HumanEvalPlus

The 'HumanEvalPlus' dataset/benchmark is an enhanced version of the HumanEval benchmark used to evaluate AI models in software engineering tasks, particularly in code generation.

Papers using HumanEvalPlus (6)

Demystifying Errors in LLM Reasoning Traces: An Empirical Study of Code Execution Simulation2025Benchmarking AI Models in Software Engineering: A Review, Search Tool, and Enhancement Protocol2025ACECODER: Acing Coder RL via Automated Test-Case Synthesis2025Dynamic Scaling of Unit Tests for Code Reward Modeling2025Uncovering Weaknesses in Neural Code Generation2024 · 2 citesMulti-Programming Language Ensemble for Code Generation in Large Language Model2024 · 1 cites