Will there be any problems replacing the <think></think> blocks in DeepSeek-R1 series output with quote markers to optimize the display effect?

[zxfishhack]

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 (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

Will replacing the blocks output by the DeepSeek-R1 series with quote markers cause problems in multi-round conversations? In the Chat Template of DeepSeek-R1, will be deleted from multi-round conversations. Will the replacement cause the thinking process to be added to the multi-round conversation? This practice is clearly prohibited in the official documentation of DeepSeek-R1.

2. Additional context or comments

3. Can you help us with this feature?

• I am interested in contributing to this feature.

[zxfishhack]

@crazywoola

[crazywoola]

Afaik, it won't

[HuChundong]

It's a breaking change commit.

This is a not-so-good upgrade. In the previous version, I was able to split R1's output into text segments, separating the 'think' part from the 'response' part. This update completely disrupted my workflow. Additionally, the content output by the LLM node is no longer the model's original output, increasing the complexity of processing for subsequent nodes. There is also no toggle or option provided to switch back or adjust this functionality.

[crazywoola]

See #13399 @HuChundong