Skip to main content
In tasks such as multi-turn reasoning, multi-hop question answering, or iterative retrieval, a single execution often fails to produce an ideal final answer.
In such cases, a loop structure can be used to repeatedly execute specific modules, enabling gradual information refinement and continuous optimization of results.

​
Usage Example

https://mintcdn.com/ultrarag/T7GffHzZitf6TThi/images/yaml.svg?fit=max&auto=format&n=T7GffHzZitf6TThi&q=85&s=69b41e79144bc908039c2ee3abbb1c3bexamples/rag_loop.yaml
# MCP Server
servers:
  benchmark: servers/benchmark
  retriever: servers/retriever
  prompt: servers/prompt
  generation: servers/generation
  evaluation: servers/evaluation
  custom: servers/custom

# MCP Client Pipeline
pipeline:
- benchmark.get_data
- retriever.retriever_init
- generation.generation_init
- retriever.retriever_search
- loop:
    times: 3
    steps:
    - prompt.gen_subq
    - generation.generate:
        output:
          ans_ls: subq_ls
    - retriever.retriever_search:
        input:
          query_list: subq_ls
        output:
          ret_psg: temp_psg
    - custom.merge_passages
- prompt.qa_rag_boxed
- generation.generate
- custom.output_extract_from_boxed
- evaluation.evaluate
In a loop structure, the Pipeline uses the following keywords to define the tools that should be executed repeatedly:
  • loop: Declares a loop block, indicating that the enclosed steps will be executed multiple times.
  • times: Specifies the maximum number of iterations.
  • steps: Defines the sequence of tool invocations within each iteration.
If you wish to dynamically control loop termination conditions, you can combine the Branch Structure (branch) with the Router Server. See Branch Structure for details.