RusTorch v0.6.28 - tensor_nd! Procedural Macro for N-Dimensional Tensors

RusTorch v0.6.28 Release

🆕 New Features

tensor_nd! Procedural Macro

Introduces a powerful new procedural macro for creating N-dimensional tensors with compile-time shape inference.

use rustorch::tensor_nd;

// 4D tensor: shape [2, 2, 2, 2]
let t4 = tensor_nd!([
    [[[1, 2], [3, 4]], [[5, 6], [7, 8]]],
    [[[9, 10], [11, 12]], [[13, 14], [15, 16]]]
]);

// 6D tensor: shape [1, 1, 1, 1, 1, 2]
let t6 = tensor_nd!([[[[[[1, 2]]]]]]);

✨ Key Features

• 📦 Support for N-dimensional tensors (1D through 8D and beyond)
• 🔍 Compile-time shape inference from nested array literals
• 🔄 Automatic numeric type conversion to f32
• 🎯 Mixed numeric types support (integers and floats)
• 🏗️ New rustorch-macros workspace member for procedural macros

🧪 Testing

• ✅ 10 integration tests covering dimensions 1D through 8D
• ✅ All 1143 library tests passing
• ✅ Full example suite verified

📚 Documentation

• Comprehensive doc tests and examples
• Updated README with usage examples
• Release workflow documentation

🔧 Technical Implementation

• Uses syn, quote, and proc-macro2 for macro implementation
• Recursive array flattening algorithm
• Re-exported in main rustorch crate for easy access

📦 Installation

Add to your Cargo.toml:

rustorch = "0.6.28"

🔗 Links

• Documentation
• Crates.io
• GitHub

🤖 Generated with Claude Code

Co-Authored-By: Claude noreply@anthropic.com