【PyTorch】深度学习与Pytorch实战笔记
学习笔记03:初学PyTorch
1. loss = ( W X + b − y ) 2 \operatorname{loss}=(W X+b-y)^{2} loss=(WX+b−y)2
∂ l ∂ w = 2 ( w x + b − y ) x \frac{\partial l}{\partial w}=2(w x+b-y) x ∂w∂l=2(wx+b−y)x
∂ l ∂ b = 2 ( w x + b − y ) \frac{\partial l}{\partial b}=2(w x+b-y) ∂b∂l=2(wx+b−y)
2. 手写数据识别
MNIST数据集
each number owns 7000 images
train/test splitting: 60k vs 10k
3. Gradient Descent 梯度下降
objective = ∑ ( pred − Y ) 2 =\sum(\text { pred }-Y)^{2} =∑( pred −Y)2
4. pytorch基本类型
5. Type check
6. shape size
271
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
