https://github.com/TIGER-AI-Lab/MMLU-Pro

The code and data for "MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark" [NeurIPS 2024]
https://github.com/TIGER-AI-Lab/MMLU-Pro

evaluation llm

Last synced: 12 months ago
JSON representation

The code and data for "MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark" [NeurIPS 2024]

• Host: GitHub
• URL: https://github.com/TIGER-AI-Lab/MMLU-Pro
• Owner: TIGER-AI-Lab
• License: apache-2.0
• Created: 2024-05-16T23:59:51.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2025-02-28T17:39:35.000Z (about 1 year ago)
• Last Synced: 2025-02-28T21:57:39.707Z (about 1 year ago)
• Topics: evaluation, llm
• Language: Python
• Homepage:
• Size: 325 MB
• Stars: 193
• Watchers: 6
• Forks: 33
• Open Issues: 7
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-llm-eval - link
• Awesome-AI-Evaluation-Guide - MMLU-Pro - Enhanced version with 10 choices per question, emphasizing reasoning over memorization (Benchmarks & Datasets / General Language Understanding)
• awesome-ai-eval - **MMLU-Pro** - AI-Lab/MMLU-Pro?style=social&label=github.com) - Harder 10-choice extension focused on reasoning-rich, low-leakage questions. (Benchmarks / General)
• Awesome-Prompt-Engineering - GitHub