Mvp Multi Task

🚀 MVP-multi-task

The MVP-multi-task model addresses the challenges in natural language generation. It offers a solution for a wide range of tasks, enabling efficient and effective text generation across various domains.

🚀 Quick Start

The detailed information and instructions can be found https://github.com/RUCAIBox/MVP.

✨ Features

• Prompt-based Design: MVP-multi-task is a prompt-based model. MVP is further equipped with prompts pre-trained using a mixture of labeled datasets. It is a variant (MVP+M) of the main MVP model, following a Transformer encoder - decoder architecture with layer-wise prompts.
• Versatile Adaptability: Specially designed for natural language generation, it can be adapted to a wide range of generation tasks, including but not limited to summarization, data - to - text generation, open - ended dialogue system, story generation, question answering, question generation, task - oriented dialogue system, commonsense generation, paraphrase generation, text style transfer, and text simplification. It can also be adapted to natural language understanding tasks such as sequence classification and (extractive) question answering.

💻 Usage Examples

Basic Usage

For summarization:

>>> from transformers import MvpTokenizer, MvpForConditionalGeneration

>>> tokenizer = MvpTokenizer.from_pretrained("RUCAIBox/mvp")
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp-multi-task")

>>> inputs = tokenizer(
...     "Summarize: You may want to stick it to your boss and leave your job, but don't do it if these are your reasons.",
...     return_tensors="pt",
... )
>>> generated_ids = model.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
["Why You Shouldn't Quit Your Job"]

For data - to - text generation:

>>> from transformers import MvpTokenizerFast, MvpForConditionalGeneration

>>> tokenizer = MvpTokenizerFast.from_pretrained("RUCAIBox/mvp")
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp-multi-task")

>>> inputs = tokenizer(
...     "Describe the following data: Iron Man | instance of | Superhero [SEP] Stan Lee | creator | Iron Man",
...     return_tensors="pt",
... )
>>> generated_ids = model.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
['Iron Man is a fictional superhero appearing in American comic books published by Marvel Comics.']

📚 Documentation

The MVP-multi-task model was proposed in MVP: Multi-task Supervised Pre-training for Natural Language Generation by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji - Rong Wen.

Related Models

MVP: https://huggingface.co/RUCAIBox/mvp.

Prompt-based models:

• MVP-multi-task: https://huggingface.co/RUCAIBox/mvp-multi-task.
• MVP-summarization: https://huggingface.co/RUCAIBox/mvp-summarization.
• MVP-open-dialog: https://huggingface.co/RUCAIBox/mvp-open-dialog.
• MVP-data-to-text: https://huggingface.co/RUCAIBox/mvp-data-to-text.
• MVP-story: https://huggingface.co/RUCAIBox/mvp-story.
• MVP-question-answering: https://huggingface.co/RUCAIBox/mvp-question-answering.
• MVP-question-generation: https://huggingface.co/RUCAIBox/mvp-question-generation.
• MVP-task-dialog: https://huggingface.co/RUCAIBox/mvp-task-dialog.

Multi-task models:

• MTL-summarization: https://huggingface.co/RUCAIBox/mtl-summarization.
• MTL-open-dialog: https://huggingface.co/RUCAIBox/mtl-open-dialog.
• MTL-data-to-text: https://huggingface.co/RUCAIBox/mtl-data-to-text.
• MTL-story: https://huggingface.co/RUCAIBox/mtl-story.
• MTL-question-answering: https://huggingface.co/RUCAIBox/mtl-question-answering.
• MTL-question-generation: https://huggingface.co/RUCAIBox/mtl-question-generation.
• MTL-task-dialog: https://huggingface.co/RUCAIBox/mtl-task-dialog.

📄 License

This project is under the Apache-2.0 license.

📚 Citation

@article{tang2022mvp,
  title={MVP: Multi-task Supervised Pre-training for Natural Language Generation},
  author={Tang, Tianyi and Li, Junyi and Zhao, Wayne Xin and Wen, Ji-Rong},
  journal={arXiv preprint arXiv:2206.12131},
  year={2022},
  url={https://arxiv.org/abs/2206.12131},
}