Anton Lozhkov

Code models

Lozhkov is first author of "StarCoder 2 and The Stack v2: The Next Generation" (February 2024), a BigCode paper whose authors span Hugging Face, ServiceNow Research, NVIDIA and other institutions; the paper lists Lozhkov's affiliation as Hugging Face. StarCoder2-3B, one of the models it introduced, was trained on 17 programming languages from The Stack v2.²⁵⁶

Data and small models

Lozhkov co-authored the FineWeb datasets paper (June 2024) on building large web-text pretraining datasets, and is second author of the SmolLM2 paper (February 2025), which describes data-centric training of a small language model.³⁴

Work and contributions

• StarCoder2-3B Open weights

  First author of the StarCoder 2 paper (2024), which introduced StarCoder2-3B, a model trained on 17 programming languages from The Stack v2.²⁶

• FineWeb Open data

  Co-author of "The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale" (2024).³

• SmolLM2 Open weights

  Second author of the SmolLM2 paper (2025), on data-centric training of a small language model.⁴

• StarCoder2-15B Open weights

  First author of the StarCoder 2 paper, which introduced StarCoder2-15B and which the model card cites.²⁷

Timeline

1. Feb 2024Posted the StarCoder 2 and The Stack v2 paper as first author.²
2. Jun 2024Co-authored the FineWeb datasets paper.³
3. Feb 2025Co-authored the SmolLM2 paper as second author.⁴

Where they work

• Hugging Face 34 people in the wiki

Related people

• Guilherme PenedoAlso at Hugging Face
• Loubna Ben AllalAlso at Hugging Face
• Yacine JerniteAlso at Hugging Face
• Hynek KydlíčekAlso at Hugging Face
• Omar SansevieroAlso at Hugging Face
• Dan SuLanguage models · Datasets and data
• Kezhi KongLanguage models · Datasets and data
• Sridhar RamaswamyLanguage models · Datasets and data
• Yuanzhi LiLanguage models · Datasets and data
• Rohan TaoriLanguage models · Datasets and data