UBio-ARCK: Model Predictions for a Biological AI Benchmark

Content-free model outputs for the ARCK v0.1 leaderboard, created by IQuestLab. The dataset includes predictions and scores for tasks across tracks like General, Co-Scientist, and Bio-Designer. It was last updated on June 9, 2026.

Use Cases

• Benchmarking model performance based on scores across different evaluation tracks.
• Comparing AI capabilities in specific biological subtasks based on the category field.
• Reproducing leaderboard results by joining predictions with regenerated upstream questions.

Strengths

• Structured by specific evaluation tracks (General, Co-Scientist, Bio-Designer) and sub-groupings.
• Explicitly licensed CC-BY-NC-4.0 for research use.
• Provides a clear link to source code for full reproduction.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Contains only model outputs; the upstream questions are not included and must be regenerated.

Provenance

Source

IQuestLab

Collection Method

Model predictions/scores generated for a benchmark leaderboard.

Freshness

Last updated 2026-06-09 08:24:20; freshness should be verified.