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 |
|
20 |
## Model Details
|
21 |
|
@@ -27,9 +27,21 @@ This model is a fine-tuned version of VietAI/envit5-translation on the mt_eng_vi
|
|
27 |
- **Model type:** T5
|
28 |
- **Finetuned from model [optional]:** [VietAI/envit5-translation](https://huggingface.co/VietAI/envit5-translation)
|
29 |
|
30 |
-
###
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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: 0.1988
|
19 |
|
20 |
## Model Details
|
21 |
|
|
|
27 |
- **Model type:** T5
|
28 |
- **Finetuned from model [optional]:** [VietAI/envit5-translation](https://huggingface.co/VietAI/envit5-translation)
|
29 |
|
30 |
+
### Hyperparameters
|
31 |
+
The following hyperparameters were used during training:
|
32 |
+
- Repo: [simpleT5](https://github.com/Shivanandroy/simpleT5)
|
33 |
+
- Dataset: [mt_eng_vietnamese](https://huggingface.co/datasets/mt_eng_vietnamese) + custom (~40%)
|
34 |
+
```
|
35 |
+
model.train(
|
36 |
+
train_df=train_df,
|
37 |
+
eval_df=test_df,
|
38 |
+
source_max_token_len=128,
|
39 |
+
target_max_token_len=128,
|
40 |
+
batch_size=32,
|
41 |
+
max_epochs=15,
|
42 |
+
use_gpu=True,
|
43 |
+
save_only_last_epoch=True,
|
44 |
+
precision=16,
|
45 |
+
dataloader_num_workers=4
|
46 |
+
)
|
47 |
+
```
|