Update README.md
Browse files
README.md
CHANGED
@@ -15,7 +15,7 @@ tags:
|
|
15 |
|
16 |
This model is a fine-tuned version of VietAI/envit5-translation on the mt_eng_vietnamese dataset. It achieves the following results on the evaluation set:
|
17 |
|
18 |
-
- Loss: 0.
|
19 |
- BLEU: 34.46
|
20 |
|
21 |
## Model Details
|
@@ -33,18 +33,10 @@ The following hyperparameters were used during training:
|
|
33 |
- Repo: [simpleT5](https://github.com/Shivanandroy/simpleT5)
|
34 |
- Dataset: [mt_eng_vietnamese](https://huggingface.co/datasets/mt_eng_vietnamese) + custom (~40%)
|
35 |
```python
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
target_max_token_len=128,
|
41 |
-
batch_size=32,
|
42 |
-
max_epochs=15,
|
43 |
-
use_gpu=True,
|
44 |
-
save_only_last_epoch=True,
|
45 |
-
precision=16,
|
46 |
-
dataloader_num_workers=4
|
47 |
-
)
|
48 |
```
|
49 |
### Examples
|
50 |
|
|
|
15 |
|
16 |
This model is a fine-tuned version of VietAI/envit5-translation on the mt_eng_vietnamese dataset. It achieves the following results on the evaluation set:
|
17 |
|
18 |
+
- Val Loss: 0.8422
|
19 |
- BLEU: 34.46
|
20 |
|
21 |
## Model Details
|
|
|
33 |
- Repo: [simpleT5](https://github.com/Shivanandroy/simpleT5)
|
34 |
- Dataset: [mt_eng_vietnamese](https://huggingface.co/datasets/mt_eng_vietnamese) + custom (~40%)
|
35 |
```python
|
36 |
+
source_max_token_len=128
|
37 |
+
target_max_token_len=128
|
38 |
+
batch_size=32
|
39 |
+
max_epochs=8
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
```
|
41 |
### Examples
|
42 |
|