Open-Weight LLM Evaluation for Screening RNA-seq Metadata

Supplementary materials for a pilot evaluation of 17 open-weight large language models screening RNA-seq metadata. The dataset includes performance metrics like AUPRC and F1 scores, runtime distributions, and reproducibility data across 150 projects per model. Mitsuo Shintani authored this CC-BY-4.0 licensed dataset, last updated in May 2026.

Use Cases

• Benchmarking LLM performance based on AUPRC and F1 scores mentioned in the description
• Analyzing inference runtime trade-offs based on elapsed time distributions
• Evaluating model reproducibility based on per-accession five-run results

Strengths

• Evaluates 17 distinct open-weight LLM models
• Includes performance metrics like AUPRC and F1 scores for each model
• Contains reproducibility data across five runs for selected models

Limitations

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

Provenance

Source

figshare

Freshness

Last updated 2026-05-05 03:01:29; freshness should be verified