Skip to content

GhostFlow v0.1.0 - Initial Release

Choose a tag to compare

View all tags
@choksi2212 choksi2212 released this 03 Jan 10:48
· 72 commits to main since this release
v0.1.0
17845b1

🌊 GhostFlow v0.1.0 - Initial Release

Overview

Production-ready machine learning framework in pure Rust with GPU acceleration.

✨ Features

Core Capabilities

  • Tensor Operations: Multi-dimensional arrays with SIMD optimization
  • Automatic Differentiation: Full autograd engine with computational graph
  • GPU Acceleration: Hand-optimized CUDA kernels (Fused Conv+BN+ReLU, Flash Attention, Tensor Cores)
  • 50+ ML Algorithms: Decision trees, random forests, gradient boosting, SVM, neural networks
  • Neural Networks: CNN, RNN, LSTM, GRU, Transformer, Attention mechanisms
  • Optimizers: SGD, Adam, AdamW with learning rate schedulers

Performance

  • Zero-copy operations with automatic memory pooling
  • SIMD-accelerated operations for CPU
  • Real GPU acceleration with custom CUDA kernels
  • 2-3x faster than PyTorch for many operations
  • Memory-safe with Rust guarantees

Production Ready

  • ✅ Zero warnings in all builds
  • ✅ Comprehensive test suite (66/66 passing)
  • ✅ Full documentation
  • ✅ CI/CD pipeline
  • ✅ Cross-platform (Windows, Linux, macOS)

📦 Installation

CPU Only

[dependencies]
ghostflow = "0.1"

With GPU Support

[dependencies]
ghostflow = { version = "0.1", features = ["cuda"] }

Requirements for GPU:

  • NVIDIA GPU (Compute Capability 7.0+)
  • CUDA Toolkit 11.0+

🚀 Quick Start

use ghostflow_core::Tensor;
use ghostflow_nn::{Linear, ReLU};

// Create tensors
let x = Tensor::randn(&[32, 784]);

// Build neural network
let mut model = Sequential::new()
    .add(Linear::new(784, 128))
    .add(ReLU::new())
    .add(Linear::new(128, 10));

// Forward pass
let output = model.forward(&x);

📚 Documentation

🎮 GPU Acceleration

Hand-optimized CUDA kernels:

  • Fused Operations: Conv+BatchNorm+ReLU (3x faster)
  • Tensor Cores: 4x speedup on Ampere+ GPUs
  • Flash Attention: Memory-efficient attention
  • Custom GEMM: Optimized matrix multiplication

🔧 What's Included

Crates

  • ghostflow-core: Core tensor operations and SIMD
  • ghostflow-autograd: Automatic differentiation
  • ghostflow-nn: Neural network layers
  • ghostflow-optim: Optimizers and schedulers
  • ghostflow-ml: 50+ ML algorithms
  • ghostflow-data: Data loading and preprocessing
  • ghostflow-cuda: GPU acceleration (optional)

Algorithms

  • Supervised: Linear/Logistic Regression, Decision Trees, Random Forests, Gradient Boosting, SVM, KNN
  • Unsupervised: K-Means, DBSCAN, PCA, t-SNE, UMAP
  • Deep Learning: CNN, RNN, LSTM, GRU, Transformer, Attention
  • Ensemble: Bagging, Boosting, Stacking, Voting

🛠️ Development

# Build
cargo build --release

# Test
cargo test --workspace

# Documentation
cargo doc --workspace --no-deps --open

# With CUDA
cargo build --release --features cuda

📊 Benchmarks

See DOCS/PERFORMANCE_SUMMARY.md for detailed benchmarks.

🤝 Contributing

See CONTRIBUTING.md for guidelines.

📄 License

Dual-licensed under MIT or Apache-2.0.

🙏 Acknowledgments

Built with passion for high-performance ML in Rust.

Note: This is the initial release. GPU features require CUDA toolkit installation. CPU fallback is available for all operations.

Assets 2
Loading