Transformers v5: Simple model definitions powering the AI ecosystem

Today, as we launch v5, Transformers is installed more than 3 million times each day via pip - up from 20,000/day in v4. Altogether, it has now surpassed 1.2 billion installs! The ecosystem has expanded from 40 model architectures in v4 to over 400 today, and the community has contributed more than 750,000 model checkpoints on the Hub compatible with Transformers. This growth is powered by the evolution of the field and the now mainstream access to AI. For v5, we wanted to work on several notable aspects: simplicity, training, inference, and production. The first focus of the team was on simplicity. We want our model integrations to be clean, so that the ecosystem may depend on our model definitions and understand what’s really happening under the hood. Over the past year, we’ve heavily pushed our modular design as a significant step forward. This allows for easier maintenance, faster integration, and better collaboration across the community. We’re sunsetting our Flax/TensorFlow support in favor of focusing on PyTorch as the sole backend. Training remains a big focus: we’ve recently done significant work to improve our support for pre-training/full-training at scale. We’re also putting a significant focus on inference for v5, with the introduction of specialized kernels, cleaner defaults, and new APIs like 'transformers serve' which deploys an OpenAI API-compatible server.