Code Generation On Turbulence
评估指标
CorrSc
评测结果
各个模型在此基准测试上的表现结果
模型名称 | CorrSc | Paper Title | Repository |
---|---|---|---|
GPT-3.5-Turbo | 0.617 | Turbulence: Systematically and Automatically Testing Instruction-Tuned Large Language Models for Code | |
CodeLlama:13B-4bit-quantised | 0.327 | Turbulence: Systematically and Automatically Testing Instruction-Tuned Large Language Models for Code | |
GPT-4 | 0.848 | Turbulence: Systematically and Automatically Testing Instruction-Tuned Large Language Models for Code | |
Command | 0.063 | Turbulence: Systematically and Automatically Testing Instruction-Tuned Large Language Models for Code | |
CodeLlama:7B-4bit-quantised | 0.289 | Turbulence: Systematically and Automatically Testing Instruction-Tuned Large Language Models for Code |
0 of 5 row(s) selected.