Working notes from applied deep-learning practice.
September 21, 2026 · LayerGenius
A search will happily optimize your validation-set noise. How we run KerasTuner: baseline first, four hyperparameters instead of twelve, executions_per_trial against spurious winners, a three-way split that matches deployment, and reading the top ten trials rather than the top one.
Read moreSeptember 21, 2026 · LayerGenius
A sigmoid output is not a probability and 0.5 is not a threshold. How to measure calibration on a Keras classifier, fix it with temperature scaling or isotonic regression, pick an operating point from the cost of each error, and keep both honest after deploy.
Read moreSeptember 21, 2026 · LayerGenius
Keras 3 can quantize a model's weights in place with model.quantize("int8") — no converter, no second toolchain. What it touches, what it skips, how to measure the accuracy cost, and when GPTQ or a calibration pass is worth the trouble.
Read moreSeptember 18, 2026 · LayerGenius
Overriding train_step used to mean a GradientTape. In Keras 3 each backend wants a different shape of code — stateless in JAX, autograd in PyTorch, tape in TensorFlow. How to write it, and how to keep it portable.
Read moreSeptember 17, 2026 · LayerGenius
Structured data is where deep learning most often loses to a 40-line GBDT. Here is how we decide: a boosted-tree baseline first, then Keras FeatureSpace for preprocessing that travels with the model, and an honest accounting of when the neural net is worth its operating cost.
Read more