Awesome AI Agents › AI Assistants

chadbrewbaker/ChadGPT

⭐ 10 Jupyter Notebook added to this list on 2025-04-19 repository created 2023-04-19

ChadGPT is a project centered around large language models (LLMs) and their self-hosting, training, and application. The repository provides insights into what LLMs are, explaining that they are data structures that predict the probability of the next token given a sequence of tokens. It highlights the scale of models like ChatGPT-4 and discusses the challenges of data availability for training such models. The project references various tools and frameworks related to LLMs, including llama.cpp, a widely used C++ codebase for hosting LLMs, and web-llm, which enables running LLMs in browsers using WebGPU. It also mentions tokenization tools like OpenAI's TickToken and smaller LLM implementations such as BabyGPT, PicoGPT, and NanoGPT. The repository includes information about the cost and complexity of training LLMs, noting that training can be expensive but fine-tuning can be done at a much lower cost. It points to hardware accelerators like Cerebras wafer-scale chips and Tenstorrent accelerators that optimize tensor operations crucial for LLM training and inference. The project also discusses practical applications of LLMs, such as transcription with whisper.cpp, autonomous agents like AutoGPT and BabyAGI, and integration with tools like GitHub Copilot. Additionally, ChadGPT provides updates and news related to Linux and open-source developments, including upcoming conferences, kernel patches, and new software releases. It envisions future Linux systems incorporating LLMs for tasks like data compression and scheduling, predicting significant changes in system architecture and software management. Overall, ChadGPT serves as a comprehensive resource for understanding, running, and leveraging large language models, with a focus on open-source tools, community news, and future trends in AI and Linux systems.

https://github.com/chadbrewbaker/ChadGPT

autogptbabyagibabygptcerebras-chipsdata-compressiondistributed-schedulersfine-tuninggithub-copilotinferencekernellarge-language-modelslinuxllama.cppllmnanogptopenai-ticktokenpicogptself-hostingtensor-operationstenstorrent-acceleratorstrainingweb-llmwebgpuwhisper.cpp