Closed wintercat1994 closed 1 day ago
training details will be released with new report soon.
looking forward to the training instructions.
looking forward to the training instructions.
If you just want some training instructions, I believe the arxiv version is comprehensive enough though it's slightly different from the implementation. (latent shuffle is cancelled, noisy training followed by a short drop training for fine-tuning.)
The new paper is much shorter and many important details are removed.
Hello, I have seen the training code, but the training steps and parameters are not provided in your instructions. Could you please supplement them?