dim/lmsys_chatbot_arena_conversations

The dataset includes multiple features such as question ID, dialogue content from two models, winner label, judge, turn count, anonymity flag, language, timestamp, and OpenAI moderation results. Each dialogue records content and role, turn number, and anonymity status. Toxicity detection results from two large models are also captured, including binary flags and probabilities. The dataset is provided as a training split with 33,000 samples.

Updated 11/8/2023

hugging_face

Dataset Overview

Dataset Information

Feature List:
- question_id: string
- model_a: string
- model_b: string
- winner: string
- judge: string
- conversation_a: list of {content: string, role: string}
- conversation_b: list of {content: string, role: string}
- turn: 64‑bit integer
- anony: boolean
- language: string
- tstamp: 64‑bit float
- openai_moderation: struct with categories (multiple booleans), category_scores (multiple 64‑bit floats), and flagged (boolean)
- toxic_chat_tag: struct with roberta-large (flagged boolean, probability float) and t5-large (flagged boolean, score float)

Dataset Split

Training Set:
- Size: 81,159,839 bytes
- Samples: 33,000

Dataset Size

Download Size: 41,573,740 bytes
Total Size: 81,159,839 bytes

Configuration

Default Config:
- Data file pattern: data/train-*

dim/lmsys_chatbot_arena_conversations

Description

Dataset Overview

Dataset Information

Dataset Split

Dataset Size

Configuration

AI studio

Access Dataset

Topics

Source