Sign in to view source links and access this dataset
Description
LabHorizon Level 2 data provides real-world experimental contexts, planning goals, and protocol-derived constraints for generating long-horizon laboratory action sequences. The dataset was created by CongLab-Research and was last updated on May 29, 2026. Each example includes an experimental context, a goal, constraints, available inputs, and an action pool to target a sequence of actions with parameters.
Use Cases
Train sequence generation models for laboratory procedures based on provided experimental contexts and goals.
Benchmark planning algorithms that must adhere to protocol-derived constraints mentioned in the description.
Develop AI assistants for experimental design using the available inputs and action pool structure.
Research on long-horizon task planning in real-world scientific domains.
Strengths
Dataset is explicitly designed for long-horizon planning with real-world experimental contexts.
Includes structured components like goals, constraints, inputs, and an action pool for each example.
Last updated on May 29, 2026, 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.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
CongLab-Research on Hugging Face
Freshness
Last updated 2026-05-29 07:17:03
License is unknown; terms of use must be verified before application.