[Bug] During the Q&A process, connect the visual model, upload the picture Q&A, and generate another answer, and the display content is incorrect.

[fit2cloud-zaopengguo]

Contact Information

[email protected]

MaxKB Version

1.10.1

Problem Description

1、工作流接入视觉模型，配置图片理解节点 2、问答过程中上传图片，第一次提问回答正常 3、点击“换个答案”，maxkb 没有把图片再次带给大模型。

Steps to Reproduce

如上描述

The expected correct result

No response

Related log output

Additional Information

No response

[baixin513]

感谢反馈，换个答案之前是针对知识库检索的，针对高级编排应用确实有问题，后续版本中迭代。

[shaohuzhang1]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

Thanks for the feedback. The answer was previously searched for the knowledge base. There are indeed problems with advanced orchestration applications, and iterated in subsequent versions.

[zyyfit]

您好，目前换个答案不支持携带图片，您可以把图片理解节点的历史聊天记录次数设置的多一点，这样换个答案的时候模型会根据第一次传的图片进行回答

[shaohuzhang1]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

Hello, the current answer does not support carrying pictures. You can set the number of historical chat records of the image understanding node a little more. In this way, when changing the answer, the model will answer based on the first transmitted picture.