All pull requests

Pull requests

• [AMD][MI355X] kimik3-fp4-mi355x-vllm-agentic-mtp a4w4 MoE + lmcache 0.5.6

  agentxAMD
  #3447·seungrokj opened Sep 26, 2026Collaborator·

  ·
  1

• [AMD] Update GLM-5.2 MI355X image to 20260924 daily / 更新 GLM-5.2 MI355X 镜像至 20260924 daily

  agentxAMDfull-sweep-enabled
  #3446·jiejingzhangamd opened Sep 25, 2026Collaborator·

  ·
  2

• TileRT: apply srt-slurm TileRT support as a patch on the pinned submodule

  #3445·cquil11 opened Sep 25, 2026Collaborator·

  ·
  1

• [Klaud Cold] Update qwen3.5-fp4-b300-sglang (+mtp) SGLang image to nightly-dev-cu13-20260925-8ca82118

  full-sweep-enabled
  #3441·functionstackx opened Sep 25, 2026Collaborator·

  ·
  4

• [Klaud Cold] Update qwen3.5-fp4-mi355x-sglang (+mtp) SGLang ROCm image to v0.5.20-rocm720-mi35x-20260924

  full-sweep-enabled
  #3439·functionstackx opened Sep 25, 2026Collaborator·

  ·
  2

• [Klaud Cold] Update qwen3.5-fp8-b200-sglang (+mtp) SGLang image to nightly-dev-cu13-20260925-8ca82118

  full-sweep-enabled
  #3438·functionstackx opened Sep 25, 2026Collaborator·

  ·
  1

• CollectiveX: graph-replay every capturable EP backend with rank alignment and the chained period

  CollectiveX
  #3437·Oseltamivir opened Sep 25, 2026Collaborator·

  ·

• Tune MiniMax-M3 AgentX on B200 and B300 with updated vLLM / 更新 B200 与 B300 上的 MiniMax-M3 AgentX 配置

  full-sweep-fail-fast
  #3435·xinli-sw opened Sep 25, 2026Collaborator·

  ·
  2

• test: validate AMD power exporters on MI300X, MI325X, and MI355X

  #3434·cquil11 opened Sep 25, 2026Collaborator·

  ·
  1

• chore(deps): bump astral-sh/setup-uv from 10.0.1 to 10.1.0 in the github-actions group across 1 directory

  dependenciesgithub_actions
  #3433·dependabot[bot] opened Sep 25, 2026Bot·

  ·

• [AMD][MI35X] 0925 DSV4 sglang mtp agentic

  agentxAMDfull-sweep-enabled
  #3430·1am9trash opened Sep 25, 2026Collaborator·

  ·
  2

• feat(srt): run AgentX on native srt-slurm

  #3428·cquil11 opened Sep 25, 2026Collaborator·

  ·
  6

• [AgentX][Dsv4] SGLang update

  full-sweep-enabled
  #3426·Oasis-Git opened Sep 25, 2026Collaborator·

  ·
  2

• config(minimaxm3): update GB200 Dynamo vLLM AgentX recipes / 更新 GB200 Dynamo vLLM AgentX 配方

  full-sweep-enabled
  #3424·RohitNagraj opened Sep 25, 2026Collaborator·

  ·
  4

• Package Mooncake H100 transfer tracing experiment / 打包 Mooncake H100 传输跟踪实验

  #3422·kimbochen opened Sep 25, 2026Collaborator·

  ·

• [AgentX][SGLang] Dsv4.1 update

  full-sweep-fail-fast
  #3421·Oasis-Git opened Sep 25, 2026Collaborator·

  ·
  4

• config(kimik3): refresh MI355X vLLM image to the 2026-09-24 ROCm 10.0 nightly / 将 Kimi-K3 MI355X vLLM 镜像刷新至 2026-09-24 的 ROCm 10.0 nightly

  agentxAMDfull-sweep-enabled
  #3419·Fangzhou-Ai opened Sep 25, 2026Collaborator·

  ·
  7

• config: add GB200 DSpark AgentX recipes / 添加 GB200 DSpark AgentX 配置

  full-sweep-enabled
  #3416·xinli-sw opened Sep 24, 2026Collaborator·

  ·
  7

• feat(agentx): bump Kimi-K3 FP4 MI355X ATOM image to 0924 and track recipe

  agentxAMDfull-sweep-enabled
  #3407·gbyu-amd opened Sep 24, 2026Collaborator·

  ·
  5

• [dnm]feat: add Kimi-K3 B300 Mooncake EFA recipes / 新增 Kimi-K3 B300 Mooncake EFA 配置

  full-sweep-enabled
  #3405·xinli-sw opened Sep 24, 2026Collaborator·

  ·
  11

• [GLM-5.2 GB300] require PowerX and isolate HiCache load waits / 启用功耗并隔离 HiCache 加载等待

  #3402·edwingao28 opened Sep 23, 2026Collaborator·

  ·
  2

• [GLM-5.2 GB200] require PowerX for draft flag-off reruns / 关闭草稿量化并要求功耗验收

  #3401·edwingao28 opened Sep 23, 2026Collaborator·

  ·
  2

• [GLM-5.2 B200] require PowerX for draft flag-off reruns / 关闭草稿量化并要求功耗验收

  #3400·edwingao28 opened Sep 23, 2026Collaborator·

  ·
  2

• Update GB300 DeepSeek V4.1 Flash vLLM nightly / 更新 GB300 DeepSeek V4.1 Flash vLLM nightly

  full-sweep-fail-fast
  #3396·xinli-sw opened Sep 23, 2026Collaborator·

  ·
  5

• Update GB200 DeepSeek V4.1 Flash vLLM nightly / 更新 GB200 DeepSeek V4.1 Flash vLLM nightly

  full-sweep-fail-fast
  #3395·xinli-sw opened Sep 23, 2026Collaborator·

  ·
  7