swarmtorch 0.2.1

Professional PyTorch library for 120+ metaheuristic optimization algorithms (Swarm, Evolutionary, Physics, Hybrid).

SwarmTorch 🐝🔥

The ultimate metaheuristic library for PyTorch

---

SwarmTorch is a high-performance, academic-grade library that brings 120 metaheuristic algorithms to the PyTorch ecosystem. It enables gradient-free neural network training and state-of-the-art hyperparameter optimization (HPO) with a single, unified API.

• 60 model-training optimizers for gradient-free weight optimization
• 60 hyperparameter-tuning searchers for intelligent HPO

Installation • Key Features • Benchmarks & Research • Usage • Citation

---

🚀 Key Features

• 120 Algorithms (60 + 60): 60 model-training optimizers + 60 HPO searchers across 6 categories: Swarm Intelligence, Evolutionary, Physics-based, Human-based, Bio-inspired, and Hybrids.
• Gradient-Free Training: Optimize weights for non-differentiable or complex loss landscapes directly as a PyTorch Optimizer.
• Deep HPO Integration: Replace Random/Grid search with intelligent, nature-inspired exploration.
• Research Ready: Includes full benchmarking suites, raw experimental data, and publication-quality visualizations.
• Highly Optimized: Leverages PyTorch's tensor operations for swarm-level parallelism.

---

📈 Benchmarks & Research

We conducted a massive-scale evaluation of 120 algorithms (60 training + 60 HPO). Our research shows that 65.5% of SwarmTorch searchers outperform the standard Random Search baseline in HPO tasks, with top performers achieving near-perfect validation accuracy.

Detailed performance analysis, convergence plots, and category reliability studies are available in the dedicated benchmarks document:

👉 View Full Research & Benchmarks Report (with Visualizations)

---

📦 Installation

Using pip:

pip install swarmtorch

Using uv (Recommended):

uv add swarmtorch

With benchmark dependencies:

pip install swarmtorch[benchmarks]

---

💻 Usage

Model Weight Optimization

import torch.nn as nn
from swarmtorch import PSO

model = nn.Sequential(nn.Linear(10, 64), nn.ReLU(), nn.Linear(64, 1))
optimizer = PSO(model.parameters(), swarm_size=30)

def closure():
    optimizer.zero_grad()
    loss = criterion(model(inputs), targets)
    return loss

optimizer.step(closure)

Hyperparameter Tuning

from swarmtorch import PSOSearch

searcher = PSOSearch(
    model_fn=build_model,
    param_space={'lr': (0.001, 0.1), 'hidden_dim': [32, 64]},
    train_fn=train_fn,
    iterations=50
)
best_params = searcher.search()

---

Why SwarmTorch?

Traditional deep learning relies on gradient-based optimization (Adam, SGD, etc.). SwarmTorch complements these with:

Scenario	Use Gradient (Adam/SGD)	Use SwarmTorch
Standard classification	✅ Best choice	Overkill
Non-differentiable loss	❌ Can't use	✅ Perfect fit
Discrete optimization	❌ Can't use	✅ Perfect fit
Multi-modal landscapes	Gets stuck	Explores well
Hyperparameter tuning	Manual search	Automated

---

🤝 Acknowledgments & References

This library was developed with reference to the pyMetaheuristic library. We are grateful for their contributions to the metaheuristic optimization community.

---

📝 Citation

@software{swarmtorch2026,
  author = {Halleluyah Darasimi Oludele},
  title = {SwarmTorch: A PyTorch Library for 120+ Metaheuristic Optimization Algorithms},
  year = {2026},
  url = {https://github.com/hallelx2/swarmtorch}
}