search

li-ronghui / LODGE

The code the CVPR2024 paper Lodge: A Coarse to Fine Diffusion Network for Long Dance Generation Guided by the Characteristic Dance Primitives
105 stars 6 forks source link

Inference error #1

Closed ForrestPi closed 6 months ago

ForrestPi commented 6 months ago

File "/home/XXX/dance/LODGE/dld/models/architectures/diffusion.py", line 277, in ddim_sample pred_noise, xstart, * = self.model_predictions(x, cond, genre, time_cond, clip_x_start = self.clip_denoised) File "/home/XXX/dance/LODGE/dld/models/architectures/diffusion.py", line 170, in model_predictions model_output = self.model.guided_forward(x, cond, genre, t, weight) File "/home/XXX/dance/LODGE/dld/models/architectures/model.py", line 480, in guided_forward unc = self.forward(x, cond_embed, genre, times, cond_drop_prob=1) File "/home/XXX/dance/LODGE/dld/models/architectures/model.py", line 506, in forward cond_tokens = torch.where(keep_mask_embed, cond_tokens, null_cond_embed) RuntimeError: The size of tensor a (768) must match the size of tensor b (1024) at non-singleton dimension 1

li-ronghui commented 6 months ago

Hi, I'm glad you ran our code! I checked the Inference code in my enviroment, and works well : )

This error show that 'The size of tensor a (768) must match the size of tensor b (1024) '. But this repo didn't set parameter 768. Maybe you can try the origin experiment yaml file and run: python infer_lodge.py --cfg exp/Local_Module/FineDance_FineTuneV2_Local/local_train.yaml --cfg_assets configs/data/assets.yaml

I hope this command can help you, if not, please let me know and provide more details. Good luck！