Comparia Fr Arena: French-Language Chatbot Conversations and Human Preferences

Compar:IA is a public chatbot arena run by the French Ministry of Culture. This dataset contains French-language conversations where users compare two anonymous models' answers and state their preference. Each row likely contains one turn of a conversation, including the two model responses and the user's indicated preference.

Use Cases

• Training preference models based on direct human feedback between model outputs.
• Benchmarking French-language LLMs based on side-by-side human evaluations.
• Analyzing conversational patterns and user preferences in chatbot interactions.
• Studying the alignment of AI responses with human judgment in a specific cultural/linguistic context.

Strengths

• Data originates from a public platform run by a government ministry, suggesting an official collection effort.
• Contains direct human preference labels between two model outputs, a valuable signal for alignment research.
• Focuses on French-language conversations, addressing a need for non-English evaluation data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license are unknown, which may limit suitability assessment.
• Data may reflect the specific user base and temporal bias inherent to the Compar:IA platform.

Provenance

Source

Compar:IA, a public chatbot arena run by the French Ministry of Culture.

Collection Method

Collected from users chatting with two anonymous models side-by-side and indicating their preferred answer.

Time Range

null

Freshness

Last updated 2026-06-11 16:31:00; freshness should be verified.

Geography

null