zhuyiyzu / msp

First, you need to download the corresponding model from Hugging Face and store it in the ./model folder. Then, modify the model path in the files fewshot.py and openprompt/pipline_base.py.

Example shell scripts:

python fewshot.py --result_file ./output_fewshot.txt --dataset cn_clickbait --template_id 0 --seed 123 --shot 10 --verbalizer manual

You can also run the entire model by executing autorun.py:

python autorun.py

To Run with Your Own Dataset

1. Format your dataset to match the style we provided and split it into training and testing sets.
2. In the scripts/TextClassification directory, create the necessary files, including verbalizer.txt and template.txt, following our examples. Please ensure the categories in the verbalizer.txt file match those in your dataset.
3. Modify openprompt/data_utils/text_classification_dataset.py according to the data processing functions we provided.

Our dataset is stored in /datasets/TextClassification, but due to the large number of images, we are unable to upload the entire dataset to GitHub. If you need the complete dataset, please contact me via email, and I will send it to you through other means. My email is: [email protected]