ultralytics-thop 0.2.7__tar.gz → 0.2.9__tar.gz

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ultralytics-thop
-Version: 0.2.7
+Version: 0.2.9
 Summary: Ultralytics THOP package for fast computation of PyTorch model FLOPs and parameters.
 Author-email: Ligeng Zhu <ligeng.zhu+github@gmail.com>
 Maintainer: Glenn Jocher
@@ -74,7 +74,7 @@ import torch
 
 model = resnet50()
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ))
+macs, params = profile(model, inputs=(input,))
 ```
 
 ### Define Custom Rules for Third-Party Modules
@@ -84,16 +84,19 @@ You can define custom rules for unsupported modules:
 ```python
 import torch.nn as nn
 
+
 class YourModule(nn.Module):
     # your definition
     pass
 
+
 def count_your_model(model, x, y):
     # your rule here
     pass
 
+
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ), custom_ops={YourModule: count_your_model})
+macs, params = profile(model, inputs=(input,), custom_ops={YourModule: count_your_model})
 ```
 
 ### Improve Output Readability
@@ -102,6 +105,7 @@ Use `thop.clever_format` for a more readable output:
 
 ```python
 from thop import clever_format
+
 macs, params = clever_format([macs, params], "%.3f")
 ```
 

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/README.md

@@ -40,7 +40,7 @@ import torch
 
 model = resnet50()
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ))
+macs, params = profile(model, inputs=(input,))
 ```
 
 ### Define Custom Rules for Third-Party Modules
@@ -50,16 +50,19 @@ You can define custom rules for unsupported modules:
 ```python
 import torch.nn as nn
 
+
 class YourModule(nn.Module):
     # your definition
     pass
 
+
 def count_your_model(model, x, y):
     # your rule here
     pass
 
+
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ), custom_ops={YourModule: count_your_model})
+macs, params = profile(model, inputs=(input,), custom_ops={YourModule: count_your_model})
 ```
 
 ### Improve Output Readability
@@ -68,6 +71,7 @@ Use `thop.clever_format` for a more readable output:
 
 ```python
 from thop import clever_format
+
 macs, params = clever_format([macs, params], "%.3f")
 ```
 

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/tests/test_matmul.py

@@ -15,7 +15,7 @@ class TestUtils:
 
     def test_matmul_case2(self):
         """Tests matrix multiplication to assert FLOPs and parameters of nn.Linear layer using random dimensions."""
-        for i in range(10):
+        for _ in range(10):
             n, in_c, out_c = torch.randint(1, 500, (3,)).tolist()
             net = nn.Linear(in_c, out_c)
             flops, params = profile(net, inputs=(torch.randn(n, in_c),))

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/tests/test_relu.py

@@ -9,8 +9,7 @@ class TestUtils:
         """Tests the ReLU activation function to ensure it has zero FLOPs and checks parameter count using THOP
         profiling.
         """
-        n, in_c, out_c = 1, 100, 200
-        data = torch.randn(n, in_c)
+        n, in_c, _out_c = 1, 100, 200
         net = nn.ReLU()
         flops, params = profile(net, inputs=(torch.randn(n, in_c),))
         print(flops, params)

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/thop/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.2.7"
+__version__ = "0.2.9"
 
 import torch
 

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/thop/fx_profile.py

@@ -33,10 +33,10 @@ def count_matmul(input_shapes, output_shapes):
 
 def count_fn_linear(input_shapes, output_shapes, *args, **kwargs):
     """Calculates total operations (FLOPs) for a linear layer given input and output shapes."""
-    mul_flops = count_matmul(input_shapes, output_shapes)
+    flops = count_matmul(input_shapes, output_shapes)
     if "bias" in kwargs:
-        add_flops = output_shapes[0].numel()
-    return mul_flops
+        flops += output_shapes[0].numel()
+    return flops
 
 
 from .vision.calc_func import calculate_conv
@@ -131,7 +131,6 @@ def fx_profile(mod: nn.Module, input: th.Tensor, verbose=False):
     information if verbose.
     """
     gm: torch.fx.GraphModule = symbolic_trace(mod)
-    g = gm.graph
     ShapeProp(gm).propagate(input)
 
     fprint = null_print
@@ -148,7 +147,6 @@ def fx_profile(mod: nn.Module, input: th.Tensor, verbose=False):
         node_flops = None
 
         input_shapes = []
-        output_shapes = []
         fprint("input_shape:", end="\t")
         for arg in node.args:
             if str(arg) not in v_maps:
@@ -157,8 +155,7 @@ def fx_profile(mod: nn.Module, input: th.Tensor, verbose=False):
             input_shapes.append(v_maps[str(arg)])
         fprint()
         fprint(f"output_shape:\t{node.meta['tensor_meta'].shape}")
-        output_shapes.append(node.meta["tensor_meta"].shape)
-
+        output_shapes = [node.meta["tensor_meta"].shape]
         if node.op in ["output", "placeholder"]:
             node_flops = 0
         elif node.op == "call_function":
@@ -194,7 +191,7 @@ def fx_profile(mod: nn.Module, input: th.Tensor, verbose=False):
                 print("weight_shape: None")
             else:
                 print(type(m))
-                print(f"weight_shape: {mod.state_dict()[node.target + '.weight'].shape}")
+                print(f"weight_shape: {mod.state_dict()[f'{node.target}.weight'].shape}")
 
         v_maps[str(node.name)] = node.meta["tensor_meta"].shape
         if node_flops is not None:

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/thop/vision/basic_hooks.py

@@ -10,9 +10,6 @@ multiply_adds = 1
 
 def count_parameters(m, x, y):
     """Calculate and update the total number of parameters in a given PyTorch model."""
-    total_params = 0
-    for p in m.parameters():
-        total_params += torch.DoubleTensor([p.numel()])
     m.total_params[0] = calculate_parameters(m.parameters())
 
 
@@ -24,10 +21,7 @@ def zero_ops(m, x, y):
 def count_convNd(m: _ConvNd, x, y: torch.Tensor):
     """Calculate and add the number of convolutional operations (FLOPs) to the model's total operations count."""
     x = x[0]
-
-    kernel_ops = torch.zeros(m.weight.size()[2:]).numel()  # Kw x Kh
-    bias_ops = 1 if m.bias is not None else 0
-
+    
     m.total_ops += calculate_conv2d_flops(
         input_size=list(x.shape),
         output_size=list(y.shape),
@@ -97,9 +91,6 @@ def count_prelu(m, x, y):
 def count_relu(m, x, y):
     """Calculate and update the total operation counts for a ReLU layer."""
     x = x[0]
-
-    nelements = x.numel()
-
     m.total_ops += calculate_relu_flops(list(x.shape))
 
 

{ultralytics_thop-0.2.7 → ultralytics_thop-0.2.9}/ultralytics_thop.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ultralytics-thop
-Version: 0.2.7
+Version: 0.2.9
 Summary: Ultralytics THOP package for fast computation of PyTorch model FLOPs and parameters.
 Author-email: Ligeng Zhu <ligeng.zhu+github@gmail.com>
 Maintainer: Glenn Jocher
@@ -74,7 +74,7 @@ import torch
 
 model = resnet50()
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ))
+macs, params = profile(model, inputs=(input,))
 ```
 
 ### Define Custom Rules for Third-Party Modules
@@ -84,16 +84,19 @@ You can define custom rules for unsupported modules:
 ```python
 import torch.nn as nn
 
+
 class YourModule(nn.Module):
     # your definition
     pass
 
+
 def count_your_model(model, x, y):
     # your rule here
     pass
 
+
 input = torch.randn(1, 3, 224, 224)
-macs, params = profile(model, inputs=(input, ), custom_ops={YourModule: count_your_model})
+macs, params = profile(model, inputs=(input,), custom_ops={YourModule: count_your_model})
 ```
 
 ### Improve Output Readability
@@ -102,6 +105,7 @@ Use `thop.clever_format` for a more readable output:
 
 ```python
 from thop import clever_format
+
 macs, params = clever_format([macs, params], "%.3f")
 ```