google-deepmind/dks
Multi-framework implementation of Deep Kernel Shaping and Tailored Activation Transformations, which are methods that modify neural network models (and their initializations) to make them easier to train.
This package helps machine learning engineers and researchers make their deep neural networks easier and faster to train, especially those without skip connections or normalization layers. It provides specialized activation function transformations, weight initializations, and optional data preprocessing. The output is a modified neural network model structure that performs as well as more complex architectures like ResNets.
No commits in the last 6 months.
Use this if you are a machine learning practitioner struggling to train deep, 'vanilla' convolutional networks efficiently and want to achieve faster convergence without adding architectural complexities like skip connections.
Not ideal if you are primarily using ReLU activation functions, as they are only partially supported, or if you are looking for a fully automated, black-box solution for network training optimization.
Stars
76
Forks
5
Language
Python
License
Apache-2.0
Category
Last pushed
Jul 01, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/google-deepmind/dks"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
explosion/thinc
🔮 A refreshing functional take on deep learning, compatible with your favorite libraries
google-deepmind/optax
Optax is a gradient processing and optimization library for JAX.
patrick-kidger/diffrax
Numerical differential equation solvers in JAX. Autodifferentiable and GPU-capable....
google/grain
Library for reading and processing ML training data.
patrick-kidger/equinox
Elegant easy-to-use neural networks + scientific computing in JAX. https://docs.kidger.site/equinox/