| # CodeBridge: Multilingual Code Understanding Framework | |
| 📑 [Paper](https://huggingface.co/papers/yyyy.yyyyy) | 🌐 [Project Page](https://codebridge.github.io/) | 💾 [Released Resources](https://huggingface.co/collections/toolevalxm/codebridge-67a978e28fd926b56a4f55a3) | 📦 [Repo](https://github.com/xmhtoolathlon/CodeBridge-Synthesis) | |
| We release the raw multilingual code data for our CodeBridge-MultiLang dataset, derived from the original CodeParrot dataset maintained by the Hugging Face team. | |
| The data format for each line in the `multilang_code_filtered.jsonl` is as follows: | |
| ``` | |
| { | |
| "code_snippet": <the original code in the source language>, | |
| "language": <programming language identifier>, | |
| "docstring": <extracted documentation>, | |
| "function_name": <the main function name>, | |
| "complexity_score": <computed cyclomatic complexity>, | |
| "source_repo": <the originating repository>, | |
| "metadata": <additional provenance information> | |
| } | |
| ``` | |
| Some entries may have incomplete docstrings due to parsing limitations in certain languages. | |
| *Note: The dataset covers 12 programming languages including Python, Java, JavaScript, Go, Ruby, Rust, C++, C#, PHP, TypeScript, Kotlin, and Swift. | |
| **License** | |
| The license for this dataset is other. |