Language Model
Definition
A language model assigns probabilities to text sequences or predicts linguistic units from context.
Modern neural language models mostly operate on tokens and are trained autoregressively or with masking objectives. A language model is not automatically a large language model.
Simply explained
Like a system that has learned which text continuations are plausible in a context.
Why it matters in practice
Basis for text generation, translation, search, and many NLP systems.
Example from business
A domain-specific model evaluates which text sequence is more likely in technical descriptions.
Limits and pitfalls
- Confusing probability with truth; equating language model and knowledge base.
Origin & history
Statistical language models have existed for decades; neural models became influential from the 2000s and Transformers from 2017 onward.
Use in mid-sized companies
- Possible application in SMEs: Creating understanding of terms for selection and limits of LLM and NLP systems.
Frequently asked questions
Is every language model an LLM?
No. Does high probability say anything about factual accuracy? Not necessarily.
Sources
- A Neural Probabilistic Language Model · Bengio et al. (2003), A Neural Probabilistic Language Model
Sources checked on 10.10.2026.
Editorially reviewed on: