未验证 提交 99cd470c 编写于 作者: G Guanghua Yu 提交者: GitHub

Update post_quant_hpo.py

上级 37c52224
......@@ -77,7 +77,7 @@ class QuantConfig(object):
activation_bits=8,
weight_quantize_type='channel_wise_abs_max',
optimize_model=False,
onnx_format=False,
onnx_format=True,
is_use_cache_file=False,
cache_dir="./temp_post_training"):
"""QuantConfig init"""
......@@ -358,7 +358,7 @@ def quant_post_hpo(
batch_size=[10, 30], ### uniform sample in list.
batch_num=[10, 30], ### uniform sample in list.
optimize_model=False,
onnx_format=False,
onnx_format=True,
is_use_cache_file=False,
cache_dir="./temp_post_training",
runcount_limit=30):
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册