yuhengliu02 / pyramid-discrete-diffusion

Official implementation of paper "Pyramid Diffusion for Fine 3D Large Scene Generation" (ECCV 2024 Oral)
https://yuheng.ink/project-page/pyramid-discrete-diffusion/
MIT License
106 stars 8 forks source link

RuntimeError: Error(s) in loading state_dict for Con_Diffusion: #9

Open peakybr1 opened 3 weeks ago

peakybr1 commented 3 weeks ago

Hello, the following error occurred while running the last piece of code

python launch.py -c configs/infinite_s_2_to_s_3.yaml -n s_2_to_s_3
/data/fgp/pyramid-discrete-diffusion/utils/args_process.py:36: UserWarning: You have chosen a specific GPU. This will completely disable DDP.
  warnings.warn('You have chosen a specific GPU. This will completely disable DDP.')
Storing logs in: infinite_generation/s_2_to_s_3
Traceback (most recent call last):
  File "/data/fgp/pyramid-discrete-diffusion/launch.py", line 106, in <module>
    main()
  File "/data/fgp/pyramid-discrete-diffusion/launch.py", line 45, in main
    launch(start, args.ngpus_per_node, args.num_node, args.node_rank, args.dist_url, args=(args,))
  File "/data/fgp/pyramid-discrete-diffusion/utils/cuda.py", line 48, in launch
    fn(local_rank, *args)
  File "/data/fgp/pyramid-discrete-diffusion/launch.py", line 103, in start
    exp.run(epochs = args.epochs)
  File "/data/fgp/pyramid-discrete-diffusion/train.py", line 53, in run
    self.resume()
  File "/data/fgp/pyramid-discrete-diffusion/train.py", line 274, in resume
    self.checkpoint_load(self.args.resume_path)
  File "/data/fgp/pyramid-discrete-diffusion/train.py", line 307, in checkpoint_load
    self.model.load_state_dict(checkpoint['model'], strict=False)
  File "/home/b1107/user/gmy/anaconda3/envs/fgp_pdd/lib/python3.9/site-packages/torch/nn/modules/module.py", line 2584, in load_state_dict
    raise RuntimeError(
RuntimeError: Error(s) in loading state_dict for Con_Diffusion:
        size mismatch for _denoise_fn.conv_in.weight: copying a param with shape torch.Size([32, 33, 1, 1, 1]) from checkpoint, the shape in current model is torch.Size([32, 34, 1, 1, 1]).