PaddlePaddle / PaddleNLP

👑 Easy-to-use and powerful NLP and LLM library with 🤗 Awesome model zoo, supporting wide-range of NLP tasks from research to industrial applications, including 🗂Text Classification, 🔍 Neural Search, ❓ Question Answering, ℹ️ Information Extraction, 📄 Document Intelligence, 💌 Sentiment Analysis etc.
https://paddlenlp.readthedocs.io
Apache License 2.0
12.11k stars 2.94k forks source link

[Question]: taskflow 文本相似度RocketQA 模型,输入同样的内容,结果都不是100%,而且比较低 #4468

Closed zhiyongLiu1114 closed 1 year ago

zhiyongLiu1114 commented 1 year ago

请提出你的问题

taskflow,基于百万量级Dureader Retrieval数据集训练RocketQA并达到前沿文本相似效果,该模型预测的结果整体偏低,即使是完全相同的句子,结果也不是100%

1649759610 commented 1 year ago

你好,可以说明下你的测试数据情况(比如领域,文本长短,数据量等),以及列出一些case看看吗?

w5688414 commented 1 year ago

rocketQA模型适合query passage这样的pair对,如果是query-query pair对,请使用simbert

zhiyongLiu1114 commented 1 year ago

![Uploading image.png…]()

rocketQA模型适合query passage这样的pair对,如果是query-query pair对,请使用simbert

了解了,谢谢