[ Tech stack ]
Hugging Face Transformers
The reference library for loading, fine-tuning and running models from the Hugging Face Hub.
Transformers is Hugging Face's open-source library that defines model architectures (LLMs, vision, audio, multimodal) and loads their pretrained weights from the Hub in one line: AutoModel, AutoTokenizer, pipeline(). The Trainer standardizes fine-tuning, with PEFT (LoRA) and quantization to fit on a modest GPU, and its model definitions serve as the reference for inference engines such as vLLM. It's our choice when a project needs to experiment with, fine-tune or evaluate open models before putting them into production.
[ Why Hugging Face Transformers at Dexon ]
What this technology does well, and why we use it.
Typical usage: Fine-tuning open models, text classification and extraction, computer vision, model prototyping.
- 01
Over a million Hub models, loaded in a single line.
- 02
Text, vision, audio and multimodal: one API, pipeline() included.
- 03
Trainer, PEFT/LoRA and quantization: fine-tuning on a modest GPU.
- 04
Reference model definitions, reused by vLLM, llama.cpp and MLX.
[ Complementary technologies ]
The building blocks we often mobilise alongside.
A stack rarely exists alone. Here are the technologies Dexon most often pairs with this one, through pipeline habits, usage similarity or internal mastery. Click on a building block to see its scope.
[ Reassurance ]
- 0+
- custom projects delivered
- 30
- engineers, designers, project managers
- 80 %
- from top French schools
- 24 h
- average reply time
[ They trust us ]
More than 100 French and European companies trust us
[ Press ]
They talk about us.
Nationwide coverage by BFM Business, Le Figaro, Challenges, La Tribune and CNews. An outside reading of our work and our innovations.
