https://github.com/timkoornstra/vgslify
Rapidly prototype TensorFlow and PyTorch models using VGSL (Variable-size Graph Specification Language) and convert them back to VGSL spec.
https://github.com/timkoornstra/vgslify
deep-learning model-generation neural-network neural-networks pytorch rapid-prototyping tensorflow tf torch vgsl vgslify
Last synced: 5 months ago
JSON representation
Rapidly prototype TensorFlow and PyTorch models using VGSL (Variable-size Graph Specification Language) and convert them back to VGSL spec.
- Host: GitHub
- URL: https://github.com/timkoornstra/vgslify
- Owner: TimKoornstra
- License: mit
- Created: 2023-10-14T06:12:16.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2025-04-16T20:06:22.000Z (6 months ago)
- Last Synced: 2025-05-07T23:37:04.379Z (5 months ago)
- Topics: deep-learning, model-generation, neural-network, neural-networks, pytorch, rapid-prototyping, tensorflow, tf, torch, vgsl, vgslify
- Language: Python
- Homepage: https://timkoornstra.github.io/VGSLify/
- Size: 5.3 MB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Support: docs/supported_layers.rst
Awesome Lists containing this project
README
# VGSLify: Variable-size Graph Specification for TensorFlow & PyTorch
VGSLify simplifies defining, training, and interpreting deep learning models using the Variable-size Graph Specification Language (VGSL). Inspired by [Tesseract's VGSL specs](https://tesseract-ocr.github.io/tessdoc/tess4/VGSLSpecs.html), VGSLify enhances and streamlines the process for both TensorFlow and PyTorch.
## Table of Contents
- [Installation](#installation)
- [How VGSL Works](#how-vgsl-works)
- [Quick Start](#quick-start)
- [Generating a Model with VGSLify](#generating-a-model-with-vgslify)
- [Creating Individual Layers with VGSLify](#creating-individual-layers-with-vgslify)
- [Converting Models to VGSL](#converting-models-to-vgsl)
- [Additional Documentation](#additional-documentation)
- [Contributing](#contributing)
- [License](#license)
- [Acknowledgements](#acknowledgements)
## Installation
### Basic Installation
To install VGSLify without any deep learning backend, run:
```bash
pip install vgslify
```
This installs only the core functionalities of VGSLify without `torch` or `tensorflow`.
### Installing with a Specific Backend
VGSLify supports both TensorFlow and PyTorch. You can install it with the required backend:
```bash
# For TensorFlow (latest compatible version)
pip install vgslify[tensorflow]
# For PyTorch (latest compatible version)
pip install vgslify[torch]
```
By default, this will install:
- `tensorflow` (latest stable version)
- `torch` (latest stable version)
### Controlling Backend Versions
If you need a specific version of `torch` or `tensorflow`, install VGSLify first and then manually install the backend:
```bash
pip install vgslify
pip install torch==2.1.0 # Example for PyTorch
pip install tensorflow==2.14 # Example for TensorFlow
```
Alternatively, you can specify the version during installation:
```bash
pip install vgslify[torch] torch==2.1.0
pip install vgslify[tensorflow] tensorflow==2.14
```
⚠ **Note**: If a different version of torch or tensorflow is already installed, pip may not downgrade it automatically. Use `--force-reinstall` or `--upgrade` if necessary:
```bash
pip install --upgrade --force-reinstall torch==2.1.0
```
### Verify installation
To check that VGSLify is installed correctly, run:
```bash
python -c "import vgslify; print(vgslify.__version__)"
```
## How VGSL Works
VGSL uses concise strings to define model architectures. For example:
```vgsl
None,None,64,1 Cr3,3,32 Mp2,2 Cr3,3,64 Mp2,2 Rc3 Fr64 D20 Lrs128 D20 Lrs64 D20 Fs92
```
Each part represents a layer: input, convolution, pooling, reshaping, fully connected, LSTM, and output. VGSL allows specifying activation functions for customization.
## Quick Start
### Generating a Model with VGSLify
```python
from vgslify import VGSLModelGenerator
# Define the VGSL specification
vgsl_spec = "None,None,64,1 Cr3,3,32 Mp2,2 Fs92"
# Choose backend: "tensorflow", "torch", or "auto" (defaults to whichever is available)
vgsl_gn = VGSLModelGenerator(backend="tensorflow")
model = vgsl_gn.generate_model(vgsl_spec, model_name="MyModel")
model.summary()
vgsl_gn = VGSLModelGenerator(backend="torch") # Switch to PyTorch
model = vgsl_gn.generate_model(vgsl_spec, model_name="MyTorchModel")
print(model)
```
### Creating Individual Layers with VGSLify
```python
from vgslify import VGSLModelGenerator
vgsl_gn = VGSLModelGenerator(backend="tensorflow")
conv2d_layer = vgsl_gn.construct_layer("Cr3,3,64")
# Integrate into an existing model:
# model = tf.keras.Sequential()
# model.add(conv2d_layer) # ...
# Example with generate_history:
history = vgsl_gn.generate_history("None,None,64,1 Cr3,3,32 Mp2,2 Fs92")
for layer in history:
print(layer)
```
### Converting Models to VGSL
```python
from vgslify import model_to_spec
import tensorflow as tf
# Or import torch.nn as nn
# TensorFlow example:
model = tf.keras.models.load_model("path_to_your_model.keras") # If loading from file
# PyTorch example:
# model = MyPyTorchModel() # Assuming MyPyTorchModel is defined elsewhere
vgsl_spec_string = model_to_spec(model)
print(vgsl_spec_string)
```
**Note:** Flatten/Reshape layers might require manual input shape adjustment in the generated VGSL.
## Additional Documentation
See the [VGSL Documentation](https://timkoornstra.github.io/VGSLify) for more details on supported layers and their specifications.
## Contributing
Contributions are welcome! Fork the repository, set up your environment, make changes, and submit a pull request. Create issues for bugs or suggestions.
## License
MIT License. See [LICENSE](LICENSE) file.
## Acknowledgements
Thanks to the creators and contributors of the original VGSL specification.