Originally created by @yuchenxie4645 on GitHub (Aug 5, 2025).
For reasoning models like Qwen3, DeepSeek, and other models from API, long reasoning for hard problems like AIME and USACO problems results in slow formatting, and lag. Lag causes token streaming to be slowed down dramatically within UI. I was thinking maybe a reasoning stripper/summarizer that google and openai uses for their reasoning models? How do we implement it?
Originally created by @yuchenxie4645 on GitHub (Aug 5, 2025).
For reasoning models like Qwen3, DeepSeek, and other models from API, long reasoning for hard problems like AIME and USACO problems results in slow formatting, and lag. Lag causes token streaming to be slowed down dramatically within UI. I was thinking maybe a reasoning stripper/summarizer that google and openai uses for their reasoning models? How do we implement it?
GiteaMirror
added the bug label 2025-11-11 16:40:13 -06:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @yuchenxie4645 on GitHub (Aug 5, 2025).
For reasoning models like Qwen3, DeepSeek, and other models from API, long reasoning for hard problems like AIME and USACO problems results in slow formatting, and lag. Lag causes token streaming to be slowed down dramatically within UI. I was thinking maybe a reasoning stripper/summarizer that google and openai uses for their reasoning models? How do we implement it?