v0.26.0
# vLLM v0.26.0 Release Notes ## Highlights This release features 411 commits from 212 contributors (61 new)! * **New Inkling model family** with a full support stack: base modeling (#48799), piecewise CUDA graph support (#48822), Hopper FA4 relative attention (#48858), MTP=1 speculative decoding (#48869), LoRA (#48884), and standard ModelOpt NVFP4 quantization (#48990). * **DeepSeek-V4 performance push** across vendors: a specialized routing kernel (2.94% E2E TPOT, #48660), `fused_topk_bias` (1.5–2x kernel, #47463), and redundant repeat/copy removal (1.8% E2E TPOT, #48137), plus ROCm two-stage compressor for HCA prefill (#47718), sparse decod
关键事实
# vLLM v0.26.0 Release Notes ## Highlights This release features 411 commits from 212 contributors (61 new)! * **New Inkling model family** with a full support stack: base modeling (#48799), piecewise CUDA graph support (#48822), Hopper FA4 relative attention (#48858), MTP=1 speculative decoding (#48869), LoRA (#48884), and standard ModelOpt NVFP4 quantization (#48990). * **DeepSeek-V4 performance push** across vendors: a specialized routing kernel (2.94% E2E TPOT, #48660), `fused_topk_bias` (1.5–2x kernel, #47463), and redundant repeat/copy removal (1.8% E2E TPOT, #48137), plus ROCm two-stage compressor for HCA prefill (#47718), sparse decod
阅读原文
自动收录自 vLLM 发布记录 的公开订阅信息；请以原始出处为准。
本站不保存或展示该来源的全文。
前往原始出处阅读 ↗