |
+
"loss": 0.3175,
|
| 2078 |
+
"step": 1725
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 30.89,
|
| 2082 |
+
"learning_rate": 3.849603540846175e-06,
|
| 2083 |
+
"loss": 0.2689,
|
| 2084 |
+
"step": 1730
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 30.98,
|
| 2088 |
+
"learning_rate": 1.0792048977780377e-07,
|
| 2089 |
+
"loss": 0.3047,
|
| 2090 |
+
"step": 1735
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 31.07,
|
| 2094 |
+
"learning_rate": 1.7199452243268694e-06,
|
| 2095 |
+
"loss": 0.2869,
|
| 2096 |
+
"step": 1740
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 31.16,
|
| 2100 |
+
"learning_rate": 8.559673257059063e-06,
|
| 2101 |
+
"loss": 0.2604,
|
| 2102 |
+
"step": 1745
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 31.25,
|
| 2106 |
+
"learning_rate": 2.0092474810602348e-05,
|
| 2107 |
+
"loss": 0.2769,
|
| 2108 |
+
"step": 1750
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 31.34,
|
| 2112 |
+
"learning_rate": 3.541688434458014e-05,
|
| 2113 |
+
"loss": 0.2263,
|
| 2114 |
+
"step": 1755
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 31.43,
|
| 2118 |
+
"learning_rate": 5.333506393059658e-05,
|
| 2119 |
+
"loss": 0.2346,
|
| 2120 |
+
"step": 1760
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 31.52,
|
| 2124 |
+
"learning_rate": 7.244643268047132e-05,
|
| 2125 |
+
"loss": 0.2531,
|
| 2126 |
+
"step": 1765
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 31.61,
|
| 2130 |
+
"learning_rate": 9.125714365012463e-05,
|
| 2131 |
+
"loss": 0.26,
|
| 2132 |
+
"step": 1770
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 31.7,
|
| 2136 |
+
"learning_rate": 0.00010829685091793499,
|
| 2137 |
+
"loss": 0.2807,
|
| 2138 |
+
"step": 1775
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 31.79,
|
| 2142 |
+
"learning_rate": 0.00012223363969730662,
|
| 2143 |
+
"loss": 0.2235,
|
| 2144 |
+
"step": 1780
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 31.88,
|
| 2148 |
+
"learning_rate": 0.0001319781359302742,
|
| 2149 |
+
"loss": 0.2793,
|
| 2150 |
+
"step": 1785
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 31.96,
|
| 2154 |
+
"learning_rate": 0.00013676865759867644,
|
| 2155 |
+
"loss": 0.2978,
|
| 2156 |
+
"step": 1790
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 32.05,
|
| 2160 |
+
"learning_rate": 0.00013623075187035093,
|
| 2161 |
+
"loss": 0.274,
|
| 2162 |
+
"step": 1795
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 32.14,
|
| 2166 |
+
"learning_rate": 0.00013040646433810576,
|
| 2167 |
+
"loss": 0.2887,
|
| 2168 |
+
"step": 1800
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 32.23,
|
| 2172 |
+
"learning_rate": 0.00011975105251098601,
|
| 2173 |
+
"loss": 0.2648,
|
| 2174 |
+
"step": 1805
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 32.32,
|
| 2178 |
+
"learning_rate": 0.00010509740044895298,
|
| 2179 |
+
"loss": 0.2468,
|
| 2180 |
+
"step": 1810
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 32.41,
|
| 2184 |
+
"learning_rate": 8.759091608374549e-05,
|
| 2185 |
+
"loss": 0.2918,
|
| 2186 |
+
"step": 1815
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 32.5,
|
| 2190 |
+
"learning_rate": 6.860000000000064e-05,
|
| 2191 |
+
"loss": 0.2799,
|
| 2192 |
+
"step": 1820
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 32.59,
|
| 2196 |
+
"learning_rate": 4.9609083916255745e-05,
|
| 2197 |
+
"loss": 0.2562,
|
| 2198 |
+
"step": 1825
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 32.68,
|
| 2202 |
+
"learning_rate": 3.2102599551048096e-05,
|
| 2203 |
+
"loss": 0.2663,
|
| 2204 |
+
"step": 1830
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 32.77,
|
| 2208 |
+
"learning_rate": 1.744894748901484e-05,
|
| 2209 |
+
"loss": 0.2774,
|
| 2210 |
+
"step": 1835
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 32.86,
|
| 2214 |
+
"learning_rate": 6.793535661894382e-06,
|
| 2215 |
+
"loss": 0.2487,
|
| 2216 |
+
"step": 1840
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 32.95,
|
| 2220 |
+
"learning_rate": 9.692481296491097e-07,
|
| 2221 |
+
"loss": 0.2686,
|
| 2222 |
+
"step": 1845
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 33.04,
|
| 2226 |
+
"learning_rate": 4.313424013235193e-07,
|
| 2227 |
+
"loss": 0.2175,
|
| 2228 |
+
"step": 1850
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 33.12,
|
| 2232 |
+
"learning_rate": 5.221864069725684e-06,
|
| 2233 |
+
"loss": 0.218,
|
| 2234 |
+
"step": 1855
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 33.21,
|
| 2238 |
+
"learning_rate": 1.4966360302693209e-05,
|
| 2239 |
+
"loss": 0.216,
|
| 2240 |
+
"step": 1860
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 33.3,
|
| 2244 |
+
"learning_rate": 2.890314908206557e-05,
|
| 2245 |
+
"loss": 0.2596,
|
| 2246 |
+
"step": 1865
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 33.39,
|
| 2250 |
+
"learning_rate": 4.5942856349876015e-05,
|
| 2251 |
+
"loss": 0.2206,
|
| 2252 |
+
"step": 1870
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 33.48,
|
| 2256 |
+
"learning_rate": 6.475356731952742e-05,
|
| 2257 |
+
"loss": 0.1793,
|
| 2258 |
+
"step": 1875
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 33.57,
|
| 2262 |
+
"learning_rate": 8.386493606940219e-05,
|
| 2263 |
+
"loss": 0.2108,
|
| 2264 |
+
"step": 1880
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 33.66,
|
| 2268 |
+
"learning_rate": 0.00010178311565541875,
|
| 2269 |
+
"loss": 0.2287,
|
| 2270 |
+
"step": 1885
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 33.75,
|
| 2274 |
+
"learning_rate": 0.00011710752518939675,
|
| 2275 |
+
"loss": 0.1961,
|
| 2276 |
+
"step": 1890
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 33.84,
|
| 2280 |
+
"learning_rate": 0.00012864032674294034,
|
| 2281 |
+
"loss": 0.2072,
|
| 2282 |
+
"step": 1895
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 33.93,
|
| 2286 |
+
"learning_rate": 0.00013548005477567306,
|
| 2287 |
+
"loss": 0.2405,
|
| 2288 |
+
"step": 1900
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 34.02,
|
| 2292 |
+
"learning_rate": 0.00013709207951022223,
|
| 2293 |
+
"loss": 0.2522,
|
| 2294 |
+
"step": 1905
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 34.11,
|
| 2298 |
+
"learning_rate": 0.00013335039645915423,
|
| 2299 |
+
"loss": 0.2533,
|
| 2300 |
+
"step": 1910
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 34.2,
|
| 2304 |
+
"learning_rate": 0.00012454747561060553,
|
| 2305 |
+
"loss": 0.2087,
|
| 2306 |
+
"step": 1915
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 34.29,
|
| 2310 |
+
"learning_rate": 0.00011137140040750936,
|
| 2311 |
+
"loss": 0.2236,
|
| 2312 |
+
"step": 1920
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 34.38,
|
| 2316 |
+
"learning_rate": 9.485208346024522e-05,
|
| 2317 |
+
"loss": 0.2418,
|
| 2318 |
+
"step": 1925
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 34.46,
|
| 2322 |
+
"learning_rate": 7.628076306068678e-05,
|
| 2323 |
+
"loss": 0.2321,
|
| 2324 |
+
"step": 1930
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 34.55,
|
| 2328 |
+
"learning_rate": 5.710907308129474e-05,
|
| 2329 |
+
"loss": 0.2204,
|
| 2330 |
+
"step": 1935
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 34.64,
|
| 2334 |
+
"learning_rate": 3.8835575496535006e-05,
|
| 2335 |
+
"loss": 0.253,
|
| 2336 |
+
"step": 1940
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 34.73,
|
| 2340 |
+
"learning_rate": 2.2888624752493407e-05,
|
| 2341 |
+
"loss": 0.1896,
|
| 2342 |
+
"step": 1945
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 34.82,
|
| 2346 |
+
"learning_rate": 1.051471993294022e-05,
|
| 2347 |
+
"loss": 0.2327,
|
| 2348 |
+
"step": 1950
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 34.91,
|
| 2352 |
+
"learning_rate": 2.6810717145663806e-06,
|
| 2353 |
+
"loss": 0.1881,
|
| 2354 |
+
"step": 1955
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 35.0,
|
| 2358 |
+
"learning_rate": 0.0,
|
| 2359 |
+
"loss": 0.2131,
|
| 2360 |
+
"step": 1960
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 35.09,
|
| 2364 |
+
"learning_rate": 2.6810717145660837e-06,
|
| 2365 |
+
"loss": 0.2035,
|
| 2366 |
+
"step": 1965
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 35.18,
|
| 2370 |
+
"learning_rate": 1.0514719932939649e-05,
|
| 2371 |
+
"loss": 0.1686,
|
| 2372 |
+
"step": 1970
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 35.27,
|
| 2376 |
+
"learning_rate": 2.28886247524926e-05,
|
| 2377 |
+
"loss": 0.1647,
|
| 2378 |
+
"step": 1975
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 35.36,
|
| 2382 |
+
"learning_rate": 3.883557549653492e-05,
|
| 2383 |
+
"loss": 0.1727,
|
| 2384 |
+
"step": 1980
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 35.45,
|
| 2388 |
+
"learning_rate": 5.710907308129464e-05,
|
| 2389 |
+
"loss": 0.1767,
|
| 2390 |
+
"step": 1985
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 35.54,
|
| 2394 |
+
"learning_rate": 7.62807630606867e-05,
|
| 2395 |
+
"loss": 0.1769,
|
| 2396 |
+
"step": 1990
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 35.62,
|
| 2400 |
+
"learning_rate": 9.485208346024514e-05,
|
| 2401 |
+
"loss": 0.1662,
|
| 2402 |
+
"step": 1995
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 35.71,
|
| 2406 |
+
"learning_rate": 0.00011137140040750926,
|
| 2407 |
+
"loss": 0.1953,
|
| 2408 |
+
"step": 2000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 35.8,
|
| 2412 |
+
"learning_rate": 0.00012454747561060548,
|
| 2413 |
+
"loss": 0.186,
|
| 2414 |
+
"step": 2005
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 35.89,
|
| 2418 |
+
"learning_rate": 0.0001333503964591542,
|
| 2419 |
+
"loss": 0.2162,
|
| 2420 |
+
"step": 2010
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 35.98,
|
| 2424 |
+
"learning_rate": 0.00013709207951022217,
|
| 2425 |
+
"loss": 0.215,
|
| 2426 |
+
"step": 2015
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 36.07,
|
| 2430 |
+
"learning_rate": 0.0001354800547756733,
|
| 2431 |
+
"loss": 0.2198,
|
| 2432 |
+
"step": 2020
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 36.16,
|
| 2436 |
+
"learning_rate": 0.00012864032674294085,
|
| 2437 |
+
"loss": 0.189,
|
| 2438 |
+
"step": 2025
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 36.25,
|
| 2442 |
+
"learning_rate": 0.00011710752518939751,
|
| 2443 |
+
"loss": 0.2029,
|
| 2444 |
+
"step": 2030
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 36.34,
|
| 2448 |
+
"learning_rate": 0.0001017831156554197,
|
| 2449 |
+
"loss": 0.1957,
|
| 2450 |
+
"step": 2035
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 36.43,
|
| 2454 |
+
"learning_rate": 8.386493606940322e-05,
|
| 2455 |
+
"loss": 0.2062,
|
| 2456 |
+
"step": 2040
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 36.52,
|
| 2460 |
+
"learning_rate": 6.47535673195285e-05,
|
| 2461 |
+
"loss": 0.2046,
|
| 2462 |
+
"step": 2045
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 36.61,
|
| 2466 |
+
"learning_rate": 4.594285634987612e-05,
|
| 2467 |
+
"loss": 0.1769,
|
| 2468 |
+
"step": 2050
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 36.7,
|
| 2472 |
+
"learning_rate": 2.8903149082065656e-05,
|
| 2473 |
+
"loss": 0.193,
|
| 2474 |
+
"step": 2055
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 36.79,
|
| 2478 |
+
"learning_rate": 1.496636030269327e-05,
|
| 2479 |
+
"loss": 0.1815,
|
| 2480 |
+
"step": 2060
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 36.88,
|
| 2484 |
+
"learning_rate": 5.2218640697257225e-06,
|
| 2485 |
+
"loss": 0.1812,
|
| 2486 |
+
"step": 2065
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 36.96,
|
| 2490 |
+
"learning_rate": 4.3134240132353453e-07,
|
| 2491 |
+
"loss": 0.177,
|
| 2492 |
+
"step": 2070
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 37.05,
|
| 2496 |
+
"learning_rate": 9.692481296490944e-07,
|
| 2497 |
+
"loss": 0.1544,
|
| 2498 |
+
"step": 2075
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 37.14,
|
| 2502 |
+
"learning_rate": 6.793535661894336e-06,
|
| 2503 |
+
"loss": 0.1606,
|
| 2504 |
+
"step": 2080
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 37.23,
|
| 2508 |
+
"learning_rate": 1.7448947489014123e-05,
|
| 2509 |
+
"loss": 0.1656,
|
| 2510 |
+
"step": 2085
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 37.32,
|
| 2514 |
+
"learning_rate": 3.210259955104718e-05,
|
| 2515 |
+
"loss": 0.1505,
|
| 2516 |
+
"step": 2090
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 37.41,
|
| 2520 |
+
"learning_rate": 4.9609083916254715e-05,
|
| 2521 |
+
"loss": 0.1481,
|
| 2522 |
+
"step": 2095
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 37.5,
|
| 2526 |
+
"learning_rate": 6.859999999999957e-05,
|
| 2527 |
+
"loss": 0.1534,
|
| 2528 |
+
"step": 2100
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 37.59,
|
| 2532 |
+
"learning_rate": 8.759091608374445e-05,
|
| 2533 |
+
"loss": 0.1499,
|
| 2534 |
+
"step": 2105
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 37.68,
|
| 2538 |
+
"learning_rate": 0.00010509740044895207,
|
| 2539 |
+
"loss": 0.1818,
|
| 2540 |
+
"step": 2110
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 37.77,
|
| 2544 |
+
"learning_rate": 0.0001197510525109853,
|
| 2545 |
+
"loss": 0.1553,
|
| 2546 |
+
"step": 2115
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 37.86,
|
| 2550 |
+
"learning_rate": 0.0001304064643381057,
|
| 2551 |
+
"loss": 0.1696,
|
| 2552 |
+
"step": 2120
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 37.95,
|
| 2556 |
+
"learning_rate": 0.0001362307518703509,
|
| 2557 |
+
"loss": 0.1848,
|
| 2558 |
+
"step": 2125
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 38.04,
|
| 2562 |
+
"learning_rate": 0.00013676865759867644,
|
| 2563 |
+
"loss": 0.1581,
|
| 2564 |
+
"step": 2130
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 38.12,
|
| 2568 |
+
"learning_rate": 0.00013197813593027424,
|
| 2569 |
+
"loss": 0.1639,
|
| 2570 |
+
"step": 2135
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 38.21,
|
| 2574 |
+
"learning_rate": 0.00012223363969730668,
|
| 2575 |
+
"loss": 0.1736,
|
| 2576 |
+
"step": 2140
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 38.3,
|
| 2580 |
+
"learning_rate": 0.00010829685091793427,
|
| 2581 |
+
"loss": 0.1716,
|
| 2582 |
+
"step": 2145
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 38.39,
|
| 2586 |
+
"learning_rate": 9.125714365012379e-05,
|
| 2587 |
+
"loss": 0.1616,
|
| 2588 |
+
"step": 2150
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 38.48,
|
| 2592 |
+
"learning_rate": 7.244643268047237e-05,
|
| 2593 |
+
"loss": 0.1674,
|
| 2594 |
+
"step": 2155
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 38.57,
|
| 2598 |
+
"learning_rate": 5.333506393059762e-05,
|
| 2599 |
+
"loss": 0.181,
|
| 2600 |
+
"step": 2160
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 38.66,
|
| 2604 |
+
"learning_rate": 3.541688434458107e-05,
|
| 2605 |
+
"loss": 0.1737,
|
| 2606 |
+
"step": 2165
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 38.75,
|
| 2610 |
+
"learning_rate": 2.009247481060311e-05,
|
| 2611 |
+
"loss": 0.1692,
|
| 2612 |
+
"step": 2170
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 38.84,
|
| 2616 |
+
"learning_rate": 8.559673257059581e-06,
|
| 2617 |
+
"loss": 0.1796,
|
| 2618 |
+
"step": 2175
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 38.93,
|
| 2622 |
+
"learning_rate": 1.719945224326892e-06,
|
| 2623 |
+
"loss": 0.149,
|
| 2624 |
+
"step": 2180
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 39.02,
|
| 2628 |
+
"learning_rate": 1.0792048977779616e-07,
|
| 2629 |
+
"loss": 0.1672,
|
| 2630 |
+
"step": 2185
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 39.11,
|
| 2634 |
+
"learning_rate": 3.849603540845817e-06,
|
| 2635 |
+
"loss": 0.1322,
|
| 2636 |
+
"step": 2190
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 39.2,
|
| 2640 |
+
"learning_rate": 1.2652524389394578e-05,
|
| 2641 |
+
"loss": 0.134,
|
| 2642 |
+
"step": 2195
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 39.29,
|
| 2646 |
+
"learning_rate": 2.5828599592490815e-05,
|
| 2647 |
+
"loss": 0.1382,
|
| 2648 |
+
"step": 2200
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 39.38,
|
| 2652 |
+
"learning_rate": 4.234791653975496e-05,
|
| 2653 |
+
"loss": 0.131,
|
| 2654 |
+
"step": 2205
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 39.46,
|
| 2658 |
+
"learning_rate": 6.0919236939313415e-05,
|
| 2659 |
+
"loss": 0.1396,
|
| 2660 |
+
"step": 2210
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 39.55,
|
| 2664 |
+
"learning_rate": 8.009092691870546e-05,
|
| 2665 |
+
"loss": 0.1475,
|
| 2666 |
+
"step": 2215
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 39.64,
|
| 2670 |
+
"learning_rate": 9.836442450346518e-05,
|
| 2671 |
+
"loss": 0.1374,
|
| 2672 |
+
"step": 2220
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 39.73,
|
| 2676 |
+
"learning_rate": 0.00011431137524750674,
|
| 2677 |
+
"loss": 0.1528,
|
| 2678 |
+
"step": 2225
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 39.82,
|
| 2682 |
+
"learning_rate": 0.0001266852800670599,
|
| 2683 |
+
"loss": 0.1468,
|
| 2684 |
+
"step": 2230
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 39.91,
|
| 2688 |
+
"learning_rate": 0.00013451892828543368,
|
| 2689 |
+
"loss": 0.1589,
|
| 2690 |
+
"step": 2235
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 40.0,
|
| 2694 |
+
"learning_rate": 0.0001372,
|
| 2695 |
+
"loss": 0.153,
|
| 2696 |
+
"step": 2240
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 40.09,
|
| 2700 |
+
"learning_rate": 0.00013451892828543387,
|
| 2701 |
+
"loss": 0.1516,
|
| 2702 |
+
"step": 2245
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 40.18,
|
| 2706 |
+
"learning_rate": 0.00012668528006706025,
|
| 2707 |
+
"loss": 0.1432,
|
| 2708 |
+
"step": 2250
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 40.27,
|
| 2712 |
+
"learning_rate": 0.00011431137524750726,
|
| 2713 |
+
"loss": 0.1714,
|
| 2714 |
+
"step": 2255
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 40.36,
|
| 2718 |
+
"learning_rate": 9.836442450346578e-05,
|
| 2719 |
+
"loss": 0.1401,
|
| 2720 |
+
"step": 2260
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 40.45,
|
| 2724 |
+
"learning_rate": 8.009092691870612e-05,
|
| 2725 |
+
"loss": 0.1529,
|
| 2726 |
+
"step": 2265
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 40.54,
|
| 2730 |
+
"learning_rate": 6.0919236939314086e-05,
|
| 2731 |
+
"loss": 0.1557,
|
| 2732 |
+
"step": 2270
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 40.62,
|
| 2736 |
+
"learning_rate": 4.234791653975558e-05,
|
| 2737 |
+
"loss": 0.1562,
|
| 2738 |
+
"step": 2275
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 40.71,
|
| 2742 |
+
"learning_rate": 2.582859959249134e-05,
|
| 2743 |
+
"loss": 0.1449,
|
| 2744 |
+
"step": 2280
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 40.8,
|
| 2748 |
+
"learning_rate": 1.2652524389394402e-05,
|
| 2749 |
+
"loss": 0.1595,
|
| 2750 |
+
"step": 2285
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 40.89,
|
| 2754 |
+
"learning_rate": 3.849603540846038e-06,
|
| 2755 |
+
"loss": 0.1353,
|
| 2756 |
+
"step": 2290
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 40.98,
|
| 2760 |
+
"learning_rate": 1.0792048977783424e-07,
|
| 2761 |
+
"loss": 0.1344,
|
| 2762 |
+
"step": 2295
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 41.07,
|
| 2766 |
+
"learning_rate": 1.7199452243265265e-06,
|
| 2767 |
+
"loss": 0.1224,
|
| 2768 |
+
"step": 2300
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 41.16,
|
| 2772 |
+
"learning_rate": 8.559673257059253e-06,
|
| 2773 |
+
"loss": 0.1157,
|
| 2774 |
+
"step": 2305
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 41.25,
|
| 2778 |
+
"learning_rate": 2.0092474810601945e-05,
|
| 2779 |
+
"loss": 0.1231,
|
| 2780 |
+
"step": 2310
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 41.34,
|
| 2784 |
+
"learning_rate": 3.541688434458049e-05,
|
| 2785 |
+
"loss": 0.1233,
|
| 2786 |
+
"step": 2315
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 41.43,
|
| 2790 |
+
"learning_rate": 5.3335063930596016e-05,
|
| 2791 |
+
"loss": 0.1129,
|
| 2792 |
+
"step": 2320
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 41.52,
|
| 2796 |
+
"learning_rate": 7.244643268047171e-05,
|
| 2797 |
+
"loss": 0.1297,
|
| 2798 |
+
"step": 2325
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 41.61,
|
| 2802 |
+
"learning_rate": 9.125714365012407e-05,
|
| 2803 |
+
"loss": 0.1117,
|
| 2804 |
+
"step": 2330
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 41.7,
|
| 2808 |
+
"learning_rate": 0.00010829685091793371,
|
| 2809 |
+
"loss": 0.117,
|
| 2810 |
+
"step": 2335
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 41.79,
|
| 2814 |
+
"learning_rate": 0.00012223363969730686,
|
| 2815 |
+
"loss": 0.1357,
|
| 2816 |
+
"step": 2340
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 41.88,
|
| 2820 |
+
"learning_rate": 0.00013197813593027397,
|
| 2821 |
+
"loss": 0.1191,
|
| 2822 |
+
"step": 2345
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 41.96,
|
| 2826 |
+
"learning_rate": 0.00013676865759867647,
|
| 2827 |
+
"loss": 0.1514,
|
| 2828 |
+
"step": 2350
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 42.05,
|
| 2832 |
+
"learning_rate": 0.00013623075187035104,
|
| 2833 |
+
"loss": 0.1221,
|
| 2834 |
+
"step": 2355
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 42.14,
|
| 2838 |
+
"learning_rate": 0.00013040646433810557,
|
| 2839 |
+
"loss": 0.1356,
|
| 2840 |
+
"step": 2360
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 42.23,
|
| 2844 |
+
"learning_rate": 0.0001197510525109864,
|
| 2845 |
+
"loss": 0.1469,
|
| 2846 |
+
"step": 2365
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 42.32,
|
| 2850 |
+
"learning_rate": 0.00010509740044895266,
|
| 2851 |
+
"loss": 0.1364,
|
| 2852 |
+
"step": 2370
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 42.41,
|
| 2856 |
+
"learning_rate": 8.759091608374603e-05,
|
| 2857 |
+
"loss": 0.146,
|
| 2858 |
+
"step": 2375
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 42.5,
|
| 2862 |
+
"learning_rate": 6.860000000000023e-05,
|
| 2863 |
+
"loss": 0.1439,
|
| 2864 |
+
"step": 2380
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 42.59,
|
| 2868 |
+
"learning_rate": 4.960908391625629e-05,
|
| 2869 |
+
"loss": 0.1339,
|
| 2870 |
+
"step": 2385
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 42.68,
|
| 2874 |
+
"learning_rate": 3.210259955104775e-05,
|
| 2875 |
+
"loss": 0.1282,
|
| 2876 |
+
"step": 2390
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 42.77,
|
| 2880 |
+
"learning_rate": 1.744894748901522e-05,
|
| 2881 |
+
"loss": 0.1341,
|
| 2882 |
+
"step": 2395
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 42.86,
|
| 2886 |
+
"learning_rate": 6.793535661894625e-06,
|
| 2887 |
+
"loss": 0.1195,
|
| 2888 |
+
"step": 2400
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 42.95,
|
| 2892 |
+
"learning_rate": 9.69248129649041e-07,
|
| 2893 |
+
"loss": 0.128,
|
| 2894 |
+
"step": 2405
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 43.04,
|
| 2898 |
+
"learning_rate": 4.3134240132345835e-07,
|
| 2899 |
+
"loss": 0.1227,
|
| 2900 |
+
"step": 2410
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 43.12,
|
| 2904 |
+
"learning_rate": 5.221864069725836e-06,
|
| 2905 |
+
"loss": 0.1054,
|
| 2906 |
+
"step": 2415
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 43.21,
|
| 2910 |
+
"learning_rate": 1.4966360302692852e-05,
|
| 2911 |
+
"loss": 0.113,
|
| 2912 |
+
"step": 2420
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 43.3,
|
| 2916 |
+
"learning_rate": 2.8903149082065897e-05,
|
| 2917 |
+
"loss": 0.1135,
|
| 2918 |
+
"step": 2425
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 43.39,
|
| 2922 |
+
"learning_rate": 4.594285634987549e-05,
|
| 2923 |
+
"loss": 0.0998,
|
| 2924 |
+
"step": 2430
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 43.48,
|
| 2928 |
+
"learning_rate": 6.475356731952782e-05,
|
| 2929 |
+
"loss": 0.1042,
|
| 2930 |
+
"step": 2435
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 43.57,
|
| 2934 |
+
"learning_rate": 8.386493606940162e-05,
|
| 2935 |
+
"loss": 0.1103,
|
| 2936 |
+
"step": 2440
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 43.66,
|
| 2940 |
+
"learning_rate": 0.0001017831156554191,
|
| 2941 |
+
"loss": 0.0937,
|
| 2942 |
+
"step": 2445
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 43.75,
|
| 2946 |
+
"learning_rate": 0.00011710752518939633,
|
| 2947 |
+
"loss": 0.1037,
|
| 2948 |
+
"step": 2450
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 43.84,
|
| 2952 |
+
"learning_rate": 0.0001286403267429405,
|
| 2953 |
+
"loss": 0.1108,
|
| 2954 |
+
"step": 2455
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 43.93,
|
| 2958 |
+
"learning_rate": 0.00013548005477567295,
|
| 2959 |
+
"loss": 0.1088,
|
| 2960 |
+
"step": 2460
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 44.02,
|
| 2964 |
+
"learning_rate": 0.0001370920795102222,
|
| 2965 |
+
"loss": 0.132,
|
| 2966 |
+
"step": 2465
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 44.11,
|
| 2970 |
+
"learning_rate": 0.00013335039645915412,
|
| 2971 |
+
"loss": 0.1115,
|
| 2972 |
+
"step": 2470
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 44.2,
|
| 2976 |
+
"learning_rate": 0.00012454747561060588,
|
| 2977 |
+
"loss": 0.1322,
|
| 2978 |
+
"step": 2475
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 44.29,
|
| 2982 |
+
"learning_rate": 0.00011137140040750903,
|
| 2983 |
+
"loss": 0.1218,
|
| 2984 |
+
"step": 2480
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 44.38,
|
| 2988 |
+
"learning_rate": 9.485208346024576e-05,
|
| 2989 |
+
"loss": 0.1266,
|
| 2990 |
+
"step": 2485
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 44.46,
|
| 2994 |
+
"learning_rate": 7.628076306068638e-05,
|
| 2995 |
+
"loss": 0.1337,
|
| 2996 |
+
"step": 2490
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 44.55,
|
| 3000 |
+
"learning_rate": 5.7109073081295306e-05,
|
| 3001 |
+
"loss": 0.1423,
|
| 3002 |
+
"step": 2495
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 44.64,
|
| 3006 |
+
"learning_rate": 3.883557549653465e-05,
|
| 3007 |
+
"loss": 0.1254,
|
| 3008 |
+
"step": 2500
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 44.73,
|
| 3012 |
+
"learning_rate": 2.2888624752493833e-05,
|
| 3013 |
+
"loss": 0.1102,
|
| 3014 |
+
"step": 2505
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 44.82,
|
| 3018 |
+
"learning_rate": 1.0514719932940006e-05,
|
| 3019 |
+
"loss": 0.1057,
|
| 3020 |
+
"step": 2510
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 44.91,
|
| 3024 |
+
"learning_rate": 2.6810717145665407e-06,
|
| 3025 |
+
"loss": 0.1127,
|
| 3026 |
+
"step": 2515
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 45.0,
|
| 3030 |
+
"learning_rate": 0.0,
|
| 3031 |
+
"loss": 0.1,
|
| 3032 |
+
"step": 2520
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 45.09,
|
| 3036 |
+
"learning_rate": 2.6810717145659236e-06,
|
| 3037 |
+
"loss": 0.0923,
|
| 3038 |
+
"step": 2525
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 45.18,
|
| 3042 |
+
"learning_rate": 1.0514719932939862e-05,
|
| 3043 |
+
"loss": 0.1096,
|
| 3044 |
+
"step": 2530
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 45.27,
|
| 3048 |
+
"learning_rate": 2.288862475249218e-05,
|
| 3049 |
+
"loss": 0.1037,
|
| 3050 |
+
"step": 2535
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 45.36,
|
| 3054 |
+
"learning_rate": 3.88355754965344e-05,
|
| 3055 |
+
"loss": 0.0967,
|
| 3056 |
+
"step": 2540
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 45.45,
|
| 3060 |
+
"learning_rate": 5.7109073081295035e-05,
|
| 3061 |
+
"loss": 0.0873,
|
| 3062 |
+
"step": 2545
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 45.54,
|
| 3066 |
+
"learning_rate": 7.628076306068611e-05,
|
| 3067 |
+
"loss": 0.1035,
|
| 3068 |
+
"step": 2550
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 45.62,
|
| 3072 |
+
"learning_rate": 9.485208346024552e-05,
|
| 3073 |
+
"loss": 0.0923,
|
| 3074 |
+
"step": 2555
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 45.71,
|
| 3078 |
+
"learning_rate": 0.00011137140040750883,
|
| 3079 |
+
"loss": 0.0814,
|
| 3080 |
+
"step": 2560
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 45.8,
|
| 3084 |
+
"learning_rate": 0.0001245474756106057,
|
| 3085 |
+
"loss": 0.0889,
|
| 3086 |
+
"step": 2565
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 45.89,
|
| 3090 |
+
"learning_rate": 0.00013335039645915404,
|
| 3091 |
+
"loss": 0.1126,
|
| 3092 |
+
"step": 2570
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 45.98,
|
| 3096 |
+
"learning_rate": 0.00013709207951022217,
|
| 3097 |
+
"loss": 0.1126,
|
| 3098 |
+
"step": 2575
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 46.07,
|
| 3102 |
+
"learning_rate": 0.00013548005477567344,
|
| 3103 |
+
"loss": 0.115,
|
| 3104 |
+
"step": 2580
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 46.16,
|
| 3108 |
+
"learning_rate": 0.00012864032674294066,
|
| 3109 |
+
"loss": 0.1107,
|
| 3110 |
+
"step": 2585
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 46.25,
|
| 3114 |
+
"learning_rate": 0.00011710752518939791,
|
| 3115 |
+
"loss": 0.1185,
|
| 3116 |
+
"step": 2590
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 46.34,
|
| 3120 |
+
"learning_rate": 0.00010178311565541935,
|
| 3121 |
+
"loss": 0.0988,
|
| 3122 |
+
"step": 2595
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 46.43,
|
| 3126 |
+
"learning_rate": 8.386493606940379e-05,
|
| 3127 |
+
"loss": 0.106,
|
| 3128 |
+
"step": 2600
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 46.52,
|
| 3132 |
+
"learning_rate": 6.475356731952809e-05,
|
| 3133 |
+
"loss": 0.1046,
|
| 3134 |
+
"step": 2605
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 46.61,
|
| 3138 |
+
"learning_rate": 4.594285634987574e-05,
|
| 3139 |
+
"loss": 0.1119,
|
| 3140 |
+
"step": 2610
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 46.7,
|
| 3144 |
+
"learning_rate": 2.890314908206612e-05,
|
| 3145 |
+
"loss": 0.1081,
|
| 3146 |
+
"step": 2615
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 46.79,
|
| 3150 |
+
"learning_rate": 1.496636030269302e-05,
|
| 3151 |
+
"loss": 0.0968,
|
| 3152 |
+
"step": 2620
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 46.88,
|
| 3156 |
+
"learning_rate": 5.221864069725935e-06,
|
| 3157 |
+
"loss": 0.1116,
|
| 3158 |
+
"step": 2625
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 46.96,
|
| 3162 |
+
"learning_rate": 4.3134240132348884e-07,
|
| 3163 |
+
"loss": 0.1083,
|
| 3164 |
+
"step": 2630
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 47.05,
|
| 3168 |
+
"learning_rate": 9.692481296489953e-07,
|
| 3169 |
+
"loss": 0.0932,
|
| 3170 |
+
"step": 2635
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 47.14,
|
| 3174 |
+
"learning_rate": 6.793535661894512e-06,
|
| 3175 |
+
"loss": 0.0854,
|
| 3176 |
+
"step": 2640
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 47.23,
|
| 3180 |
+
"learning_rate": 1.7448947489013744e-05,
|
| 3181 |
+
"loss": 0.0843,
|
| 3182 |
+
"step": 2645
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 47.32,
|
| 3186 |
+
"learning_rate": 3.210259955104752e-05,
|
| 3187 |
+
"loss": 0.0891,
|
| 3188 |
+
"step": 2650
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 47.41,
|
| 3192 |
+
"learning_rate": 4.960908391625416e-05,
|
| 3193 |
+
"loss": 0.0791,
|
| 3194 |
+
"step": 2655
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 47.5,
|
| 3198 |
+
"learning_rate": 6.859999999999997e-05,
|
| 3199 |
+
"loss": 0.09,
|
| 3200 |
+
"step": 2660
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 47.59,
|
| 3204 |
+
"learning_rate": 8.75909160837439e-05,
|
| 3205 |
+
"loss": 0.0897,
|
| 3206 |
+
"step": 2665
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 47.68,
|
| 3210 |
+
"learning_rate": 0.00010509740044895241,
|
| 3211 |
+
"loss": 0.0856,
|
| 3212 |
+
"step": 2670
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 47.77,
|
| 3216 |
+
"learning_rate": 0.00011975105251098491,
|
| 3217 |
+
"loss": 0.0932,
|
| 3218 |
+
"step": 2675
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 47.86,
|
| 3222 |
+
"learning_rate": 0.00013040646433810547,
|
| 3223 |
+
"loss": 0.0998,
|
| 3224 |
+
"step": 2680
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 47.95,
|
| 3228 |
+
"learning_rate": 0.000136230751870351,
|
| 3229 |
+
"loss": 0.1009,
|
| 3230 |
+
"step": 2685
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 48.04,
|
| 3234 |
+
"learning_rate": 0.00013676865759867652,
|
| 3235 |
+
"loss": 0.0975,
|
| 3236 |
+
"step": 2690
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 48.12,
|
| 3240 |
+
"learning_rate": 0.00013197813593027408,
|
| 3241 |
+
"loss": 0.1072,
|
| 3242 |
+
"step": 2695
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 48.21,
|
| 3246 |
+
"learning_rate": 0.00012223363969730703,
|
| 3247 |
+
"loss": 0.105,
|
| 3248 |
+
"step": 2700
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 48.3,
|
| 3252 |
+
"learning_rate": 0.00010829685091793393,
|
| 3253 |
+
"loss": 0.102,
|
| 3254 |
+
"step": 2705
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 48.39,
|
| 3258 |
+
"learning_rate": 9.125714365012433e-05,
|
| 3259 |
+
"loss": 0.0928,
|
| 3260 |
+
"step": 2710
|
| 3261 |
+
},
|
| 3262 |
+
{
|
| 3263 |
+
"epoch": 48.48,
|
| 3264 |
+
"learning_rate": 7.244643268047198e-05,
|
| 3265 |
+
"loss": 0.117,
|
| 3266 |
+
"step": 2715
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 48.57,
|
| 3270 |
+
"learning_rate": 5.333506393059818e-05,
|
| 3271 |
+
"loss": 0.0985,
|
| 3272 |
+
"step": 2720
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 48.66,
|
| 3276 |
+
"learning_rate": 3.541688434458072e-05,
|
| 3277 |
+
"loss": 0.1006,
|
| 3278 |
+
"step": 2725
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 48.75,
|
| 3282 |
+
"learning_rate": 2.0092474810603514e-05,
|
| 3283 |
+
"loss": 0.0912,
|
| 3284 |
+
"step": 2730
|
| 3285 |
+
},
|
| 3286 |
+
{
|
| 3287 |
+
"epoch": 48.84,
|
| 3288 |
+
"learning_rate": 8.559673257059383e-06,
|
| 3289 |
+
"loss": 0.0824,
|
| 3290 |
+
"step": 2735
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 48.93,
|
| 3294 |
+
"learning_rate": 1.7199452243270216e-06,
|
| 3295 |
+
"loss": 0.0914,
|
| 3296 |
+
"step": 2740
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 49.02,
|
| 3300 |
+
"learning_rate": 1.07920489777819e-07,
|
| 3301 |
+
"loss": 0.103,
|
| 3302 |
+
"step": 2745
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 49.11,
|
| 3306 |
+
"learning_rate": 3.849603540845946e-06,
|
| 3307 |
+
"loss": 0.0818,
|
| 3308 |
+
"step": 2750
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 49.2,
|
| 3312 |
+
"learning_rate": 1.265252438939425e-05,
|
| 3313 |
+
"loss": 0.083,
|
| 3314 |
+
"step": 2755
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 49.29,
|
| 3318 |
+
"learning_rate": 2.5828599592491126e-05,
|
| 3319 |
+
"loss": 0.0802,
|
| 3320 |
+
"step": 2760
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 49.38,
|
| 3324 |
+
"learning_rate": 4.2347916539754424e-05,
|
| 3325 |
+
"loss": 0.0776,
|
| 3326 |
+
"step": 2765
|
| 3327 |
+
},
|
| 3328 |
+
{
|
| 3329 |
+
"epoch": 49.46,
|
| 3330 |
+
"learning_rate": 6.091923693931382e-05,
|
| 3331 |
+
"loss": 0.0897,
|
| 3332 |
+
"step": 2770
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 49.55,
|
| 3336 |
+
"learning_rate": 8.00909269187049e-05,
|
| 3337 |
+
"loss": 0.0723,
|
| 3338 |
+
"step": 2775
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 49.64,
|
| 3342 |
+
"learning_rate": 9.836442450346554e-05,
|
| 3343 |
+
"loss": 0.0913,
|
| 3344 |
+
"step": 2780
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 49.73,
|
| 3348 |
+
"learning_rate": 0.00011431137524750631,
|
| 3349 |
+
"loss": 0.0869,
|
| 3350 |
+
"step": 2785
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 49.82,
|
| 3354 |
+
"learning_rate": 0.00012668528006706012,
|
| 3355 |
+
"loss": 0.0839,
|
| 3356 |
+
"step": 2790
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 49.91,
|
| 3360 |
+
"learning_rate": 0.00013451892828543352,
|
| 3361 |
+
"loss": 0.0867,
|
| 3362 |
+
"step": 2795
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 50.0,
|
| 3366 |
+
"learning_rate": 0.0001372,
|
| 3367 |
+
"loss": 0.0961,
|
| 3368 |
+
"step": 2800
|
| 3369 |
+
},
|
| 3370 |
+
{
|
| 3371 |
+
"epoch": 50.09,
|
| 3372 |
+
"learning_rate": 0.00013451892828543404,
|
| 3373 |
+
"loss": 0.0873,
|
| 3374 |
+
"step": 2805
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 50.18,
|
| 3378 |
+
"learning_rate": 0.00012668528006706004,
|
| 3379 |
+
"loss": 0.0834,
|
| 3380 |
+
"step": 2810
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 50.27,
|
| 3384 |
+
"learning_rate": 0.00011431137524750768,
|
| 3385 |
+
"loss": 0.0971,
|
| 3386 |
+
"step": 2815
|
| 3387 |
+
},
|
| 3388 |
+
{
|
| 3389 |
+
"epoch": 50.36,
|
| 3390 |
+
"learning_rate": 9.836442450346542e-05,
|
| 3391 |
+
"loss": 0.0968,
|
| 3392 |
+
"step": 2820
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 50.45,
|
| 3396 |
+
"learning_rate": 8.009092691870668e-05,
|
| 3397 |
+
"loss": 0.0985,
|
| 3398 |
+
"step": 2825
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 50.54,
|
| 3402 |
+
"learning_rate": 6.0919236939313686e-05,
|
| 3403 |
+
"loss": 0.092,
|
| 3404 |
+
"step": 2830
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 50.62,
|
| 3408 |
+
"learning_rate": 4.2347916539756105e-05,
|
| 3409 |
+
"loss": 0.0898,
|
| 3410 |
+
"step": 2835
|
| 3411 |
+
},
|
| 3412 |
+
{
|
| 3413 |
+
"epoch": 50.71,
|
| 3414 |
+
"learning_rate": 2.582859959249102e-05,
|
| 3415 |
+
"loss": 0.0918,
|
| 3416 |
+
"step": 2840
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 50.8,
|
| 3420 |
+
"learning_rate": 1.2652524389395302e-05,
|
| 3421 |
+
"loss": 0.0936,
|
| 3422 |
+
"step": 2845
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 50.89,
|
| 3426 |
+
"learning_rate": 3.849603540845908e-06,
|
| 3427 |
+
"loss": 0.0775,
|
| 3428 |
+
"step": 2850
|
| 3429 |
+
},
|
| 3430 |
+
{
|
| 3431 |
+
"epoch": 50.98,
|
| 3432 |
+
"learning_rate": 1.0792048977781139e-07,
|
| 3433 |
+
"loss": 0.086,
|
| 3434 |
+
"step": 2855
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 51.07,
|
| 3438 |
+
"learning_rate": 1.719945224326618e-06,
|
| 3439 |
+
"loss": 0.071,
|
| 3440 |
+
"step": 2860
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 51.16,
|
| 3444 |
+
"learning_rate": 8.559673257059451e-06,
|
| 3445 |
+
"loss": 0.0789,
|
| 3446 |
+
"step": 2865
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 51.25,
|
| 3450 |
+
"learning_rate": 2.0092474810602236e-05,
|
| 3451 |
+
"loss": 0.0844,
|
| 3452 |
+
"step": 2870
|
| 3453 |
+
},
|
| 3454 |
+
{
|
| 3455 |
+
"epoch": 51.34,
|
| 3456 |
+
"learning_rate": 3.541688434458084e-05,
|
| 3457 |
+
"loss": 0.0717,
|
| 3458 |
+
"step": 2875
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 51.43,
|
| 3462 |
+
"learning_rate": 5.333506393059641e-05,
|
| 3463 |
+
"loss": 0.0714,
|
| 3464 |
+
"step": 2880
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 51.52,
|
| 3468 |
+
"learning_rate": 7.244643268047212e-05,
|
| 3469 |
+
"loss": 0.0735,
|
| 3470 |
+
"step": 2885
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 51.61,
|
| 3474 |
+
"learning_rate": 9.125714365012261e-05,
|
| 3475 |
+
"loss": 0.0773,
|
| 3476 |
+
"step": 2890
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 51.7,
|
| 3480 |
+
"learning_rate": 0.00010829685091793405,
|
| 3481 |
+
"loss": 0.0718,
|
| 3482 |
+
"step": 2895
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 51.79,
|
| 3486 |
+
"learning_rate": 0.0001222336396973059,
|
| 3487 |
+
"loss": 0.0782,
|
| 3488 |
+
"step": 2900
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 51.88,
|
| 3492 |
+
"learning_rate": 0.00013197813593027413,
|
| 3493 |
+
"loss": 0.0685,
|
| 3494 |
+
"step": 2905
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 51.96,
|
| 3498 |
+
"learning_rate": 0.00013676865759867633,
|
| 3499 |
+
"loss": 0.0792,
|
| 3500 |
+
"step": 2910
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 52.05,
|
| 3504 |
+
"learning_rate": 0.000136230751870351,
|
| 3505 |
+
"loss": 0.076,
|
| 3506 |
+
"step": 2915
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 52.14,
|
| 3510 |
+
"learning_rate": 0.00013040646433810625,
|
| 3511 |
+
"loss": 0.0802,
|
| 3512 |
+
"step": 2920
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 52.23,
|
| 3516 |
+
"learning_rate": 0.00011975105251098613,
|
| 3517 |
+
"loss": 0.0753,
|
| 3518 |
+
"step": 2925
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 52.32,
|
| 3522 |
+
"learning_rate": 0.0001050974004489523,
|
| 3523 |
+
"loss": 0.0935,
|
| 3524 |
+
"step": 2930
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 52.41,
|
| 3528 |
+
"learning_rate": 8.759091608374565e-05,
|
| 3529 |
+
"loss": 0.0924,
|
| 3530 |
+
"step": 2935
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 52.5,
|
| 3534 |
+
"learning_rate": 6.859999999999984e-05,
|
| 3535 |
+
"loss": 0.0846,
|
| 3536 |
+
"step": 2940
|
| 3537 |
+
},
|
| 3538 |
+
{
|
| 3539 |
+
"epoch": 52.59,
|
| 3540 |
+
"learning_rate": 4.96090839162559e-05,
|
| 3541 |
+
"loss": 0.0729,
|
| 3542 |
+
"step": 2945
|
| 3543 |
+
},
|
| 3544 |
+
{
|
| 3545 |
+
"epoch": 52.68,
|
| 3546 |
+
"learning_rate": 3.210259955104741e-05,
|
| 3547 |
+
"loss": 0.0807,
|
| 3548 |
+
"step": 2950
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 52.77,
|
| 3552 |
+
"learning_rate": 1.7448947489014953e-05,
|
| 3553 |
+
"loss": 0.0955,
|
| 3554 |
+
"step": 2955
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 52.86,
|
| 3558 |
+
"learning_rate": 6.793535661894451e-06,
|
| 3559 |
+
"loss": 0.0765,
|
| 3560 |
+
"step": 2960
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 52.95,
|
| 3564 |
+
"learning_rate": 9.692481296493e-07,
|
| 3565 |
+
"loss": 0.0808,
|
| 3566 |
+
"step": 2965
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 53.04,
|
| 3570 |
+
"learning_rate": 4.3134240132350404e-07,
|
| 3571 |
+
"loss": 0.0749,
|
| 3572 |
+
"step": 2970
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 53.12,
|
| 3576 |
+
"learning_rate": 5.221864069725242e-06,
|
| 3577 |
+
"loss": 0.0756,
|
| 3578 |
+
"step": 2975
|
| 3579 |
+
},
|
| 3580 |
+
{
|
| 3581 |
+
"epoch": 53.21,
|
| 3582 |
+
"learning_rate": 1.4966360302693102e-05,
|
| 3583 |
+
"loss": 0.0691,
|
| 3584 |
+
"step": 2980
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 53.3,
|
| 3588 |
+
"learning_rate": 2.8903149082064643e-05,
|
| 3589 |
+
"loss": 0.0656,
|
| 3590 |
+
"step": 2985
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 53.39,
|
| 3594 |
+
"learning_rate": 4.5942856349875866e-05,
|
| 3595 |
+
"loss": 0.0701,
|
| 3596 |
+
"step": 2990
|
| 3597 |
+
},
|
| 3598 |
+
{
|
| 3599 |
+
"epoch": 53.48,
|
| 3600 |
+
"learning_rate": 6.475356731952822e-05,
|
| 3601 |
+
"loss": 0.0626,
|
| 3602 |
+
"step": 2995
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 53.57,
|
| 3606 |
+
"learning_rate": 8.386493606940203e-05,
|
| 3607 |
+
"loss": 0.066,
|
| 3608 |
+
"step": 3000
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 53.66,
|
| 3612 |
+
"learning_rate": 0.00010178311565541946,
|
| 3613 |
+
"loss": 0.0758,
|
| 3614 |
+
"step": 3005
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 53.75,
|
| 3618 |
+
"learning_rate": 0.00011710752518939663,
|
| 3619 |
+
"loss": 0.0757,
|
| 3620 |
+
"step": 3010
|
| 3621 |
+
},
|
| 3622 |
+
{
|
| 3623 |
+
"epoch": 53.84,
|
| 3624 |
+
"learning_rate": 0.00012864032674294072,
|
| 3625 |
+
"loss": 0.0673,
|
| 3626 |
+
"step": 3015
|
| 3627 |
+
},
|
| 3628 |
+
{
|
| 3629 |
+
"epoch": 53.93,
|
| 3630 |
+
"learning_rate": 0.00013548005477567304,
|
| 3631 |
+
"loss": 0.0775,
|
| 3632 |
+
"step": 3020
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 54.02,
|
| 3636 |
+
"learning_rate": 0.00013709207951022217,
|
| 3637 |
+
"loss": 0.0768,
|
| 3638 |
+
"step": 3025
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 54.11,
|
| 3642 |
+
"learning_rate": 0.00013335039645915464,
|
| 3643 |
+
"loss": 0.0761,
|
| 3644 |
+
"step": 3030
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 54.2,
|
| 3648 |
+
"learning_rate": 0.00012454747561060564,
|
| 3649 |
+
"loss": 0.0757,
|
| 3650 |
+
"step": 3035
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 54.29,
|
| 3654 |
+
"learning_rate": 0.00011137140040751024,
|
| 3655 |
+
"loss": 0.0755,
|
| 3656 |
+
"step": 3040
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 54.38,
|
| 3660 |
+
"learning_rate": 9.48520834602454e-05,
|
| 3661 |
+
"loss": 0.0803,
|
| 3662 |
+
"step": 3045
|
| 3663 |
+
},
|
| 3664 |
+
{
|
| 3665 |
+
"epoch": 54.46,
|
| 3666 |
+
"learning_rate": 7.628076306068793e-05,
|
| 3667 |
+
"loss": 0.0809,
|
| 3668 |
+
"step": 3050
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 54.55,
|
| 3672 |
+
"learning_rate": 5.71090730812949e-05,
|
| 3673 |
+
"loss": 0.0821,
|
| 3674 |
+
"step": 3055
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 54.64,
|
| 3678 |
+
"learning_rate": 3.8835575496536036e-05,
|
| 3679 |
+
"loss": 0.0792,
|
| 3680 |
+
"step": 3060
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 54.73,
|
| 3684 |
+
"learning_rate": 2.288862475249353e-05,
|
| 3685 |
+
"loss": 0.073,
|
| 3686 |
+
"step": 3065
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 54.82,
|
| 3690 |
+
"learning_rate": 1.0514719932939786e-05,
|
| 3691 |
+
"loss": 0.0754,
|
| 3692 |
+
"step": 3070
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 54.91,
|
| 3696 |
+
"learning_rate": 2.6810717145664263e-06,
|
| 3697 |
+
"loss": 0.0837,
|
| 3698 |
+
"step": 3075
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 55.0,
|
| 3702 |
+
"learning_rate": 0.0,
|
| 3703 |
+
"loss": 0.0772,
|
| 3704 |
+
"step": 3080
|
| 3705 |
+
},
|
| 3706 |
+
{
|
| 3707 |
+
"epoch": 55.09,
|
| 3708 |
+
"learning_rate": 2.681071714566038e-06,
|
| 3709 |
+
"loss": 0.0661,
|
| 3710 |
+
"step": 3085
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 55.18,
|
| 3714 |
+
"learning_rate": 1.0514719932940075e-05,
|
| 3715 |
+
"loss": 0.0746,
|
| 3716 |
+
"step": 3090
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 55.27,
|
| 3720 |
+
"learning_rate": 2.2888624752492478e-05,
|
| 3721 |
+
"loss": 0.0666,
|
| 3722 |
+
"step": 3095
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 55.36,
|
| 3726 |
+
"learning_rate": 3.883557549653476e-05,
|
| 3727 |
+
"loss": 0.0662,
|
| 3728 |
+
"step": 3100
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 55.45,
|
| 3732 |
+
"learning_rate": 5.710907308129351e-05,
|
| 3733 |
+
"loss": 0.0673,
|
| 3734 |
+
"step": 3105
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 55.54,
|
| 3738 |
+
"learning_rate": 7.628076306068652e-05,
|
| 3739 |
+
"loss": 0.0648,
|
| 3740 |
+
"step": 3110
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 55.62,
|
| 3744 |
+
"learning_rate": 9.485208346024408e-05,
|
| 3745 |
+
"loss": 0.0616,
|
| 3746 |
+
"step": 3115
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 55.71,
|
| 3750 |
+
"learning_rate": 0.00011137140040750914,
|
| 3751 |
+
"loss": 0.066,
|
| 3752 |
+
"step": 3120
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 55.8,
|
| 3756 |
+
"learning_rate": 0.00012454747561060483,
|
| 3757 |
+
"loss": 0.0587,
|
| 3758 |
+
"step": 3125
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 55.89,
|
| 3762 |
+
"learning_rate": 0.00013335039645915415,
|
| 3763 |
+
"loss": 0.0635,
|
| 3764 |
+
"step": 3130
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 55.98,
|
| 3768 |
+
"learning_rate": 0.0001370920795102222,
|
| 3769 |
+
"loss": 0.0683,
|
| 3770 |
+
"step": 3135
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 56.07,
|
| 3774 |
+
"learning_rate": 0.00013548005477567333,
|
| 3775 |
+
"loss": 0.0734,
|
| 3776 |
+
"step": 3140
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 56.16,
|
| 3780 |
+
"learning_rate": 0.00012864032674294044,
|
| 3781 |
+
"loss": 0.0734,
|
| 3782 |
+
"step": 3145
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 56.25,
|
| 3786 |
+
"learning_rate": 0.00011710752518939762,
|
| 3787 |
+
"loss": 0.0714,
|
| 3788 |
+
"step": 3150
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 56.34,
|
| 3792 |
+
"learning_rate": 0.00010178311565541898,
|
| 3793 |
+
"loss": 0.0763,
|
| 3794 |
+
"step": 3155
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 56.43,
|
| 3798 |
+
"learning_rate": 8.38649360694034e-05,
|
| 3799 |
+
"loss": 0.076,
|
| 3800 |
+
"step": 3160
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 56.52,
|
| 3804 |
+
"learning_rate": 6.47535673195277e-05,
|
| 3805 |
+
"loss": 0.065,
|
| 3806 |
+
"step": 3165
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 56.61,
|
| 3810 |
+
"learning_rate": 4.594285634987719e-05,
|
| 3811 |
+
"loss": 0.0772,
|
| 3812 |
+
"step": 3170
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 56.7,
|
| 3816 |
+
"learning_rate": 2.8903149082065792e-05,
|
| 3817 |
+
"loss": 0.0719,
|
| 3818 |
+
"step": 3175
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 56.79,
|
| 3822 |
+
"learning_rate": 1.4966360302693987e-05,
|
| 3823 |
+
"loss": 0.0661,
|
| 3824 |
+
"step": 3180
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 56.88,
|
| 3828 |
+
"learning_rate": 5.2218640697257835e-06,
|
| 3829 |
+
"loss": 0.0754,
|
| 3830 |
+
"step": 3185
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 56.96,
|
| 3834 |
+
"learning_rate": 4.313424013236564e-07,
|
| 3835 |
+
"loss": 0.0747,
|
| 3836 |
+
"step": 3190
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 57.05,
|
| 3840 |
+
"learning_rate": 9.69248129649064e-07,
|
| 3841 |
+
"loss": 0.067,
|
| 3842 |
+
"step": 3195
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 57.14,
|
| 3846 |
+
"learning_rate": 6.793535661893841e-06,
|
| 3847 |
+
"loss": 0.0616,
|
| 3848 |
+
"step": 3200
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 57.23,
|
| 3852 |
+
"learning_rate": 1.744894748901401e-05,
|
| 3853 |
+
"loss": 0.0606,
|
| 3854 |
+
"step": 3205
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 57.32,
|
| 3858 |
+
"learning_rate": 3.2102599551047865e-05,
|
| 3859 |
+
"loss": 0.0688,
|
| 3860 |
+
"step": 3210
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 57.41,
|
| 3864 |
+
"learning_rate": 4.9609083916254546e-05,
|
| 3865 |
+
"loss": 0.0625,
|
| 3866 |
+
"step": 3215
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 57.5,
|
| 3870 |
+
"learning_rate": 6.860000000000037e-05,
|
| 3871 |
+
"loss": 0.0574,
|
| 3872 |
+
"step": 3220
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 57.59,
|
| 3876 |
+
"learning_rate": 8.75909160837443e-05,
|
| 3877 |
+
"loss": 0.0555,
|
| 3878 |
+
"step": 3225
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 57.68,
|
| 3882 |
+
"learning_rate": 0.00010509740044895277,
|
| 3883 |
+
"loss": 0.0597,
|
| 3884 |
+
"step": 3230
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 57.77,
|
| 3888 |
+
"learning_rate": 0.00011975105251098518,
|
| 3889 |
+
"loss": 0.0566,
|
| 3890 |
+
"step": 3235
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 57.86,
|
| 3894 |
+
"learning_rate": 0.00013040646433810563,
|
| 3895 |
+
"loss": 0.0627,
|
| 3896 |
+
"step": 3240
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 57.95,
|
| 3900 |
+
"learning_rate": 0.00013623075187035074,
|
| 3901 |
+
"loss": 0.0736,
|
| 3902 |
+
"step": 3245
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 58.04,
|
| 3906 |
+
"learning_rate": 0.00013676865759867647,
|
| 3907 |
+
"loss": 0.07,
|
| 3908 |
+
"step": 3250
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 58.12,
|
| 3912 |
+
"learning_rate": 0.00013197813593027467,
|
| 3913 |
+
"loss": 0.0627,
|
| 3914 |
+
"step": 3255
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 58.21,
|
| 3918 |
+
"learning_rate": 0.00012223363969730676,
|
| 3919 |
+
"loss": 0.0709,
|
| 3920 |
+
"step": 3260
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 58.3,
|
| 3924 |
+
"learning_rate": 0.0001082968509179352,
|
| 3925 |
+
"loss": 0.0699,
|
| 3926 |
+
"step": 3265
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 58.39,
|
| 3930 |
+
"learning_rate": 9.125714365012395e-05,
|
| 3931 |
+
"loss": 0.0791,
|
| 3932 |
+
"step": 3270
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 58.48,
|
| 3936 |
+
"learning_rate": 7.244643268047157e-05,
|
| 3937 |
+
"loss": 0.0732,
|
| 3938 |
+
"step": 3275
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 58.57,
|
| 3942 |
+
"learning_rate": 5.333506393059779e-05,
|
| 3943 |
+
"loss": 0.0646,
|
| 3944 |
+
"step": 3280
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 58.66,
|
| 3948 |
+
"learning_rate": 3.541688434458037e-05,
|
| 3949 |
+
"loss": 0.0675,
|
| 3950 |
+
"step": 3285
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 58.75,
|
| 3954 |
+
"learning_rate": 2.0092474810603233e-05,
|
| 3955 |
+
"loss": 0.0586,
|
| 3956 |
+
"step": 3290
|
| 3957 |
+
},
|
| 3958 |
+
{
|
| 3959 |
+
"epoch": 58.84,
|
| 3960 |
+
"learning_rate": 8.559673257059193e-06,
|
| 3961 |
+
"loss": 0.0643,
|
| 3962 |
+
"step": 3295
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 58.93,
|
| 3966 |
+
"learning_rate": 1.7199452243269301e-06,
|
| 3967 |
+
"loss": 0.0693,
|
| 3968 |
+
"step": 3300
|
| 3969 |
+
},
|
| 3970 |
+
{
|
| 3971 |
+
"epoch": 59.02,
|
| 3972 |
+
"learning_rate": 1.0792048977784186e-07,
|
| 3973 |
+
"loss": 0.0652,
|
| 3974 |
+
"step": 3305
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 59.11,
|
| 3978 |
+
"learning_rate": 3.849603540845436e-06,
|
| 3979 |
+
"loss": 0.058,
|
| 3980 |
+
"step": 3310
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 59.2,
|
| 3984 |
+
"learning_rate": 1.2652524389394478e-05,
|
| 3985 |
+
"loss": 0.0575,
|
| 3986 |
+
"step": 3315
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 59.29,
|
| 3990 |
+
"learning_rate": 2.5828599592489917e-05,
|
| 3991 |
+
"loss": 0.0572,
|
| 3992 |
+
"step": 3320
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 59.38,
|
| 3996 |
+
"learning_rate": 4.234791653975481e-05,
|
| 3997 |
+
"loss": 0.0532,
|
| 3998 |
+
"step": 3325
|
| 3999 |
+
},
|
| 4000 |
+
{
|
| 4001 |
+
"epoch": 59.46,
|
| 4002 |
+
"learning_rate": 6.0919236939312284e-05,
|
| 4003 |
+
"loss": 0.057,
|
| 4004 |
+
"step": 3330
|
| 4005 |
+
},
|
| 4006 |
+
{
|
| 4007 |
+
"epoch": 59.55,
|
| 4008 |
+
"learning_rate": 8.00909269187053e-05,
|
| 4009 |
+
"loss": 0.0541,
|
| 4010 |
+
"step": 3335
|
| 4011 |
+
},
|
| 4012 |
+
{
|
| 4013 |
+
"epoch": 59.64,
|
| 4014 |
+
"learning_rate": 9.836442450346415e-05,
|
| 4015 |
+
"loss": 0.0572,
|
| 4016 |
+
"step": 3340
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 59.73,
|
| 4020 |
+
"learning_rate": 0.00011431137524750662,
|
| 4021 |
+
"loss": 0.0581,
|
| 4022 |
+
"step": 3345
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 59.82,
|
| 4026 |
+
"learning_rate": 0.00012668528006705928,
|
| 4027 |
+
"loss": 0.0531,
|
| 4028 |
+
"step": 3350
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 59.91,
|
| 4032 |
+
"learning_rate": 0.00013451892828543363,
|
| 4033 |
+
"loss": 0.0639,
|
| 4034 |
+
"step": 3355
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 60.0,
|
| 4038 |
+
"learning_rate": 0.0001372,
|
| 4039 |
+
"loss": 0.0604,
|
| 4040 |
+
"step": 3360
|
| 4041 |
}
|
| 4042 |
],
|
| 4043 |
+
"max_steps": 3360,
|
| 4044 |
+
"num_train_epochs": 60,
|
| 4045 |
+
"total_flos": 3472571105280000.0,
|
| 4046 |
"trial_name": null,
|
| 4047 |
"trial_params": null
|
| 4048 |
}
|
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 sha256:9513d7ccc6c6518d7961d6a29e81891e29d0f23c89e6554823a7f17f43c8c556
|
| 3 |
size 2671
|