I got different results using thop and torchinfo
zz-2024 opened this issue · 2 comments
zz-2024 commented
Describe the bug
my code:
import torch
from PIL import Image
from thop import profile, clever_format
from torchinfo import summary
import cn_clip.clip as clip
from cn_clip.clip import load_from_name, available_models
print("Available models:", available_models())
# Available models: ['ViT-B-16', 'ViT-L-14', 'ViT-L-14-336', 'ViT-H-14', 'RN50']
device = "cuda" if torch.cuda.is_available() else "cpu"
model, preprocess = load_from_name("RN50", device=device, download_root='./')
model.eval()
image = preprocess(Image.open("examples/pokemon.jpeg")).unsqueeze(0).to(device)
text = clip.tokenize(["杰尼龟", "妙蛙种子", "小火龙", "皮卡丘"]).to(device)
print(f"device:{device}")
with torch.no_grad():
image_features = model.encode_image(image)
text_features = model.encode_text(text)
# 对特征进行归一化，请使用归一化后的图文特征用于下游任务
image_features /= image_features.norm(dim=-1, keepdim=True)
text_features /= text_features.norm(dim=-1, keepdim=True)
logits_per_image, logits_per_text = model.get_similarity(image, text)
probs = logits_per_image.softmax(dim=-1).cpu().numpy()
print("Label probs:", probs) # [[1.268734e-03 5.436878e-02 6.795761e-04 9.436829e-01]]
# 计算FLOPs
print(f"image:{image.shape}, text:{text.shape}")
flops, params = profile(model, inputs=(image, text), verbose=False)
flops, params = clever_format([flops, params], "%.3f")
print(f"Model FLOPs: {flops}, Model Params:{params}")
print("========seperate==========")
flops, params = profile(model.visual, inputs=(image, ), verbose=False)
flops, params = clever_format([flops, params], "%.3f")
print(f"Model FLOPs: {flops}, Model Params:{params}")
print(f"=========another method=============")
summary(model, input_data=[image, text])
my results:
Model FLOPs: 9.839G, Model Params:44.792M
========seperate==========
Model FLOPs: 5.418G, Model Params:23.527M
=========another method=============
=========================================================================================================
Layer (type:depth-idx) Output Shape Param #
=========================================================================================================
CLIP [1, 1024] 786,433
├─ModifiedResNet: 1-1 [1, 1024] --
│ └─Conv2d: 2-1 [1, 32, 112, 112] 864
│ └─BatchNorm2d: 2-2 [1, 32, 112, 112] 64
│ └─ReLU: 2-3 [1, 32, 112, 112] --
│ └─Conv2d: 2-4 [1, 32, 112, 112] 9,216
│ └─BatchNorm2d: 2-5 [1, 32, 112, 112] 64
│ └─ReLU: 2-6 [1, 32, 112, 112] --
│ └─Conv2d: 2-7 [1, 64, 112, 112] 18,432
│ └─BatchNorm2d: 2-8 [1, 64, 112, 112] 128
│ └─ReLU: 2-9 [1, 64, 112, 112] --
│ └─AvgPool2d: 2-10 [1, 64, 56, 56] --
│ └─Sequential: 2-11 [1, 256, 56, 56] --
│ │ └─Bottleneck: 3-1 [1, 256, 56, 56] 75,008
│ │ └─Bottleneck: 3-2 [1, 256, 56, 56] 70,400
│ │ └─Bottleneck: 3-3 [1, 256, 56, 56] 70,400
│ └─Sequential: 2-12 [1, 512, 28, 28] --
│ │ └─Bottleneck: 3-4 [1, 512, 28, 28] 379,392
│ │ └─Bottleneck: 3-5 [1, 512, 28, 28] 280,064
│ │ └─Bottleneck: 3-6 [1, 512, 28, 28] 280,064
│ │ └─Bottleneck: 3-7 [1, 512, 28, 28] 280,064
│ └─Sequential: 2-13 [1, 1024, 14, 14] --
│ │ └─Bottleneck: 3-8 [1, 1024, 14, 14] 1,512,448
│ │ └─Bottleneck: 3-9 [1, 1024, 14, 14] 1,117,184
│ │ └─Bottleneck: 3-10 [1, 1024, 14, 14] 1,117,184
│ │ └─Bottleneck: 3-11 [1, 1024, 14, 14] 1,117,184
│ │ └─Bottleneck: 3-12 [1, 1024, 14, 14] 1,117,184
│ │ └─Bottleneck: 3-13 [1, 1024, 14, 14] 1,117,184
│ └─Sequential: 2-14 [1, 2048, 7, 7] --
│ │ └─Bottleneck: 3-14 [1, 2048, 7, 7] 6,039,552
│ │ └─Bottleneck: 3-15 [1, 2048, 7, 7] 4,462,592
│ │ └─Bottleneck: 3-16 [1, 2048, 7, 7] 4,462,592
│ └─AttentionPool2d: 2-15 [1, 1024] 14,789,632
├─BertModel: 1-2 [4, 52, 768] --
│ └─BertEmbeddings: 2-16 [4, 52, 768] --
│ │ └─Embedding: 3-17 [4, 52, 768] 16,226,304
│ │ └─Embedding: 3-18 [4, 52, 768] 393,216
│ │ └─Embedding: 3-19 [4, 52, 768] 1,536
│ │ └─LayerNorm: 3-20 [4, 52, 768] 1,536
│ │ └─Dropout: 3-21 [4, 52, 768] --
│ └─BertEncoder: 2-17 [4, 52, 768] --
│ │ └─ModuleList: 3-22 -- 21,263,616
=========================================================================================================
Total params: 76,989,537
Trainable params: 76,989,537
Non-trainable params: 0
Total mult-adds (G): 5.52
=========================================================================================================
Input size (MB): 0.60
Forward/backward pass size (MB): 123.19
Params size (MB): 122.93
Estimated Total Size (MB): 246.73
=========================================================================================================
Why the number of parameters varies so much？ （44.792M VS 77M）
Why are mult-adds such different? (9.839G VS 5.52G)
To Reproduce
Steps to reproduce the behavior:
- Go to '...'
- Click on '....'
- Scroll down to '....'
- See error
Expected behavior
A clear and concise description of what you expected to happen.
Screenshots
If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
- OS: [e.g. iOS]
- Browser [e.g. chrome, safari]
- Version [e.g. 22]
Smartphone (please complete the following information):
- Device: [e.g. iPhone6]
- OS: [e.g. iOS8.1]
- Browser [e.g. stock browser, safari]
- Version [e.g. 22]
Additional context
Add any other context about the problem here.
atonyo11 commented
Same with me! Have you found the cause yet?
6DammK9 commented
I have found that torchinfo, ultralytics-thop, and common approach
pytorch_total_params = sum(p.numel() for p in model.parameters())They are likely yield different results for newer models.