PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
fix: recreate calibrator for secondary diffusion transformers (#2335)
Wang, Changcea5abd75414
diff --git a/auto_round/compressors/diffusion_mixin.py b/auto_round/compressors/diffusion_mixin.pyindex 1b285138c..5228c80a6 100644--- a/auto_round/compressors/diffusion_mixin.py+++ b/auto_round/compressors/diffusion_mixin.py@@ -352,6 +352,9 @@ class DiffusionMixin:self.model_context.model = transformerself.model_context.quantized = Falseself._post_init_done = False+ # Calibrators snapshot the active model at construction time. Recreate+ # it so transformer_2 block hooks are not installed on the primary model.+ self.calibration = None# Dispatch the pipeline for the secondary transformer without recasting it.self._align_device_and_dtype_for_secondary(comp_name)