Analyzing the Performance of Large Language Models on Code Summarization

Keyphrases

• BERTScore 33%
• Code Generation 33%
• Code Structure 33%
• Code Summarization 100%
• Code-based 33%
• Evaluation Metrics 33%
• Function Names 66%
• Highly Similar 33%
• Language Codes 33%
• Large Language Models 100%
• Llama 2 33%
• Multiple Evaluation 33%
• Natural Language 33%
• Natural Language Description 33%
• Natural Sources 33%
• Relative Performance 33%
• Source Code 33%
• Subword 33%
• Token 100%

Computer Science

• Code Generation 50%
• Evaluation Metric 50%
• Large Language Model 100%
• Relative Performance 50%
• Source Codes 50%