模型:
microsoft/DialogRPT-updown
请尝试这个 ➤➤➤ Colab Notebook Demo (click me!)
Context | Response | updown score |
---|---|---|
I love NLP! | Here’s a free textbook (URL) in case anyone needs it. | 0.613 |
I love NLP! | Me too! | 0.111 |
上下得分(updown score)预测回应被点赞的可能性有多大。
对话回应有多大可能被点赞?或者得到回复??
这是 DialogRPT 学会预测的东西。它是由 Microsoft Research NLP Group 提出的一组对话回应排序模型,使用了1亿多条人类反馈数据进行训练。可以通过重新排序生成的回应候选集来提高现有的对话生成模型(例如 DialoGPT )。
快速链接:
我们考虑了以下任务,并提供了相应的预训练模型。本页是针对 updown 任务的,其他模型卡片可以在下表中找到。
Task | Description | Pretrained model |
---|---|---|
Human feedback | given a context and its two human responses, predict... | |
updown | ... which gets more upvotes? | this model |
width | ... which gets more direct replies? | 1238321 |
depth | ... which gets longer follow-up thread? | 1239321 |
Human-like (human vs fake) | given a context and one human response, distinguish it with... | |
human_vs_rand | ... a random human response | 12310321 |
human_vs_machine | ... a machine generated response | 12311321 |
请在 our repo 上创建一个问题。
@inproceedings{gao2020dialogrpt, title={Dialogue Response RankingTraining with Large-Scale Human Feedback Data}, author={Xiang Gao and Yizhe Zhang and Michel Galley and Chris Brockett and Bill Dolan}, year={2020}, booktitle={EMNLP} }