Pinned Loading

1. LLM_guided_sokoban_planner LLM_guided_sokoban_planner Public

   an AI-powered Sokoban puzzle solver that combines classical search algorithms (Weighted A*, Beam Search, MCTS) with Large Language Model (LLM) guidance to solve complex pushing puzzles.

   Python

2. Inference-Time-Optimization-for-Open-Source-LLMs Inference-Time-Optimization-for-Open-Source-LLMs Public

   A Controlled Experiment: Vanilla → vLLM → vLLM + Deep Kernel Fusion

   Python

3. LlmPostTraining LlmPostTraining Public

   A comprehensive, hands-on guide to LLM Post-Training. Implements SFT, LoRA, and the full RLHF pipeline (Reward Modeling & PPO) using GPT-2. Includes InstructGPT-style alignment and reasoning distil…

   Jupyter Notebook

4. ProdBot ProdBot Public

   🤖 From Zero to Production AI Chatbot — Self-hosted, multimodal, agentic chatbot platform with MCP tools, real-time voice, hybrid memory, and full observability. Built with FastAPI, Next.js, Ollama …

   Python