EB-Man: Embodied AI Agent Trajectories from Multiple Models

A trajectory dataset collected from EmbodiedBench using several closed-source and open-source models. The dataset is released by EmbodiedBench and was last updated on June 4, 2025. It is intended to support the development of more capable embodied agents with improved perception, reasoning, and planning abilities.

Use Cases

• Training embodied agents based on recorded trajectory data.
• Evaluating agent performance across different EmbodiedBench subsets.
• Analyzing planning and reasoning patterns from model-generated trajectories.
• Benchmarking closed-source and open-source models on embodied tasks.

Strengths

• Data collected from the EmbodiedBench platform, suggesting a structured evaluation environment.
• Includes trajectories from multiple model types, potentially allowing for comparative analysis.
• Last updated on June 4, 2025, indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The description recommends separating training and evaluation sets, implying potential data leakage concerns if not handled properly.

Provenance

Source

EmbodiedBench

Collection Method

Trajectories collected from the EmbodiedBench platform using several models.

Time Range

null

Freshness

Last updated 2025-06-04 01:43:07.

Geography

null