Name: Jeff Johnson
Type: User
Company: Meta Fundamental AI Research (FAIR)
Bio: SIMD + GPU + FPGA + ASIC stuff for AI/ML. I wrote the original PyTorch GPU backend, GPU Faiss, and many other GPU things broadly in use across the industry.
Location: Jackson, WY
Blog: https://www.linkedin.com/in/jeff-johnson-698a822/
Jeff Johnson's Projects
A CUDA backend for Torch7
GPU implementation of a fast generalized ANS (asymmetric numeral system) entropy encoder and decoder, with extensions for lossless compression of numerical and other data types in HPC/ML applications.
A library for efficient similarity search and clustering of dense vectors.
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Threads for Lua and LuaJIT. Transparent exchange of data between threads is allowed thanks to torch serialization.