Add torch.compile support for neural network optimization

[Copilot]

This PR adds support for PyTorch's torch.compile feature to autoBOT's neural networks, enabling JIT compilation for improved training and inference performance.

Changes Made

Core Implementation

• Added compile_model parameter to the SFNN class with default False for backward compatibility
• Updated torch_sparse_nn_ff_basic hyperparameter configuration to include compile_model: [False, True]
• Applied torch.compile() to models in SFNN.fit() when compilation is enabled
• Added comprehensive docstring documentation for the new parameter

Bug Fixes

• Fixed existing bug in predict_proba() method that caused unpacking errors when iterating over datasets without labels

Testing & Examples

• Created comprehensive test suite in tests/test_torch_compile.py with 4 test cases covering:
  • Default behavior (compilation disabled)
  • Compilation enabled scenarios
  • Hyperparameter optimization integration
  • Environment compatibility checks
• Added examples/minimal_torch_compile.py demonstrating usage with both compiled and non-compiled models
• Created docs/TORCH_COMPILE.md with detailed documentation

Usage

Through Hyperparameter Configuration

import autoBOTLib

custom_hyperparams = {
    "compile_model": [True]  # Enable torch.compile
}

autobot = autoBOTLib.GAlearner(
    texts, labels,
    framework="torch",
    custom_hyperparameters=custom_hyperparams
)

Direct Usage

from autoBOTLib.learning.torch_sparse_nn import SFNN

model = SFNN(compile_model=True)  # Enable compilation
model.fit(X_train, y_train)

Benefits

• Performance: Significant speedup during training and inference via JIT compilation
• Automatic: Works transparently with existing hyperparameter optimization
• Backward Compatible: Default behavior unchanged, existing code continues to work

Verification

All tests pass successfully, and compiled models show as OptimizedModule indicating torch.compile is working correctly. The hyperparameter optimization automatically explores compiled configurations when available.

Fixes #14.

---

💬 Share your feedback on Copilot coding agent for the chance to win a $200 gift card! Click here to start the survey.