neural networks - All else equal, why would switching from Glorot_Uniform to He initializers cause my loss function to blow up? - Cross Validated
Understanding the difficulty of training deep feedforward neural networks
python - ¿Cómo puedo obtener usando la misma seed exactamente los mismos resultados usando inicializadores "manualmente" y con keras? - Stack Overflow en español
Hyper-parameters in Action! Part II — Weight Initializers | by Daniel Godoy | Towards Data Science
Understanding the difficulty of training deep feedforward neural networks
Priming neural networks with an appropriate initializer. | by Ahmed Hosny | Becoming Human: Artificial Intelligence Magazine
Train and test average loss of ResNet-50 trained from Glorot uniform... | Download Scientific Diagram