OrchestrateBench is a research benchmark for evaluating the orchestration
meta-decision layer of multi-agent LLM systems: routing, failure recovery,
cascade propagation, and decomposition quality.
This dataset repository contains the committed measured inputs from
anote-ai/research-orchestratebench, plus normalized JSONL viewer tables so
the Hugging Face Dataset Viewer can render each experiment schema separately.
exp1_baselines:… See the full description on the dataset page:
https://huggingface.co/datasets/anote-ai/Orchestratebench.
