VainF / Diff-Pruning

[NeurIPS 2023] Structural Pruning for Diffusion Models
Apache License 2.0
141 stars 9 forks source link

Error in prune_LDM for CelebaHQ Dataset #10

Open Satyab23 opened 1 month ago

Satyab23 commented 1 month ago

Dear Authors, Thank you for sharing the amazing and inspiring pruning work!

As mentioned in celebahq-ldm-vq-4.yaml line59, CelebaHQ is unconditional stage_config While running the pruning, I am getting the following error:

Pruning ... Traceback (most recent call last): File "prune_ldm_celeba.py", line 75, in uc = model.get_learned_conditioning( File "/raid/akshay/satyabrat/DiffPruneLDM/Diff-Pruning/ldm_exp/ldm/models/diffusion/ddpm.py", line 561, in get_learned_conditioning c = self.cond_stage_model(c) TypeError: 'NoneType' object is not callable

Command used to run: CUDA_VISIBLE_DEVICES=1 python prune_ldm.py --sparsity 0.3 --pruner random

Could you please share your code and config for CelebaHQ Dataset as reported in the paper.

Satyab23 commented 1 month ago

Dear @VainF ,

Did you get the chance to look into it. Kindly look into it.