rename @legacy to @c2_ops

Reviewed By: sstsai-adl

Differential Revision: D32216605

fbshipit-source-id: bebee1edae85e940c7dcc6a64dbe341a2fde36a2

d2go/modeling/meta_arch/rcnn.py

@@ -64,7 +64,7 @@ class GeneralizedRCNNPatch:
 @RCNN_PREPARE_FOR_EXPORT_REGISTRY.register()
 def default_rcnn_prepare_for_export(self, cfg, inputs, predictor_type):
     if (
-        "@legacy" in predictor_type
+        "@c2_ops" in predictor_type
         or "caffe2" in predictor_type
         or "onnx" in predictor_type
     ):
@@ -90,7 +90,7 @@ def default_rcnn_prepare_for_export(self, cfg, inputs, predictor_type):
             # Caffe2MetaArch takes a single tuple as input (which is the return of
             # preprocess_func), data_generator requires all positional args as a tuple.
             data_generator=lambda x: (preprocess_func(x),),
-            model_export_method=predictor_type.replace("@legacy", "", 1),
+            model_export_method=predictor_type.replace("@c2_ops", "", 1),
             model_export_kwargs=model_export_kwargs,
             preprocess_info=preprocess_info,
             postprocess_info=postprocess_info,

tests/modeling/test_meta_arch_rcnn.py

@@ -29,9 +29,9 @@ class TestFBNetV3MaskRCNNFP32(RCNNBaseTestCases.TemplateTestCase):
 
     @RCNNBaseTestCases.expand_parameterized_test_export(
         [
-            ["torchscript@legacy", True],
+            ["torchscript@c2_ops", True],
             ["torchscript", True],
-            ["torchscript_int8@legacy", False],
+            ["torchscript_int8@c2_ops", False],
             ["torchscript_int8", False],
         ]
     )
@@ -60,7 +60,7 @@ class TestFBNetV3MaskRCNNQATEager(RCNNBaseTestCases.TemplateTestCase):
 
     @RCNNBaseTestCases.expand_parameterized_test_export(
         [
-            ["torchscript_int8@legacy", False],  # TODO: fix mismatch
+            ["torchscript_int8@c2_ops", False],  # TODO: fix mismatch
             ["torchscript_int8", False],  # TODO: fix mismatch
         ]
     )
@@ -87,7 +87,7 @@ class TestFBNetV3KeypointRCNNFP32(RCNNBaseTestCases.TemplateTestCase):
 
     @RCNNBaseTestCases.expand_parameterized_test_export(
         [
-            ["torchscript_int8@legacy", False],  # TODO: fix mismatch
+            ["torchscript_int8@c2_ops", False],  # TODO: fix mismatch
             ["torchscript_int8", False],  # TODO: fix mismatch
         ]
     )

tests/modeling/test_rcnn_export_example.py

@@ -2,7 +2,6 @@
 # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved
 
 import os
-import tempfile
 import unittest
 
 import torch
@@ -13,7 +12,7 @@ from d2go.utils.testing.rcnn_helper import get_quick_test_config_opts
 from mobile_cv.common.misc.file_utils import make_temp_directory
 
 
-def maskrcnn_export_legacy_vs_new_format_example():
+def maskrcnn_export_caffe2_vs_torchvision_opset_format_example():
     with make_temp_directory("export_demo") as tmp_dir:
         # use a fake dataset for ci
         dataset_name = create_local_dataset(tmp_dir, 5, 224, 224)
@@ -31,45 +30,49 @@ def maskrcnn_export_legacy_vs_new_format_example():
         cfg.merge_from_list(config_list)
 
         # equivalent to running:
-        #   exporter.par --runner GeneralizedRCNNRunner --config-file config.yaml --predictor-types torchscript tourchscript@legacy --output-dir tmp_dir
+        #   exporter.par --runner GeneralizedRCNNRunner --config-file config.yaml --predictor-types torchscript tourchscript@c2_ops --output-dir tmp_dir
         _ = main(
             cfg,
             tmp_dir,
             runner,
-            predictor_types=["torchscript@legacy", "torchscript"],
+            predictor_types=["torchscript@c2_ops", "torchscript"],
         )
 
         # the path can be fetched from the return of main, here just use hard-coded values
-        new_path = os.path.join(tmp_dir, "torchscript", "model.jit")
-        lagacy_path = os.path.join(tmp_dir, "torchscript@legacy", "model.jit")
-        new_model = torch.jit.load(new_path)
-        legacy_model = torch.jit.load(lagacy_path)
+        torchvision_ops_model = torch.jit.load(
+            os.path.join(tmp_dir, "torchscript", "model.jit")
+        )
+        caffe2_ops_model = torch.jit.load(
+            os.path.join(tmp_dir, "torchscript@c2_ops", "model.jit")
+        )
 
-        # Running inference using new format
+        # Running inference using torchvision-style format
         image = torch.zeros(1, 64, 96)  # chw 3D tensor
-        new_outputs = new_model(image)  # suppose N instances are detected
+        torchvision_style_outputs = torchvision_ops_model(
+            image
+        )  # suppose N instances are detected
         # NOTE: the output are flattened tensors of the real output (which is a dict), they're
         # ordered by the key in dict, which is deterministic for the given model, but it might
         # be difficult to figure out just from model.jit file. The predictor_info.json from
         # the same directory contains the `outputs_schema`, which indicate how the final output
         # is constructed from flattened tensors.
-        pred_boxes = new_outputs[0]  # torch.Size([N, 4])
-        pred_classes = new_outputs[1]  # torch.Size([N])
-        pred_masks = new_outputs[2]  # torch.Size([N, 1, Hmask, Wmask])
-        scores = new_outputs[3]  # torch.Size([N])
+        pred_boxes = torchvision_style_outputs[0]  # torch.Size([N, 4])
+        pred_classes = torchvision_style_outputs[1]  # torch.Size([N])
+        pred_masks = torchvision_style_outputs[2]  # torch.Size([N, 1, Hmask, Wmask])
+        scores = torchvision_style_outputs[3]  # torch.Size([N])
 
-        # Running inference using legacy caffe2 format
+        # Running inference using caffe2-style format
         data = torch.zeros(1, 1, 64, 96)
         im_info = torch.tensor([[64, 96, 1.0]])
-        legacy_outputs = legacy_model([data, im_info])
+        caffe2_style_outputs = caffe2_ops_model([data, im_info])
         # NOTE: the output order is determined in the order of creating the tensor during
         # forward function, it's also follow the order of original Caffe2 model.
-        roi_bbox_nms = legacy_outputs[0]  # torch.Size([N, 4])
-        roi_score_nms = legacy_outputs[1]  # torch.Size([N])
-        roi_class_nms = legacy_outputs[2]  # torch.Size([N])
-        mask_fcn_probs = legacy_outputs[3]  # torch.Size([N, Cmask, Hmask, Wmask])
+        roi_bbox_nms = caffe2_style_outputs[0]  # torch.Size([N, 4])
+        roi_score_nms = caffe2_style_outputs[1]  # torch.Size([N])
+        roi_class_nms = caffe2_style_outputs[2]  # torch.Size([N])
+        mask_fcn_probs = caffe2_style_outputs[3]  # torch.Size([N, Cmask, Hmask, Wmask])
 
-        # relations between legacy outputs and new outputs
+        # relations between torchvision-style outputs and caffe2-style outputs
         torch.testing.assert_allclose(pred_boxes, roi_bbox_nms)
         torch.testing.assert_allclose(pred_classes, roi_class_nms)
         torch.testing.assert_allclose(
@@ -81,4 +84,4 @@ def maskrcnn_export_legacy_vs_new_format_example():
 
 class TestOptimizer(unittest.TestCase):
     def test_maskrcnn_export_legacy_vs_new_format_example(self):
-        maskrcnn_export_legacy_vs_new_format_example()
+        maskrcnn_export_caffe2_vs_torchvision_opset_format_example()