AI BENCHMARK PROFILE

PHITSBench

General AIKnowledge & Reasoning

PHITSBench evaluates AI-assisted generation of PHITS radiation-transport input via natural language across 282 tasks in three workflows: Edit, Repair, and Reproduce. Scoring uses a Composite Metric Score combining execution success and agreement with reference transport observables.

Released: 2026-07-08
Readiness: Paper only
Primary field: General AI

Why it matters

Provides an execution-grounded benchmark for a niche task, highlighting the need for machine-readable knowledge bases and curated training data in AI-assisted radiation-transport modeling.

Motivation

We introduce PHITSBench, an execution-scored benchmark for the Monte Carlo Particle and Heavy Ion Transport code System (PHITS).

Primary resources

Paper ↗Hugging Face ↗

Benchmark Radar records only publicly supported details and links back to primary sources for verification.