12月1月2月3月4月5月6月7月8月9月10月11月
周一周三周五
一年内 264223 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.26.0 分支比较 10 提交 »568afb3a13 [CI/Build] Refresh tags before building macOS wheel (#49901)f2654939e6 [ROCm] [Release] [Bugfix] Fix the per commit wheel release pipeline. (#49245)ffd46bfab2 [Bugfix] Register axk1 config to fix A.X-K1 init (#49727)ffd6ee4bcc [CI] Bump PyTorch Compilation Unit Tests timeout to 150 min (#49606)bb26ce8e93 [CI] Increase timeout of pytorch-compilation-unit-tests (#49450)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.25.1 分支比较 2 提交 »752a3a5044 [Bugfix] Guard mixed-dtype allreduce RMSNorm quant fusions (#48330)3c31722d6d [Bugfix] Avoid blocking model launching when no system ffmpeg available for TorchCodec (#47888)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.24.0 分支比较 10 提交 »ee0da84ab9 [KV-Offloading] Fix tensors_per_block stride (#46888)217c64a976 [CI] Raise gsm8k startup timeout for MoE Refactor Qwen3 NVFP4 configs (#46882)cfe8a4d063 [CI] Raise gsm8k startup timeout for Qwen3 NVFP4 trtllm configs (#46881)6d37570a1c Fix P/D with DP Supervisor (#46628)f85a9f112a [Bugfix] FLASHINFER_MLA_SPARSE_SM120 compatibility with GLM-5 NVFP4 (#46506)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.23.0 分支比较 6 提交 »0fc695fc6d [Bugfix][Frontend] Cap fastapi < 0.137 to avoid prometheus-fastapi-instrumentator crash on serve startup91df0fad4d [Bugfix][CPU] Don't build triton-cpu on arm64 release image (#45401)78743ab5bf [Docker] Fix CUTLASS DSL cu13 install order in Dockerfile (#45204)b2d7294b0f [ROCm][Bugfix] Make intermediate_pad TP-aware in rocm_aiter_fused_experts (#44679)741ba421d8 [Bugfix] [DSV4] [ROCm] Pin apache-tvm-ffi version to `0.1.10` (#45169)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.22.1 分支比较 8 提交 »0decac0d96 fix: resolve CUTLASS fmin compatibility for DeepSeek-V4 initfd56c57bde Fix `OlmoHybridForCausalLM` not initialising (#43846)7285178622 [Bugfix] Fix HyperCLOVAX CI failure after upstream removed remote code (#43860)27509c8dde [Bugfix][CI] Normalize NIXL connector CUDA wheel installs (#44266)b284862ea9 [docker] Stop using extra-index-url for flashinfer-jit-cache (#44366)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.21.0 分支比较 10 提交 »ad7125a431 [Bugfix] Fix DeepSeek V4 MTP HC state handling (#42320)9da56fd18b [Bugfix] Add swiglu limits to deepgemm fp8 methods (#41986)800604bf53 [MLA Attention Backend] Add TOKENSPEED_MLA backend for DSR1/Kimi K25 prefill + decode on Blackwell (#41778)75a7914326 pin cutlass-dsl to 4.4.23b581add43 [PD] Fix broken NIXL EP installation (#42542)
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.20.2 分支比较 6 提交 »bc150f5029 [CI] Automate Docker Hub release image publishing (#40415)9bc5a0d664 [Bugfix] Remove invalid deepstack boundary check for Qwen3-VL (#40932)fa8accac62 [Bugfix] Fix failure to allocate KV blocks error (#41282)637495c1dc [Bugfix] Plumb hidden_dim_unpadded through moe_forward fake to fix gpt-oss MXFP4 + torch.compile (#42002)fbd51e3dfc [Bugfix] Fix condition to clear persistent topk so that it can be captured regardless
- frostbyte_neo 推送了仓库 github-featured/vllm-project--vllm 的 releases/v0.20.1-python-from-source 分支比较 4 提交 »7c36acf7fd Fix Python version regex quoting in Dockerfile9f876610de Build Python from source instead of using deadsnakes PPA747e379546 retry 10 times