hww123/MADBench-eval
MADBench-Eval: Multi-Agent System Anomaly Detection Benchmark (Evaluation Split) MADBench-Eval is the fixed evaluation split of the MADBench benchmark family. It contains 1,000 execution traces from five LLMs solving procedurally generated escape room puzzles across four difficulty levels at temperature 0.0. It is intended as a standardized held-out benchmark for evaluating anomaly detection models on multi-agent execution traces. Dataset Overview Property… See the full description on the dataset page: https://huggingface.co/datasets/hww123/MADBench-eval.
| Author / organization | hww123 |
|---|---|
| License (as declared) | mit |
| Downloads | 128 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 6 May 2026 |
| Last updated at source | 6 May 2026 |
Popularity, last 2 days
128 now (-1 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Recent changes
- 16 Oct 2026 Description: MADBench-Eval: Multi-Agent System Anomaly Detection Benchmark (Evaluation Split) MADBench-Eval is the fixed evaluation split of the MADBench benchmark family. It contains 1,000 execution traces from five LLMs solving procedurally generated → MADBench-Eval: Multi-Agent System Anomaly Detection Benchmark (Evaluation Split) MADBench-Eval is the fixed evaluation split of the MADBench benchmark family. It contains 1,000 execution traces from five LLMs solving procedurally generated
Changes detected by Silytics between refreshes of the source.
More from hww123
- hww123/MAADBench-fullDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.