Nous Research announces new open source model beyond Mixtral Instruct
Nous Research announced that a new model based on Mixtral8x7B, Nous-Hermes-2Mixtral8x7B, surpasses Mixtral Instruct on multiple benchmarks. The model was trained on more than 1 million GPT-4 generated data and other high-quality data, divided into SFT
2025-01-30