A occasional issue that Tensor size do not match

[XuanjiaZ]

While evaluating InstantDrag on the DragBench Dataset, an issue occurred when processing an image located at "DragBench/art_work/PJC_2023-09-15-1052-01".
The detail error info as follows:

Traceback (most recent call last):
  File "/home/InstantDrag-main/demo/run_drag_diffusion_instantdrag.py", line 120, in <module>
    out_image = pipeline.run(original_image,
  File "/home/InstantDrag-main/demo/demo_utils.py", line 240, in run
    edited_image = self.drag(original_image, selected_points, save_results)
  File "/home/InstantDrag-main/demo/demo_utils.py", line 200, in drag
    edited_image = self.flowdiffusion(**kwargs).images[0]
  File "/root/miniconda3/envs/instantdrag/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/InstantDrag-main/demo/../flowdiffusion/pipeline.py", line 247, in __call__
    image_latents = self.prepare_image_latents(
  File "/home/InstantDrag-main/demo/../flowdiffusion/pipeline.py", line 470, in prepare_image_latents
    image_latents_flow_cond = torch.cat([image_latents, flow.to(device)], dim=1)
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 63 but got size 64 for tensor number 1 in the list.

Other images prior to this one were successfully dragged.