name: Self-hosted runner scale set (AMD mi325 scheduled CI caller)

# Note: For every job in this workflow, the name of the runner scale set is finalized in the runner yaml i.e. huggingface/hf-workflows/.github/workflows/transformers_amd_ci_scheduled_arc_scale_set.yaml
# For example, 1gpu scale set: amd-mi325-ci-1gpu
#              2gpu scale set: amd-mi325-ci-2gpu
# Important: Do not pin the reusable workflow ref to a SHA. AMD runner groups only route jobs for
# workflows referenced at @main; a pinned SHA causes jobs to wait for a runner until the 24h timeout.

on:
  workflow_run:
    workflows: ["Self-hosted runner (AMD scheduled CI caller)"]
    branches: ["main"]
    types: [completed]
  push:
    branches:
      - run_amd_scheduled_ci_caller*

permissions:
  contents: read

jobs:
  model-ci:
    name: Model CI
    uses: huggingface/hf-workflows/.github/workflows/transformers_amd_ci_scheduled_arc_scale_set.yaml@main
    with:
      job: run_models_gpu
      slack_report_channel: "#amd-hf-ci"
      runner_group: amd-mi325
      docker: huggingface/transformers-pytorch-amd-gpu
      ci_event: Scheduled CI (AMD) - mi325
      report_repo_id: optimum-amd/transformers_daily_ci
      env_file: /etc/podinfo/gha-gpu-isolation-settings
    secrets: inherit

  torch-pipeline:
    name: Torch pipeline CI
    uses: huggingface/hf-workflows/.github/workflows/transformers_amd_ci_scheduled_arc_scale_set.yaml@main
    with:
      job: run_pipelines_torch_gpu
      slack_report_channel: "#amd-hf-ci"
      runner_group: amd-mi325
      docker: huggingface/transformers-pytorch-amd-gpu
      ci_event: Scheduled CI (AMD) - mi325
      report_repo_id: optimum-amd/transformers_daily_ci
      env_file: /etc/podinfo/gha-gpu-isolation-settings
    secrets: inherit

  example-ci:
    name: Example CI
    uses: huggingface/hf-workflows/.github/workflows/transformers_amd_ci_scheduled_arc_scale_set.yaml@main
    with:
      job: run_examples_gpu
      slack_report_channel: "#amd-hf-ci"
      runner_group: amd-mi325
      docker: huggingface/transformers-pytorch-amd-gpu
      ci_event: Scheduled CI (AMD) - mi325
      report_repo_id: optimum-amd/transformers_daily_ci
      env_file: /etc/podinfo/gha-gpu-isolation-settings
    secrets: inherit

  deepspeed-ci:
    name: DeepSpeed CI
    uses: huggingface/hf-workflows/.github/workflows/transformers_amd_ci_scheduled_arc_scale_set.yaml@main
    with:
      job: run_torch_cuda_extensions_gpu
      slack_report_channel: "#amd-hf-ci"
      runner_group: amd-mi325
      docker: huggingface/transformers-pytorch-deepspeed-amd-gpu
      ci_event: Scheduled CI (AMD) - mi325
      report_repo_id: optimum-amd/transformers_daily_ci
      env_file: /etc/podinfo/gha-gpu-isolation-settings
    secrets: inherit