Scenic: A Jax Library for Computer Vision Research and Beyond
Role in this project:
ML Engineer Contributions:7 reviews, 168 commits, 23 PRs in 1 year 5 months
Contributions summary:Hadi contributed code for a Vision Transformer (ViT) model, including the definition of an `Encoder1DBlock`, `Encoder`, and the main `ViT` model class. The commits focused on adding baseline models for computer vision tasks within the Scenic framework. The code additions include components for processing image patches and building a multi-label classification model based on ViT.
computer-visionjaxdeep-learningtransformersvision-transformer
Flax is a neural network library for JAX that is designed for flexibility.
Role in this project:
ML Engineer Contributions:5 reviews, 8 commits, 6 PRs in 3 months
Contributions summary:Hadi contributed to the Flax library, focusing on improving documentation and example code. Their work included adding FAQs and an annotated MNIST example notebook for ease of use. The user also addressed bugs in the testing framework and example code, specifically fixing typos and model instantiation issues. Finally, they corrected errors in the model creation process within the flax_intro notebook and the moco example.
jaxneural-network