Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-08 05:48:42 +02:00
Code Issues Packages Projects Releases Wiki Activity

Use transformers QuantoConfig for TE and LLM

Browse Source
This commit is contained in:
Disty0
2025-04-16 19:16:35 +03:00
parent 68d1262836
commit a6ea26fb8d
1 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/model_quant.py
+7 -3
View File
@@ -88,9 +88,13 @@ def create_quanto_config(kwargs = None, allow_quanto: bool = True, module: str =
load_quanto(silent=True)
if optimum_quanto is None:
return kwargs
quanto_config = diffusers.QuantoConfig(weights_dtype=shared.opts.quanto_quantization_type)
quanto_config.activations = None # patch so it works with transformers
quanto_config.weights = quanto_config.weights_dtype
if module in {'TE', 'LLM'}:
quanto_config = transformers.QuantoConfig(weights=shared.opts.quanto_quantization_type)
quanto_config.weights_dtype = quanto_config.weights
else:
quanto_config = diffusers.QuantoConfig(weights_dtype=shared.opts.quanto_quantization_type)
quanto_config.activations = None # patch so it works with transformers
quanto_config.weights = quanto_config.weights_dtype
log.debug(f'Quantization: module="{module}" type=quanto dtype={shared.opts.quanto_quantization_type}')
if kwargs is None:
return quanto_config