[ExecuTorch][WebGPU] Flatten landed-op test dirs to test/ops/test_<op>.py

Pull Request resolved: #20435

Pure test-file relocation: moves the already-landed ops' tests from nested `test/ops/<op>/test_<op>.py` to flat `test/ops/test_<op>.py`, matching the ExecuTorch convention (XNNPACK uses flat `test/ops/test_<op>.py`; Vulkan uses flat `test/test_*.py`) and completing the flatten applied to the new ops in the stack below. Drops the per-op `__init__.py`; the parent `test/ops/__init__.py` is kept. Ops: `add`, `rms_norm`, `sdpa` (`test_sdpa` + `test_update_cache`), `dispatch_order`, `quantized_linear`, `embedding_q4gsw`, `rope`, `prepack`.

No behavior change — the test modules and their export/golden functions are unchanged; only their path moves. Every reference to the old paths is updated: the `cases.py` op-test imports (`add`, `rms_norm`), `test/TARGETS` (`test_add` srcs), `test/ops/test_dispatch_order.py`'s internal `rms_norm` import, and the build/CI scripts that import the per-op export functions (`test/test_build_webgpu.sh`, `scripts/test_webgpu_native_ci.sh`).

Nothing required the per-op subdirectory: the codegen framework imports only `cases.py`, the one buck target uses a literal path, and the native-golden scripts import the modules by path — each resolves identically at the flat path.
ghstack-source-id: 397534695
@exported-using-ghexport

@diff-train-skip-merge

Differential Revision: [D109349894](https://our.internmc.facebook.com/intern/diff/D109349894/)

Author: JCNTH

backends/webgpu/README.md

@@ -140,11 +140,10 @@ backends/webgpu/
     ├── test_webgpu_native.cpp     # C++ native test runner
     ├── test_wgsl_codegen.py       # Shader codegen check
     ├── native/                    # C++ operator tests
-    └── ops/                       # Python export tests
-        ├── add/
-        │   └── test_add.py        # add export tests
-        └── rms_norm/
-            └── test_rms_norm.py   # rms_norm export tests
+    └── ops/                       # Python op test suites (flat: test_<op>.py)
+        ├── test_add.py
+        ├── test_rms_norm.py
+        └── ...                    # one test_<op>.py per op
 ```
 
 ## Requirements

backends/webgpu/scripts/test_webgpu_native_ci.sh

@@ -65,36 +65,36 @@ PREPACK_TIED_MODEL="/tmp/webgpu_prepack_tied_const.pte"
 PREPACK_TIED_GOLDEN="/tmp/webgpu_prepack_tied_const_golden.bin"
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.quantized_linear.test_quantized_linear import export_all_quantized_linear_models
+from executorch.backends.webgpu.test.ops.test_quantized_linear import export_all_quantized_linear_models
 export_all_quantized_linear_models('/tmp')
 " || echo "WARN: q4gsw export failed; required configs will FAIL in webgpu_native_test"
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.embedding_q4gsw.test_embedding_q4gsw import export_embedding_q4gsw_model
+from executorch.backends.webgpu.test.ops.test_embedding_q4gsw import export_embedding_q4gsw_model
 export_embedding_q4gsw_model('${EMBEDDING_MODEL}', '${EMBEDDING_GOLDEN}', '${EMBEDDING_INDICES}')
 export_embedding_q4gsw_model('${EMBEDDING_LLAMA1B_MODEL}', '${EMBEDDING_LLAMA1B_GOLDEN}', '${EMBEDDING_LLAMA1B_INDICES}', 'llama1b')
 " || echo "WARN: embedding_q4gsw export failed; embedding configs will FAIL in webgpu_native_test"
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.rope.test_rope import export_rope_model
+from executorch.backends.webgpu.test.ops.test_rope import export_rope_model
 export_rope_model('${ROPE_MODEL}', '${ROPE_XQ_GOLDEN}', '${ROPE_XK_GOLDEN}')
 export_rope_model('${ROPE_DECODE_MODEL}', '${ROPE_DECODE_XQ_GOLDEN}', '${ROPE_DECODE_XK_GOLDEN}', 'decode')
 " || echo "WARN: rope export failed; apply_rotary_emb configs will FAIL in webgpu_native_test"
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.prepack.test_prepack import export_prepack_model, export_prepack_two_const_model, export_prepack_tied_const_model
+from executorch.backends.webgpu.test.ops.test_prepack import export_prepack_model, export_prepack_two_const_model, export_prepack_tied_const_model
 export_prepack_model('${PREPACK_MODEL}', '${PREPACK_GOLDEN}')
 export_prepack_two_const_model('${PREPACK2_MODEL}', '${PREPACK2_GOLDEN}')
 export_prepack_tied_const_model('${PREPACK_TIED_MODEL}', '${PREPACK_TIED_GOLDEN}')
 " || echo "WARN: prepack export failed; prepack configs will FAIL in webgpu_native_test"
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.dispatch_order.test_dispatch_order import export_dispatch_order_cases
+from executorch.backends.webgpu.test.ops.test_dispatch_order import export_dispatch_order_cases
 export_dispatch_order_cases('${DISPATCH_ORDER_DIR}')
 " || { echo "WARN: dispatch_order export failed; skipping dispatch_order native test"; DISPATCH_ORDER_OK=0; }
 
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.sdpa.test_update_cache import (
+from executorch.backends.webgpu.test.ops.test_update_cache import (
     export_update_cache_cases,
     export_update_cache_replay,
     export_update_cache_negative,
@@ -107,7 +107,7 @@ export_update_cache_negative('${UPDATE_CACHE_DIR}')
 # Non-fatal: a failed sdpa export makes the required 4k/8k configs hard-fail in
 # webgpu_native_test below (precise per-config error), so don't exit/mask here.
 $PYTHON_EXECUTABLE -c "
-from executorch.backends.webgpu.test.ops.sdpa.test_sdpa import (
+from executorch.backends.webgpu.test.ops.test_sdpa import (
     export_all_sdpa_models,
     export_replay_sequences,
     export_dynamic_decode,

backends/webgpu/test/TARGETS

@@ -7,7 +7,7 @@ oncall("executorch")
 python_unittest(
     name = "test_add",
     srcs = [
-        "ops/add/test_add.py",
+        "ops/test_add.py",
     ],
     deps = [
         "//caffe2:torch",

backends/webgpu/test/op_tests/cases.py

@@ -25,17 +25,11 @@
     WebGPUTestSuite,
     XS,
 )
-from executorch.backends.webgpu.test.ops.add.test_add import (
+from executorch.backends.webgpu.test.ops.test_add import (
     AddChainedModule,
     AddModule,
     AddSelfModule,
 )
-from executorch.backends.webgpu.test.ops.rms_norm.test_rms_norm import (
-    _CASES,
-    _linspace_weight,
-    _ramp,
-    RmsNormModule,
-)
 from executorch.backends.webgpu.test.ops.test_cat import (
     CatModule,
     CONFIGS as _CAT_CONFIGS,
@@ -48,6 +42,12 @@
     CONFIGS as _PERMUTE_CONFIGS,
     PermuteModule,
 )
+from executorch.backends.webgpu.test.ops.test_rms_norm import (
+    _CASES,
+    _linspace_weight,
+    _ramp,
+    RmsNormModule,
+)
 from executorch.backends.webgpu.test.ops.test_select import (
     CONFIGS as _SELECT_CONFIGS,
     SelectModule,
@@ -92,7 +92,7 @@ def _add_factory(variant: str = "regular") -> torch.nn.Module:
 @register_op_test("add")
 def _add_suite() -> WebGPUTestSuite:
     # Same-shape numeric coverage only: broadcast adds stay export-smoke in
-    # ops/add/test_add.py because the kernel can't broadcast.
+    # ops/test_add.py because the kernel can't broadcast.
     return WebGPUTestSuite(
         module_factory=_add_factory,
         cases=[