Starcoder Dataset (The Stack - Sub-sampled)

This dataset is derived from the "Starcoder" version of The Stack, a 6.4 TB dataset of permissively licensed source code in 384 programming languages.

This repository contains the data organized into subsets, one for each programming language or data type.

How to Use

You can load any language-specific subset of the data using the datasets library. You must specify the name parameter with the desired language.

For example, to load the Python or Java subsets:

from datasets import load_dataset

# Load the Python subset
python_data = load_dataset("Sam-Shin/starcoder", name="python", split="train")

# Load the C++ subset
cpp_data = load_dataset("Sam-Shin/starcoder", name="cpp", split="train")

# Load the C# subset
csharp_data = load_dataset("Sam-Shin/starcoder", name="c-sharp", split="train")

print(python_data)