Pinned Loading

1. Neural-Edge-Distiller Neural-Edge-Distiller Public

   Synthetic CoT distillation pipeline — transfers structured reasoning from Llama 3.2 8B to 3B via LoRA fine-tuning on Apple Silicon using MLX. Includes data generation, training, benchmarking, and a…

   Python

2. RAG-Projects RAG-Projects Public

   A multi-stage RAG (Retrieval-Augmented Generation) system featuring local LLMs (Ollama/Gemma), hybrid search, and semantic re-ranking with Cross-Encoders.

   Python

3. beverina-safety-rag beverina-safety-rag Public

   End-to-end LLM system for ingredient safety analysis using RAG, FAISS, and QLoRA with local GGUF deployment.

   Python 1

4. speculative-llm-engine speculative-llm-engine Public

   Low-latency LLM inference engine using speculative decoding, optimized for multilingual QA workloads.

   Python