The Stack v3 becomes the largest open code dataset at 114 TB and 5T tokens
anton_lozhkov · x · 2026-07-24
The Stack v3 has been released as the largest open code dataset yet, expanding the open-source corpus for training and analysis.
- Scale: 114 TB raw, 15.9 TB processed, about 5T tokens after deduplication and filtering.
- Coverage: 770 languages, 224M repositories, and no restrictively licensed code included.
- Compared with v2, the dataset adds more recent open-source material and fixes a deduplication bug that affected earlier filtering.
- The post also notes major category shifts, including larger shares for C++, TypeScript, Rust, Python, and even COBOL.
Related event: Hugging Face Releases The Stack v3: 114TB Largest Open Source Code Dataset(13 posts)→
More from Research
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- The 'sweet lesson' is real: two kinds of ML problems need opposite strategies — mike64_t · 2026-10-27
- PixGR: Pixel-space generative decoding for fast feed-forward 3D reconstruction — kwangmoo_yi · 2026-10-27
- UniAction: unified action conditioning lets one world model handle mixed action types — ZeYanjie · 2026-10-27
- Infini-AI Lab open-sources Hermit, a framework for agent harness evolution and continual learning — BeidiChen · 2026-10-27
- Meta FAIR's Self-Direction turns agent failures into reusable LoRA skills, +13.8pp on Terminal-Bench — karen_ullrich · 2026-10-27