This is a 3-way classifier judge model fine-tuned on the Chatbot Arena human preference dataset. The base model is llama 13B. More details can be found in the Appendix. F of this paper.
This is a 3-way classifier judge model fine-tuned on the Chatbot Arena human preference dataset. The base model is llama 13B. More details can be found in the Appendix. F of this paper.