SantaCoder: don't reach for the stars!

Loubna Ben Allal, Raymond Li, Denis Kocetkov, Chenghao Mou, Christopher Akiki, Carlos Munoz Ferrandis, Niklas Muennighoff, Mayank Mishra, Alex Gu, Manan Dey, Logesh Kumar Umapathi, Carolyn Jane Anderson, Yangtian Zi, Joel Lamy Poirier, Hailey Schoelkopf, Sergey Troshin, Dmitry Abulkhanov, Manuel Romero, Michael Lappert, Francesco De Toni, Bernardo García del Río, Qian Liu, Shamik Bose, Urvashi Bhattacharyya, Terry Yue Zhuo, Ian Yu, Paulo Villegas, Marco Zocca, Sourab Mangrulkar, David Lansky, Huu Nguyen, Danish Contractor, Luis Villa, Jia Li, Dzmitry Bahdanau, Yacine Jernite, Sean Hughes, Daniel Fried, Arjun Guha, Harm de Vries, and Leandro von Werra
Deep Learning for Code Workshop (DL4C), 2023
Best Paper Award

The BigCode project is an open-scientific collaboration working on the responsible development of large language models for code. This tech report describes the progress of the collaboration until December 2022, outlining the current state of the Personally Identifiable Information (PII) redaction pipeline, the experiments conducted to de-risk the model architecture, and the experiments investigating better preprocessing methods for the training data. We train 1.1B parameter models on the Java, JavaScript, and Python subsets of The Stack and evaluate them on the MultiPL-E text-to-code benchmark. We find that more aggressive filtering of near-duplicates can further boost performance and, surprisingly, that selecting files from repositories with 5+ GitHub stars deteriorates performance significantly. Our best model outperforms previous open-source multilingual code generation models (InCoder-6.7B and CodeGen-Multi-2.7B) in both left-to-right generation and infilling on the Java, JavaScript, and Python portions of MultiPL-E, despite being a substantially smaller model. All models are released under an OpenRAIL license.

PDF available on arXiv

  author = {Allal, Loubna Ben and Li, Raymond and Kocetkov, Denis and 
    Mou, Chenghao and Akiki, Christopher and Ferrandis, Carlos Munoz and 
    Muennighoff, Niklas and Mishra, Mayank and Gu, Alex and Dey, Manan and
    Umapathi, Logesh Kumar and Anderson, Carolyn Jane and Zi, Yangtian and
    Poirier, Joel Lamy and Schoelkopf, Hailey and Troshin, Sergey and
    Abulkhanov, Dmitry and Romero, Manuel and Lappert, Michael and
    De Toni, Francesco and del Río, Bernardo García and Liu, Qian and
    Bose, Shamik and Bhattacharyya, Urvashi and Zhuo, Terry Yue and
    Yu, Ian and Villegas, Paulo and Zocca, Marco and Mangrulkar, Sourab and
    Lansky, David and Nguyen, Huu and Contractor, Danish and Villa, Luis and
    Li, Jia and Bahdanau, Dzmitry and Jernite, Yacine and Hughes, Sean and
    Fried, Daniel and Guha, Arjun and de Vries, Harm and von Werra, Leandro},
  title = {SantaCoder: don't reach for the stars!},
  booktitle = "Deep Learning for Code Workshop (DL4C)",
  year = {2023},