Hudson Forge Iqr V2: Multi-Model Reasoning Benchmark with Peer Critique

200 questions were answered by multiple models in a structured five-round deliberation process. The dataset likely contains explicit reasoning chains, symbolic logic, confidence scores, and final answers from each model, along with peer critiques. It was created by Billyrdavis1985 and last updated on Hugging Face in May 2026.

Use Cases

• Analyzing model reasoning consistency based on structured reasoning chain outputs
• Studying the effect of anonymous peer critique on model answer revision
• Benchmarking model confidence calibration using explicit confidence scores
• Evaluating multi-agent deliberation protocols in AI systems

Strengths

• 200 questions provide a fixed benchmark for evaluation
• Five-round deliberation process suggests structured, multi-step interaction data
• Pre-registered peer assignments before data collection indicate a controlled experimental design

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Freshness should be verified as the last update timestamp is in the future (2026-05-17)

Provenance

Source

Hugging Face user Billyrdavis1985

Collection Method

Structured five-round deliberation protocol involving independent response, anonymous peer critique, and revision rounds.

Freshness

Last updated 2026-05-17 22:26:35