AFlow is used for workflow construction;
Sela handles the tasks related to AutoML;
FACT is responsible for multi-fact incremental retrieval.
It seems that a verifier,
that all of its indicators tops arxiv metrics just like AFlow Sela and FACT, is needed,
to verify whether the result of execution of code ,
meets the user's requirement documentation.
Then Programer will only need to ask questions....
FACT will help to search useful information,
AFlow generate good workflow , based on PRD and FACT's searching,
Sela generate useful ml agents code , based on workflow, PRD and FACT's searching.
Finally "verifier" will run code and will review result , and will update PRD , to start next loop.
AFlow is used for workflow construction; Sela handles the tasks related to AutoML; FACT is responsible for multi-fact incremental retrieval.
It seems that a verifier, that all of its indicators tops arxiv metrics just like AFlow Sela and FACT, is needed,
to verify whether the result of execution of code , meets the user's requirement documentation.
Then Programer will only need to ask questions.... FACT will help to search useful information,
AFlow generate good workflow , based on PRD and FACT's searching, Sela generate useful ml agents code , based on workflow, PRD and FACT's searching. Finally "verifier" will run code and will review result , and will update PRD , to start next loop.