Hugging Face's Transformers library now supports Quanto, GaLore, and LLAVa-NEXT (LLAVa-1.6) models. The integration brings new capabilities like MLX support and OS Vision Language mode. The updates offer enhanced functionality and ease of use for developers.
π πππππ-ππππ (πππππ-π.π) is now added to transformers library! LLAVa1.6 is a powerful OS Vision Language modeπ With just a few lines of code using @HuggingFace Transformers, you can now harness the power of this cutting-edge model. π» https://t.co/mEtqNSRdXk
It's happening: π€ Hugging Face's Transformers got some MLX support! - Tokenize directly to MLX arrays - Load MLX formatted safetensors to use with Transformers Release notes: https://t.co/2a4CyxQILI Example: https://t.co/JRXpVvgZ1x
Galore officially supported in @huggingface transformers ! check out the blogpost below for more details ! https://t.co/oJfF6wTBSD
Try GaLore in @huggingface transformers! https://t.co/JnsL0tu8KQ Thanks for the quick support @huggingface @younesbelkada More integration and upgrades are underway! GaLore repo: https://t.co/tH3NToifFu https://t.co/Q5f3tdRm53
Quanto library can now be used with @huggingface transformers ! check out all the details below ! https://t.co/g66uUn4TE7