BearX is my personal try to create self-sufficient Deep-Learning library which contains the best aspects (in my opinion) of both Keras and Pytorch frameworks, but most importantly I am working on this project to get in-depth knownledge about maths behind neural network and learn how to properly structure larger projects in Python.
from bearx.layers import Linear
from bearx.models import Sequential
model = Sequential()
model.add(Linear(2, 2), activation='relu')
# Before training we have to compile model
model.compile(loss='mse',
optimizer='sgd')- Sequential
- Linear (Dense)
- RNN
- (testing)
- (testing)
- Relu
- Tanh
- Sigmoid (testing)
- Softmax(testing)
- MSE
- Cross-Entropy (testing)
- SGD
- Zeros
- Ones
- Normal Distribution
- Uniform Distribution
- RNNinit
BearX supports only python 3.x
- Convolutional
- RMSprop
- Adam