TinyGPT-V: Efficient Multimodal Large Language Model via Small Backbones

Learning LLM by doing

Tiny C++ LLM inference implementation from scratch

A tiny version of GPT fully implemented in Python with zero dependencies

🎈 A series of lightweight GPT models featuring TinyGPT Base (~51M params) and TinyGPT2 (~95M params). Fast, creative text generation trained on whimsical stories.

Simple Implementation of TinyGPTV in super simple Zeta lego blocks

No description available

A pure Rust GPT implementation from scratch.

A lightweight deep language model

TinyGPT - 9M Parameters

Tinystories version of gpt with custom inference engine

implementation of decoder-only GPT model for text generation

tinygpt allows users to train small (~2GB) language models based on GPT-2

Local GPU-powered console AI chatbot built with Ollama and Rich

nanoGPT in tinygrad

A tiny GPT model

A lightweight AI text generator based on distilgpt2

A tiny and fast GPT implement with JAX

No description available

My own super super tiny GPT

Tiny GPT model for simple text generation experiments.

No description available

No description available

In this repo , i have implemented a tiny gpt model (including attention layer , decoder blocks) from scratch.

No description available

No description available

整理学习版，博客依赖该仓库，不可删除

Inference for GPT.c models in microcontrollers.

A minimal implementation of a GPT-style language model built from scratch using PyTorch. This project demonstrates the core concepts of transformer architecture and language modeling.

TinyChatGPT is a from-scratch, decoder-only Transformer (GPT-style) chatbot project built to learn the full training pipeline end-to-end. It starts small on CPU (laptop-friendly) and scales to GPU/cloud later using the same architecture and codebase.