Read:

• Attention Is All You Need
• https://github.com/karpathy/nanogpt
• https://github.com/karpathy/llm.c
• https://github.com/karpathy/llama2.c

• Small projects:

  • Implement Dice Loss and train a U-Net that uses it

  • Move cursor on screen through gaze analysis

    • Write a blog article on how I did it

  • Real time pose estimation with custom BlazePose implementation

    • Write a blog article on how I did it

  • Real time facial expression classification

  • Learn RL and add an env to pufferlib

  • Build a model to predict next period date from past period dates

  • Model that recommends which movies and books to read

  • Automatic differentiation engine like Pytorch or fix a bug in tinygrad.

  • Model that takes live driving footage from Waymo's dataset and builds a map of the city roads in real time (SLAM?)

  • RL for noise cancellation

  • Contribute to Pytorch

  • Kaggle competitions

• Implement papers:

  • Something interesting from here

  • Attention Is All You Need

    • Implement multi-head attention, positional encoding, encoder, decoder.

  • Variational autoencoders

  • Gaussian splatting

    • 3D Gaussian Splatting for Real-Time Radiance Field Rendering

  • CLIP

    • Learn multimodel learning, contrastive learning, embeddings, zero-shot classification.

  • SegFormer

    • Learn segmentation and dense prediction.

  • Stable diffusion

  • Speech to text and text to speech

    • Tacotron
    • Whisper

  • AlphaZero