Hugging Face's logo

Datasets:
bigcode
/
starcoder2data-extras

Modalities:
Tabular
Text
Formats:
parquet
Size:
10M - 100M
ArXiv:
Libraries:
Datasets
Dask
Dataset card Data Studio Files Community
2
starcoder2data-extras / README.md
loubnabnl's picture
loubnabnl HF staff
update config
1ba0d4f verified
preview code
|
raw
history blame contribute delete
7.02 kB
metadata
dataset_info:
  - config_name: arxiv
    features:
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 89223183645
        num_examples: 1558306
    download_size: 40911186876
    dataset_size: 89223183645
  - config_name: documentation
    features:
      - name: project
        dtype: string
      - name: source
        dtype: string
      - name: language
        dtype: string
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 5421472234
        num_examples: 59733
    download_size: 1853451922
    dataset_size: 5421472234
  - config_name: ir_cpp
    features:
      - name: __index_level_0__
        dtype: string
      - name: id
        dtype: string
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 102081135272
        num_examples: 2916655
    download_size: 26047978422
    dataset_size: 102081135272
  - config_name: ir_low_resource
    features:
      - name: __index_level_0__
        dtype: string
      - name: id
        dtype: string
      - name: content
        dtype: string
      - name: size
        dtype: int64
    splits:
      - name: train
        num_bytes: 10383382043
        num_examples: 393988
    download_size: 2464513603
    dataset_size: 10383382043
  - config_name: ir_python
    features:
      - name: id
        dtype: string
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 12446664464
        num_examples: 154507
    download_size: 3039297625
    dataset_size: 12446664464
  - config_name: ir_rust
    features:
      - name: __index_level_0__
        dtype: string
      - name: id
        dtype: string
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 4764927851
        num_examples: 32720
    download_size: 1254786199
    dataset_size: 4764927851
  - config_name: issues
    features:
      - name: repo_name
        dtype: string
      - name: content
        dtype: string
      - name: issue_id
        dtype: string
    splits:
      - name: train
        num_bytes: 31219575534.38484
        num_examples: 15549682
    download_size: 16483899047
    dataset_size: 31219575534.38484
  - config_name: kaggle
    features:
      - name: content
        dtype: string
      - name: file_id
        dtype: string
    splits:
      - name: train
        num_bytes: 5228745262
        num_examples: 580195
    download_size: 2234440007
    dataset_size: 5228745262
  - config_name: lhq
    features:
      - name: content
        dtype: string
      - name: metadata
        struct:
          - name: difficulty
            dtype: string
          - name: field
            dtype: string
          - name: topic
            dtype: string
    splits:
      - name: train
        num_bytes: 751273849
        num_examples: 7037500
    download_size: 272913202
    dataset_size: 751273849
  - config_name: owm
    features:
      - name: url
        dtype: string
      - name: date
        dtype: timestamp[s]
      - name: metadata
        dtype: string
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 56294728333
        num_examples: 6315233
    download_size: 27160071916
    dataset_size: 56294728333
  - config_name: stackoverflow
    features:
      - name: date
        dtype: string
      - name: nb_tokens
        dtype: int64
      - name: text_size
        dtype: int64
      - name: content
        dtype: string
    splits:
      - name: train
        num_bytes: 35548199612
        num_examples: 10404628
    download_size: 17008831030
    dataset_size: 35548199612
  - config_name: wikipedia
    features:
      - name: content
        dtype: string
      - name: meta
        dtype: string
      - name: red_pajama_subset
        dtype: string
    splits:
      - name: train
        num_bytes: 21572720540
        num_examples: 6630651
    download_size: 12153445493
    dataset_size: 21572720540
configs:
  - config_name: arxiv
    data_files:
      - split: train
        path: arxiv/train-*
  - config_name: documentation
    data_files:
      - split: train
        path: documentation/train-*
  - config_name: ir_cpp
    data_files:
      - split: train
        path: ir_cpp/train-*
  - config_name: ir_low_resource
    data_files:
      - split: train
        path: ir_low_resource/train-*
  - config_name: ir_python
    data_files:
      - split: train
        path: ir_python/train-*
  - config_name: ir_rust
    data_files:
      - split: train
        path: ir_rust/train-*
  - config_name: issues
    data_files:
      - split: train
        path: issues/train-*
  - config_name: kaggle
    data_files:
      - split: train
        path: kaggle/train-*
  - config_name: lhq
    data_files:
      - split: train
        path: lhq/train-*
  - config_name: owm
    data_files:
      - split: train
        path: owm/train-*
  - config_name: stackoverflow
    data_files:
      - split: train
        path: stackoverflow/train-*
  - config_name: wikipedia
    data_files:
      - split: train
        path: wikipedia/train-*

StarCoder2 Extras

This is the dataset of extra sources (besides Stack v2 code data) used to train the StarCoder2 family of models. It contains the following subsets:

  • Kaggle (kaggle): Kaggle notebooks from Meta-Kaggle-Code dataset, converted to scripts and prefixed with information on the Kaggle datasets used in the notebook. The file headers have a similar format to Jupyter Structured but the code content is only one single script.
  • StackOverflow (stackoverflow): stackoverflow conversations from this StackExchange dump.
  • Issues (issues): processed GitHub issues, same as the Stack v1 issues.
  • OWM (owm): the Open-Web-Math dataset.
  • LHQ (lhq): Leandro's High quality dataset, it is a compilation of high quality code files from: APPS-train, CodeContests, GSM8K-train, GSM8K-SciRel, DeepMind-Mathematics, Rosetta-Code, MultiPL-T, ProofSteps, ProofSteps-lean.
  • Wiki (wikipedia): the English subset of the Wikipedia dump in RedPajama.
  • ArXiv (arxiv): the ArXiv subset of RedPajama dataset, further processed the dataset only to retain latex source files and remove preambles, comments, macros, and bibliographies from these files.
  • IR_language (ir_cpp, ir_low_resource, ir_python, ir_rust): these are intermediate representations of Python, Rust, C++ and other low resource languages.
  • Documentation (documentation): documentation of popular libraries.

For more details on the processing of each subset, check the StarCoder2 paper or The Stack v2 GitHub repository.

Usage 

from datasets import load_dataset

# replace `kaggle` with one of the config names listed above
ds = load_dataset("bigcode/starcoder2data-extras", "kaggle", split="train")

Citation 

@article{lozhkov2024starcoder,
  title={Starcoder 2 and the stack v2: The next generation},
  author={Lozhkov, Anton and Li, Raymond and Allal, Loubna Ben and Cassano, Federico and Lamy-Poirier, Joel and Tazi, Nouamane and Tang, Ao and Pykhtar, Dmytro and Liu, Jiawei and Wei, Yuxiang and others},
  journal={arXiv preprint arXiv:2402.19173},
  year={2024}
}