Closed agentj01 closed 8 months ago
Same with BLIP captioning
Batch size 1: RuntimeError: The size of tensor a (10) must match the size of tensor b (100) at non-singleton dimension 0
Batch size 5: RuntimeError: The size of tensor a (50) must match the size of tensor b (500) at non-singleton dimension 0
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ c:\AI\kohya_ss\finetune\make_captions.py:200 in
I get the same error. When you set batch_size and number of beams to 1 the error isn´t thrown
Well found the issue, not even sure why it is a thing. If you leave the default name "last" and inference the lora directly you get that error i posted. Just renaming the LoRA fix it. Tested and reproduced on 21.8.5 and lastest commit (21.8.8).
Well found the issue, not even sure why it is a thing. If you leave the default name "last" and inference the lora directly you get that error i posted. Just renaming the LoRA fix it. Tested and reproduced on 21.8.5 and lastest commit (21.8.8).
I'm not sure this fixed it I just tried your fix and it didn't work for me I still got this message I used images of varying sizes though, because some of the images had some stuff that needed to be removed so I cropped some of the images. None of the images are traditional image sizes, I'm sure most of them weren't multiples of 2, could this possible be the problem?
Here is the Traceback:
Traceback (most recent call last):
File "H:\StableDiffusion\stable-diffusion-webui\modules\call_queue.py", line 58, in f
res = list(func(*args, kwargs))
File "H:\StableDiffusion\stable-diffusion-webui\modules\call_queue.py", line 37, in f
res = func(*args, *kwargs)
File "H:\StableDiffusion\stable-diffusion-webui\modules\txt2img.py", line 62, in txt2img
processed = processing.process_images(p)
File "H:\StableDiffusion\stable-diffusion-webui\modules\processing.py", line 677, in process_images
res = process_images_inner(p)
File "H:\StableDiffusion\stable-diffusion-webui\modules\processing.py", line 794, in process_images_inner
samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts)
File "H:\StableDiffusion\stable-diffusion-webui\modules\processing.py", line 1054, in sample
samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x))
File "H:\StableDiffusion\stable-diffusion-webui\modules\sd_samplers_kdiffusion.py", line 464, in sample
samples = self.launch_sampling(steps, lambda: self.func(self.model_wrap_cfg, x, extra_args={
File "H:\StableDiffusion\stable-diffusion-webui\modules\sd_samplers_kdiffusion.py", line 303, in launch_sampling
return func()
File "H:\StableDiffusion\stable-diffusion-webui\modules\sd_samplers_kdiffusion.py", line 464, in
I'm also getting this error when attempting BLIP captioning: RuntimeError: The size of tensor a (12) must match the size of tensor b (144) at non-singleton dimension 0
. Probably not a coincidence that the size of tensor a (12) is equal to the beam count I chose.
Edit: Seems this is likely a dupe of #1236. There is a workaround for this noted there for BLIP captioning, though this may or may not work for SDXL training.
I trained other SDXL LoRA on previous commit and were working fine. I was on version 21.8.5. I haven't git pull but then LoRA trained on it was giving this error. The only thing i remember is a LoRA/Lycoris module updated when i started the Gui. I am also having this error whatever the version i am trying, so git checkout another commit or downloading another version not gonna fix it.