lsq 训练量化error: name 'grad_factor' is not defined
Closed this issue · 2 comments
wangshankun commented
name 'grad_factor' is not defined
这份代码不论is_tracing_state是否为true都会报错
if is_tracing_state():
X = FakeQuantizeLearnablePerchannelAffine.apply(
X, self.scale, self.zero_point, self.ch_axis,
self.quant_min, self.quant_max, grad_factor)
else:
X = _fake_quantize_learnable_per_channel_affine_training(
X, self.scale, self.zero_point, self.ch_axis,
self.quant_min, self.quant_max, grad_factor)
wimh966 commented
您好,按理不会报这个错,在调用grad_factor之前grad_factor都被定义了。
可否提供进一步的错误信息
github-actions commented
This issue has not received any updates in 120 days. Please reply to this issue if this still unresolved!