PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
fix bug of zero accuracy for mx-fp (#709)
Heng Guo34051b3d2b92
diff --git a/auto_round/script/llm.py b/auto_round/script/llm.pyindex 748b6ef69..7f151c629 100644--- a/auto_round/script/llm.py+++ b/auto_round/script/llm.py@@ -178,7 +178,7 @@ class BasicArgumentParser(argparse.ArgumentParser):self.add_argument("--act_bits", default=16, type=int, help="activation bits")- self.add_argument("--act_group_size", default=0, type=int, help="activation group size")+ self.add_argument("--act_group_size", default=None, type=int, help="activation group size")self.add_argument("--fp_layers", default="", type=str, help="list of Layer names to maintain original data type"