canqin001

Hi, When I edit the code, I find a potential bug that might cause an error. In line 28 of utils/utils.py, you have returned optimizer_g, optimizer_c as the outputs of...

No module named 'torch.cuda.amp'

2

When I run "python run.py", there is an error as No module named 'torch.cuda.amp'. Does anyone know how to address it?

No such file or directory: 'configs/first_stage_models/vq-f4/model.yaml'

7

Hi, When I am trying to train a LDM model as "CUDA_VISIBLE_DEVICES=1 python main.py --base configs/latent-diffusion/lsun_bedrooms-ldm-vq-4.yaml -t --gpus 0", there is an error massage as "No such file or directory:...

Evaluation Codes on COCO dataset

9

Dear authors, I noticed that COCO is an essential benchmark for evaluating text-to-image generation. May I ask for the COCO dataset's evaluation code for computing IS and FID? Thank you...

Error of running inference on multiple GPUs

5

Hi, @Doubiiu thank you for the impressive work! I met an error when running `sh scripts/run_mp.sh 1024`: `RuntimeError: Expected all tensors to be on the same device, but found at...

OOM for 16x512x512 when bs is 8

1

I noticed the default bs for 16x512x512 is 8 (https://github.com/hpcaitech/Open-Sora/blob/b4f47c8771b8b7d8036f5a2f6c9de5f65c2735a9/configs/opensora/train/16x512x512.py#L52) . But I can only run it with the bs as 2. I use the H100 with 80G for each...

Large gap on "usps -> mnist"

2

I find that there is a big gap on u2m. I have tried some tricks, but the performance is still below the paper by 4%-5%. Do you have any ideas...

Which BDD100K data is used for training?

2

Hi @pixeli99 , thank you for your amazing work! I just read your code and wondering which subset of bdd100 data (I have not found the "bdd100k/images/track/mini" in their official...

Loss become nan after several epoches

5

Thank you for the amazing open-source project! I recently applied this code to train a model on ucf101, but the loss becomes nan after several epochs. Does anyone have similar...