search

JoePenna / Dreambooth-Stable-Diffusion

Implementation of Dreambooth (https://arxiv.org/abs/2208.12242) by way of Textual Inversion (https://arxiv.org/abs/2208.01618) for Stable Diffusion (https://arxiv.org/abs/2112.10752). Tweaks focused on training faces, objects, and styles.
MIT License
3.19k stars 558 forks source link

Error: HeaderTooLarge #202

Open Vizzana opened 4 months ago

Vizzana commented 4 months ago

the pretrained stable difussion model can not be loaded sucessfully, the returned error is as below:

with safe_open(checkpoint_file, framework="pt") as f: safetensors_rust.SafetensorError: Error while deserializing header: HeaderTooLarge

yushan777 commented 4 months ago

JoePenna's trainer doesn't support safetensors. it expects a ckpt format. You will need to convert it or use a more up-to-date trainer such as kohya

Vizzana commented 4 months ago

JoePenna's trainer doesn't support safetensors. it expects a ckpt format. You will need to convert it or use a more up-to-date trainer such as kohya

but, in fact, the format i used is just 'ckpt' ! I don't kown why this program always occurs this problem. it has exhaused me for one afternoon

yushan777 commented 4 months ago

I haven't used it in a very long time. are you running it locally or on vast.ai? also which model are you trying to train on top of a finetuned model or the base SD1.5?

if all else fails you can try using kohya_ss or onetrainer which is actively maintained.

Vizzana commented 4 months ago

i have solved this problem by re-download all pretrained models used in this program. and now the program is wording on my server!