# Info Dump * Gradient vanishing and blowing up for sigmoid, ReLU * Gradient Clipping * Stable Sigmoid * Leaky ReLU * Learning Rate needed for ReLU is roughly `10x` smaller than for sigmoid due to bounds of derivative * Flipping the kernel in backpropagation * He and Xavier Initialization + reasons * Importance of weight initialization * im2col * Importance of Non-Linearity (via activation functions) * Batch Normalization (mean 0, std 1) vs `/= 255.0` * CNNs * Translation Invariance * FC NN vs CNN for MNIST * Dead Kernels (constantly `0`) * Learning Rate for kernels is much smaller than for the FC NN at the end, due to their size --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://ir0nstone.gitbook.io/ai-ml/machine-learning/info-dump.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.