Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

fix triton issue (#148)

wenhuach2104ea869402a9
diff --git a/auto_round_extension/cuda/qliner_triton.py b/auto_round_extension/cuda/qliner_triton.py
index e307ace38..7686522c8 100644
--- a/auto_round_extension/cuda/qliner_triton.py
+++ b/auto_round_extension/cuda/qliner_triton.py
@@ -47,7 +47,7 @@ from auto_round_extension.cuda.triton_utils.mixin import TritonModuleMixin
logger = getLogger(__name__)
try:
- from auto_round_extension.cuda.triton_utils import (
+ from auto_round_extension.cuda.triton_utils.kernels import (
QuantLinearFunction,
QuantLinearInferenceOnlyFunction,
quant_matmul_248,