Sotopia-RL: Utterance-Level Reward Annotations for Social Intelligence

A dataset from the paper 'Sotopia-RL: Reward Design for Social Intelligence' by ulab-ai, last updated on August 7, 2025. It contains annotations for refining episode-level feedback into utterance-level, multi-dimensional rewards to train socially intelligent agents via reinforcement learning.

Use Cases

• Training socially intelligent agents via reinforcement learning based on utterance-level rewards.
• Evaluating reward design frameworks for multi-agent social scenarios based on the described multi-dimensional feedback.
• Benchmarking agent performance in partially observable social environments based on the described annotation structure.

Strengths

• Dataset is associated with a published research paper, suggesting a defined methodology.
• Last updated on 2025-08-07, indicating recent maintenance.

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.

Provenance

Source

ulab-ai via Hugging Face

Collection Method

Likely created as part of the research for the paper 'Sotopia-RL: Reward Design for Social Intelligence'.

Time Range

null

Freshness

Last updated 2025-08-07 18:20:47.

Geography

null