Hermes 2 Theta Llama 3 8B

About

Hermes-2-Theta-Llama-3-8B is an advanced large language model designed by NousResearch in partnership with Charles Goddard from Arcee. This model merges the capabilities of NousResearch's Hermes 2 Pro and Meta's Llama-3 Instruct, further refined with Reinforcement Learning from Human Feedback (RLHF). Notably, it adopts the ChatML prompt format, facilitating structured multi-turn dialogues and system prompts. Hermes-2-Theta-Llama-3-8B excels in diverse areas such as general conversation, creative writing, and executing function calls, alongside generating structured JSON outputs. Its notable design aims for low hallucination rates, positioning it as an efficient and reliable AI tool. The model is available on Hugging Face and other platforms, with various quantized versions for different hardware.

Capabilities

Providers(1)