Understand this story
On November 2, 2026, Ziang Li and others from humans& released a GPU-accelerated weight transfer solution, and have open-sourced the related implementation in Miles and SGLang. Under Ampersand's production RL configuration, this solution reduces sampler pause time to under 0.23 seconds and end-to-end weight synchronization time to under 4 seconds. The solution is based on a bitwise XOR mask of incremental weights, combined with GPU parallel compression, two-layer Snappy plus Zstd compression, the Blackwell decompression engine, and Triton kernels for in-place weight updates, and overlaps decompression with SM computation. There are currently no further developments.
AI-generated from public reports
Developments
1 confirmed developments
- Source:English text is being prepared.
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 2
Hacker News Trending
humans& proposes a GPU-accelerated weight transfer solution, reducing the RL weight synchronization pause to 0.23 seconds
Ziang Li and others at humans& have released a GPU-accelerated weight transfer solution that, under the Ampersand production RL configuration, reduces sampler pause time to under 0.23 seconds and end-to-end weight synchronization time to under 4 seconds. The solution is based on a bitwise XOR mask of incremental weights, combined with GPU-parallel compression, two-layer Snappy plus Zstd compression, the Blackwell decompression engine, and Triton kernels for in-place weight updates, and it overlaps decompression with SM computation. The related implementation has been open-sourced in Miles and SGLang.