lyhue1991/torchkeras
A unified PyTorch training template library that provides a clean, ~200-line Keras-style training loop with visualization and callbacks.
Not currently ranked — collecting fresh signals.
star history
torchkeras is a general-purpose PyTorch model training template designed to standardize the training workflow across different model libraries. It abstracts away the complexity of writing custom training loops by providing a clean, modular interface with built-in support for progress bars, evaluation metrics, early-stopping, and visualization. The tool includes callbacks for TensorBoard and Weights & Biases integration, aiming to replace the inconsistent training implementations found across various PyTorch ecosystems.
Frequently asked
- What is lyhue1991/torchkeras?
- A unified PyTorch training template library that provides a clean, ~200-line Keras-style training loop with visualization and callbacks.
- Is torchkeras open source?
- Yes — lyhue1991/torchkeras is open source, released under the Apache-2.0 license.
- What language is torchkeras written in?
- lyhue1991/torchkeras is primarily written in Jupyter Notebook.
- How popular is torchkeras?
- lyhue1991/torchkeras has 2k stars on GitHub.
- Where can I find torchkeras?
- lyhue1991/torchkeras is on GitHub at https://github.com/lyhue1991/torchkeras.