Pre-training
སྔོན་ཚུད་སྦྱོང་བརྡར།
Tibetan source: Community contribution
预训练
Pre-training · སྔོན་ཚུད་སྦྱོང་བརྡར། · 预训练 · pree-TRAY-ning · pretraining · foundation model training · Artificial Intelligence · training · llm · mlArtificial Intelligencepretraining foundation model training/pree-TRAY-ning/
Also known as: pretraining, foundation model training
Last updated September 16, 2026
Definition
Pre-training is the first, large-scale training stage of a model on a broad dataset — often unlabeled text or images — so it learns general patterns before anyone fine-tunes it for a specific job.
Real-World Example
An LLM is pre-trained on a huge crawl of public text so it can predict the next word; a clinic later fine-tunes a copy on medical notes.
References
Related terms
discuss://translation · public
Translation discussion
Anyone can read and write here. Pick a display name — no email or account. Please keep the thread constructive. Thumb up a rendering you agree with — that helps the editor see consensus; it does not publish the term by itself. You can edit or delete your own comments from this browser; another device or cleared site data cannot.
Community lists who wrote, on which term, and when.
Loading discussion…