Dataset · Hugging Face

Thorsu/miha-bench

Name: miha-bench
Creator: Thorsu
License: apache-2.0
Keywords: task_categories:question-answering, task_categories:text-classification, language:en, license:apache-2.0, size_categories:n<1K, format:json, modality:text, library:datasets, library:pandas, library:polars, library:mlcroissant, region:us, benchmark, deterministic-evaluation, evidence-grounding

MIHA-Bench MIHA-Bench is the deterministic holdout benchmark for MIHA-style evidence-path reasoning. It scores structured answers, cited source IDs, required search-space coverage, dead-end handling, temporal cutoffs, access boundaries, revision supersession, graph integrity, and dual-readback claims without an LLM judge. The benchmark is deliberately separate from live MIHA hunts. Live hunts may export to Thorsu/miha-evidence; they must not automatically become MIHA-Bench gold… See the full description on the dataset page: https://huggingface.co/datasets/Thorsu/miha-bench.

View on Hugging FaceReport an issue

Author / organization	Thorsu
License (as declared)	apache-2.0
Downloads	202 (as reported by Hugging Face)
Likes	1
Created	5 October 2026
Last updated at source	6 October 2026

benchmarkdeterministic-evaluationevidence-groundingagent-trajectoriesmiha

Embed this entry

<iframe src="https://silytics.si/embed/item/f1717895-7e09-4743-af0c-89cb080568a5" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

More from Thorsu

Thorsu/miha-evidenceDataset
Thorsu/sovereign-shadow-inference-benchDataset
Thorsu/evidence-bound-agent-evalsDataset
Thorsu/miha-train-atomicDataset
Thorsu/satoshi-evidence-ledgerDataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.