Vision-Language Agent Trajectories Across 17 Interactive Environments

VisGym consists of 17 diverse, long-horizon environments for evaluating Vision-Language Models on interactive tasks. The dataset contains agent trajectories where actions are conditioned on past actions and observation history, challenging multimodal sequence handling.

Use Cases

• Systematically evaluate VLMs on long-horizon interactive tasks using the provided 17 distinct environments.
• Diagnose agent performance by analyzing trajectories conditioned on past actions and observation history.
• Train agents on complex, multimodal sequences from visually interactive tasks.
• Benchmark VLM capabilities in robotics and embodied AI scenarios using the diverse environment set.

Strengths

• Contains data from 17 distinct, diverse environments for systematic evaluation.
• Focuses on long-horizon, visually interactive tasks requiring multimodal reasoning.
• Designed to challenge agents with sequences conditioned on past actions and observation history.

Limitations

• Specific dataset size, row count, and column details are unknown.
• Data format, file structure, and sample data are not provided in the input.
• The scope of individual trajectories and interaction details is unspecified.

Provenance

Source

VisGym

Collection Method

Contains trajectories and interaction data from agents operating in designed environments.

Freshness

Last updated on 2026-02-05.