LlamaGPT

LlamaGPT is a chatbot that delivers an experience similar to ChatGPT, with no data leaving your device. Note: The download size for LlamaGPT is approximately 5.5GB. Depending on your internet speed, installation may take some time. LlamaGPT consumes about 5GB of RAM. As a result, it is not suitable for users running Umbrel OS on a Raspberry Pi 4 with 4GB of RAM. For the best user experience, at least 8GB of RAM is recommended. LlamaGPT is optimized for Umbrel Home, generating words at a speed of ~3 words per second. On a Raspberry Pi 4 with 8GB of RAM, it produces words at ~1 word per second. Performance may vary depending on other applications installed on your Umbrel. LlamaGPT is powered by the state-of-the-art Hermes Llama 2 7B language model, fine-tuned on over 300,000 instructions to deliver longer responses and lower hallucination rates. LlamaGPT benefits from the incredible open-source work of various developers and teams. We acknowledge McKay Wrigley for building the Chatbot UI, Georgi Gerganov for implementing Llama.cpp, Andrei for developing Python bindings for Llama.cpp, NousResearch for fine-tuning the model, and Tom Jobbins for quantizing the Llama 2 model, as well as Meta for making Llama 2 available under a permissive license.

LlamaGPT is a chatbot that offers a ChatGPT-like experience based on the Llama 2 model, with the core value of keeping data on the user's device and providing longer, less hallucinatory responses.