Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-20 01:31:13 +02:00
Code Issues Packages Projects Releases Wiki Activity

Use inference_context with SDNQ

Browse Source
This commit is contained in:
Disty0
2025-07-14 13:07:44 +03:00
parent 3087b8d8cd
commit 30af1f8fb0
1 changed files with 4 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/sdnq/__init__.py
+4 -1
View File
@@ -14,6 +14,7 @@ from .dequantizer import dequantizer_dict
from .forward import get_forward_func
@devices.inference_context()
def sdnq_quantize_layer(layer, weights_dtype="int8", torch_dtype=None, group_size=0, quant_conv=False, use_quantized_matmul=False, use_quantized_matmul_conv=False, dequantize_fp32=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:
@@ -262,7 +263,8 @@ class SDNQQuantizer(DiffusersQuantizer):
return True
else:
return True
param_value.data = param_value.clone() # safetensors is unable to release the cpu memory without this
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:
@@ -271,6 +273,7 @@ class SDNQQuantizer(DiffusersQuantizer):
"""
return self.check_if_quantized_param(*args, **kwargs)
@devices.inference_context()
def create_quantized_param( # pylint: disable=arguments-differ
self,
model,