Liuhui Wang comments

Results 13 comments of


                                            Liuhui Wang

A bug shows when the batch_size sets 1

ok, I got it. The batchnorm layer needs more than one sample to calculate the parameters in it.

pre-trained model

@leaderj1001 Could you please provide a pre-trained model?

ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.

I can't install version 0.4.1 of Pytorch. How can I solve this problem?

Hello, in the training of the following mistakes, how to solve it

x, y are determined by the fmap_size. You should set the fmap_size according to your dataset.

Single GPU

I have the same question. If I use a single GPU, the code will warn that 'You have chosen a specific GPU. This will completely disable data parallelism.' And I...

CUDA error: unknown error

> @liuhui0401 - this seems like a cuda error, or a bad state that the GPUs are in. If you power cycle the machine, does nvidia-smi work? Yes. But if...

When using the train_ucf101.py file for model training, the model evaluation part cannot enter the loop.

I think you should check out your evaluation dataloader. It may be empty.

How to optimize B, which is the all zero matrix in the Lora method?

I have the same problem. When I perform gradient backpropagation, the weight of A can be updated, but the weight of B is always 0. Please tell me how should...

Fine-tune with pre-trained model

You can use torch.save()

Where to start from a Beginner's point of view ?

Firstly, you should read README.md to find the correct way to run this model. Secondly, you should find out the input and output format. By the way, pdb is a...