Hermes 3 技術報告

Hermes 3 Technical Report

August 15, 2024

作者: Ryan Teknium, Jeffrey Quesnelle, Chen Guang

cs.AI

摘要

指導（或“聊天”）微調模型已成為大多數人與大型語言模型互動的主要方式。與“基礎”或“基礎”模型相反，指導微調模型被優化以回應命令性陳述。我們提出了Hermes 3，一個中立對齊的通用指導和工具使用模型，具有強大的推理和創造能力。其最大版本，Hermes 3 405B，在幾個公共基準測試中實現了開放權重模型的最新性能。

English

Instruct (or "chat") tuned models have become the primary way in which most people interact with large language models. As opposed to "base" or "foundation" models, instruct-tuned models are optimized to respond to imperative statements. We present Hermes 3, a neutrally-aligned generalist instruct and tool use model with strong reasoning and creative abilities. Its largest version, Hermes 3 405B, achieves state of the art performance among open weight models on several public benchmarks.