- 09 Feb, 2026 1 commit
-
-
wujl5 authored
-
- 05 Feb, 2026 1 commit
-
-
wanglong3 authored
feat: support moe sum when topk==9 bugfix: Fix mtp model load error when eable shared experts fusion.
-
- 22 Jan, 2026 1 commit
-
-
王敏 authored
-
- 21 Jan, 2026 1 commit
-
-
laibao authored
新增 VLLM_USE_FUSED_RMS_ROPE 分支,走 fused 路径 注册 torch.ops.vllm.rms_rotary_embedding_fuse(direct_register_custom_op) cos_sin_cache 自动转 device/dtype 并缓存,避免每次重复拷贝
-
- 16 Jan, 2026 2 commits
-
-
zhuwenwen authored
add VLLM_USE_FUSED_CACHE_QUANT_BMM_MLA to use fused rmsnorm + contiguous + rope(for dpsk-v3) + concat_and_cache_mla + q quant, control bmm(todo) + cat +mla (fp8)
-
laibao authored
新增环境变量 VLLM_MOE_ROUTER_CAPTURE / DIR / RANK / MAX_LAYERS / NUM_TOKENS_* 用于开关与过滤控制 新增 router_capture.py,支持按 num_tokens 分桶抓取 router logits 并落盘 在 qwen3_moe 中接入抓取逻辑,默认关闭,仅在开启时记录 固定 skip_profile / skip_stack_funcs 为默认启用,避免抓到 warmup/profile 形状 统一配置入口到 vllm.envs,作为运行时基准
-
- 15 Jan, 2026 2 commits
- 12 Jan, 2026 2 commits
- 08 Jan, 2026 1 commit
-
-
wanglong3 authored
-
- 07 Jan, 2026 1 commit
-
-
wujl5 authored
-
- 05 Jan, 2026 2 commits
- 04 Jan, 2026 1 commit
-
-
wujl5 authored
-
- 23 Dec, 2025 2 commits
- 22 Dec, 2025 3 commits
- 20 Dec, 2025 1 commit
-
-
zhuwenwen authored
-
- 18 Dec, 2025 2 commits
- 17 Dec, 2025 1 commit
-
-
zhuwenwen authored
-
- 16 Dec, 2025 2 commits
- 15 Dec, 2025 1 commit
-
-
王敏 authored
-
- 14 Dec, 2025 1 commit
-
-
laibao authored
- 新增rms_rotary_embedding_fuse自定义操作 - 添加内核配置文件E=160,N=320 - 通过VLLM_USE_FUSED_RMS_ROPE环境变量控制融合路径
-
- 10 Dec, 2025 1 commit
-
-
王敏 authored
-
- 08 Dec, 2025 1 commit
-
-
王敏 authored
-
- 04 Dec, 2025 1 commit
-
-
zhuwenwen authored
add VLLM_USE_LIGHTOP_RMS_ROPE_CONCAT when use USE_FUSED_RMS_QUANT and USE_FUSED_CUSTOM_ALL_REDUCE_RMS_QUANT
-
- 02 Dec, 2025 2 commits
- 01 Dec, 2025 1 commit
-
-
zhuwenwen authored
-
- 26 Nov, 2025 1 commit
-
-
wujl5 authored
-
- 20 Nov, 2025 2 commits
- 18 Nov, 2025 1 commit
-
-
wujl5 authored
-
- 13 Nov, 2025 2 commits