…_type variant

# Conflicts:
#	src/winml/modelkit/loader/config.py
#	src/winml/modelkit/models/auto.py

…c shapes

- Add missing docstrings / return-type annotations and drop dead noqa directives across qwen3_export_ops.py, qwen3_modeling.py and the transformer-only registration so 'ruff check src/ tests/' (CI lint) passes.
- build/hf.py: re-persist config.json after winml_finalize_quant_config runs, so the saved metadata reflects the actually-applied w8a16 scheme (int8/uint16/symmetry + GQA nodes_to_exclude) rather than the pre-finalize policy dtypes.
- qwen_transformer_only_quant._graph_shapes: treat a non-positive dim_value (symbolic/dynamic axis) as a hard error instead of silently returning a zero-length shape.

…_type-override test

- build_hf_model: look up winml_finalize_quant_config on type(pytorch_model) instead of the instance, and call it with explicit self. Fixes the mypy 'Tensor not callable' error (getattr yields Any) and stops the hook firing on raw HF models / MagicMock test doubles (whose attributes are instance-synthesized), which was serializing a MagicMock into config.json.
- test_resolve_loader_config: replace the obsolete 'never mutated' test with one asserting the intended explicit-model_type override (needed for variants like qwen3_transformer_only).

Relocate the model-specific transformer-only calibration/quant logic out of
models/hf (an export-only package) into a new quant/calibration/ subpackage,
dispatched via a model_type-keyed registry that mirrors COMPOSITE_MODEL_REGISTRY.

- Add quant/calibration/{base,registry}.py: QuantConfigFinalizer protocol +
  register_quant_finalizer / get_quant_finalizer (lazy, torch-free import).
- git mv qwen_transformer_only_quant.py -> quant/calibration/qwen3_transformer_only.py
  and register Qwen3TransformerOnlyQuantFinalizer for 'qwen3_transformer_only'.
- build/hf.py: replace the winml_finalize_quant_config wrapper hook with explicit
  registry dispatch keyed on config.model_type; unregistered types fall back to
  the default DatasetCalibrationReader. Preserve the model_id/_name_or_path
  fallback (now model-agnostic in the build layer).
- Remove the hook from the export wrapper (back to export-only).
- Relocate unit tests to tests/unit/quant/calibration/ and add test_registry.py.

w8a16 scheme unchanged; CPU e2e (quantized-graph + GQA-exclusion + FP-parity)
and 86 build/quant unit tests pass.

…transformer-only into subpackage

The CLI-only _build_hf_pipeline did not pass loader.model_type to
_load_model, so a config requesting qwen3_transformer_only was silently
loaded as native qwen3 and crashed at export (embedding got HalfTensor).
It also skipped the model-type quant finalizer, producing the default
uint8/uint16 minmax scheme instead of the registered int8-sym /
GQA-excluded policy. Both gaps existed only in the CLI path; the library
build_hf_model already handled them. Mirror that logic so winml build
produces the verified w8a16 graph (985 Q / 1294 DQ / 28 GQA / 0
QDQ-touching-GQA) end-to-end.

Also move qwen3_export_ops, qwen3_modeling and qwen_transformer_only into
a models/hf/qwen3/ subpackage and add regression tests for both fixes.

…rt script

Reparent WinMLQwen3TransformerOnlyModel from WinMLDecoderOnlyModel to the
plain WinMLCompositeModel. The decoder-only base wires a generation runtime
from the eager KV signature (past_0_key) in __init__, which the
transformer-only graph (past_keys_0 + symbolic axes) lacks, so from_pretrained
crashed while constructing the handle even though both sub-model ONNX built
fine. The build is export-only, so the plain composite base (which just stores
the built sub-models) is the correct parent.

Add a from_pretrained override that injects model_type=qwen3_transformer_only
for every sub-model, so omitting model_type no longer silently builds the
native (full) qwen3 architecture.

Add scripts/export_qwen3_transformer_only.py to export the prefill + decode
transformer-only pair in one call, with optional --output-dir copy.

Add tests/unit/models/qwen3/test_transformer_only_composite.py covering the
reparent, registry entry, and model_type injection.

# Conflicts:
#	tests/unit/commands/test_build.py

…gistry, non-mutating model_type override

- registry: replace decorator/register API with a plain QUANT_FINALIZERS dict; get_quant_finalizer lazily imports + instantiates (review #1).
- quantizer: resolve+apply the model-type-specific quant policy inside quantize_onnx from config.model_type, a single seam shared by all callers; drop the duplicated dispatch blocks in commands/build.py and build/hf.py (review #2).
- loader: thread an explicit model_type as model_type_override through resolve_task instead of mutating hf_config.model_type, so exporters/patchers keep the architecture's native type while the loader config surfaces the build variant (review #4).

# Conflicts:
#	src/winml/modelkit/config/build.py
#	src/winml/modelkit/models/hf/__init__.py

…ype check