Pinned Loading

1. llm-audit-bench llm-audit-bench Public

   A modular pipeline that audits 5 small HuggingFace LLMs across transparency, fairness, robustness, explainability, and privacy. Produces per-pillar scores across the 5 pillars of AI trustworthiness.

   Jupyter Notebook

2. llm-stress-tester llm-stress-tester Public

   A framework to evaluate how open-source language models handle adversarial prompts. Tests across hallucination traps and jailbreak scenarios.

   Jupyter Notebook

3. interpreting-transformer-hallucinations interpreting-transformer-hallucinations Public

   Mechanistic interpretability of transformer hallucinations via attention flow, residual stream geometry, and head-level attribution analysis.

   HTML

4. rag-from-scratch rag-from-scratch Public

   A minimal, fully local Retrieval-Augmented Generation pipeline built from scratch. Covers document chunking, embedding with sentence-transformers, vector storage via ChromaDB, and LLM generation th…

   Python

5. xai-credit-risk-analysis xai-credit-risk-analysis Public

   A credit risk prediction project that compares multiple machine learning models and applies explainable AI methods to interpret and analyze model decisions.

   Jupyter Notebook

6. deepfractals deepfractals Public

   Visualizing deep learning dynamics as fractals in the complex plane

   Python