USC-FORTIS/AD-LLM

[ACL Findings 2025] A benchmark for anomaly detection using large language models. It supports zero-shot detection, data augmentation, and model selection, with scripts and data for GPT-4 and Llama experiments.

42
/ 100
Emerging

This project helps machine learning researchers and practitioners evaluate how Large Language Models (LLMs) perform in anomaly detection tasks involving natural language. It takes various LLMs (like GPT-4 or Llama) and natural language datasets, providing benchmark results for zero-shot detection, data augmentation, and model selection. Researchers working on advancing NLP-driven anomaly detection would find this useful.

No commits in the last 6 months.

Use this if you are a machine learning researcher or engineer looking to benchmark LLMs for natural language anomaly detection, explore their ability to generate synthetic data, or recommend suitable unsupervised models.

Not ideal if you are an end-user seeking a ready-to-use anomaly detection application for your specific business data, as this is a research benchmark.

natural-language-processing anomaly-detection machine-learning-research large-language-models model-benchmarking
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 7 / 25
Maturity 16 / 25
Community 17 / 25

How are scores calculated?

Stars

41

Forks

8

Language

Python

License

MIT

Last pushed

Oct 09, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/USC-FORTIS/AD-LLM"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.