Open RohollahHS opened 1 month ago
Hi, thank you for your interest. I attached the logs here VQVAE_HML3D_run.log VQVAE_KIT_run.log
Thanks for sharing. Could you please explain why the FID metric in these two log files looks random? There is no sign of continuous decrease in FID, and I think it's a weird. For example, in VQVAE_KIT_run.log, it always fluctuates around 0.5 for 300,000 iterations.
Great work. Thanks for sharing the code. Could you please share the log for training the VQ-VAE for both KIT-ML dataset and HumanML3D (t2m) dataset? Thanks