tenyx/Llama3-TenyxChat-70B

Llama3-TenyxChat-70B is a 70 billion parameter instruction-tuned language model developed by Tenyx Research, fine-tuned from Meta's Llama3-70B using Direct Preference Optimization (DPO) on the UltraFeedback dataset. This model is designed to function as a useful assistant, excelling in multi-turn chat scenarios by mitigating catastrophic forgetting. It achieves a high MT-Bench score of 8.15, making it a top-ranked open-source model for conversational AI at its release.