…r-quantization linear backend overrides ( 51204), AutoRound 2/3/5/6/7-bit on CUDA ( 52890), and DeepSelect top-k for the DSA sparse indexer ( 56464). Breaking changes : s

… on stock vLLM 0.27.1 Measured on the plain Intel AutoRound checkpoint without the hybrid conversion. These do not replace the headline numbers, which are the full v2 sta

…ge. - Live chat requests passed with a Qwen3.5 9B AutoRound INT4 model from the same family in both thinking modes. - The official 122B index registered all 785 MTP mappi

…df61d1f7cc)) by AlpinDale - [sync] [Quantization][Autoround][XPU] Add W4A16(moe) / MXFP4(linear/moe) Support ( 47124) ([ 439a390a9 ](https://github.com/dphnAI/sonar/commi

…er v1.3.7 Blackwell-only cleanup release. The two AutoRound INT4 snapshots for the RTX 5090 are gone; NVFP4 is now the sole 5090 path. Ampere/Ada (3090/4090) are unaffect