LabHorizon: Protocol-Conditioned Laboratory Action Sequences

Stanford-CongLab's Level 2 split for protocol-conditioned long-horizon laboratory action-sequence planning. Each example provides a real-world experimental context, a planning goal, protocol-derived constraints, available inputs, and an action pool. The dataset was last updated on June 4, 2026.

Use Cases

• Training models for laboratory action-sequence planning based on experimental context and goals.
• Benchmarking AI agents on long-horizon planning tasks with protocol-derived constraints.
• Developing systems that integrate planning goals with available inputs and action pools.

Strengths

• Data is structured for a specific, complex task: protocol-conditioned long-horizon planning.
• Examples include multiple components: experimental context, goals, constraints, inputs, and an action pool.

Limitations

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

Provenance

Source

Stanford-CongLab

Freshness

Last updated 2026-06-04 04:16:00; freshness should be verified.