- Added
square_cost
loss function inloss.py
- Upadated loss functions names in for passing arguments
- Don't know why
square_cost
is giving6.0
as loss for quite often (the weights, biases are random, so the loss should be random too, but why!!!) - Nothing else for now.
- Added loss functions dcfc6d8a1a8e7881aef93a98ed4c46093ed5cb97
- Binary Cross Entropy
- Categorical Cross Entropy
- Added fitting and loss calculations at
Model
class ce8d8cd052457c284f73c95c45a70b91b5bf1c27
- Loss can be
inf
due tolog(0)
inCategoricalCrossEntropy
loss function. Added a small value(EPSILON
)1e-10
to avoid this.
- Neuron class added 907fae19b830c1244786a905cd80dbfd7ec0d0c7
- Layers class added 907fae19b830c1244786a905cd80dbfd7ec0d0c7
- Activation functions added fca8cbb332b3b33c45a27d5319abdc067330b0f2
- Softmax Layer added ff1787c3ee27b426dc48a89b6bb57eef2d89d0db