The Stack v3, a 15.9 TB dataset comprising source code from 173 million repositories across 713 programming languages, has been released. This dataset, compiled by Hugging Face, was scraped from GitHub in 2025. Concerns have been raised regarding the use of personal and organizational projects, including potentially copyrighted material like PhD theses and algorithms, for training large language models. AI
IMPACT Raises questions about data provenance and copyright for LLM training datasets.
RANK_REASON Release of a large dataset for AI training. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
