I am Sacha Wendling, a MSc graduate in Applied and Industrial Mathematics from Grenoble INP - ENSIMAG. Self-taught for +10 years in computer science, I specialize in Artificial Intelligence research, focusing on machine learning and deep learning. On this website, you’ll find my Curriculum Vitae, experimental projects, and articles exploring recent advances in AI and Data Science.
Featured
-
Muon Optimizer: From Conception to Modern Adaptations
Introduced in October 2024, Muon is a geometry-aware optimizer treating weight matrices as structured objects rather than flat vectors. In this article we study the evolution of Muon through recent advancements.
Recent Posts
-
FMMI: Estimating Mutual Information with Flow Matching
A presentation of a method that estimates Mutual Information by learning a probability flow between independence and the true joint distribution. Using Flow Matching and a divergence-based identity, MI becomes a simple expectation that we compute with a neural velocity field. A small experiment illustrates the approach and confirms its accuracy.
-
Geodesic Calculus on Latent Space
A review of Geodesic Calculus on Latent Spaces (Hartwig et al., 2025), explaining its mathematical foundations in differential geometry, demonstrating a VAE-based implementation on MNIST with a learned projection operator, and visualizing how geodesic paths yield more realistic, manifold-aware interpolations than linear ones.
-
Less is More: Recursive Reasoning with Tiny Networks
A paper showing how small neural models can achieve powerful reasoning through iterative self-correction, deep supervision, and elegant simplicity, challenging the idea that intelligence requires scale.
-
From GRPO to ExGRPO: Learning to Reason from Experience
A compact reproduction of ExGRPO, showing how reusing confident past reasoning trajectories can make reinforcement learning more stable and sample-efficient than standard GRPO.