Apertus 1.5, a new series of open multimodal reasoning models, has been released, extending the Apertus family beyond text to include image and audio understanding. These models feature a significantly increased context window of 262,144 tokens and incorporate supervised fine-tuning, long-context fine-tuning, and reinforcement learning. The 8B model demonstrates leading performance among open models in its class for knowledge and instruction following, and is competitive in multimodal tasks, particularly document understanding and speech recognition. AI
IMPACT Extends open-source multimodal reasoning capabilities and sets new benchmarks for instruction following and context window length.
RANK_REASON The cluster describes a new research paper detailing the release of an open multimodal reasoning model. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
