Skip to content

Dataset · lmarena-ai

Chatbot Arena leaderboard historical snapshots

lmarena-ai/leaderboard-dataset

Millions of records with language model rating history from the popular Arena benchmark.

What's inside

Large tabular-text dataset ranging from 1 to 10 million records containing historical slices of the Arena leaderboard. Includes splits for analysis with text style control (text_style_control) and category filtering.

What you can do with it

  • Analyzing quality dynamics and development progress of modern LLMs
  • Researching the impact of response style on user ratings in benchmarks
Size
1M–10M rows
Downloads this month
361K
Likes
53

How to load it

from datasets import load_dataset

ds = load_dataset("lmarena-ai/leaderboard-dataset")