software

This is an overview of the most interesting software I've written in the past few years both for research and for fun.

• Research code releases
• Open-source contributions
• Paper reimplementations
• Libraries
• Pedagogical implementations
• Miscellaneous

Research code releases

• robopianist:
• xirl:
• x-magical:
• form2fit:

Open-source contributions

I actively contribute to open-source projects:

• MuJoCo: A general purpose physics simulator
  • My PRs here
  • 
• dm_control: A software stack for physics-based simulation and reinforcement learning
  • My PRs here
  • 
• MuJoCo WASM
  • MY PRs here
  • 

Paper reimplementations

These are machine learning paper reimplementations:

• torchnca: A PyTorch implementation of Neighbourhood Components Analysis
  • 
• ibc: A PyTorch implementation of Implicit Behavioral Cloning
  • 
• recurrent-visual-attention: A PyTorch implementation of Recurrent Models of Visual Attention
  • 
• spatial-transformer-networks: A TensorFlow implementation of Spatial Transformer Networks
  • 

Libraries

Software and libraries that I've written for my research and for fun:

• mujoco_menagerie: A collection of high-quality models for the MuJoCo physics engine, curated by DeepMind
  • 
• clip-playground: An ever-growing playground of notebooks showcasing CLIP's impressive zero-shot capabilities
  • 
• nanorl: A tiny reinforcement learning codebase for continuous control, built on top of JAX
  • 
• dm_env_wrappers: Standalone library of frequently-used wrappers for dm_env environments
  • 
• obj2mjcf: A CLI for processing composite Wavefront OBJ files for use in MuJoCo
  • 
• mujoco_scanned_objects: MuJoCo Models for Google's Scanned Objects Dataset
  • 
• torchkit: A lightweight library containing PyTorch utilities useful for day-to-day research
  • 

Pedagogical implementations

Purely pedagogical implementations of algorithms and concepts:

• learn-linalg: Reinforcing my linear algebra and numerical analysis by reimplementing basic, fundamental algorithms in Python
  • 
• learn-blur: Learning about various blurring filters and how to optimize them by exploiting caching and separability
  • 
• volumetric-fusion: An implementation of Volumetric TSDF fusion in pure numpy for pedagogical purposes
  • 
• learn-ransac: Learning about the different uses of RANSAC: Ransom Sample Consensus
  • 

Miscellaneous

Miscellaneous other repositories:

• pytorch-goodies: A collection of PyTorch code snippets
  • 
• mujoco_tips_and_tricks: A collection of tips and tricks for working with MuJoCo
  •