rank_strategy.cpp计算逻辑问题
Open
- Dominant language
- C++
- Stars
- 2.6k
- Forks
- 657
- PR merge metrics
- No merged PRs in 30d
Description
大致阅读了框架代码,在run_strategy函数里有两个细节有点困惑,希望能够得到解答。谢谢
1)添加了粗排特征、分数和过滤的计算,然后取top_k的候选。 这块不是太懂,召回就充当了粗排了吧, 为什么需要在排序的时候添加这块设计?
2)在计算精排特征、分数和排序之后,对候选进行去重, 这块也很懵, 召回的时候应该已经对各路分支召回的汇总结果进行的去重,这块设计有什么额外的点我没有get到吗?
Contributor guide
No contributing guide indexed for this repository
Research direction
Start in rank_strategy.cpp at the run_strategy function and trace where coarse-ranking features, scores, filtering, top_k selection, and final deduplication are applied. Compare these stages with the retrieval flow, then document the distinct purpose of each step and answer both questions with references to the relevant code paths.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- search
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100