All pull requests

Pull requests

• Update MI355X DSV4 image and C1 mono kernel / 更新 MI355X DSV4 镜像及 C1 mono kernel

  #4602·Duyi-Wang opened Oct 28, 2026Collaborator·

  ·
  1

• chore: bump srt-slurm to v2.65.2 / chore：将 srt-slurm 升级到 v2.65.2

  #4601·weireweire opened Oct 28, 2026Collaborator·6 of 9 tasks·

  ·
  2

• [AgentX] MiniMax-M3.1-Flash GB300 SGLang: store DSpark draft KV in FP8 / [AgentX] MiniMax-M3.1-Flash GB300 SGLang：DSpark draft KV 改用 FP8

  full-sweep-fail-fast
  #4600·zijiexia opened Oct 28, 2026Collaborator·

  ·
  2

• [AgentX] MiniMax-M3.1-Flash GB200 SGLang: store DSpark draft KV in FP8 / [AgentX] MiniMax-M3.1-Flash GB200 SGLang：DSpark draft KV 改用 FP8

  full-sweep-fail-fast
  #4599·zijiexia opened Oct 28, 2026Collaborator·

  ·
  3

• [AgentX] MiniMax-M3.1-Flash B200 SGLang: store DSpark draft KV in FP8 / [AgentX] MiniMax-M3.1-Flash B200 SGLang：DSpark draft KV 改用 FP8

  full-sweep-fail-fast
  #4598·zijiexia opened Oct 28, 2026Collaborator·

  ·
  3

• [Klaud Cold] Update DeepSeek-V4.1-Flash FP4 H200 SGLang AgentX DSpark image to nightly-dev-cu13-20261028-586355a8 / [Klaud Cold] 将 DeepSeek-V4.1-Flash FP4 H200 SGLang AgentX DSpark 镜像更新至 nightly-dev-cu13-20261028-586355a8

  full-sweep-fail-fastklaud-parked
  #4597·Klaud-Cold opened Oct 28, 2026Collaborator·

  ·
  6

• [AMD][MI35X] 1028 DSV4 sglang mtp agentic

  agentxAMDnon-canary-full-sweep-enabled
  #4596·1am9trash opened Oct 28, 2026Collaborator·

  ·

• [NVIDIA][AgentX] Add GB300 DSV4.1-Flash SGLang DEP16 disagg points at c7480 with prefill HiCache / [NVIDIA][AgentX] 新增 GB300 DSV4.1-Flash SGLang DEP16 分离式 c7480 点（prefill HiCache）

  #4595·weireweire opened Oct 28, 2026Collaborator·

  ·
  2

• CollectiveX: restore nccl-ep zero-copy under CUDA graphs and FP8 in the default matrix / CollectiveX：恢复 nccl-ep 在 CUDA graph 下的零拷贝，并让 FP8 重新进入默认矩阵

  #4594·Oseltamivir opened Oct 28, 2026Collaborator·

  ·

• [AMD][AgentX] Add GLM-5.3 MXFP4 MI355X SGLang MTP AgentX recipe

  full-sweep-fail-fast
  #4593·xiaofei-zheng opened Oct 28, 2026Collaborator·6 of 11 tasks·

  ·
  6

• CollectiveX: ask InferenceX-app to ingest results that land on main / CollectiveX：结果合入 main 时通知 InferenceX-app 导入

  #4592·Oseltamivir opened Oct 28, 2026Collaborator·

  ·
  2

• [Klaud Cold] Update DeepSeek-V4.1-Flash FP4 GB200 SGLang AgentX DSpark image to nightly-dev-cu13-20261027-5884e4fe / [Klaud Cold] 将 DeepSeek-V4.1-Flash FP4 GB200 SGLang AgentX DSpark 镜像更新至 nightly-dev-cu13-20261027-5884e4fe

  full-sweep-fail-fastklaud-finished
  #4591·Klaud-Cold opened Oct 28, 2026Collaborator·

  ·
  11

• [Klaud Cold] Update Muse-Spark-Open FP4 GB200 SGLang AgentX DFlash image to nightly-dev-20261027-5884e4fe / [Klaud Cold] 将 Muse-Spark-Open FP4 GB200 SGLang AgentX DFlash 镜像更新为 nightly-dev-20261027-5884e4fe

  full-sweep-fail-fastklaud-finished
  #4590·Klaud-Cold opened Oct 28, 2026Collaborator·

  ·
  8

• feat(ci): add report-only AgentX acceptance gate / feat(ci): 添加仅作报告的 AgentX 接受长度门禁

  #4586·faradawn opened Oct 27, 2026·

  ·
  1

• Add B300 Dynamo vLLM DSpark recipes on nightly / 新增基于 nightly 的 B300 Dynamo vLLM DSpark 配置

  non-canary-full-sweep-enabled
  #4585·RohitNagraj opened Oct 27, 2026Collaborator·7 of 11 tasks·

  ·
  8

• [Klaud Cold] Update DeepSeek-V4-Pro FP4 GB300 Dynamo-TRT AgentX MTP image to TensorRT-LLM 1.4.0rc1 / [Klaud Cold] 将 DeepSeek-V4-Pro FP4 GB300 Dynamo-TRT AgentX MTP 镜像更新至 TensorRT-LLM 1.4.0rc1

  full-sweep-fail-fastklaud-parked
  #4583·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  26

• CollectiveX: publish the hand-run mi300x and mi325x results / CollectiveX：发布手动运行的 mi300x 与 mi325x 结果

  #4581·Oseltamivir opened Oct 27, 2026Collaborator·

  ·

• [Klaud Cold] Update Qwen3.5 FP4 MI355X SGLang AgentX MTP image to v0.5.22 / [Klaud Cold] 将 Qwen3.5 FP4 MI355X SGLang AgentX MTP 镜像更新至 v0.5.22

  full-sweep-fail-fastklaud-parked
  #4579·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  7

• [Klaud Cold] Update DeepSeek-V4-Pro FP4 MI355X vLLM AgentX MTP image to nightly-5d35ce39 / [Klaud Cold] 将 DeepSeek-V4-Pro FP4 MI355X vLLM AgentX MTP 镜像更新至 nightly-5d35ce39

  full-sweep-fail-fastklaud-finished
  #4578·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  8

• [Klaud Cold] Update DeepSeek-V4-Pro FP4 MI355X ATOM AgentX DSpark image to nightly_202610261558 / [Klaud Cold] 将 DeepSeek-V4-Pro FP4 MI355X ATOM AgentX DSpark 镜像更新至 nightly_202610261558

  full-sweep-fail-fastklaud-parked
  #4577·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  12

• [Klaud Cold] Update GLM-5.2 FP4 B200 SGLang AgentX MTP image to nightly-dev-cu13-20261027-5884e4fe / [Klaud Cold] 将 GLM-5.2 FP4 B200 SGLang AgentX MTP 镜像更新至 nightly-dev-cu13-20261027-5884e4fe

  full-sweep-fail-fastklaud-finished
  #4573·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  9

• [Klaud Cold] Update Muse-Spark-Open FP4 B200 SGLang AgentX DFlash image to nightly-dev-20261027-5884e4fe / [Klaud Cold] 将 Muse-Spark-Open FP4 B200 SGLang AgentX DFlash 镜像更新至 nightly-dev-20261027-5884e4fe

  full-sweep-fail-fastklaud-finished
  #4572·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  9

• [Klaud Cold] Update DeepSeek-V4.1-Flash FP4 H100 vLLM AgentX DSpark image to nightly-00b6c4c / [Klaud Cold] 将 DeepSeek-V4.1-Flash FP4 H100 vLLM AgentX DSpark 镜像更新至 nightly-00b6c4c

  full-sweep-fail-fastklaud-finished
  #4569·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  9

• Refactor/vng benchmark

  #4564·tandao0909 opened Oct 27, 2026·

  ·

• [Klaud Cold] Update Muse-Spark-Open FP4 GB300 Dynamo-SGLang AgentX DFlash image to nightly-dev-20261026-8252e3af / [Klaud Cold] 将 Muse-Spark-Open FP4 GB300 Dynamo-SGLang AgentX DFlash 镜像更新到 nightly-dev-20261026-8252e3af

  full-sweep-fail-fastklaud-finished
  #4561·Klaud-Cold opened Oct 27, 2026Collaborator·

  ·
  10