Winnie Xu

Code for "Infinitely Deep Bayesian Neural Networks with Stochastic Differential Equations"

🌱 personal site

An app that runs a generative model for employing personalized solutions based on irregularly sampled habitual usage data to quitting vaping.

A library with extensible implementations of DPO, KTO, PPO, and other human-aware loss functions (HALOs).

Configs, setups, etc

Personal Site winniexu.ca

An automatic evaluator for instruction-following language models. Human-validated, high-quality, cheap, and fast.

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

Fitting a recognition model (VAE) to do approximate inference on intractable posteriors of probabilistic models using an ELBO estimator.

Berkeley CS285 Deep RL assignments exploring theory and implementations of model free algorithms.

No description provided.

Lab Exercises and Assignments

Rules for building and handling Docker images with Bazel

[NeurIPS'21] Shape As Points: A Differentiable Poisson Solver

Google Research

Official Pytorch and JAX implementation of "Efficient-VDVAE: Less is more"

Generating structured sparsity in neural networks using weight and unit pruning techniques.

hello :)

Pytorch implementation of COIN, a framework for compression with implicit neural representations 🌸

A reository for various configuration files

Official code for Score-Based Generative Modeling through Stochastic Differential Equations (ICLR 2021, Oral)

PyTorch implementation of soft actor critic

Google Foo-Bar Challenge

A toolkit to optimize ML models for deployment for Keras and TensorFlow, including quantization and pruning.

Docker + Bazel Runs

An Open Source Machine Learning Framework for Everyone

A system and visualization tool to track and summarize daily social encounters, i.e. a social "FitBIT".

Automatic skeletal muscle fibre masking, segmentation, and analysis with ImageJ integration for confocal microscopy image formats.

A tool that processes language data in real time and applies the expectation maximization algorithm to infer medical diagnoses based on symptoms.

Predicting transcription factor binding sites using next-generation sequencing genomic data and Segway.