Merge-Stable-Diffusion-models-without-distortion
Merge-Stable-Diffusion-models-without-distortion copied to clipboard
ogkalu2
Adaptation of the merging method described in the paper - Git Re-Basin: Merging Models modulo Permutation Symmetries (https://arxiv.org/abs/2209.04836) for Stable Diffusion
First of all, great work, used this back then to screw around with models. Got back into SD, tried to merge 2.x models, got this error: `KeyError: 'cond_stage_model.model.transformer.text_model.embeddings.position_ids'` Any 2.x...
Do you have any plans to Develop a PermutationSpec for SDXL? If not, would you be so kind as to explain a bit more how you arrived at the one...
From the paper, I see Algorithm 1 and 2 are implemented. Using `scipy` for the SOLVELAP look nice to me. However in the bottom session, there is one more algorithm...
Hi, I'm interested in understanding what the code does. ``` **easyblock("model.diffusion_model.output_blocks.6.0", "P_bg208","P_bg209"), **conv("model.diffusion_model.output_blocks.6.0.skip_connection","P_bg210","P_bg211"), **norm("model.diffusion_model.output_blocks.6.1.norm", "P_bg212"), **conv("model.diffusion_model.output_blocks.6.1.proj_in", "P_bg212", "P_bg213"), **dense("model.diffusion_model.output_blocks.6.1.transformer_blocks.0.attn1.to_q", "P_bg214", "P_bg215", bias=False), ``` in some sections, the bg's are disconnected...
KeyError
on line: 55 ` theta_0 = {key: (1 - (new_alpha)) * theta_0[key] + (new_alpha) * value for key, value in theta_1.items() if "model" in key and key in theta_1}` and...
Error: ``` /content/merger Using half precision --------------------- ITERATION 1 --------------------- new alpha = 0.045 Traceback (most recent call last): File "SD_rebasin_merge.py", line 55, in theta_0 = {key: (1 - (new_alpha))...
I installed torch==1.11.0 when trying to merge i get this Using half precision --------------------- ITERATION 1 --------------------- new alpha = 0.05 FINDING PERMUTATIONS P_bg337: -0.5 P_bg324: -0.5 P_bg358: -0.125 P_bg324:...
Using half precision --------------------- ITERATION 1 --------------------- new alpha = 0.05 Traceback (most recent call last): File "I:\Dreambooth-Stable-Diffusion\Merge-Stable-Diffusion-models-without-distortion\SD_rebasin_merge.py", line 55, in theta_0 = {key: (1 - (new_alpha)) * theta_0[key] +...
Lot of people have trouble running it, this would make it easier , also providing a colab notebook that was tested and working would be great
I have merged two dreambooth models (A.ckpt and B.ckpt). The default parameters (alpha and ITERATION) are used. The size of the merged file (merged. ckpt) is the same as A...
