PyTorch Conference + Bay Area AI Meetup
PyTorch Conference + Bay Area AI Meetup was a developer community event in San Francisco.
Talks & recordings
4 connected sessions
AI for Science with Neural Operators
Applying AI to science problems using neural operators, which generalize traditional deep learning to learn mappings between function spaces. Applications include weather forecasting and solving partial differential equations, running orders of magnitude faster than traditional simulators.
Efficiently Serving LLMs at Scale
Challenges of serving LLMs efficiently in highly concurrent, multi-user contexts, including continuous batching of heterogeneous requests and careful manipulation of tensors with PyTorch.
Tensor and 2D Parallelism in PyTorch
DTensor and Tensor Parallel APIs in PyTorch Distributed, including Sequence Parallel and contributions to TCPStore and ProcessGroup for large training clusters.
What's New for Python Developer Infrastructure in PyTorch
Updates on the state of the world for PyTorch Developer Infrastructure including future projects, Data- and Model-Parallel training support for large models, and automating PyTorch's package release infrastructure.
Connections
5 relationships