Shulou(Shulou.com)06/01 Report--

How to analyze the principle of torchkeras, in view of this problem, this article introduces the corresponding analysis and solution in detail, hoping to help more partners who want to solve this problem to find a more simple and feasible method.

Torchkeras is a high-level Model interface imitating keras implemented on pytorch. With it, you can summary,compile,fit,evaluate and predict the model built by pytorch, just like Keras. Everything is as natural as running water.

It sounds like torchkeras is very powerful. But in fact, its implementation is very simple, with less than 300 lines of source code. If you want to understand some of the details of how it works, or modify its functionality, don't hesitate to read and modify the project source code.

To install it, you only need to run:

Pip install torchkeras

Official account background reply key words: torchkeras. Get the project git source code and all the source code of this article!

The following is a complete example of using torchkeras to train the model. We design a three-layer neural network to solve the classification problem of positive and negative samples according to the distribution of concentric circles.

Import numpy as np

Import pandas as pd

From matplotlib import pyplot as plt

Import torch

From torch import nn

Import torch.nn.functional as F

From torch.utils.data import Dataset,DataLoader,TensorDataset

From torchkeras import Model,summary # Attention this line! First, prepare the data

Construct positive and negative sample data according to concentric circle distribution.

% matplotlib inline

% config InlineBackend.figure_format = 'svg'

# number of samples

Positive negative = 2000

# positive samples

Rsqup = 5. 0 + torch.normal (0. 0. 0. 0. 0. 0. 0)

Theta_p = 2*np.pi*torch.rand ([nasty positive1])

Xp = torch.cat ([r_p*torch.cos (theta_p), r_p*torch.sin (theta_p)], axis = 1)

Yp = torch.ones_like (rascp)

# negative samples

Rang n = 8 + torch.normal (0. 0. 0. 0)

Theta_n = 2*np.pi*torch.rand ([nasty negative1])

Xn = torch.cat ([r_n*torch.cos (theta_n), r_n*torch.sin (theta_n)], axis = 1)

Yn = torch.zeros_like (rnsn)

# concat positive and negative samples

X = torch.cat ([Xp,Xn], axis = 0)

Y = torch.cat ([Yp,Yn], axis = 0)

# visual samples

Plt.figure (figsize = (6, 6))

Plt.scatter (Xp [:, 0], Xp [:, 1], c = "r")

Plt.scatter (Xn [:, 0], Xn [:, 1], c = "g")

Plt.legend (["positive", "negative"])

# split samples into train and valid data.

Ds = TensorDataset (XQuery Y)

Ds_train,ds_valid = torch.utils.data.random_split (ds, [int (len (ds) * 0.7), len (ds)-int (len (ds) * 0.7)])

Dl_train = DataLoader (ds_train,batch_size = 100 minutes shuffle True numb workersgiving 2)

Dl_valid = DataLoader (ds_valid,batch_size = 100 numb workers2) II, build the model

We build the model by subclassing torchkeras.Model, not by subclassing torch.nn.Module. Torchkeras.Model is actually a subclass of torch.nn.Moduled.

Class DNNModel (Model): # Attention here

Def _ init__ (self):

Super (DNNModel, self). _ _ init__ ()

Self.fc1 = nn.Linear (2jue 4)

Self.fc2 = nn.Linear (4. 8)

Self.fc3 = nn.Linear (8pm 1)

Def forward (self,x):

X = F.relu (self.fc1 (x))

X = F.relu (self.fc2 (x))

Y = nn.Sigmoid () (self.fc3 (x))

Return y

Model = DNNModel ()

Model.summary (input_shape = (2,))

Third, training model

We need to first use compile to bind the loss function, optimizer, and evaluation metrics to the model. Then the fit method can be used for model training.

The answer to the question on how to analyze the principle of torchkeras is shared here. I hope the above content can be of some help to you. If you still have a lot of doubts to be solved, you can follow the industry information channel for more related knowledge.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.