derwind https://blog.hatena.ne.jp/derwind/ らんだむな記憶 https://randommemory.hatenablog.com/ machine_learning 畳み込み層の重みの初期化のテクニックっぽい。 torch.nn.init — PyTorch 1.11.0 documentation の torch.nn.init.kaiming_uniform_(tensor, a=0, mode='fan_in', nonlinearity='leaky_relu') が利用できそう。使い方とかは python - How to initialize weights in PyTorch? - Stack Overflow Initializing pytorch layers weight with kaiming | Kaggle 深層強化学習ライ… 190 <iframe src="https://hatenablog-parts.com/embed?url=https%3A%2F%2Frandommemory.hatenablog.com%2Fentry%2F2021%2F07%2F10%2F180459" title="Kaiming He の初期化 - らんだむな記憶" class="embed-card embed-blogcard" scrolling="no" frameborder="0" style="display: block; width: 100%; height: 190px; max-width: 500px; margin: 10px 0px;"></iframe> Hatena Blog https://hatena.blog 2021-07-10 18:04:59 rich https://randommemory.hatenablog.com/entry/2021/07/10/180459 1.0 100%