Repository Issue Activity (beta)

#362 Enable Conditioning in DDPM/DDIM Diffusion code and Add Sample Representation

#121 Sharing answer from the question "How many iterations do I need?"

#349 Conditional diffusion

#361 Is it applicable to time series data？

#310 No available kernel

#357 Why not interleave sin/cos as in the original Transformer positional encoding?

#359 DDPM or DDIM or Both

#360 Loss nan

#309 Fast attention in Windows possible?

#100 Did anyone get good CIFAR10 results?

#289 Size of dataset

#291 when do we use objective "pred_x_start"?

#354 A800 not working

#162 Where does the predict_noise_from_start function and the predict_noise_from_start function correspond to the paper？

#348 How to load the model i trained and denoise from a given step?

#353 How to generate images through trained model files？

#30 Question about DDPM: Meaning of L_{0} (in loss) , and how it gets simplified into L_{simple}?

#355 Implementations to denoise 3d images

#65 clip_denoised purpose?

#47 differences on sampling between the paper and your implementation

#347 Clarification on T in DDPM paper and train_num_steps in the code

#344 Is there workaround for 3090

#58 "pred_x0" preforms better than "pred_noise"

#201 solved #201

#332 sampling loop time step