Closed RAgNarokBear closed 11 months ago
Thanks for your observation regarding the CUDA and CPU conversion in our code. While we developed and tested on an A100 GPU with 40 GB memory and didn't encounter memory-related challenges, we understand the potential implications of this in different setups like GPU with less memory. We'll make the necessary adjustments. Thank you for bringing this to our attention.
I found a bug in your code. In pivot_tuning.py, line 79, it is supposed to be self.noisesave.append(noise), instead of self.noise_save.append(noises), which leads to continuously increasing cuda usage.