Generative AI

Large Language Models (LLMs)

Models

Proprietary

Open-weights

Evaluation and Benchmarks

LLM benchmark results:

• AI benchmarking hub (Epoch AI)

Other overviews of LLM benchmarks:

• llm benchmarks (GitHub)
• Awesome Human Preference Datasets for LLM (GitHub)

Summarization:

• CNN / Dailymail news articles
• Summarization dataset used by OpenAI for training model

Question answering / explaining concepts / trivia questions:

• Explaining concepts / question answering: https://huggingface.co/datasets/Hello-SimpleAI/HC3
• TriviQA

Multiple choice:

• sciq (HuggingFace
  • Multiple choice questions on science
• Measuring Massive Multitask Language Understanding (MMLU)
  • Multiple choice questions across 57 different subjects, ranging from STEM to social sciences