Yunhao Chen

GAN！ GAN！ GAN！

I encounter the same problem today. And I solve it by installing torchinfo(pip install torchinfo)

> ```python > # losses.sliced_sm > def sliced_score_estimation(score_net, samples, n_particles=1): > dup_samples = samples.unsqueeze(0).expand(n_particles, *samples.shape).contiguous().view(-1, *samples.shape[1:]) > dup_samples.requires_grad_(True) > vectors = torch.randn_like(dup_samples) > vectors = vectors / torch.norm(vectors, dim=-1, keepdim=True)...

> I noticed another question. The multiplication of vectors and grad1/2 is element-wise but in the paper, it is matrix multiplication. Or I misunderstand the theory? They are equivalent. Flatten...

You can refer to CLIPAG https://arxiv.org/abs/2306.16805. The paper has demonstrated this phenomenon. It might be helpful. @realfolkcode