Yoann Poupart's Projects
Interpreting how transformers simulate reinforcement learning algorithms
Tools for studying developmental interpretability in neural networks.
Adversarial Generation of Extreme Samples
Supplementary materials to my internship's report.
Build and share delightful machine learning apps, all in Python. 🌟 Star to support our work!
TransformerLens + HuggingFace
Experiment on Iterative Summarisation, potentially linked to Natural Abstraction. Made for the Alignment jam #4.
A Bulletproof Way to Generate Structured JSON from Language Models
⚡ Building applications with LLMs through composability ⚡
🔬 Interpretability for Leela Chess Zero networks.
Reimplementation of the interpretability tool Logit Lens describe GPT-2 behaviour ([Lesswrong](https://www.lesswrong.com/posts/AcKRB8wDpdaN6v6ru/interpreting-gpt-the-logit-lens)).
Simple playground wrapper around Mistral API.
Projects realised for the ML Engineer path on OpenClassrooms
:lock_with_ink_pen: Safe inference using representation engineering.
Application de gestion pour l'association ronrhone
Scrapy middleware to handle javascript pages using selenium
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Little study of the Twitch network from the point of vie of complex systems and graphs.
Wasserstein Generative Models for Patch-based Texture Synthesis
Zennit is a high-level framework in Python using PyTorch for explaining/exploring neural networks using attribution methods like LRP.
An eXplainable AI toolkit with Concept Relevance Propagation and Relevance Maximization