HW3_4 / README-4.md
jimboHsueh's picture
Upload README-4.md
a4ab28e
# **HW2**
## **Environment**
```
pip install -r requirements.txt
```
## **Download**
Download adapter checkpoints
```
bash ./download.sh
```
## **Fine tuning mT5 model**
```
python train.py --base_model_path jimboHsueh/save_hw3 \
--peft_path ./PETH \
--train_data_path train.json \
--per_device_train_batch_size 2 \
--gradient_accumulation_steps 8 \
--test_data_path public_test.json \
--output_dir HW3_4 \
--push_to_hub
```
-**base_model_path**: Path to Taiwan_llama.
-**peft_path**: Path to peft model.
-**train_data_path**: Path to train.json.
-**test_data_path**: Path to test.json.
## **Generate prediction.json**
```
bash ./download.sh
bash ./run.sh python /path/to/Taiwan-LLaMa-folder /path/to/adapter_checkpoint \
/path/to/input.json /path/to/output.json
```