Sunsincer97/trellm-audit-corpora
TreLLM audit corpora Faulted agent trajectories used to measure whether an LLM overseer reports every safety-relevant defect when more than one is present. Each task is a complete trading-agent trajectory into which defects were injected programmatically, plus the prompt an auditor model was given. The answer key records exactly what was injected. Scoring is recall against that key, not against a human judgement. Code, analysis scripts, scored result rows, answer keys and the… See the full description on the dataset page: https://huggingface.co/datasets/Sunsincer97/trellm-audit-corpora.
| Author / organization | Sunsincer97 |
|---|---|
| License (as declared) | mit |
| Downloads | 142 (as reported by Hugging Face) |
| Likes | 1 |
| Created | 14 August 2026 |
| Last updated at source | 14 August 2026 |
Popularity, last 3 days
142 now (-4 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
Related papers
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.