[Bug]: Missing Characters in Thought Chain When Using API for Inference

[SimZun6]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (Language Policy).
• Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) (Language Policy).
• Please do not modify this template :) and fill in all the required fields.

RAGFlow workspace code commit ID

bf0d516

RAGFlow image version

v17.0.0

Other environment information

Actual behavior

When using the API to call the inference model in RAGFlow, the output's thought chain section is missing some characters. However, this issue does not affect other parts of the output. Please investigate this issue.

Expected behavior

When using the API to call the inference model in RAGFlow, the output should include the complete thought chain without missing any characters. Other sections of the output should also display correctly without issues.

Steps to reproduce

1. Use the RAGFlow API to call the inference model.
2. Ensure that the output includes a thought chain.
3. Check the output and observe that some characters in the thought chain are missing.
4. Compare this with other sections of the output, which should be complete and correct.

Additional information

No response

[KevinHuSh]

Pull the nightly version of docker image.

[SimZun6]

Pull the nightly version of docker image.

I tried 0.17.2 version before, but I found it also has a bug that I can't use local ollama embedding model service.