Detect anomalies in text embeddings using PyOD's IsolationForest.
npx skills add https://github.com/wuyoscar/ISC-Bench --skill aiml-pyod-detection
Please help me install this Agent Skill. Skill: aiml-pyod-detection Source: https://github.com/wuyoscar/ISC-Bench/tree/main/templates/aiml_pyod Command: npx skills add https://github.com/wuyoscar/ISC-Bench --skill aiml-pyod-detection
ISC PyOD detection template provides a guided workflow to validate anomaly detection on text embeddings by using IsolationForest to distinguish normal samples from outliers in an AI safety evaluation dataset.
Run the anomaly-detection template on a sample dataset to validate the PyOD integration.
High-intent search queries and answers about installing and using this skill.
To detect anomalies in text embeddings, you can use PyOD's IsolationForest within an AI safety evaluation workflow to distinguish normal samples from outliers. This template provides script scaffolding to validate the integration and produce verifiable outliers.
Yes, you need SentenceTransformers to generate the text embeddings required by the PyOD IsolationForest model. The template requires Python, PyOD, SentenceTransformers, and a minimal embedding dataset to produce deterministic results.
The semantic polarity mechanism verifies outliers by ensuring they differ from normal samples in the evaluation dataset. This polarity-based verification confirms that the flagged anomalous samples are semantically distinct from the baseline.
Yes, the template is designed for researchers and engineers validating frontier-model evaluation pipelines across text data and prompts. It provides reproducible script scaffolding and sample prompts for consistent testing.
The best way to reproduce anomaly detection results is by using the template's reproducible evaluation scaffolding and sample prompts. It employs PyOD's IsolationForest on textual embeddings to deliver deterministic results for outlier detection.
cinienceSmoke test text embedding models in Alibaba Cloud Model Studio.
tbroadleyExtract samples, messages, events, and metrics from Inspect AI .eval logs.
idanmoradarthasEncode multi-label sets and text into sklearn pipeline features.
eren726290Apply descriptive statistics, trend analysis, outlier detection, and hypothesis testing to data using Python.
