The Hugging Face Transformers library has released version 5.24.0, introducing several new models. These include Inkling 2 Mini, a multimodal model from Thinking Machines Lab with a unique looping mechanism for enhanced quality, and Apertus 1.5, a Swiss AI Initiative model that processes text, images, and audio. The release also features WavTokenizer for audio encoding and decoding, Dory, a looped language model from NVIDIA, and HY4-VL, a vision-language model from Alibaba combining a mixture-of-experts text backbone with a vision encoder. Additionally, Omnilingual ASR models for multilingual speech recognition and Lara Mini, a multilingual base language model from Translated, are now available. AI
IMPACT Expands the range of available models for developers, particularly in multimodal and multilingual capabilities.
RANK_REASON This is a software library release with multiple model additions, not a frontier model release from a core AI lab.
Read on Transformers — Releases →
- Apertus 1.5
- Chameleon
- HY4-VL
- HYV4
- Inkling 2 Mini
- Inkling 2 Small
- Lara Mini
- Meta FAIR
- Nvidia
- Omnilingual ASR
- Qwen2-VL
- RaphaelKreft
- Swiss AI Initiative
- Thinking Machines Lab
- WavTokenizer
- EMU3.5
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →