Open wangjing60755 opened 3 years ago
Hello. I will check it ....09:12, 1 апреля 2021 г., wangjing60755 @.***>: from MobileNetV3
Could you share the model you working with? I have not MobileNetV3. I have MobileNetV3Small and MobileNetV3Large.
You can check with MobileNetV3Small .
---Original--- From: @.> Date: Thu, Apr 1, 2021 15:35 PM To: @.>; Cc: @.**@.>; Subject: Re: [azeme1/keras2ncnn] Unknown activation function:_hard_swish (#17)
Could you share the model you working with? I have not MobileNetV3. I have MobileNetV3Small and MobileNetV3Large.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
Segment model is here! Use MobileNetV3Small as encode.So, has hard_sigmoid , _hard_swish activations
------------------ 原始邮件 ------------------ 发件人: "azeme1/keras2ncnn" @.>; 发送时间: 2021年4月1日(星期四) 下午3:35 @.>; @.**@.>; 主题: Re: [azeme1/keras2ncnn] Unknown activation function:_hard_swish (#17)
Could you share the model you working with? I have not MobileNetV3. I have MobileNetV3Small and MobileNetV3Large.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
从QQ邮箱发来的超大附件
weights_200-0.8080.zip (6.92M, 无限期)进入下载页面:http://mail.qq.com/cgi-bin/ftnExs_download?k=293538621679a9c79f7c41fe1f330a1701530e5a58000f0b4f0d0b530a1e0c0d5556155b5f0a011500005a5a0e010c5a040d01553925384f075c5f0a4d40670a52051552170b0800521b420b493305&t=exs_ftn_download&code=b58b9388
Is there any progress?
def _hard_swish(self, x): return x * K.relu(x + 3.0, max_value=6.0) / 6.0
from MobileNetV3