calculate-flops.pytorch
calculate-flops.pytorch copied to clipboard
MrYxJ
The calflops is designed to calculate FLOPs、MACs and Parameters in all various neural networks, such as Linear、 CNN、 RNN、 GCN、Transformer(Bert、LlaMA etc Large Language Model)
I tried to calculate the FLOPs of EfficientNet, but there were some errors. Can you help me? ` File "/.conda/envs/test/lib/python3.8/site-packages/calflops/pytorch_ops.py", line 361, in newFunc flops, macs = funcFlopCompute(*args, **kwds) File...
https://github.com/MrYxJ/calculate-flops.pytorch/blob/29ca58443fc15252888de377b842fa0308a5cfe8/calflops/pytorch_ops.py#L248 we calculate the upsample flops but never set them to `flops = `. Does this become an issue and lead to incorrect numbers for subsequent layers?
What is the difference between setting the parameter 'forward_mode' in function calculate_flops() as 'generate' and setting it as 'forward' (default)?
You can add a CITATION.cff file to the root of a repository to let others know how you would like them to cite your work.
I am trying to run: ```python from calflops import calculate_flops_hf model_name = "EleutherAI/pythia-70m" batch_size = 1 max_seq_length = 10 flops, macs, params = calculate_flops_hf(model_name, input_shape=(batch_size, max_seq_length)) ``` Here is the...
For models with two inputs, can I wrapper them like this? ```python class Wrapper(nn.Module): def __init__(self, model): super(Wrapper, self).__init__() self.model = model def forward(self, x): class_labels = torch.eye(1000, device=device)[torch.randint(0, 1000,...
I was trying to calculate the flops of InstructPix2Pix from huggingface using the method described in the README, but I got this error: ``` ValueError: Library `diffusers` is not supported...
Hi, the input shape of point cloud is not the shape of image. And what should I input?
Though 是FLOPs is around twice as much as MACS, isn't the proportion of each submodule supposed to be equal.
