Mario Gemoll

Machine learning notes and experiments

• Pick and Place with the SO-101 Robot Arm
  An end-to-end robotics project covering inverse kinematics, motion planning, computer vision, sim-to-real transfer, and ACT imitation learning.
• Reinforcement Learning
  Foundations of reinforcement learning, from Markov decision processes and value functions to dynamic programming, Monte Carlo methods, temporal-difference learning, and policy gradients.
• LLM Posttraining
  A guide to LLM posttraining basics: LoRA, SFT, RLHF, RLVR.
• Flow Matching
  An overview of flow matching techniques for generative modeling.
• Diffusion
  Extending the flow matching framework to diffusion models.
• Normalizing Flows
  A page exploring the basics of normalizing flows.
• Variational Autoencoders
  An interactive exploration of Variational Autoencoders (VAEs) with theory, implementation details, and live demonstrations using a synthetic dataset.
• Attention Is All You Need
  Describing the training of an encoder-decoder transformer, following the "Attention Is All You Need" paper.

Other

• Parallel Corpus Viewer
  A viewer for the datasets used to train the model for the Attention Is All You Need page.
• gen3d.chat
  An agent to help with (basic) 3D CAD modeling.

GitHub | LinkedIn

RSS | Atom