Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-19 09:14:35 +02:00
Code Issues Packages Projects Releases Wiki Activity

Add warning for and skip post-load quant of pre-quants

Browse Source
This commit is contained in:
Disty0
2026-06-07 18:02:28 +03:00
parent f0c93defb8
commit 03fa90ff35
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/model_quant.py
+8
View File
@@ -356,6 +356,14 @@ def sdnq_quantize_model(model, op=None, sd_model=None, do_gc: bool = True, weigh
from modules.sdnq import sdnq_post_load_quant
from modules.sdnq.common import use_torch_compile as sdnq_use_torch_compile
if (
hasattr(model, "quantization_config")
or (hasattr(model, "config") and hasattr(model.config, "quantization_config"))
or (hasattr(model, "config") and isinstance(model.config, dict) and "quantization_config" in model.config)
):
log.warning(f'Quantization: Trying to quantize a pre-quantized model. Skipping quantization of module="{op if op is not None else model.__class__}"')
return model
if shared.opts.sdnq_use_quantized_matmul and not sdnq_use_torch_compile:
log.warning('SDNQ Quantized MatMul requires a working Triton install. Disabling Quantized MatMul.')
shared.opts.sdnq_use_quantized_matmul = False