ocannl
ocannl copied to clipboard
Support quantization for optimizers: low-bit optimizers
https://github.com/pytorch/ao/tree/main/torchao/prototype/low_bit_optim https://x.com/marksaroufim/status/1809398186198593566