Pivotal Token Search

Adversarial Manipulation of CoT

mech-interp suite for Granite4 models that use Mamba-2 architecture

Analysed determinism, faithfulness, reasoning patterns, & steering. Developed and tested methods to enhance control and fail-safes

Official implementation of the 'Uncovering Competency Gaps in Large Language Models and Their Benchmarks' paper

Detect safety degradation during LLM fine-tuning before it becomes behavioral

All code, stimuli, and results for a mechanistic interpretability study investigating how large language models internally represent emotional content

Implementation and analysis of Sparse Autoencoders for neural network interpretability research. Features interactive visualization dashboard and W&B integration.

Collection and learnings of my journey in Artificial Intelligence

Unofficial implementation to reproduce the experiments from "Superposition as a Phase Change" of "Toy Models of Superposition".

Local agent-driven mechanistic interpretability research platform for Apple Silicon

Cross-architecture mechanistic interpretability toolkit — first OSS Mamba SSM state extraction. Works on transformer + SSM + hybrid models with unified API.

Automated Forensic Discovery of Reasoning Circuits in Transformers