AI Tools Jun 11, 2026
Profiling in PyTorch (Part 2): From nn.Linear to a Fused MLP
Part 2 of Hugging Face's PyTorch profiling series shows how to read torch.profiler traces from nn.Linear through compiled and hand-tuned MLP kernels, and what fuse decisions actually improve.
AI Infrastructure May 14, 2026
Unlocking Asynchronicity in Continuous Batching
Hugging Face shows how asynchronous continuous batching, using CUDA streams, events, and double-buffered slots, lifted GPU utilization from 76 to 99.4 percent and cut generation time by 22 percent.