vibebin: code and host inside Incus containers on your own VPS/server.

Cross-platform toolkit to enhance Claude Code with multi-LLM consensus, 8 specialist agents, semantic knowledge search, and one-command install.

nanoGPT for Diffusion Language Models

Decoder-only Transformer language model implemented from scratch in PyTorch, trained on Tiny Shakespeare. / Tiny Shakespeare 데이터셋으로 character-level 텍스트 생성을 학습하는, PyTorch로 직접 구현한 Decoder-only Transformer 언어모델.

Provide specialized AI agents that develop, review, debug, and deploy production-ready code efficiently across various programming tasks.

Transformer Based Learning of Fundamental Linear Algebra Operations

From-scratch LLM inference engine in PyTorch. MHA/GQA/MQA in one module, RoPE (NeoX/Llama half-rotation), append-only KV cache with numerical parity vs F.scaled_dot_product_attention, RMSNorm + SwiGLU decoder blocks, KV-cached greedy/top-k sampling. Trains a 1M-param GQA transformer on Tiny Shakespeare end-to-end.

nanoGPT - A simple GPT-Style Transformer from Scratch in PyTorch

A minimal GPT implementation created as practice alongside Andrej Karpathy’s NanoGPT lessons.

A lightweight GPT-based model that learns to generate concise news headlines given the full body text of news articles. Inspired by karpathy/nanoGPT, this project fine-tunes a Transformer on a real-world dataset to perform text-to-title generation.

🛠️ Enhance VSCode with syntax highlighting, LaTeX rendering, and AI response copying. Improve your coding experience with practical features.