State-of-the-art transformers for Ruby

A Ruby-native interface for fine-tuning machine learning models, starting with image classification using SigLIP2

A Ruby port of mlx-lm providing large language model inference, quantization, LoRA fine-tuning, and an OpenAI-compatible server built on the mlx gem. Supports Llama, Gemma, Qwen2, Phi3, Mixtral, DeepSeek, and many more architectures.

Training loop, LoRA, and optimization utilities for fine-tuning HuggingFace transformer models using torch-rb and transformers-rb. Supports full fine-tuning, LoRA adapters, learning rate scheduling, callbacks, and model serialization via safetensors.