- Deep Learning with PyTorch
- Vishnu Subramanian
- 182字
- 2021-06-24 19:16:25
Network implementation
As we have all the parameters (x, w, b, and y) required to implement the network, we perform a matrix multiplication between w and x. Then, sum the result with b. That will give our predicted y. The function is implemented as follows:
def simple_network(x):
y_pred = torch.matmul(x,w)+b
return y_pred
PyTorch also provides a higher-level abstraction in torch.nn called layers, which will take care of most of these underlying initialization and operations associated with most of the common techniques available in the neural network. We are using the lower-level operations to understand what happens inside these functions. In later chapters, that is Chapter 5, Deep Learning for Computer Vision and Chapter 6, Deep Learning with Sequence Data and Text, we will be relying on the PyTorch abstractions to build complex neural networks or functions. The previous model can be represented as a torch.nn layer, as follows:
f = nn.Linear(17,1) # Much simpler.
Now that we have calculated the y values, we need to know how good our model is, which is done in the loss function.
- Intel FPGA/CPLD設計(基礎篇)
- 零點起飛學Xilinx FPG
- Learning Game Physics with Bullet Physics and OpenGL
- VCD、DVD原理與維修
- 微型計算機系統原理及應用:國產龍芯處理器的軟件和硬件集成(基礎篇)
- 單片機技術及應用
- Python Machine Learning Blueprints
- 單片微機原理及應用
- 3D Printing Blueprints
- 微控制器的應用
- Angular 6 by Example
- The Reinforcement Learning Workshop
- 詳解FPGA:人工智能時代的驅動引擎
- PIC系列單片機的流碼編程
- Hands-On Markov Models with Python