awesome-hallucination-detection

Why we included this project

Hallucination detection is one of those research areas where the field moves fast enough that a good paper index saves real time. This one, maintained by the Edinburgh NLP group, collects work on detecting when LLMs state things confidently but falsely, and it does the sorting for you: each entry lists the metrics the authors used, the datasets they evaluated on, and a short summary of the method in plain language. The summaries are the useful part, since they go past the title to note things like whether a detector needs training, which benchmarks it was checked against, and where it openly fails. If your team is trying to survey the field before picking a detection approach or a benchmark to test against, this is a solid place to start, though it is a reading list rather than something you install.