Skip to content

v0.2.2

Choose a tag to compare

@iMMIQ iMMIQ released this 24 Oct 14:29
· 281 commits to main since this release

CUDA Graph reuse now separates packet shape from request ownership across Qwen, Flash Next and ASR. Ordinary decode, mixed prefill and MTP share a common operand-table contract and native binding code. Replacing request owners updates addresses without capturing another graph of the same shape.

All supported model execution packages are rebuilt for 0.2.2. Model-specific GDN, attention, expert and audio arithmetic is retained. The release also includes native execution packages for Nemotron-3.5-Embed-1B and Qwen3-Reranker-0.6B, using their complete-sequence graph paths.

Assets include the Linux aarch64 engine, SM87 execution packages, matching model metadata, source and license notices, and SHA256SUMS. Existing quantized weight payloads can be reused with the matching metadata and execution package.