LCC_csharp is a C# code comment dataset released by Microsoft in 2023, derived from the Large Code Comments (LCC) project. It aims to provide a high-quality corpus for tasks such as code understanding, code generation, and code comment analysis.

The dataset comprises 120,000 C# code and comment samples, split into training, validation, and test sets. Each sample primarily includes code context and the corresponding target comment, which can be used for pre-training, fine-tuning, and evaluating NLP models related to code.

Dataset Composition

The dataset consists of the following three parts:

• Training set (train): 100,000 samples, approximately 1.85 GB.
• Validation set (validation): 10,000 samples, approximately 136 MB.
• Test set (test): 10,000 samples, approximately 136 MB.

Each sample primarily contains the following fields:

• context: The C# code context.
• gt: The corresponding target comment or label.