[Quant] Add FX support in quantization examples

Summary: Previously, the quantization examples use only eager
mode quantization. This commit adds support for FX mode
quantization as well. TODO: provide accuracy comparison.

Test Plan:

python train_quantization.py
  --device='cpu'
  --post-training-quantize
  --backend='fbgemm'
  --model='$MODEL'

model: $MODEL is one of googlenet, inception_v3, resnet18, resnet50,
resnext101_32x8d, shufflenet_v2_x0_5 and shufflenet_v2_x1_0

Reviewers: jerryzh168, vkuzo

Subscribers: jerryzh168, vkuzo

ghstack-source-id: 50b3f482794f42c742e9e61ff3f4fcbf2b040703
Pull Request resolved: #5797

references/classification/train_quantization.py

@@ -2,16 +2,23 @@
 import datetime
 import os
 import time
+from enum import Enum
 
 import torch
 import torch.ao.quantization
+import torch.ao.quantization.quantize_fx
 import torch.utils.data
 import torchvision
 import utils
 from torch import nn
 from train import train_one_epoch, evaluate, load_data
 
 
+class QuantizationWorkflowType(Enum):
+    EAGER_MODE_QUANTIZATION = 1
+    FX_GRAPH_MODE_QUANTIZATION = 2
+
+
 def main(args):
     if args.output_dir:
         utils.mkdir(args.output_dir)
@@ -22,6 +29,15 @@ def main(args):
     if args.post_training_quantize and args.distributed:
         raise RuntimeError("Post training quantization example should not be performed on distributed mode")
 
+    # Validate quantization workflow type
+    quantization_workflow_type = args.quantization_workflow_type.upper()
+    if quantization_workflow_type not in QuantizationWorkflowType.__members__:
+        raise RuntimeError(
+            "Unknown workflow type '%s', please choose from: %s"
+            % (args.quantization_workflow_type, str(tuple([t.lower() for t in QuantizationWorkflowType.__members__])))
+        )
+    use_fx_graph_mode_quantization = quantization_workflow_type == QuantizationWorkflowType.FX_GRAPH_MODE_QUANTIZATION
+
     # Set backend engine to ensure that quantized model runs on the correct kernels
     if args.backend not in torch.backends.quantized.supported_engines:
         raise RuntimeError("Quantized backend not supported: " + str(args.backend))
@@ -45,14 +61,23 @@ def main(args):
     )
 
     print("Creating model", args.model)
+    if use_fx_graph_mode_quantization:
+        model_namespace = torchvision.models
+    else:
+        model_namespace = torchvision.models.quantization
     # when training quantized models, we always start from a pre-trained fp32 reference model
-    model = torchvision.models.quantization.__dict__[args.model](weights=args.weights, quantize=args.test_only)
+    model = model_namespace.__dict__[args.model](weights=args.weights, quantize=args.test_only)
     model.to(device)
 
     if not (args.test_only or args.post_training_quantize):
-        model.fuse_model(is_qat=True)
-        model.qconfig = torch.ao.quantization.get_default_qat_qconfig(args.backend)
-        torch.ao.quantization.prepare_qat(model, inplace=True)
+        qconfig = torch.ao.quantization.get_default_qat_qconfig(args.backend)
+        if use_fx_graph_mode_quantization:
+            qconfig_dict = {"": qconfig}
+            model = torch.ao.quantization.quantize_fx.prepare_qat_fx(model, qconfig_dict)
+        else:
+            model.fuse_model(is_qat=True)
+            model.qconfig = qconfig
+            torch.ao.quantization.prepare_qat(model, inplace=True)
 
         if args.distributed and args.sync_bn:
             model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
@@ -84,13 +109,21 @@ def main(args):
             ds, batch_size=args.batch_size, shuffle=False, num_workers=args.workers, pin_memory=True
         )
         model.eval()
-        model.fuse_model(is_qat=False)
-        model.qconfig = torch.ao.quantization.get_default_qconfig(args.backend)
-        torch.ao.quantization.prepare(model, inplace=True)
+        qconfig = torch.ao.quantization.get_default_qconfig(args.backend)
+        if use_fx_graph_mode_quantization:
+            qconfig_dict = {"": qconfig}
+            model = torch.ao.quantization.quantize_fx.prepare_fx(model, qconfig_dict)
+        else:
+            model.fuse_model(is_qat=False)
+            model.qconfig = qconfig
+            torch.ao.quantization.prepare(model, inplace=True)
         # Calibrate first
         print("Calibrating")
         evaluate(model, criterion, data_loader_calibration, device=device, print_freq=1)
-        torch.ao.quantization.convert(model, inplace=True)
+        if use_fx_graph_mode_quantization:
+            model = torch.ao.quantization.quantize_fx.convert_fx(model)
+        else:
+            torch.ao.quantization.convert(model, inplace=True)
         if args.output_dir:
             print("Saving quantized model")
             if utils.is_main_process():
@@ -125,7 +158,10 @@ def main(args):
             quantized_eval_model = copy.deepcopy(model_without_ddp)
             quantized_eval_model.eval()
             quantized_eval_model.to(torch.device("cpu"))
-            torch.ao.quantization.convert(quantized_eval_model, inplace=True)
+            if use_fx_graph_mode_quantization:
+                quantized_eval_model = torch.ao.quantization.quantize_fx.convert_fx(quantized_eval_model)
+            else:
+                torch.ao.quantization.convert(quantized_eval_model, inplace=True)
 
             print("Evaluate Quantized model")
             evaluate(quantized_eval_model, criterion, data_loader_test, device=torch.device("cpu"))
@@ -233,6 +269,12 @@ def get_args_parser(add_help=True):
         help="Post training quantize the model",
         action="store_true",
     )
+    parser.add_argument(
+        "--quantization-workflow-type",
+        default="eager_mode_quantization",
+        type=str,
+        help="The quantization workflow type to use, either 'eager_mode_quantization' (default) or 'fx_graph_mode_quantization'",
+    )
 
     # distributed training parameters
     parser.add_argument("--world-size", default=1, type=int, help="number of distributed processes")