[qti-mmadhava]

Summary

• Split the legacy combined base.model_class selector into:
  • base.model_family for architecture and export-path
    selection.
  • base.model_id for specific model selection, checkpoint download and artifact
    naming.
• Add model identity resolution and validation:
  • Infer a family from known model IDs.
  • model_family and model_id are mutually exclusive.
  • base.params required for every model_id, including legacy model_class that resolves to an id.
  • Reject mixing legacy model_class with either new field.
• Update Llama export, CLI parsing, DictConfig compatibility,
  model loading, and documentation to use the new identity fields.
• Add coverage for config resolution, CLI compatibility, model
  routing, checkpoint download selection, checkpoint override
  behavior, and legacy OmegaConf input.
• Default artifact names use model_id, or model_family when no model_id is selected. Legacy model_class=static_llama now defaults to llama3_2.pte instead of static_llama.pte.

This is a follow up for RFC #20402

Test plan

  python -m unittest \
    executorch.examples.models.llama.config.test_llm_config \
    executorch.examples.models.llama.tests.test_export_llama_lib

  python -m extension.llm.export.export_llm \
    base.model_id=qwen3_0_6b \
    base.params=examples/models/qwen3/config/0_6b_config.json \
    model.use_kv_cache=false \
    export.max_seq_length=4 \
    export.max_context_length=4 \
    export.output_dir=/tmp \
    export.output_name=qwen3_model_id

 python -m extension.llm.export.export_llm \
   base.model_family=llama3_2 \
   base.params=/path/to/Llama-3.2-1B-Instruct/original/params.json \
   base.checkpoint=/path/to/Llama-3.2-1B-Instruct/original/consolidated.00.pth \
   model.use_kv_cache=false \
   export.max_seq_length=4 \
   export.max_context_length=4 \
   export.output_dir=/tmp \
   export.output_name=llama3_2_1b_family

cc @psiddh @JacobSzwejbka @digantdesai

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24762

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

⚠️ 28 Awaiting Approval

As of commit e221b7a with merge base f2564fb ():

AWAITING APPROVAL - The following workflows need approval before CI can run:

• Build Aarch64 Linux CUDA Wheels (gh)
• Build Aarch64 Linux Native Wheels (gh)
• Build Aarch64 Linux Wheels (gh)
• Build documentation (gh)
• Build Linux CUDA Wheels (gh)
• Build Linux Native Wheels (gh)
• Build Linux Wheels (gh)
• Build macOS Wheels (gh)
• Build Presets (gh)
• Build Windows Wheels (gh)
• Cadence Build & Test (gh)
• Lint (gh)
• MLX (gh)
• periodic (gh)
• pull (gh)
• Test ARM Backend (gh)
• Test CoreML Backend (gh)
• Test Cortex-M Backend (gh)
• Test CPU Backend (gh)
• Test Metal Backend (gh)
• Test Native Engines (gh)
• Test NXP Backend (gh)
• Test QNN Backend (gh)
• Test QNN Windows MSVC build (gh)
• Test Vulkan Backend (gh)
• Test Vulkan Backend (specialized runners) (gh)
• Test WebGPU Backend (gh)
• Test XNNPACK Backend (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.