QualityFlow: An Agentic Workflow for Program Synthesis Controlled by LLM Quality Checks
Open paper
Benchmark
Model
Rank
Results
code-generation-on-humaneval
QualityFlow (Sonnet-3.5)
#1
Pass@1: 98.8
code-generation-on-mbpp
QualityFlow (Sonnet-3.5)
–
Accuracy: 94.2
Rank counts only results with a code link.