TensorFlow Fold
A library for creating TensorFlow models that handle structured data with dynamic computation graphs using dynamic batching.
What is TensorFlow Fold?
TensorFlow Fold is a library for creating TensorFlow models that consume structured data, where the computation graph dynamically adapts to the input's structure. It solves the problem of handling arbitrarily shaped data, such as parse trees in natural language processing, by using dynamic batching to transform variable graphs into static ones for efficient execution. This enables deep learning on non-uniform data without manual padding or fixed-size constraints.
Machine learning researchers and developers working with structured or variable-sized data, such as natural language parse trees, graphs, or recursive structures, who need to build models in TensorFlow.
Developers choose TensorFlow Fold for its ability to seamlessly integrate structured data into TensorFlow with dynamic computation graphs and batching, offering flexibility and performance without the overhead of manual graph management. Its unique approach allows training on arbitrary data shapes while leveraging TensorFlow's optimized execution.
Overview
Deep learning with dynamic computation graphs in TensorFlow
Use Cases
Best For
- Implementing TreeLSTMs for sentiment analysis on parse trees
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
