✨ Meet #ResiDual, a novel perspective on the alignment of multimodal latent spaces! Think of it as a spectral "panning for gold" along the residual stream. It improves text-image alignment by simply amplifying task-related directions! 🌌🔍 arxiv.org/abs/2411.00246 [1/6]

I just landed in Vancouver to present NeurIPS Conference the findings of our new work! Few-shot learning and fine-tuning change the hidden layers of LLMs in a dramatically different way, even when they perform equally well on multiple-choice question-answering tasks. 🧵1/6

🚨🚨 Excited to share our latest paper, now on arXiv.org! 🖼️ We studied how unified VLMs, trained to generate both text and images (e.g., @MetaAI's Chameleon), exchange information between modalities, comparing them to standard VLMs. Deep dive:👇

Thrilled to announce that our paper is accepted at #NeurIPS 2025!! See you in San Diego! 🇺🇸

Excited to share that 2/2 papers from our Lab (LADE) were accepted to #NeurIPS2025 (one spotlight 🎉) Great work from all the students and collaborators involved! Area Science Park alexserra Lorenzo Basile francesco ortu Lucrezia Valeriani Diego alessio ansuini Francesco Locatello