Code Search On Codesearchnet
評価指標
Go
JS
Java
Overall
PHP
Python
Ruby
評価結果
このベンチマークにおける各モデルのパフォーマンス結果
モデル名 | Go | JS | Java | Overall | PHP | Python | Ruby | Paper Title | Repository |
|---|---|---|---|---|---|---|---|---|---|
| GraphCodeBERT | 84.1 | 71.1 | 75.7 | 77.4 | 72.5 | 87.9 | 73.2 | GraphCodeBERT: Pre-training Code Representations with Data Flow | |
| CodeT5+ 220M | 92.4 | 70..8 | 76.1 | 77.1 | 69.8 | 75.6 | 77.7 | CodeT5+: Open Code Large Language Models for Code Understanding and Generation | |
| CodeBERT | 69.3 | 74.8 | 86.8 | 76.0 | 70.6 | 84.0 | 70.6 | CodeBERT: A Pre-Trained Model for Programming and Natural Languages | |
| cpt-code S | 97.7 | 86.0 | 94.0 | 93.4 | 96.7 | 99.8 | 86.3 | Text and Code Embeddings by Contrastive Pre-Training | |
| cpt-code M | 97.5 | 86.5 | 94.4 | 93.5 | 97.2 | 99.9 | 85.5 | Text and Code Embeddings by Contrastive Pre-Training | |
| CodeT5+ 770M | 92.7 | 71.3 | 76.2 | 77.4 | 70.1 | 75.8 | 78.0 | CodeT5+: Open Code Large Language Models for Code Understanding and Generation |
0 of 6 row(s) selected.