LLM
acronymcloud
Stands for: Large Language Model
A neural network trained on vast text that generates and understands language.
A Large Language Model is a transformer-based model trained on huge corpora to predict text, powering chat assistants, code tools, and retrieval systems. Its scale is what gives it broad capability.
A large language model predicts likely continuations of text, having been trained on a very large corpus. That framing matters because it explains both the capability and the characteristic failure: fluency is what it optimizes for, and fluency is not accuracy.
For technical work the practical consequences are specific. Output is plausible by construction, so a wrong answer looks exactly like a right one and must be verified against a source rather than assessed by how confident it sounds. It reflects its training data, including the outdated and the incorrect. And anything typed into a hosted model is disclosed to that service, which is a straightforward data handling question that policy has generally been slower to address than usage has been to spread.
Also known as: llm