PR title

Workflow runs for PR title

• Retain CED encoder endpoints for extended prompts

  requires action with the application:
  PR title#720:Pull request#1054editedbychristopherowen
  christopherowen:feat/prefix-encoder-endpoint

  Nov 6, 2026Action required
  More actions for workflow run 720

• Retain CED encoder endpoints for extended prompts

  requires action with the application:
  PR title#719:Pull request#1054editedbychristopherowen
  christopherowen:feat/prefix-encoder-endpoint

  Nov 5, 2026Action required
  More actions for workflow run 719

• Retain CED encoder endpoints for extended prompts

  requires action with the application:
  PR title#718:Pull request#1054editedbychristopherowen
  christopherowen:feat/prefix-encoder-endpoint

  Nov 5, 2026Action required
  More actions for workflow run 718

• KVarN MLA cache: single-plane folded-latent codec (kvarn_mla_k4_g256) for GLM-5.3-Flash on SM120

  skipped:
  PR title#717:Pull request#1038editedbycoderabbitai[bot]Bot
  JMPSequeira:upstream/kvarn

  Nov 3, 20262s
  More actions for workflow run 717

• KVarN MLA cache: single-plane folded-latent codec (kvarn_mla_k4_g256) for GLM-5.3-Flash on SM120

  skipped:
  PR title#716:Pull request#1038synchronizebyJMPSequeira
  JMPSequeira:upstream/kvarn

  Nov 3, 20268s
  More actions for workflow run 716

• [KV Offload] Backport prefix-cacheable group offloading and align hits for groups left out

  requires action with the application:
  PR title#715:Pull request#1040synchronizebyjavieratomic
  javieratomic:pr/offload-multinode-hybrid

  Nov 2, 2026Action required
  More actions for workflow run 715

• [KV Offload] Mirror CPU offload region slots across nodes

  requires action with the application:
  PR title#714:Pull request#1042synchronizebyjavieratomic
  javieratomic:pr/offload-multinode-mirror

  Nov 2, 2026Action required
  More actions for workflow run 714

• [KV Offload] Bound staged loads by primary tier capacity

  requires action with the application:
  PR title#713:Pull request#1041synchronizebyjavieratomic
  javieratomic:pr/offload-staged-load-bound

  Nov 2, 2026Action required
  More actions for workflow run 713

• [KV Offload] Report failed transfers instead of asserting

  requires action with the application:
  PR title#712:Pull request#1043synchronizebyjavieratomic
  javieratomic:pr/offload-transfer-failures

  Nov 2, 2026Action required
  More actions for workflow run 712

• [KV Offload] Backport prefix-cacheable group offloading and align hits for groups left out

  requires action with the application:
  PR title#711:Pull request#1040synchronizebyjavieratomic
  javieratomic:pr/offload-multinode-hybrid

  Nov 2, 2026Action required
  More actions for workflow run 711

• [KV Offload] Backport prefix-cacheable group offloading and align hits for groups left out

  requires action with the application:
  PR title#710:Pull request#1040editedbyjavieratomic
  javieratomic:pr/offload-multinode-hybrid

  Nov 2, 2026Action required
  More actions for workflow run 710

• [KV Offload] Bound staged loads by primary tier capacity

  requires action with the application:
  PR title#709:Pull request#1041editedbyjavieratomic
  javieratomic:pr/offload-staged-load-bound

  Nov 2, 2026Action required
  More actions for workflow run 709

• [KV Offload] Mirror CPU offload region slots across nodes

  requires action with the application:
  PR title#708:Pull request#1042editedbyjavieratomic
  javieratomic:pr/offload-multinode-mirror

  Nov 2, 2026Action required
  More actions for workflow run 708

• [KV Offload] Report failed transfers instead of asserting

  requires action with the application:
  PR title#707:Pull request#1043editedbyjavieratomic
  javieratomic:pr/offload-transfer-failures

  Nov 2, 2026Action required
  More actions for workflow run 707

• [KV Offload] Add fs_multinode: a filesystem tier for TP across nodes with node-local disks

  requires action with the application:
  PR title#706:Pull request#1051synchronizebyzengjiajun0623
  zengjiajun0623:kv-offload/fs-multinode-tier

  Oct 30, 2026Action required
  More actions for workflow run 706

• [KV Offload] Treat host-only secondary tiers as VMM-safe

  requires action with the application:
  PR title#705:Pull request#1050synchronizebyzengjiajun0623
  zengjiajun0623:kv-offload/vmm-safe-host-tiers

  Oct 30, 2026Action required
  More actions for workflow run 705

• [KV Offload] Gather every worker's output only while transfers are outstanding

  requires action with the application:
  PR title#704:Pull request#1049synchronizebyzengjiajun0623
  zengjiajun0623:kv-offload/gather-all-ranks-on-demand

  Oct 30, 2026Action required
  More actions for workflow run 704

• [KV Offload] Gather every worker's output only while transfers are outstanding

  requires action with the application:
  PR title#703:Pull request#1049synchronizebyzengjiajun0623
  zengjiajun0623:kv-offload/gather-all-ranks-on-demand

  Oct 30, 2026Action required
  More actions for workflow run 703

• Size-classed prefill lanes: keep short work flowing beside long prefills (stacked on #1037)

  requires action with the application:
  PR title#702:Pull request#1048editedbycoderabbitai[bot]Bot
  hughmadden:qos/size-classed-prefill-lanes

  Oct 29, 2026Action required
  More actions for workflow run 702

• csf: read each packed expert weight in one pass (~10x faster expert loading on ZFS)

  requires action with the application:
  PR title#701:Pull request#1047editedbypierloh
  pierloh:perf/csf-whole-tensor-read

  Oct 28, 2026Action required
  More actions for workflow run 701

• csf: read each packed expert weight in one pass (~10x faster expert loading on ZFS)

  requires action with the application:
  PR title#700:Pull request#1047synchronizebypierloh
  pierloh:perf/csf-whole-tensor-read

  Oct 28, 2026Action required
  More actions for workflow run 700

• csf: read each packed expert weight in one pass (~10x faster expert loading on ZFS)

  requires action with the application:
  PR title#699:Pull request#1047synchronizebypierloh
  pierloh:perf/csf-whole-tensor-read

  Oct 28, 2026Action required
  More actions for workflow run 699

• csf: read each packed expert weight in one pass (~10x faster expert loading on ZFS)

  requires action with the application:
  PR title#698:Pull request#1047editedbycoderabbitai[bot]Bot
  pierloh:perf/csf-whole-tensor-read

  Oct 28, 2026Action required
  More actions for workflow run 698

• Advisory disk row-cache prefetch for Qwen4Exp PLE and DeepSeek V4.1 Engram, with row-cache metrics

  requires action with the application:
  PR title#697:Pull request#1044editedbyrandomvariable
  randomvariable:ple-prefetch

  Oct 25, 2026Action required
  More actions for workflow run 697

• Advisory disk row-cache prefetch for Qwen4Exp PLE and DeepSeek V4.1 Engram, with row-cache metrics

  requires action with the application:
  PR title#696:Pull request#1044editedbycoderabbitai[bot]Bot
  randomvariable:ple-prefetch

  Oct 25, 2026Action required
  More actions for workflow run 696