CI pipeline

Workflow runs for CI pipeline

• Skip nightly Slack notification on the public mirror

  skipped:
  CI pipeline#2619:Commit8454d8cpushedbyaimet-bot
  develop

  Oct 14, 20262s
  More actions for workflow run 2619

• Fix nearest-even rounding of prequantized ONNX constants

  requires action with the application:
  CI pipeline#2618:Pull request#4125openedbyvera-rubin
  vera-rubin:fix/prequantize-rounding

  Oct 14, 2026Action required
  More actions for workflow run 2618

• Enable subgraph quantization for ONNX Quantsim in GenAI Lab

  skipped:
  CI pipeline#2617:Commit594ebe8pushedbyaimet-bot
  develop

  Oct 13, 20266s
  More actions for workflow run 2617

• Use configure_llm API in GenAILab

  skipped:
  CI pipeline#2616:Commitbbb2c97pushedbyaimet-bot
  develop

  Oct 13, 20261s
  More actions for workflow run 2616

• Support Scan in AdaScale's ONNX to torch conversion

  skipped:
  CI pipeline#2615:Commite4f5c5cpushedbyaimet-bot
  develop

  Oct 13, 202610s
  More actions for workflow run 2615

• Extract SeqMSE subgraphs with onnx-ir to support Scan models

  skipped:
  CI pipeline#2614:Commit96708bepushedbyaimet-bot
  develop

  Oct 13, 202611s
  More actions for workflow run 2614

• Remove ByteSize() gate for externalizing tensors (#7937)

  skipped:
  CI pipeline#2613:Commitfe922f8pushedbyaimet-bot
  develop

  Oct 13, 20261s
  More actions for workflow run 2613

• Use uv-managed Python for Windows x64 CI envs (#7968)

  skipped:
  CI pipeline#2612:Commitb7bc045pushedbyaimet-bot
  develop

  Oct 12, 20268s
  More actions for workflow run 2612

• Clarify nightly Slack notifications (#7967)

  skipped:
  CI pipeline#2611:Commit1b71932pushedbyaimet-bot
  develop

  Oct 12, 20262s
  More actions for workflow run 2611

• Refactor _finalize_graph_changes() to use onnx_ir graph (#7867)

  skipped:
  CI pipeline#2610:Commitb2705cfpushedbyaimet-bot
  develop

  Oct 12, 20261s
  More actions for workflow run 2610

• Preserve runtime reduction axes in AdaScale ONNX conversion

  failed:
  CI pipeline#2609:Pull request#4123openedbytaking-lying-flat
  taking-lying-flat:fix/adascale-reduction-axes

  Oct 11, 20264d 17h 40m 15s
  More actions for workflow run 2609

• Use parent forward inputs for mixed attention block sampling

  requires action with the application:
  CI pipeline#2608:Pull request#4122openedbytaking-lying-flat
  taking-lying-flat:fix/blockwise-sampler-layer-kwargs

  Oct 11, 2026Action required
  More actions for workflow run 2608

• Preserve zero-point shift in SeqMSE candidate encodings

  failed:
  CI pipeline#2607:Pull request#4121openedbytaking-lying-flat
  taking-lying-flat:fix/seqmse-zero-point-shift

  Oct 11, 20264d 22h 52m 11s
  More actions for workflow run 2607

• Fix SpinQuant visual merger rotation compensation

  failed:
  CI pipeline#2606:Pull request#4120openedbytaking-lying-flat
  taking-lying-flat:fix/spinquant-visual-merger-rotation

  Oct 11, 20264d 21h 22m 1s
  More actions for workflow run 2606

• Fix stale ONNX quantization axes when loading encodings

  failed:
  CI pipeline#2605:Pull request#4119openedbytaking-lying-flat
  taking-lying-flat:fix/onnx-load-encoding-axis

  Oct 11, 20264d 20h 25m 30s
  More actions for workflow run 2605

• Fix clipped QDQ scale gradients with nonzero zero_point_shift

  failed:
  CI pipeline#2604:Pull request#4118openedbytaking-lying-flat
  taking-lying-flat:fix/qdq-zero-point-shift-scale-grad

  Oct 10, 20267d 2h 15m 21s
  More actions for workflow run 2604

• Move torch adascale and spinquant out of experimental (#7961)

  skipped:
  CI pipeline#2603:Commit6360e0cpushedbyaimet-bot
  develop

  Oct 10, 20261s
  More actions for workflow run 2603

• Move ONNX adascale, spinquant, llm_topology, llm_configurator out of …

  skipped:
  CI pipeline#2602:Commit94a2319pushedbyaimet-bot
  develop

  Oct 10, 20267s
  More actions for workflow run 2602

• Relax AIMETRegression accuracy-drop limits after dataset change (#7958)

  skipped:
  CI pipeline#2601:Commitf5b5980pushedbyaimet-bot
  develop

  Oct 9, 20261s
  More actions for workflow run 2601

• Separate calibration and eval datasets in AIMETRegression (#7948)

  skipped:
  CI pipeline#2600:Commitc8558eapushedbyaimet-bot
  develop

  Oct 9, 20261s
  More actions for workflow run 2600

• Rename GenAILab linear-attention state I/O to conv_state / recurrent_…

  skipped:
  CI pipeline#2599:Commite078c16pushedbyaimet-bot
  develop

  Oct 8, 20267s
  More actions for workflow run 2599

• Support Mamba2 in analyze_llm_topology (#7919)

  skipped:
  CI pipeline#2598:Commitd62b91bpushedbyaimet-bot
  develop

  Oct 8, 20261s
  More actions for workflow run 2598

• Tie Scan boundary quantizers in QuantizationSimModel

  skipped:
  CI pipeline#2597:Commitb1f9dcdpushedbyaimet-bot
  develop

  Oct 7, 20267s
  More actions for workflow run 2597

• Export dynamo backbones with one torch.export mode per torch version …

  skipped:
  CI pipeline#2596:Commit7d13097pushedbyaimet-bot
  develop

  Oct 7, 20262s
  More actions for workflow run 2596

• Support export of quantized Scan bodies in QuantizationSimModel

  skipped:
  CI pipeline#2595:Commit0b7790cpushedbyaimet-bot
  develop

  Oct 7, 20267s
  More actions for workflow run 2595