1οΌIntroduction
The torchkeras library is a simple tool for training neural network in pytorch jusk like in a keras style.
With torchkeras, You need not to write your training loop with many lines of code, all you need to do is just
like this three steps as below:
(i) create your network and wrap it with torchkeras.Model like this: model = torchkeras.Model(net)
(ii) compile your model to bind the loss function, the optimizer and the metrics function.
(iii) fit your model with the training data and validate data.
This project seems somehow powerful, but the source code is very simple.
Actually, less than 200 lines of Python code.
If you want to understand or modify some details of this project, feel free to read and change the source code!!!
it's more powerful than torchkeras.Model and with more flexibility and easier to use!
the tutorial of torchkeras.LightModel is here:
Use Pytorch-Lightning Like Keras
2, Use example
You can install torchkeras using pip:
pip install torchkeras
Here is a complete examples using torchkeras!
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
import torchkeras #Attention this line
(1) prepare data
%matplotlib inline
%config InlineBackend.figure_format = 'svg'
#number of samples
n_positive,n_negative = 2000,2000
#positive samples
r_p = 5.0 + torch.normal(0.0,1.0,size = [n_positive,1])
theta_p = 2*np.pi*torch.rand([n_positive,1])
Xp = torch.cat([r_p*torch.cos(theta_p),r_p*torch.sin(theta_p)],axis = 1)
Yp = torch.ones_like(r_p)
#negative samples
r_n = 8.0 + torch.normal(0.0,1.0,size = [n_negative,1])
theta_n = 2*np.pi*torch.rand([n_negative,1])
Xn = torch.cat([r_n*torch.cos(theta_n),r_n*torch.sin(theta_n)],axis = 1)
Yn = torch.zeros_like(r_n)
#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(X,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,shuffle=True,num_workers=2)
dl_valid = DataLoader(ds_valid,batch_size = 100,num_workers=2)(2) create the model
class Net(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(2,4)
self.fc2 = nn.Linear(4,8)
self.fc3 = nn.Linear(8,1)
def forward(self,x):
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
y = nn.Sigmoid()(self.fc3(x))
return y
net = Net()
### Attention here
model = torchkeras.Model(net)
model.summary(input_shape =(2,))----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Linear-1 [-1, 4] 12
Linear-2 [-1, 8] 40
Linear-3 [-1, 1] 9
================================================================
Total params: 61
Trainable params: 61
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.000008
Forward/backward pass size (MB): 0.000099
Params size (MB): 0.000233
Estimated Total Size (MB): 0.000340
----------------------------------------------------------------
(3) train the model
# define metric
def accuracy(y_pred, y_true):
y_pred = torch.where(y_pred>0.5,torch.ones_like(y_pred,dtype = torch.float32),
torch.zeros_like(y_pred,dtype = torch.float32))
acc = torch.mean(1-torch.abs(y_true-y_pred))
return acc
def mse(y_pred, y_true):
return torch.sqrt(torch.mean((y_true - y_pred) ** 2))
# if gpu is available, use gpu
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model.compile(loss_func = nn.BCELoss(),optimizer= torch.optim.Adam(model.parameters(),lr = 0.01),
metrics_dict={accuracy, mse},device = device)
dfhistory=model.fit(epochs=10, train_data=dl_train, val_data=dl_valid, patience=5, monitor="val_loss", save_path="save_model.pkl", verbose=True)Epoch 1 / 10
[========================================] 100% loss: 0.6570 accuracy: 0.5525 mse: 0.4824 val_loss: 0.6188 val_accuracy: 0.6167 val_mse: 0.4638
Validation loss decreased (inf --> 0.618847). Saving model ...
Epoch 2 / 10
[========================================] 100% loss: 0.5877 accuracy: 0.6857 mse: 0.4476 val_loss: 0.5518 val_accuracy: 0.6950 val_mse: 0.4315
Validation loss decreased (0.618847 --> 0.551810). Saving model ...
Epoch 3 / 10
[========================================] 100% loss: 0.4949 accuracy: 0.8079 mse: 0.3984 val_loss: 0.4342 val_accuracy: 0.8558 val_mse: 0.3645
Validation loss decreased (0.551810 --> 0.434237). Saving model ...
Epoch 4 / 10
[========================================] 100% loss: 0.3819 accuracy: 0.8682 mse: 0.3359 val_loss: 0.3284 val_accuracy: 0.9117 val_mse: 0.3023
Validation loss decreased (0.434237 --> 0.328433). Saving model ...
Epoch 5 / 10
[========================================] 100% loss: 0.2942 accuracy: 0.9007 mse: 0.2882 val_loss: 0.2541 val_accuracy: 0.9092 val_mse: 0.2649
Validation loss decreased (0.328433 --> 0.254060). Saving model ...
Epoch 6 / 10
[========================================] 100% loss: 0.2441 accuracy: 0.9104 mse: 0.2627 val_loss: 0.2311 val_accuracy: 0.9125 val_mse: 0.2561
Validation loss decreased (0.254060 --> 0.231079). Saving model ...
Epoch 7 / 10
[========================================] 100% loss: 0.2247 accuracy: 0.9100 mse: 0.2542 val_loss: 0.2218 val_accuracy: 0.9083 val_mse: 0.2546
Validation loss decreased (0.231079 --> 0.221847). Saving model ...
Epoch 8 / 10
[========================================] 100% loss: 0.2091 accuracy: 0.9164 mse: 0.2441 val_loss: 0.2084 val_accuracy: 0.9192 val_mse: 0.2441
Validation loss decreased (0.221847 --> 0.208386). Saving model ...
Epoch 9 / 10
[========================================] 100% loss: 0.1972 accuracy: 0.9218 mse: 0.2366 val_loss: 0.2032 val_accuracy: 0.9175 val_mse: 0.2435
Validation loss decreased (0.208386 --> 0.203234). Saving model ...
Epoch 10 / 10
[========================================] 100% loss: 0.1940 accuracy: 0.9204 mse: 0.2367 val_loss: 0.2058 val_accuracy: 0.9167 val_mse: 0.2445
EarlyStopping counter: 1 out of 5
# visual the results
fig, (ax1,ax2) = plt.subplots(nrows=1,ncols=2,figsize = (12,5))
ax1.scatter(Xp[:,0],Xp[:,1], c="r")
ax1.scatter(Xn[:,0],Xn[:,1],c = "g")
ax1.legend(["positive","negative"]);
ax1.set_title("y_true")
Xp_pred = X[torch.squeeze(model.forward(X)>=0.5)]
Xn_pred = X[torch.squeeze(model.forward(X)<0.5)]
ax2.scatter(Xp_pred[:,0],Xp_pred[:,1],c = "r")
ax2.scatter(Xn_pred[:,0],Xn_pred[:,1],c = "g")
ax2.legend(["positive","negative"]);
ax2.set_title("y_pred")
(4) evaluate the model
%matplotlib inline
%config InlineBackend.figure_format = 'svg'
import matplotlib.pyplot as plt
def plot_metric(dfhistory, metric):
train_metrics = dfhistory[metric]
val_metrics = dfhistory['val_'+metric]
epochs = range(1, len(train_metrics) + 1)
plt.plot(epochs, train_metrics, 'bo--')
plt.plot(epochs, val_metrics, 'ro-')
plt.title('Training and validation '+ metric)
plt.xlabel("Epochs")
plt.ylabel(metric)
plt.legend(["train_"+metric, 'val_'+metric])
plt.show()plot_metric(dfhistory,"loss")
plot_metric(dfhistory,"accuracy")
model.evaluate(dl_valid){'val_loss': 0.13576620258390903, 'val_accuracy': 0.9441666702429453}
(5) use the model
model.predict(dl_valid)[0:10]tensor([[0.8767],
[0.0154],
[0.9976],
[0.9990],
[0.9984],
[0.0071],
[0.3529],
[0.4061],
[0.9938],
[0.9997]])
for features,labels in dl_valid:
with torch.no_grad():
predictions = model.forward(features)
print(predictions[0:10])
breaktensor([[0.9979],
[0.0011],
[0.9782],
[0.9675],
[0.9653],
[0.9906],
[0.1774],
[0.9994],
[0.9178],
[0.9579]])
(6) save the model
# save the model parameters
model_clone = torchkeras.Model(Net())
model_clone.load_state_dict(torch.load("save_model.pkl"))
model_clone.compile(loss_func = nn.BCELoss(),optimizer= torch.optim.Adam(model.parameters(),lr = 0.01),
metrics_dict={"accuracy":accuracy})
model_clone.evaluate(dl_valid){'val_loss': 0.17422042911251387, 'val_accuracy': 0.9358333299557368}