#171171st of 206 in ML frameworks and libraries
morph-net
TensorFlow library that learns smaller CNN layer widths under FLOP or size limits
- Stars
- 1k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Language
- Python
1,033 on 21 Apr 20261,043 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
MorphNet adds a resource regularizer (FLOPs, model size or device-specific latency) to the training loss of a TensorFlow convolutional network. Channels whose influence falls below an alive threshold are marked for removal, and a StructureExporter writes the proposed per-layer channel counts as JSON. You then rebuild the model with those widths and retrain from scratch. The recommended LogisticSigmoid (FiGS) regularizer requires inserting gating ops after prunable layers.
Who it is for: ML engineers shrinking TensorFlow CNNs to meet FLOP, size or latency budgets
Strengths
- Targets specific costs: FLOPs, model size or hardware-based latency estimates
- Keeps layer count and connectivity; only output channel counts change
- Gamma and GroupLasso regularizers work without modifying the model, though GroupLasso is deprecated
- Exports learned structure as JSON for retraining without the regularizer
Weaknesses
- Examples use TF1 graph APIs (tf.Session, tf.layers); the Keras example is a TODO
- Two-stage workflow: structure learning, then manual rebuild and full retrain
- Recommended regularizer requires adding gating ops to the model
- No tagged release, and the README only covers convolutional image classification
What it needs
- Needs TensorFlow
- Models: convolutional neural networks (TensorFlow)
Also in ML frameworks and libraries
See all 206| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | ultralyticsPython library and CLI for training and running YOLO vision models | 94 out of 100 |
| 3 | pytorchPython tensor library with GPU support and tape-based autograd | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and training scripts | 93 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 92 out of 100 |