code format for example. (#2865)

* code format for example 1st round.
* Reformat llm-related example script files

---------

Co-authored-by: Xia, Weiwen <weiwen.xia@intel.com>

Author: xuhancn

examples/cpu/features/fast_bert/fast_bert_inference_bf16.py

@@ -12,6 +12,7 @@
 
 #################### code changes ####################  # noqa F401
 import intel_extension_for_pytorch as ipex
+
 model = ipex.fast_bert(model, dtype=torch.bfloat16)
 ######################################################  # noqa F401
 

examples/cpu/features/fast_bert/fast_bert_training_bf16.py

@@ -1,7 +1,9 @@
 import torch
 from transformers import BertForSequenceClassification
 
-model = BertForSequenceClassification.from_pretrained('bert-base-uncased', return_dict=True)
+model = BertForSequenceClassification.from_pretrained(
+    "bert-base-uncased", return_dict=True
+)
 model.train()
 optimizer = torch.optim.Adam(model.parameters(), lr=1e-5)
 
@@ -13,14 +15,15 @@
 
 #################### code changes ####################  # noqa F401
 import intel_extension_for_pytorch as ipex
+
 model, optimizer = ipex.fast_bert(model, optimizer=optimizer, dtype=torch.bfloat16)
 ######################################################  # noqa F401
 
 with torch.cpu.amp.autocast(dtype=torch.bfloat16):
-  labels = torch.tensor(1)
-  outputs = model(data, labels=labels)
-  loss = outputs.loss
-  loss.backward()
-  optimizer.step()
+    labels = torch.tensor(1)
+    outputs = model(data, labels=labels)
+    loss = outputs.loss
+    loss.backward()
+    optimizer.step()
 
 print("Execution finished")

examples/cpu/features/graph_capture.py

@@ -1,12 +1,13 @@
 import torch
 import torchvision.models as models
 
-model = models.resnet50(weights='ResNet50_Weights.DEFAULT')
+model = models.resnet50(weights="ResNet50_Weights.DEFAULT")
 model.eval()
 data = torch.rand(1, 3, 224, 224)
 
 #################### code changes ####################  # noqa F401
 import intel_extension_for_pytorch as ipex
+
 model = ipex.optimize(model, graph_mode=True)
 ######################################################  # noqa F401
 

examples/cpu/features/graph_optimization/folding.py

@@ -1,7 +1,7 @@
 import torch
 import torchvision.models as models
 
-model = models.resnet50(weights='ResNet50_Weights.DEFAULT')
+model = models.resnet50(weights="ResNet50_Weights.DEFAULT")
 model.eval()
 x = torch.randn(4, 3, 224, 224)
 

examples/cpu/features/graph_optimization/fp32_bf16.py

@@ -4,7 +4,7 @@
 # Import the Intel Extension for PyTorch
 import intel_extension_for_pytorch as ipex
 
-model = models.resnet50(weights='ResNet50_Weights.DEFAULT')
+model = models.resnet50(weights="ResNet50_Weights.DEFAULT")
 model.eval()
 
 # Apply some fusions at the front end

examples/cpu/features/graph_optimization/int8.py

@@ -4,36 +4,38 @@
 from intel_extension_for_pytorch.quantization import prepare, convert
 
 # construct the model
-model = models.resnet50(weights='ResNet50_Weights.DEFAULT')
+model = models.resnet50(weights="ResNet50_Weights.DEFAULT")
 qconfig = ipex.quantization.default_static_qconfig
 model.eval()
 example_inputs = torch.rand(1, 3, 224, 224)
 prepared_model = prepare(model, qconfig, example_inputs=example_inputs, inplace=False)
 
 ##### Example Dataloader #####  # noqa F401
 import torchvision
-DOWNLOAD = True
-DATA = 'datasets/cifar10/'
 
-transform = torchvision.transforms.Compose([
-    torchvision.transforms.Resize((224, 224)),
-    torchvision.transforms.ToTensor(),
-    torchvision.transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
-])
+DOWNLOAD = True
+DATA = "datasets/cifar10/"
+
+transform = torchvision.transforms.Compose(
+    [
+        torchvision.transforms.Resize((224, 224)),
+        torchvision.transforms.ToTensor(),
+        torchvision.transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+    ]
+)
 train_dataset = torchvision.datasets.CIFAR10(
     root=DATA,
     train=True,
     transform=transform,
     download=DOWNLOAD,
 )
 calibration_data_loader = torch.utils.data.DataLoader(
-    dataset=train_dataset,
-    batch_size=128
+    dataset=train_dataset, batch_size=128
 )
 
 with torch.no_grad():
     for batch_idx, (d, target) in enumerate(calibration_data_loader):
-        print(f'calibrated on batch {batch_idx} out of {len(calibration_data_loader)}')
+        print(f"calibrated on batch {batch_idx} out of {len(calibration_data_loader)}")
         prepared_model(d)
 ##############################  # noqa F401
 

examples/cpu/features/hypertune/resnet50.py

@@ -1,6 +1,7 @@
 import torch
 import torchvision.models as models
 
+
 def inference(model, data):
     with torch.no_grad():
         # warm up
@@ -9,6 +10,7 @@ def inference(model, data):
 
         # measure
         import time
+
         measure_iter = 100
         start = time.time()
         for _ in range(measure_iter):
@@ -19,8 +21,13 @@ def inference(model, data):
         latency = duration / measure_iter
         throughput = measure_iter / duration
 
-        print("@hypertune {'name': 'latency (ms)'}")  # Add print statement of the form @hypertune {'name': str, 'higher_is_better': bool, 'target_val': int or float}`
-        print(latency)  # Print the objective(s) you want to optimize. Make sure this is just an int or float to be minimzied or maximized.
+        print(
+            "@hypertune {'name': 'latency (ms)'}"
+        )  # Add print statement of the form @hypertune {'name': str, 'higher_is_better': bool, 'target_val': int or float}`
+        print(
+            latency
+        )  # Print the objective(s) you want to optimize. Make sure this is just an int or float to be minimzied or maximized.
+
 
 def main(args):
     model = models.resnet50(pretrained=False)
@@ -30,9 +37,9 @@ def main(args):
 
     import intel_extension_for_pytorch as ipex
 
-    if args.dtype == 'float32':
+    if args.dtype == "float32":
         model = ipex.optimize(model, dtype=torch.float32)
-    elif args.dtype == 'bfloat16':
+    elif args.dtype == "bfloat16":
         model = ipex.optimize(model, dtype=torch.bfloat16)
     else:  # int8
         from intel_extension_for_pytorch.quantization import prepare, convert
@@ -47,18 +54,22 @@ def main(args):
 
         model = convert(model)
 
-    with torch.cpu.amp.autocast(enabled=args.dtype == 'bfloat16'):
+    with torch.cpu.amp.autocast(enabled=args.dtype == "bfloat16"):
         if args.torchscript:
             with torch.no_grad():
                 model = torch.jit.trace(model, data)
                 model = torch.jit.freeze(model)
 
         inference(model, data)
 
-if __name__ == '__main__':
+
+if __name__ == "__main__":
     import argparse
+
     parser = argparse.ArgumentParser()
-    parser.add_argument('--dtype', default='float32', choices=['float32', 'bfloat16', 'int8'])
+    parser.add_argument(
+        "--dtype", default="float32", choices=["float32", "bfloat16", "int8"]
+    )
     parser.add_argument("--torchscript", default=False, action="store_true")
 
     main(parser.parse_args())