IndexError: index 999 is out of bounds for dimension 0 with size 51

[xalteropsx]

File "Z:\software\python11\Lib\site-packages\diffusers\pipelines\stable_diffusion_xl\pipeline_stable_diffusion_xl.py", line 1230, in __call__
   latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "Z:\software\python11\Lib\site-packages\diffusers\schedulers\scheduling_dpmsolver_multistep.py", line 972, in step
   model_output = self.convert_model_output(model_output, sample=sample)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "Z:\software\python11\Lib\site-packages\diffusers\schedulers\scheduling_dpmsolver_multistep.py", line 561, in convert_model_output
   sigma = self.sigmas[self.step_index]
           ~~~~~~~~~~~^^^^^^^^^^^^^^^^^
IndexError: index 999 is out of bounds for dimension 0 with size 51

always getting this issue in batch img

[a-r-r-o-w]

Please refer to #10003 for the time being.

Maybe better warnings for possible configurations should be raised in the future? cc @yiyixuxu

[yiyixuxu]

@a-r-r-o-w oh but he says it's an issue with batched input,

@xalteropsx can you provide more details? which scheduler/pipeline you are running? possible to share a reproducible code example?

[xalteropsx]

not sure it feel like my graphic card driver issuing it i will let u know once i check and will try to reproduce with code if it didnt go away

[xalteropsx]

i closing this for now this problem arrise may be due resizable bar setting not appear in my gigabyte bios , due to exceed of memory causing image currption then this error appear itself on the next run ,not sure how long this gigabyte support ticket take but have to wait around for weeks or days

not entierly sure if this is the cause of error but i will try to find it little better meanwhile i close this issue and open it later on seems like dealing with gpu issue need to fix some currpt image issue first before going on this one