DialoGPT-medium
DialoGPT is a SOTA large-scale pretrained dialogue response generation model for multiturn conversations. The human evaluation results indicate that the response generated from DialoGPT is comparable to human response quality under a single-turn conversation Turing test. The model is trained on 147M multi-turn dialogue from Reddit discussion thread. Please find the information about preprocessing, training and full details of the DialoGPT in the original DialoGPT repository ArXiv paper: https://arxiv.org/abs/1911.00536
译文尚未就绪,显示来源原文
DialoGPT is a SOTA large-scale pretrained dialogue response generation model for multiturn conversations. The human evaluation results indicate that the response generated from DialoGPT is comparable to human response quality under a single-turn conversation Turing test. The model is trained on 147M multi-turn dialogue from Reddit discussion thread. Please find the information about preprocessing, training and full details of the DialoGPT in the original DialoGPT repository ArXiv paper: https://arxiv.org/abs/1911.00536
AI NEWS.cc