Validate checkpoint selection

Workflow runs for Validate checkpoint selection

• fix(train): select XCCL as the XPU distributed backend instead of NCCL

  completed successfully:
  Validate checkpoint selection#11:Pull request#3786openedbyli-lizhe
  li-lizhe:fix/xpu-dist-backend-xccl

  Oct 21, 20265m 12s
  More actions for workflow run 11

• fix(train): bind distributed training to the active accelerator (#3771)

  completed successfully:
  Validate checkpoint selection#10:Commite5be9c9pushedbyLauraGPT
  main

  Oct 17, 202652s
  More actions for workflow run 10

• fix(train): bind distributed training to the active accelerator

  completed successfully:
  Validate checkpoint selection#9:Pull request#3771synchronizebyli-lizhe
  li-lizhe:fix/distributed-device-agnostic

  Oct 17, 20261m 8s
  More actions for workflow run 9

• docs: fix SenseVoice images and Android SDK guide link (#3775)

  completed successfully:
  Validate checkpoint selection#8:Commitdddd5dcpushedbyLauraGPT
  main

  Oct 16, 20261m 8s
  More actions for workflow run 8

• docs: fix SenseVoice README images and Android SDK guide link

  completed successfully:
  Validate checkpoint selection#7:Pull request#3775openedbypratikgx
  pratikgx:docs-fix-sensevoice-images

  Oct 16, 20261m 8s
  More actions for workflow run 7

• fix(train): bind distributed training to the active accelerator

  completed successfully:
  Validate checkpoint selection#6:Pull request#3771synchronizebyli-lizhe
  li-lizhe:fix/distributed-device-agnostic

  Oct 12, 20261m 9s
  More actions for workflow run 6

• fix(train): bind distributed training to the active accelerator

  completed successfully:
  Validate checkpoint selection#5:Pull request#3771openedbyli-lizhe
  li-lizhe:fix/distributed-device-agnostic

  Oct 11, 20261m 10s
  More actions for workflow run 5

• fix: avoid deep-copying the checkpoint in load_pretrained_model (#3729)

  completed successfully:
  Validate checkpoint selection#4:Commit9e7f6afpushedbyLauraGPT
  main

  Sep 26, 20261m 6s
  More actions for workflow run 4

• fix: avoid deep-copying the checkpoint in load_pretrained_model

  failed:
  Validate checkpoint selection#3:Pull request#3729openedbyyydhYYDH
  yydhYYDH:perf/avoid-deepcopy-in-load-pretrained-model

  Sep 25, 20261d 0h 44m 59s
  More actions for workflow run 3

• Merge pull request #3677: preserve checkpoint metric availability

  completed successfully:
  Validate checkpoint selection#2:Commit3bfa94cpushedbyLauraGPT
  main

  Sep 8, 20261m 4s
  More actions for workflow run 2

• fix(training): exclude unavailable metrics from checkpoint ranking

  completed successfully:
  Validate checkpoint selection#1:Pull request#3677openedbyLauraGPT
  codex/checkpoint-selection-20260909

  Sep 8, 20261m 5s
  More actions for workflow run 1