[Bug] The output is incomplete after inputting long text

[dllucyboy]

Contact Information

No response

MaxKB Version

专业版 v1.10.3-lts

Problem Description

创建了一个文本重排应用，主要是要将输入的文本，按照提示词设定的格式，重新编排并输出。文本总字数大概2000多，输出的时候不全。如果每次输入1000左右字数的，那么输出是完整的。通过xinference部署的本地模型，在xinference客户端的WEB UI界面测试，也能复现问题，但是调整xinference Web UI界面的max tokens后就正常了。但是返回MaxKB中，调整模型参数的最大输出tokens没有效果。感觉在MaxKB中设置这个max tokens参数没有效果。

Steps to Reproduce

xinference Web UI界面调整参数后就正常了。

The expected correct result

No response

Related log output

Additional Information

No response

[Shenguobin0102]

你好，请问你是在MaxKB的哪个位置调整模型的max_tokens参数的？实在应用里，还是在模型管理中？

[shaohuzhang1]

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

---

Hello, where are you adjusting the max_tokens parameter of the model at MaxKB? In real application, or in model management?

[zyyfit]

请问具体使用的是哪个模型呢？

[shaohuzhang1]

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

---

Which model are you using?

[dllucyboy]

你好，请问你是在MaxKB的哪个位置调整模型的max_tokens参数的？实在应用里，还是在模型管理中？

在工作流中。

[shaohuzhang1]

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

---

Hello, where do you adjust the max_tokens parameter of the model at which point do you adjust the model? In real application, or in model management?

in the workflow.

[dllucyboy]

请问具体使用的是哪个模型呢？

Qwen2.5 32B，deepseek-r1-distill-qwen-32b，glm4-chat-1m试过这几个

[shaohuzhang1]

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

---

Which model are you using?

Qwen2.5 32B, deepseek-r1-distill-qwen-32b, glm4-chat-1m tried these

[dllucyboy]

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

Which model are you using?

Qwen2.5 32B，deepseek-r1-distill-qwen-32b，glm4-chat-1m试过这几个

[Shenguobin0102]

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

Which model are you using?

Qwen2.5 32B，deepseek-r1-distill-qwen-32b，glm4-chat-1m试过这几个

你好，问题已确认，会在后续版本修复

[baixin513]

V1.10.5已发布，请更新试用。

[shaohuzhang1]

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

---

V1.10.5 has been released, please update and try it out.