- 28 Jan, 2026 1 commit
-
-
zhuwenwen authored
fix: 解决原版0消耗chunk-prefill崩溃问题 See merge request dcutoolkit/deeplearing/vllm!393
-
- 27 Jan, 2026 6 commits
- 26 Jan, 2026 2 commits
- 23 Jan, 2026 2 commits
- 22 Jan, 2026 4 commits
- 21 Jan, 2026 9 commits
-
-
zhuwenwen authored
feat(moe/marlin): Marlin W16A16 MoE 自动探测并预打包（去掉手动开关） See merge request dcutoolkit/deeplearing/vllm!382
-
-
zhuwenwen authored
for qwen3, VLLM_USE_FUSED_RMS_ROPE=1 (default)
-
laibao authored
- 移除 VLLM_USE_MARLIN_W16A16_MOE 环境变量 - 初始化阶段基于 lightop 探测并缓存 _marlin_w16a16_moe_enabled，满足条件强制 use_nn_moe=False - 权重加载后按缓存结果一次性 Marlin pack；运行时按 packed 标记走 Marlin fast path
-
zhuwenwen authored
perf(qwen3): 融合 q/k RMSNorm + RoPE See merge request dcutoolkit/deeplearing/vllm!381
-
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
-
zhuwenwen authored
fused_moe_fp8接入lmslim See merge request dcutoolkit/deeplearing/vllm!379
-
SAC_fanth authored
-
zhuwenwen authored
feat: Support w8a8-fp8 GEMM backend. See merge request dcutoolkit/deeplearing/vllm!375
-
- 20 Jan, 2026 2 commits
- 19 Jan, 2026 2 commits
- 17 Jan, 2026 7 commits
- 16 Jan, 2026 5 commits
-
-
zhuwenwen authored
-
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)
-
zhuwenwen authored
-
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
-
zhuwenwen authored
-
