Text-To-Video-Finetuning icon indicating copy to clipboard operation
Text-To-Video-Finetuning copied to clipboard

init_video problem

Open Yuuki-00 opened this issue 1 year ago • 0 comments
trafficstars

when running inference with an init_video, runtime error happen. スクリーンショット 2023-12-12 19 50 20

timesteps = timesteps[0] (inference.py line194) → for timestep in timesteps(diffusers/schedulers/scheduling_dpmsolver_multistep.py line900) This error, "iteration over a 0-d tensor," occurs when attempting to iterate over a 0-dimensional tensor using a for loop. A 0-dimensional tensor represents a scalar value, effectively having no elements to iterate through. Consequently, attempting iteration on such a tensor is invalid and results in an error because there are no elements within the tensor to iterate over.

It's possible that updates to the Git repository might be causing this issue. Are there any potential solutions to this problem?

Yuuki-00 avatar Dec 12 '23 10:12 Yuuki-00