Skip to content

Code and data of "MultiSkill: Evaluating Large Multimodal Models for Fine-grained Alignment Skills" (EMNLP 2024 Findings).

Notifications You must be signed in to change notification settings

HITsz-TMG/MultiSkill

Repository files navigation

MultiSkill: Evaluating Large Multimodal Models for Fine-grained Alignment Skills

✨ Overview

This repository contains the code and data for "MultiSkill: Evaluating Large Multimodal Models for Fine-grained Alignment Skills" (EMNLP 2024 Findings).

🚨 Usage

Data

Download data from here and unzip images.zip.

MultiSkill.jsonl
skillset_description_multimodal.json
images
result
temp
MultiSkill_inference.py
MultiSkill_score.py
utils.py

Inference

  1. Set your api_key and choose the base model.
  2. Run MultiSkill_inference.py.

Score

  1. Set your api_key and choose the base model, evaluation file.
  2. Run MultiSkill_score.py.

About

Code and data of "MultiSkill: Evaluating Large Multimodal Models for Fine-grained Alignment Skills" (EMNLP 2024 Findings).

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages