HuggingFace Transformers Drop JAX & TensorFlow Support | PyTorch Focus

Recently, a significant pull request was merged into HuggingFace’s transformers repository — they removed about 50,000 lines of code in one go. How did it happen? It’s quite simple: the developers decided they will no longer support integration with JAX and TensorFlow. Currently, the library supports only PyTorch.

They justify this by wanting to make the project simpler and to eliminate unnecessary complexity, avoiding bloating the functionality without real need. As a result, support for JAX disappears, and although TensorFlow is not cut as sharply, it is also being reduced.

Honestly, it’s very unfortunate to lose support for JAX (the situation with TensorFlow is a bit easier, but still not very pleasing).

Page view /ai-blog/huggingface-transformers-drop-jax-tensorflow-support-pytorch-focus/ 25.04 12:45 Page view 25.04 12:39 Page view /ai-blog/free-opus-4-5-competitive-pricing-trends-by-google/ 25.04 12:32 Page view 25.04 12:29 Page view /ai-blog/ai-video-generator-open-source-t2v-model-by-brothers/ 25.04 12:17 Page view /ai-blog/trump-reveals-secret-sound-weapon-in-venezuela-exclusive-insight/ 25.04 12:17 Page view /ai-blog/pentagon-announces-xai-grok-integration-military-systems-boost/ 25.04 12:16 Page view /ai-blog/robots-in-retail-real-life-applications-safety-concerns/ 25.04 12:12 Page view /ai-blog/hunyuanimage-3-0-instruct-advanced-open-source-image-editing/ 25.04 12:08 Page view 25.04 11:58