Commit ·
2e64eea
1
Parent(s): 62cf5a8
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# **HW2**
|
| 2 |
+
|
| 3 |
+
## **Environment**
|
| 4 |
+
```
|
| 5 |
+
pip install -r requirements.txt
|
| 6 |
+
```
|
| 7 |
+
|
| 8 |
+
## **Download**
|
| 9 |
+
Download training and validation data, as well as fine-tuned mT5 model.
|
| 10 |
+
```
|
| 11 |
+
bash ./download.sh
|
| 12 |
+
```
|
| 13 |
+
|
| 14 |
+
## **Fine tuning mT5 model**
|
| 15 |
+
```
|
| 16 |
+
python ./code/run_summarization.py \
|
| 17 |
+
--model_name_or_path google/mt5-small \
|
| 18 |
+
--train_file train.jsonl \
|
| 19 |
+
--validation_file public.jsonl \
|
| 20 |
+
--dataset_config "3.0.0" \
|
| 21 |
+
--source_prefix "summarize: " \
|
| 22 |
+
--output_dir ./HW2 \
|
| 23 |
+
--num_train_epochs 3 \
|
| 24 |
+
--num_beams 1 \
|
| 25 |
+
--do_sample False \
|
| 26 |
+
--push_to_hub
|
| 27 |
+
```
|
| 28 |
+
-**model_name_or_path**: Path to pretrained model.
|
| 29 |
+
|
| 30 |
+
-**output_dir**: Path to directory which saves the model outputs.
|
| 31 |
+
|
| 32 |
+
-**train_file**: Path to train.jsonl.
|
| 33 |
+
|
| 34 |
+
-**validation_file**: Path to public.jsonl.
|
| 35 |
+
|
| 36 |
+
## **Generate output.jsonl**
|
| 37 |
+
```
|
| 38 |
+
bash ./download.sh
|
| 39 |
+
bash ./run.sh /path/to/input.jsonl /path/to/output.jsonl
|
| 40 |
+
```
|