Ego-R1: Chain-of-Tool-Thought Examples for Egocentric Video Reasoning

A collection of 25,000 Chain-of-Tool-Thought examples for training large language models in tool-augmented reasoning and reinforcement learning. The dataset was created by Ego-R1 and last updated on July 31, 2025. It is designed for use with the Ego-R1 codebase presented in the associated research paper.

Use Cases

• Supervised fine-tuning of LLMs for tool-augmented reasoning based on the described Chain-of-Tool-Thought examples.
• Training reinforcement learning agents for sequential decision-making based on the dataset's structure.
• Benchmarking model performance on ultra-long egocentric video reasoning tasks as referenced in the paper.

Strengths

• Contains 25,000 structured examples for Chain-of-Tool-Thought training.
• Explicitly designed for a specific, advanced research area: tool-augmented reasoning for LLMs.
• Directly associated with a published research paper and codebase, providing context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown for the full dataset, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Ego-R1

Collection Method

Likely generated or curated for the associated research paper.

Time Range

null

Freshness

Last updated 2025-07-31 06:12:31; freshness should be verified.

Geography

null