cfeng16 comments

Results 7 comments of


                                            cfeng16

[Bug] Predict method does not finish

Hi @ArlindKadra, I face the same problem now, and I use the AMI corpus dataset. During the training, I concatenate all the training samples as a single numpy matrix, and...

One of the variables needed for gradient computation has been modified by an inplace operation

Hi, I met the same problem, Have you solved it?

Keep previous settings by default?

Same question ^^

How to handle gradient accumulation with multiple models ?

Does it now support gradient accumulation for multiple models?

How to handle gradient accumulation with multiple models ?

Can we use gradient accumulation for multiple models in distributed training?

is somebody knows how many validation file there are?

btw where can i find the tokenizer?

Multi-node Multi-gpu Wan1.3b training

@Artiprocher Would you mind sharing some instructions about how to do multi-node training for WAN-1.3b by DiffSynth-Studio? I really appreciate any help you can provide.