choose_qparams_and_quantize_affine_hqq

torchao.quantization.choose_qparams_and_quantize_affine_hqq(tensor: ~torch.Tensor, nbits: float = 4, group_size: int = 64, optimize: bool = True, axis: int = 1, compute_dtype: ~torch.dtype = torch.float16, device: str = 'cuda', verbose: bool = False, raw_output: bool = False, optimize_weights: ~typing.Callable = <function optimize_weights_proximal_legacy>) → tuple