Update on "[executorch] Add coreml quant recipes"

abhinaykukkadapu · abhinaykukkadapu · commit d95714f73c80 · 2025-08-13T14:49:29.000-07:00
Fixing tests for stack that got reverted: #13265 Adds coreml quant recipes after FP32/16 recipes added in #13121 Recipes added: PT2E_INT8_STATIC PT2E_INT8_WEIGHT_ONLY INT4_WEIGHT_ONLY_PER_CHANNEL INT4_WEIGHT_ONLY_PER_GROUP INT8_WEIGHT_ONLY_PER_CHANNEL INT8_WEIGHT_ONLY_PER_GROUP CODEBOOK_WEIGHT_ONLY Differential Revision: [D80206542](https://our.internmc.facebook.com/intern/diff/D80206542/) [ghstack-poisoned]
diff --git a/backends/xnnpack/test/recipes/test_xnnpack_recipes.py b/backends/xnnpack/test/recipes/test_xnnpack_recipes.py
@@ -155,7 +155,7 @@ def forward(self, x) -> torch.Tensor:
                     session, example_inputs, 1e-3
                 )
                 self._compare_eager_unquantized_model_outputs(
-                    session, model, example_inputs
+                    session, model, example_inputs, 15
                 )
 
     def _get_recipe_for_quant_type(self, quant_type: QuantType) -> XNNPackRecipeType:

Original file line number	Diff line number	Diff line change
`@@ -155,7 +155,7 @@ def forward(self, x) -> torch.Tensor:`
`155`	`155`	`session, example_inputs, 1e-3`
`156`	`156`	`)`
`157`	`157`	`self._compare_eager_unquantized_model_outputs(`
`158`		`- session, model, example_inputs`
	`158`	`+ session, model, example_inputs, 15`
`159`	`159`	`)`
`160`	`160`
`161`	`161`	`def _get_recipe_for_quant_type(self, quant_type: QuantType) -> XNNPackRecipeType:`