# pylint: disable=redefined-builtin,no-member,protected-access from typing import Any, Dict, List, Tuple, Optional, Union from dataclasses import dataclass from enum import Enum import re import torch from transformers.quantizers import HfQuantizer from diffusers.quantizers.base import DiffusersQuantizer from diffusers.quantizers.quantization_config import QuantizationConfigMixin from diffusers.utils import get_module_from_name from modules import devices, shared from .common import dtype_dict, accepted_weights, use_tensorwise_fp8_matmul, allowed_types, conv_types, conv_transpose_types, use_contiguous_mm from .dequantizer import dequantizer_dict, dequantize_sdnq_model from .forward import get_forward_func class QuantizationMethod(str, Enum): SDNQ = "sdnq" def get_scale_asymmetric(weight: torch.FloatTensor, reduction_axes: Union[int, List[int]], weights_dtype: str) -> Tuple[torch.FloatTensor, torch.FloatTensor]: zero_point = torch.amin(weight, dim=reduction_axes, keepdims=True) scale = torch.amax(weight, dim=reduction_axes, keepdims=True).sub_(zero_point).div_(dtype_dict[weights_dtype]["max"] - dtype_dict[weights_dtype]["min"]) if dtype_dict[weights_dtype]["min"] != 0: zero_point.sub_(torch.mul(scale, dtype_dict[weights_dtype]["min"])) return scale, zero_point def get_scale_symmetric(weight: torch.FloatTensor, reduction_axes: Union[int, List[int]], weights_dtype: str) -> torch.FloatTensor: return torch.amax(weight.abs(), dim=reduction_axes, keepdims=True).div_(dtype_dict[weights_dtype]["max"]) def quantize_weight(weight: torch.FloatTensor, reduction_axes: Union[int, List[int]], weights_dtype: str) -> Tuple[torch.Tensor, torch.FloatTensor, torch.FloatTensor]: if dtype_dict[weights_dtype]["is_unsigned"]: scale, zero_point = get_scale_asymmetric(weight, reduction_axes, weights_dtype) quantized_weight = torch.sub(weight, zero_point).div_(scale) else: scale = get_scale_symmetric(weight, reduction_axes, weights_dtype) quantized_weight = torch.div(weight, scale) zero_point = None if dtype_dict[weights_dtype]["is_integer"]: quantized_weight.round_() else: quantized_weight.nan_to_num_() quantized_weight = quantized_weight.clamp_(dtype_dict[weights_dtype]["min"], dtype_dict[weights_dtype]["max"]).to(dtype_dict[weights_dtype]["torch_dtype"]) return quantized_weight, scale, zero_point def apply_svdquant(weight: torch.FloatTensor, rank: int = 32) -> Tuple[torch.FloatTensor, torch.FloatTensor, torch.FloatTensor]: reshape_weight = False if weight.ndim > 2: # convs reshape_weight = True weight_shape = weight.shape weight = weight.flatten(1,-1) U, S, svd_down = torch.svd_lowrank(weight, q=rank) svd_up = torch.mul(U, S.unsqueeze(0)) svd_down = svd_down.t_() weight = weight.sub_(torch.mm(svd_up, svd_down)) if reshape_weight: weight = weight.unflatten(-1, (*weight_shape[1:],)) # pylint: disable=possibly-used-before-assignment return weight, svd_up, svd_down @devices.inference_context() def sdnq_quantize_layer(layer, weights_dtype="int8", torch_dtype=None, group_size=0, svd_rank=32, use_svd=False, quant_conv=False, use_quantized_matmul=False, use_quantized_matmul_conv=False, dequantize_fp32=False, non_blocking=False, quantization_device=None, return_device=None, param_name=None): # pylint: disable=unused-argument layer_class_name = layer.__class__.__name__ if layer_class_name in allowed_types: num_of_groups = 1 is_conv_type = False is_conv_transpose_type = False is_linear_type = False result_shape = None original_shape = layer.weight.shape if torch_dtype is None: torch_dtype = layer.weight.dtype if layer_class_name in conv_types: if not quant_conv: return layer if dtype_dict[weights_dtype]["num_bits"] < 4: weights_dtype = "uint4" is_conv_type = True reduction_axes = 1 output_channel_size, channel_size = layer.weight.shape[:2] group_channel_size = channel_size // layer.groups use_quantized_matmul = False if use_quantized_matmul_conv: use_quantized_matmul = group_channel_size >= 32 and output_channel_size >= 32 if use_quantized_matmul and not dtype_dict[weights_dtype]["is_integer"]: use_quantized_matmul = output_channel_size % 16 == 0 and group_channel_size % 16 == 0 if use_quantized_matmul and dtype_dict[weights_dtype]["num_bits"] == 8: result_shape = layer.weight.shape layer.weight.data = layer.weight.flatten(1,-1) reduction_axes = -1 elif layer_class_name in conv_transpose_types: if not quant_conv: return layer if dtype_dict[weights_dtype]["num_bits"] < 4: weights_dtype = "uint4" is_conv_transpose_type = True reduction_axes = 0 channel_size, output_channel_size = layer.weight.shape[:2] use_quantized_matmul = False else: is_linear_type = True reduction_axes = -1 try: output_channel_size, channel_size = layer.weight.shape except Exception as e: raise ValueError(f"SDNQ: param_name={param_name} layer_class_name={layer_class_name} layer_weight_shape={layer.weight.shape} weights_dtype={weights_dtype} unsupported") from e if use_quantized_matmul: use_quantized_matmul = channel_size >= 32 and output_channel_size >= 32 if use_quantized_matmul: if dtype_dict[weights_dtype]["is_integer"]: use_quantized_matmul = output_channel_size % 8 == 0 and channel_size % 8 == 0 else: use_quantized_matmul = output_channel_size % 16 == 0 and channel_size % 16 == 0 layer.weight.requires_grad = False if return_device is None: return_device = layer.weight.device if quantization_device is not None: layer.weight.data = layer.weight.to(quantization_device, non_blocking=non_blocking) if layer.weight.dtype != torch.float32: layer.weight.data = layer.weight.to(dtype=torch.float32) if use_svd: layer.weight.data, svd_up, svd_down = apply_svdquant(layer.weight, rank=svd_rank) if use_quantized_matmul: svd_up = svd_up.t_() svd_down = svd_down.t_() else: svd_up, svd_down = None, None if group_size == 0: if use_quantized_matmul and dtype_dict[weights_dtype]["num_bits"] >= 6: group_size = -1 elif is_linear_type: group_size = 2 ** ((2 if not use_svd else 3) + dtype_dict[weights_dtype]["num_bits"]) else: group_size = 2 ** ((1 if not use_svd else 2) + dtype_dict[weights_dtype]["num_bits"]) elif use_quantized_matmul and dtype_dict[weights_dtype]["num_bits"] == 8: group_size = -1 # override user value, re-quantizing 8bit into 8bit is pointless elif group_size != -1 and not is_linear_type: group_size = max(group_size // 2, 1) if group_size > 0: if group_size >= channel_size: group_size = channel_size num_of_groups = 1 else: num_of_groups = channel_size // group_size while num_of_groups * group_size != channel_size: # find something divisible num_of_groups -= 1 if num_of_groups <= 1: group_size = channel_size num_of_groups = 1 break group_size = channel_size // num_of_groups group_size = int(group_size) num_of_groups = int(num_of_groups) if num_of_groups > 1: if result_shape is None: result_shape = layer.weight.shape new_shape = list(result_shape) if is_conv_type: # output_channel_size, channel_size, X, X # output_channel_size, num_of_groups, group_size, X, X new_shape[1] = group_size new_shape.insert(1, num_of_groups) reduction_axes = 2 elif is_conv_transpose_type: #channel_size, output_channel_size, X, X #num_of_groups, group_size, output_channel_size, X, X new_shape[0] = group_size new_shape.insert(0, num_of_groups) reduction_axes = 1 elif is_linear_type: # output_channel_size, channel_size # output_channel_size, num_of_groups, group_size last_dim_index = layer.weight.ndim new_shape[last_dim_index - 1 : last_dim_index] = (num_of_groups, group_size) layer.weight.data = layer.weight.reshape(new_shape) layer.weight.data, scale, zero_point = quantize_weight(layer.weight, reduction_axes, weights_dtype) if not dequantize_fp32 and not (use_quantized_matmul and not dtype_dict[weights_dtype]["is_integer"] and not use_tensorwise_fp8_matmul): scale = scale.to(dtype=torch_dtype) if zero_point is not None: zero_point = zero_point.to(dtype=torch_dtype) if svd_up is not None: svd_up = svd_up.to(dtype=torch_dtype) svd_down = svd_down.to(dtype=torch_dtype) re_quantize_for_matmul = (num_of_groups > 1 or zero_point is not None) if use_quantized_matmul and not re_quantize_for_matmul: scale.t_() layer.weight.t_() if use_contiguous_mm: layer.weight.data = layer.weight.contiguous() elif layer.weight.is_contiguous(): layer.weight.data = layer.weight.t_().contiguous().t_() if not use_tensorwise_fp8_matmul and not dtype_dict[weights_dtype]["is_integer"]: scale = scale.to(dtype=torch.float32) scale = scale.to(return_device, non_blocking=non_blocking) layer.scale = torch.nn.Parameter(scale, requires_grad=False) if zero_point is not None: zero_point = zero_point.to(return_device, non_blocking=non_blocking) layer.zero_point = torch.nn.Parameter(zero_point, requires_grad=False) else: layer.zero_point = None if svd_up is not None: svd_up = svd_up.to(return_device, non_blocking=non_blocking) svd_down = svd_down.to(return_device, non_blocking=non_blocking) layer.svd_up = torch.nn.Parameter(svd_up, requires_grad=False) layer.svd_down = torch.nn.Parameter(svd_down, requires_grad=False) else: layer.svd_up, layer.svd_down = None, None layer.sdnq_dequantizer = dequantizer_dict[weights_dtype]( quantized_weight_shape=layer.weight.shape, result_dtype=torch_dtype, result_shape=result_shape, original_shape=original_shape, weights_dtype=weights_dtype, use_quantized_matmul=use_quantized_matmul, re_quantize_for_matmul=re_quantize_for_matmul, ) layer.weight.data = layer.sdnq_dequantizer.pack_weight(layer.weight).to(return_device, non_blocking=non_blocking) layer.forward = get_forward_func(layer_class_name, use_quantized_matmul, dtype_dict[weights_dtype]["is_integer"], use_tensorwise_fp8_matmul) layer.forward = layer.forward.__get__(layer, layer.__class__) return layer def apply_sdnq_to_module(model, weights_dtype="int8", torch_dtype=None, group_size=0, svd_rank=32, use_svd=False, quant_conv=False, use_quantized_matmul=False, use_quantized_matmul_conv=False, dequantize_fp32=False, non_blocking=False, quantization_device=None, return_device=None, modules_to_not_convert: List[str] = None, modules_dtype_dict: Dict[str, List[str]] = None, full_param_name="", op=None): # pylint: disable=unused-argument has_children = list(model.children()) if not has_children: return model if modules_to_not_convert is None: modules_to_not_convert = [] if modules_dtype_dict is None: modules_dtype_dict = {} for param_name, module in model.named_children(): if param_name == "sdnq_dequantizer": continue if full_param_name: param_name = full_param_name + "." + param_name if hasattr(module, "weight") and module.weight is not None: param_name = param_name + ".weight" split_param_name = param_name.split(".") if ( param_name in modules_to_not_convert or any(param in split_param_name for param in modules_to_not_convert) or any("*" in param and re.match(param.replace(".*", "\\.*").replace("*", ".*"), param_name) for param in modules_to_not_convert) ): continue layer_class_name = module.__class__.__name__ if layer_class_name in allowed_types: if (layer_class_name in conv_types or layer_class_name in conv_transpose_types) and not quant_conv: continue else: continue if len(modules_dtype_dict.keys()) > 0: for key, value in modules_dtype_dict.items(): if ( param_name in value or any(param in split_param_name for param in value) or any("*" in param and re.match(param.replace(".*", "\\.*").replace("*", ".*"), param_name) for param in value) ): key = key.lower() if key in {"8bit", "8bits"}: if dtype_dict[weights_dtype]["num_bits"] != 8: weights_dtype = "int8" elif key.startswith("minimum_"): minimum_bits_str = key.removeprefix("minimum_").removesuffix("bits").removesuffix("bit") minimum_bits = int(minimum_bits_str) if dtype_dict[weights_dtype]["num_bits"] < minimum_bits: weights_dtype = "int" + minimum_bits_str if minimum_bits <= 4: weights_dtype = "u" + weights_dtype else: weights_dtype = key module = sdnq_quantize_layer( module, weights_dtype=weights_dtype, torch_dtype=torch_dtype, group_size=group_size, svd_rank=svd_rank, use_svd=use_svd, quant_conv=quant_conv, use_quantized_matmul=use_quantized_matmul, use_quantized_matmul_conv=use_quantized_matmul_conv, dequantize_fp32=dequantize_fp32, non_blocking=non_blocking, quantization_device=quantization_device, return_device=return_device, param_name=param_name, ) module = apply_sdnq_to_module( module, weights_dtype=weights_dtype, torch_dtype=torch_dtype, group_size=group_size, svd_rank=svd_rank, use_svd=use_svd, quant_conv=quant_conv, use_quantized_matmul=use_quantized_matmul, use_quantized_matmul_conv=use_quantized_matmul_conv, dequantize_fp32=dequantize_fp32, non_blocking=non_blocking, quantization_device=quantization_device, return_device=return_device, modules_to_not_convert=modules_to_not_convert, modules_dtype_dict=modules_dtype_dict, full_param_name=param_name, op=op, ) return model def add_module_skip_keys(model, modules_to_not_convert: List[str] = None, modules_dtype_dict: Dict[str, List[str]] = None): if modules_to_not_convert is None: modules_to_not_convert = [] if modules_dtype_dict is None: modules_dtype_dict = {} if getattr(model, "_keep_in_fp32_modules", None) is not None: modules_to_not_convert.extend(model._keep_in_fp32_modules) # pylint: disable=protected-access if getattr(model, "_skip_layerwise_casting_patterns", None) is not None: modules_to_not_convert.extend(model._skip_layerwise_casting_patterns) # pylint: disable=protected-access if model.__class__.__name__ == "ChromaTransformer2DModel": modules_to_not_convert.append("distilled_guidance_layer") elif model.__class__.__name__ == "QwenImageTransformer2DModel": modules_to_not_convert.extend(["transformer_blocks.0.img_mod.1.weight", "time_text_embed", "img_in", "txt_in", "proj_out", "norm_out", "pos_embed"]) if "minimum_6bit" not in modules_dtype_dict.keys(): modules_dtype_dict["minimum_6bit"] = ["img_mod"] else: modules_dtype_dict["minimum_6bit"].append("img_mod") return model, modules_to_not_convert, modules_dtype_dict def sdnq_post_load_quant( model, weights_dtype="int8", torch_dtype: torch.dtype = None, group_size: int = 0, svd_rank: int = 32, use_svd: bool = False, quant_conv: bool = False, use_quantized_matmul: bool = False, use_quantized_matmul_conv: bool = False, dequantize_fp32: bool = False, non_blocking: bool = False, add_skip_keys:bool = True, quantization_device: torch.device = None, return_device: torch.device = None, modules_to_not_convert: List[str] = None, modules_dtype_dict: Dict[str, List[str]] = None, op=None, ): if add_skip_keys: model, modules_to_not_convert, modules_dtype_dict = add_module_skip_keys(model, modules_to_not_convert, modules_dtype_dict) model.eval() model = apply_sdnq_to_module( model, weights_dtype=weights_dtype, torch_dtype=torch_dtype, group_size=group_size, svd_rank=svd_rank, use_svd=use_svd, quant_conv=quant_conv, use_quantized_matmul=use_quantized_matmul, use_quantized_matmul_conv=use_quantized_matmul_conv, dequantize_fp32=dequantize_fp32, non_blocking=non_blocking, quantization_device=quantization_device, return_device=return_device, modules_to_not_convert=modules_to_not_convert, modules_dtype_dict=modules_dtype_dict.copy(), op=op, ) model.quantization_config = SDNQConfig( weights_dtype=weights_dtype, group_size=group_size, svd_rank=svd_rank, use_svd=use_svd, quant_conv=quant_conv, use_quantized_matmul=use_quantized_matmul, use_quantized_matmul_conv=use_quantized_matmul_conv, dequantize_fp32=dequantize_fp32, non_blocking=non_blocking, quantization_device=quantization_device, return_device=return_device, modules_to_not_convert=modules_to_not_convert, modules_dtype_dict=modules_dtype_dict.copy(), ) try: model.config.quantization_config = model.quantization_config except Exception: pass model.quantization_method = QuantizationMethod.SDNQ return model class SDNQQuantizer(DiffusersQuantizer, HfQuantizer): r""" Diffusers and Transformers Quantizer for SDNQ """ requires_parameters_quantization = True use_keep_in_fp32_modules = True requires_calibration = False required_packages = None torch_dtype = None def __init__(self, quantization_config, **kwargs): super().__init__(quantization_config, **kwargs) self.modules_to_not_convert = [] self.updated_expected_keys = False def check_if_quantized_param( self, model, param_value: "torch.Tensor", param_name: str, *args, **kwargs, # pylint: disable=unused-argument ): if self.pre_quantized and self.updated_expected_keys and (param_name.endswith(".scale") or param_name.endswith(".zero_point") or param_name.endswith(".svd_up") or param_name.endswith(".svd_down")): return True if param_name.endswith(".weight"): split_param_name = param_name.split(".") if ( param_name not in self.modules_to_not_convert and not any(param in split_param_name for param in self.modules_to_not_convert) and not any("*" in param and re.match(param.replace(".*", "\\.*").replace("*", ".*"), param_name) for param in self.modules_to_not_convert) ): layer_class_name = get_module_from_name(model, param_name)[0].__class__.__name__ if layer_class_name in allowed_types: if layer_class_name in conv_types or layer_class_name in conv_transpose_types: if self.quantization_config.quant_conv: return True else: return True if param_value is not None: with devices.inference_context(): param_value.data = param_value.clone() # safetensors is unable to release the cpu memory without this return False def check_quantized_param(self, *args, **kwargs) -> bool: """ needed for transformers compatibilty, returns self.check_if_quantized_param """ return self.check_if_quantized_param(*args, **kwargs) def param_needs_quantization(self, model, param_name: str, *args, **kwargs) -> bool: """ needed for transformers compatibilty, returns self.check_if_quantized_param """ return self.check_if_quantized_param(model, None, param_name, *args, **kwargs) @devices.inference_context() def create_quantized_param( # pylint: disable=arguments-differ self, model, param_value: torch.FloatTensor, param_name: str, target_device: torch.device, state_dict: Dict[str, Any], *args, **kwargs, # pylint: disable=unused-argument ): if self.pre_quantized and self.updated_expected_keys and (param_name.endswith(".scale") or param_name.endswith(".zero_point") or param_name.endswith(".svd_up") or param_name.endswith(".svd_down")): layer, tensor_name = get_module_from_name(model, param_name) return_dtype = torch.float32 if self.quantization_config.dequantize_fp32 else self.torch_dtype if self.torch_dtype is not None else param_value.dtype if param_value is not None: if param_value.dtype == return_dtype and devices.same_device(param_value.device, target_device): param_value = param_value.clone() else: param_value = param_value.to(target_device, dtype=return_dtype) param_value = torch.nn.Parameter(param_value, requires_grad=False) setattr(layer, tensor_name, param_value) return weights_dtype = self.quantization_config.weights_dtype if len(self.quantization_config.modules_dtype_dict.keys()) > 0: split_param_name = param_name.split(".") for key, value in self.quantization_config.modules_dtype_dict.items(): if ( param_name in value or any(param in split_param_name for param in value) or any("*" in param and re.match(param.replace(".*", "\\.*").replace("*", ".*"), param_name) for param in value) ): key = key.lower() if key in {"8bit", "8bits"}: if dtype_dict[weights_dtype]["num_bits"] != 8: weights_dtype = "int8" elif key.startswith("minimum_"): minimum_bits_str = key.removeprefix("minimum_").removesuffix("bits").removesuffix("bit") minimum_bits = int(minimum_bits_str) if dtype_dict[weights_dtype]["num_bits"] < minimum_bits: weights_dtype = "int" + minimum_bits_str if minimum_bits <= 4: weights_dtype = "u" + weights_dtype else: weights_dtype = key if self.quantization_config.return_device is not None: return_device = self.quantization_config.return_device else: return_device = target_device if self.quantization_config.quantization_device is not None: target_device = self.quantization_config.quantization_device if not self.pre_quantized: torch_dtype = param_value.dtype if self.torch_dtype is None else self.torch_dtype if param_value.dtype == torch.float32 and devices.same_device(param_value.device, target_device): param_value = param_value.clone() else: param_value = param_value.to(target_device, non_blocking=self.quantization_config.non_blocking).to(dtype=torch.float32) layer, _ = get_module_from_name(model, param_name) layer.weight = torch.nn.Parameter(param_value, requires_grad=False) layer = sdnq_quantize_layer( layer, weights_dtype=weights_dtype, torch_dtype=torch_dtype, group_size=self.quantization_config.group_size, svd_rank=self.quantization_config.svd_rank, use_svd=self.quantization_config.use_svd, quant_conv=self.quantization_config.quant_conv, use_quantized_matmul=self.quantization_config.use_quantized_matmul, use_quantized_matmul_conv=self.quantization_config.use_quantized_matmul_conv, dequantize_fp32=self.quantization_config.dequantize_fp32, non_blocking=self.quantization_config.non_blocking, quantization_device=None, return_device=return_device, param_name=param_name, ) else: from accelerate import init_empty_weights layer, _ = get_module_from_name(model, param_name) torch_dtype = layer.weight.dtype if self.torch_dtype is None else self.torch_dtype if self.updated_expected_keys: # prevent overwrite to meta scale, zero_point, svd_up, svd_down = None, None, None, None if hasattr(layer, "scale") and layer.scale.device.type != "meta": scale = layer.scale del layer.scale if hasattr(layer, "zero_point") and layer.zero_point.device.type != "meta": zero_point = layer.zero_point del layer.zero_point if hasattr(layer, "svd_up") and layer.svd_up.device.type != "meta": svd_up = layer.svd_up del layer.svd_up if hasattr(layer, "svd_down") and layer.svd_down.device.type != "meta": svd_down = layer.svd_down del layer.svd_down with init_empty_weights(): # add sdnq_dequantizer layer = sdnq_quantize_layer( layer, weights_dtype=weights_dtype, torch_dtype=torch_dtype, group_size=self.quantization_config.group_size, svd_rank=self.quantization_config.svd_rank, use_svd=self.quantization_config.use_svd, quant_conv=self.quantization_config.quant_conv, use_quantized_matmul=self.quantization_config.use_quantized_matmul, use_quantized_matmul_conv=self.quantization_config.use_quantized_matmul_conv, dequantize_fp32=self.quantization_config.dequantize_fp32, non_blocking=self.quantization_config.non_blocking, quantization_device="meta", return_device="meta", param_name=param_name, ) layer.weight = torch.nn.Parameter(param_value.clone().to(target_device), requires_grad=False) if self.updated_expected_keys: # Transformers if scale is not None: layer.scale = torch.nn.Parameter(scale, requires_grad=False) if zero_point is not None: layer.zero_point = torch.nn.Parameter(zero_point, requires_grad=False) if svd_up is not None: layer.svd_up = torch.nn.Parameter(svd_up, requires_grad=False) if svd_down is not None: layer.svd_down = torch.nn.Parameter(svd_down, requires_grad=False) else: # Diffusers doesn't have the API for updating expected keys layer_key = param_name.removesuffix(".weight") layer.scale = torch.nn.Parameter(state_dict[layer_key + ".scale"].clone().to(target_device), requires_grad=False) if layer.zero_point is not None: layer.zero_point = torch.nn.Parameter(state_dict[layer_key + ".zero_point"].clone().to(target_device), requires_grad=False) if layer.svd_up is not None: layer.svd_up = torch.nn.Parameter(state_dict[layer_key + ".svd_up"].clone().to(target_device), requires_grad=False) layer.svd_down = torch.nn.Parameter(state_dict[layer_key + ".svd_down"].clone().to(target_device), requires_grad=False) def adjust_max_memory(self, max_memory: Dict[str, Union[int, str]]) -> Dict[str, Union[int, str]]: max_memory = {key: val * 0.80 for key, val in max_memory.items()} return max_memory def adjust_target_dtype(self, target_dtype: torch.dtype) -> torch.dtype: # pylint: disable=unused-argument,arguments-renamed return dtype_dict[self.quantization_config.weights_dtype]["target_dtype"] def update_torch_dtype(self, torch_dtype: torch.dtype = None) -> torch.dtype: self.torch_dtype = torch_dtype return torch_dtype def _process_model_before_weight_loading( # pylint: disable=arguments-differ self, model, device_map, # pylint: disable=unused-argument keep_in_fp32_modules: List[str] = None, **kwargs, # pylint: disable=unused-argument ): if self.quantization_config.add_skip_keys: if keep_in_fp32_modules is not None: self.modules_to_not_convert.extend(keep_in_fp32_modules) model, self.modules_to_not_convert, self.quantization_config.modules_dtype_dict = add_module_skip_keys( model, self.modules_to_not_convert, self.quantization_config.modules_dtype_dict ) self.modules_to_not_convert.extend(self.quantization_config.modules_to_not_convert) self.quantization_config.modules_to_not_convert = self.modules_to_not_convert if hasattr(model, "config"): model.config.quantization_config = self.quantization_config model.quantization_config = self.quantization_config if self.pre_quantized and hasattr(model, "get_parameter_or_buffer"): from functools import wraps @wraps(model.get_parameter_or_buffer) def get_parameter_or_buffer(self, target: str): try: return self.original_get_parameter_or_buffer(target) except Exception as e: if target.endswith(".scale") or target.endswith(".zero_point") or target.endswith(".svd_up") or target.endswith(".svd_down"): return None raise e model.original_get_parameter_or_buffer = model.get_parameter_or_buffer model.get_parameter_or_buffer = get_parameter_or_buffer.__get__(model, model.__class__) def _process_model_after_weight_loading(self, model, **kwargs): # pylint: disable=unused-argument if shared.opts.diffusers_offload_mode != "none": model = model.to(devices.cpu) devices.torch_gc(force=True, reason="sdnq") return model def get_accelerator_warm_up_factor(self): return 32 // dtype_dict[self.quantization_config.weights_dtype]["num_bits"] def get_cuda_warm_up_factor(self): """ needed for transformers compatibilty, returns self.get_accelerator_warm_up_factor """ return self.get_accelerator_warm_up_factor() def update_unexpected_keys(self, model, unexpected_keys: List[str], *args, **kwargs) -> List[str]: # pylint: disable=unused-argument if not self.pre_quantized: return unexpected_keys new_unexpected_keys = [] for key in unexpected_keys: if not (key.endswith(".scale") or key.endswith(".zero_point") or key.endswith(".svd_up") or key.endswith(".svd_down")): new_unexpected_keys.append(key) return new_unexpected_keys def update_expected_keys(self, model, expected_keys: List[str], loaded_keys: list[str], *args, **kwargs) -> List[str]: # pylint: disable=unused-argument if not self.pre_quantized: return expected_keys self.updated_expected_keys = True for key in loaded_keys: if key.endswith(".scale") or key.endswith(".zero_point") or key.endswith(".svd_up") or key.endswith(".svd_down"): expected_keys.append(key) return expected_keys def update_tp_plan(self, config, *args, **kwargs): # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return config def update_ep_plan(self, config, *args, **kwargs): # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return config def update_missing_keys_after_loading(self, model, missing_keys: List[str], *args, **kwargs) -> List[str]: # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return missing_keys def update_state_dict_with_metadata(self, state_dict: dict, *args, **kwargs) -> dict: # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return state_dict def update_param_name(self, param_name: str, *args, **kwargs) -> str: # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return param_name def update_dtype(self, dtype: torch.dtype, *args, **kwargs) -> torch.dtype: # pylint: disable=unused-argument """ needed for transformers compatibilty, no-op function """ return dtype def _dequantize(self, model): model = dequantize_sdnq_model(model) if hasattr(model, "quantization_method"): del model.quantization_method if hasattr(model, "quantization_config"): del model.quantization_config if hasattr(model, "config") and hasattr(model.config, "quantization_config"): del model.config.quantization_config return model def is_serializable(self, *args, **kwargs) -> bool: # pylint: disable=unused-argument, invalid-overridden-method return True @property def is_trainable(self): return False @property def is_compileable(self): return True @dataclass class SDNQConfig(QuantizationConfigMixin): """ This is a wrapper class about all possible attributes and features that you can play with a model that has been loaded using `sdnq`. Args: weights_dtype (`str`, *optional*, defaults to `"int8"`): The target dtype for the weights after quantization. Supported values are: ("int8", "int7", "int6", "int5", "int4", "int3", "int2", "uint8", "uint7", "uint6", "uint5", "uint4", "uint3", "uint2", "uint1", "bool", "float8_e4m3fn", "float8_e4m3fnuz", "float8_e5m2", "float8_e5m2fnuz") group_size (`int`, *optional*, defaults to `0`): Used to decide how many elements of a tensor will share the same quantization group. group_size = 0 will automatically select a group size based on weights_dtype. svd_rank (`int`, *optional*, defaults to `32`): The rank size used for the SVDQuant algorithm. use_svd (`bool`, *optional*, defaults to `False`): Enabling this option will use SVDQuant algorithm on top of SDNQ quantization. quant_conv (`bool`, *optional*, defaults to `False`): Enabling this option will quantize the convolutional layers in UNet models too. use_quantized_matmul (`bool`, *optional*, defaults to `False`): Enabling this option will use quantized INT8 or FP8 MatMul instead of BF16 / FP16. use_quantized_matmul_conv (`bool`, *optional*, defaults to `False`): Same as use_quantized_matmul_conv but for the convolutional layers with UNets like SDXL. dequantize_fp32 (`bool`, *optional*, defaults to `False`): Enabling this option will use FP32 on the dequantization step. non_blocking (`bool`, *optional*, defaults to `False`): Enabling this option will use non blocking ops when moving layers between the quantization device and the return device. add_skip_keys (`bool`, *optional*, defaults to `True`): Disabling this option won't add model specific modules_to_not_convert and modules_dtype_dict keys. quantization_device (`torch.device`, *optional*, defaults to `None`): Used to set which device will be used for the quantization calculation on model load. return_device (`torch.device`, *optional*, defaults to `None`): Used to set which device will the quantized weights be sent back to. modules_to_not_convert (`list`, *optional*, default to `None`): The list of modules to not quantize, useful for quantizing models that explicitly require to have some modules left in their original precision (e.g. Whisper encoder, Llava encoder, Mixtral gate layers). modules_dtype_dict (`dict`, *optional*, default to `None`): The dict of dtypes and list of modules, useful for quantizing some modules with a different dtype. """ def __init__( # pylint: disable=super-init-not-called self, weights_dtype: str = "int8", group_size: int = 0, svd_rank: int = 32, use_svd: bool = False, quant_conv: bool = False, use_quantized_matmul: bool = False, use_quantized_matmul_conv: bool = False, dequantize_fp32: bool = False, non_blocking: bool = False, add_skip_keys: bool = True, quantization_device: Optional[torch.device] = None, return_device: Optional[torch.device] = None, modules_to_not_convert: Optional[List[str]] = None, modules_dtype_dict: Optional[Dict[str, List[str]]] = None, **kwargs, # pylint: disable=unused-argument ): self.weights_dtype = weights_dtype self.quant_method = QuantizationMethod.SDNQ self.group_size = group_size self.svd_rank = svd_rank self.use_svd = use_svd self.quant_conv = quant_conv self.use_quantized_matmul = use_quantized_matmul self.use_quantized_matmul_conv = use_quantized_matmul_conv self.dequantize_fp32 = dequantize_fp32 self.non_blocking = non_blocking self.add_skip_keys = add_skip_keys self.quantization_device = quantization_device self.return_device = return_device self.modules_to_not_convert = modules_to_not_convert self.modules_dtype_dict = modules_dtype_dict self.post_init() self.is_integer = dtype_dict[self.weights_dtype]["is_integer"] def post_init(self): r""" Safety checker that arguments are correct """ if self.weights_dtype not in accepted_weights: raise ValueError(f"SDNQ only support weights in {accepted_weights} but found {self.weights_dtype}") if self.modules_to_not_convert is None: self.modules_to_not_convert = [] elif isinstance(self.modules_to_not_convert, str): self.modules_to_not_convert = [self.modules_to_not_convert] elif isinstance(self.modules_to_not_convert, tuple): self.modules_to_not_convert = list(self.modules_to_not_convert) elif not isinstance(self.modules_to_not_convert, list): raise ValueError(f"modules_to_not_convert must be a list but got {type(self.modules_to_not_convert)}") if self.modules_dtype_dict is None: self.modules_dtype_dict = {} elif not isinstance(self.modules_dtype_dict, dict): raise ValueError(f"modules_dtype_dict must be a dict but got {type(self.modules_dtype_dict)}") elif len(self.modules_dtype_dict.keys()) > 0: for key, value in self.modules_dtype_dict.items(): if isinstance(value, str): value = [value] self.modules_dtype_dict[key] = value elif isinstance(value, tuple): value = list(value) self.modules_dtype_dict[key] = value if not isinstance(key, str) or not isinstance(value, list): raise ValueError(f"modules_dtype_dict must be a dictionary of strings and lists but got {type(key)} and {type(value)}") def to_dict(self): dct = self.__dict__.copy() # make serializable dct["quantization_device"] = str(dct["quantization_device"]) if dct["quantization_device"] is not None else None dct["return_device"] = str(dct["return_device"]) if dct["return_device"] is not None else None return dct import diffusers.quantizers.auto # noqa: E402,RUF100 # pylint: disable=wrong-import-order diffusers.quantizers.auto.AUTO_QUANTIZER_MAPPING["sdnq"] = SDNQQuantizer diffusers.quantizers.auto.AUTO_QUANTIZATION_CONFIG_MAPPING["sdnq"] = SDNQConfig import transformers.quantizers.auto # noqa: E402,RUF100 # pylint: disable=wrong-import-order transformers.quantizers.auto.AUTO_QUANTIZER_MAPPING["sdnq"] = SDNQQuantizer transformers.quantizers.auto.AUTO_QUANTIZATION_CONFIG_MAPPING["sdnq"] = SDNQConfig