cpu-torch-latest

Workflow runs for cpu-torch-latest

• [AutoSP] Align cross entropy targets after an in-model causal label shift

  requires action with the application:
  cpu-torch-latest#11199:Pull request#8811synchronizebyjinyouzhi
  jinyouzhi:autosp-label-shift

  Oct 26, 2026Action required
  More actions for workflow run 11199

• Fix lost TP all-reduce results for non-contiguous tensors

  requires action with the application:
  cpu-torch-latest#11198:Pull request#8837synchronizebyjinyouzhi
  jinyouzhi:fix-tp-contiguous-allreduce

  Oct 26, 2026Action required
  More actions for workflow run 11198

• XPU: register pinned host memory via Level Zero ext_memmap

  completed successfully:
  cpu-torch-latest#11197:Pull request#8887synchronizebydelock
  delock:gma/xpu-pin-memory-l0

  Oct 26, 202633m 40s
  More actions for workflow run 11197

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11196:Merge group checks requested
  Oct 26, 202631m 55s
  More actions for workflow run 11196

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11195:Merge group checks requested
  Oct 26, 202635m 28s
  More actions for workflow run 11195

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11194:Merge group checks requested
  Oct 26, 202632m 51s
  More actions for workflow run 11194

• Fix performance regression caused by comms logger and bug in comms_logger.enabled

  completed successfully:
  cpu-torch-latest#11193:Pull request#8909openedbyhwchen2017
  hongwei/fix_log_timer

  Oct 26, 202631m 35s
  More actions for workflow run 11193

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11192:Merge group checks requested
  Oct 26, 202631m 56s
  More actions for workflow run 11192

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11191:Merge group checks requested
  Oct 26, 202625m 14s
  More actions for workflow run 11191

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11190:Merge group checks requested
  Oct 26, 202628m 24s
  More actions for workflow run 11190

• Fix ZeRO-3 NVMe module-only checkpoint weight refresh

  requires action with the application:
  cpu-torch-latest#11189:Pull request#8908openedbyLevius-Fubuki
  Levius-Fubuki:fix/zero3-nvme-module-only-refresh

  Oct 26, 2026Action required
  More actions for workflow run 11189

• Emit intra_op_parallelism and JSON booleans from ds_nvme_tune

  requires action with the application:
  cpu-torch-latest#11188:Pull request#8907openedbySonica-B
  Sonica-B:fix-nvme-tune-aio-config-keys

  Oct 26, 2026Action required
  More actions for workflow run 11188

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11187:Scheduled
  master

  Oct 26, 202631m 23s
  More actions for workflow run 11187

• Skip accelerator sync for unprofiled comm ops

  completed successfully:
  cpu-torch-latest#11186:Pull request#8795synchronizebyguptaishaan
  guptaishaan:fix-8756

  Oct 26, 202628m 3s
  More actions for workflow run 11186

• fix: inference norm missing threadsPerGroup==32 dispatch + ROCm 7.2.2 bf16 ABI

  completed successfully:
  cpu-torch-latest#11185:Pull request#8906openedbyumechand-amd
  umechand-amd:fix/inference-norm-dispatch-and-rocm-bf16-abi

  Oct 25, 202630m 49s
  More actions for workflow run 11185

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11184:Merge group checks requested
  Oct 25, 202631m 51s
  More actions for workflow run 11184

• ZeRO-3: reuse the weight partition as the master copy when it already has the master dtype

  completed successfully:
  cpu-torch-latest#11183:Pull request#8842synchronizebypengdurice
  pengdurice:peng-fix-zero3-master-copy-device-v1

  Oct 25, 202631m 49s
  More actions for workflow run 11183

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11182:Merge group checks requested
  Oct 25, 202633m 25s
  More actions for workflow run 11182

• cpu-torch-latest

  completed successfully:
  cpu-torch-latest#11181:Merge group checks requested
  Oct 25, 202628m 31s
  More actions for workflow run 11181

• Honor fp16 hysteresis in the ZeRO stage 0 fp16 optimizers

  completed successfully:
  cpu-torch-latest#11180:Pull request#8647synchronizebypengdurice
  vineethsaivs:fix-stage0-fp16-hysteresis-20260923

  Oct 25, 202632m 52s
  More actions for workflow run 11180

• Forward the per-head Muon tag at every muon_update call site

  failed:
  cpu-torch-latest#11179:Pull request#8612synchronizebypengdurice
  vineethsaivs:fix-muon-per-head-tags-20260920

  Oct 25, 202630m 37s
  More actions for workflow run 11179

• [AutoEP] Ensure consistent order of partition groups in all-gather when ZeRO-3

  completed successfully:
  cpu-torch-latest#11178:Pull request#8876synchronizebypengdurice
  jinyouzhi:zero3-gather-group-order

  Oct 25, 202627m 33s
  More actions for workflow run 11178

• [Feature] Extend affine live-state transfer to CPU-offloaded optimizer storage

  requires action with the application:
  cpu-torch-latest#11177:Pull request#8905synchronizebyArjunPakhan
  ArjunPakhan:feature/cpu-offload-live-transfer

  Oct 25, 2026Action required
  More actions for workflow run 11177

• [Feature] Extend affine live-state transfer to CPU-offloaded optimizer storage

  requires action with the application:
  cpu-torch-latest#11176:Pull request#8905synchronizebyArjunPakhan
  ArjunPakhan:feature/cpu-offload-live-transfer

  Oct 25, 2026-1s
  More actions for workflow run 11176

• [Feature] Extend affine live-state transfer to CPU-offloaded optimizer storage

  requires action with the application:
  cpu-torch-latest#11175:Pull request#8905openedbyArjunPakhan
  ArjunPakhan:feature/cpu-offload-live-transfer

  Oct 25, 2026Action required
  More actions for workflow run 11175