GitLab

deepseek-r1-w4a8 mlp/moe调用silu-mul-quant融合

See merge request dcutoolkit/deeplearing/vllm!209

修改P等待问题

See merge request dcutoolkit/deeplearing/vllm!210

fix the performance issue of tbo pd separation

See merge request dcutoolkit/deeplearing/vllm!208

接入新的concat算子，包含decode和prefill，并根据size的不同进行选择

See merge request dcutoolkit/deeplearing/vllm!207

deepseek-r1-w4a8使用rmsquant融合算子及横向融合

See merge request dcutoolkit/deeplearing/vllm!205

V0.9.2 dev pd tbo

See merge request dcutoolkit/deeplearing/vllm!204

[fix]修复eagle 创建cu_num_tokens类型错误问题

See merge request dcutoolkit/deeplearing/vllm!203

update the default values of VLLM_USE_TRITON_CAT and VLLM_USE_LIGHT_OP to True

fix: w4a8 marlin 中 weight重排接入lightop算子

See merge request dcutoolkit/deeplearing/vllm!202