root@fr:~$ lexique/inference

Inférence

Models

Running an already trained model: it receives a request and produces its answer, token by token. It is the phase that runs day after day and carries the real bill of an AI system; its cost keeps falling year after year, which makes self-hosting capable models economically viable.

Also called Inference

Related terms

Entraînement

Models

The phase where a model learns its parameters from very large volumes of data, a compute effort reserved to a handful of global players. Owning your AI does...

Token

Models

The unit of text a model works with: a short word, a fragment of a word or a punctuation mark. It is also the billing unit of inference; the volume of...

Modèle open-weight

Models

An AI model whose weights are released and can be downloaded, run and adapted on your own infrastructure. It enables self-hosting and execution sovereignty,...

GPU

Architecture

A massively parallel processor, originally designed for graphics, now the standard hardware for training and running AI models. Its production is...

Does this notion touch your business?

Go from definition to deployment: a first agent on your data, in France, on a model you control end to end.

The right model Let's talk about your AI