google/neural-tangents
A JAX-based library for defining finite and infinite-width neural networks with theoretical guarantees from neural network kernel theory.
Not currently ranked — collecting fresh signals.
star history
Neural Tangents enables researchers to specify complex, hierarchical neural networks of both finite and infinite width through a high-level API. It leverages neural network kernel theory to provide insights into training dynamics and correspondence with Gaussian processes. The library is built on JAX for automatic differentiation and scalable computation, supporting gradient descent analysis and Bayesian inference in deep networks.
Frequently asked
- What is google/neural-tangents?
- A JAX-based library for defining finite and infinite-width neural networks with theoretical guarantees from neural network kernel theory.
- Is neural-tangents open source?
- Yes — google/neural-tangents is open source, released under the Apache-2.0 license.
- What language is neural-tangents written in?
- google/neural-tangents is primarily written in Jupyter Notebook.
- How popular is neural-tangents?
- google/neural-tangents has 2.4k stars on GitHub.
- Where can I find neural-tangents?
- google/neural-tangents is on GitHub at https://github.com/google/neural-tangents.