🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

↗️Task vectors? More like gradients🔽! We show that, under certain assumptions, they’re actually deeply related.🔗 ATM: A game-changing framework for multi-task model merging with no hidden fees💵!

Tomorrow Maria Sofia Bucarelli will give a presentation on “Large Language Models and Agents: A Challenging Symbiosis”, showing our research agenda on the hottest topic of the moment in #AI and #LLMs. See you at AIxIA | AI-Conference , MLDM24 workshop, from 2.30 to 2.50 PM CET, room F0.03!

Maria Sofia Bucarelli today has presented at AIxIA | AI-Conference “Large Language Models and Agents: A Challenging Symbiosis”: this is a presentation showing our #research agenda on the hottest topic of the moment in #AI and #LLM's. Reach us out to continue the #chat.

*Task Singular Vectors: Reducing Task Interference in Model Merging* with Emanuele Rodolà Fabrizio Silvestri Maria Sofia Bucarelli Donato Crisostomi @ ICML We show that task vectors are inherently low-rank, and we propose a merging method that significantly improves SOTA. arxiv.org/abs/2412.00081

🚨 Call for Submissions 🚨 The IJCNN 2025 Special Session on Neural Methods for IR and RecSys is open for papers!📄 🔑Topics: Neural architectures, RAG, scalability & more. 📅Deadline: Jan 15, 2025 🔗Website: coda.io/@rstless-group… #IJCNN2025 #AI #RecSys #InformationRetrieval

🎉Check out our new work on Timbre Transfer, which has been accepted at #ICASSP2025! 🎶 Latent Diffusion Bridges for Unsupervised Musical Audio Timbre Transfer 🔥 arXiv: arxiv.org/abs/2409.06096 Demo: sony.github.io/diffusion-timb… Code: github.com/sony/diffusion… 🙌Thanks to our

🎉 Hyped that 3 of our papers are accepted at #ICASSP2025! Huge congrats to my co-authors, and thanks for their efforts! 👏 - COCOLA: Coherence-Oriented Contrastive Learning of Musical Audio Representations arxiv.org/pdf/2404.16969 - Latent Diffusion Bridges for Unsupervised

📢Prepend “Singular” to “Task Vectors” and get +15% average accuracy for free! 1. Perform a low-rank approximation of layer-wise task vectors. 2. Minimize task interference by orthogonalizing inter-task singular vectors. 🧵(1/6)

*Task Singular Vectors: Reducing Task Interference in Model Merging* by Gargiulo Donato Crisostomi @ CVPR Emanuele Rodolà et al. Happy to share our work was accepted at #CVPR2025! We provide a SOTA model merging method based on the SVD of task vectors. arxiv.org/abs/2412.00081

Will present this at #CVPR2025 ✈️ See you in Nashville🇺🇸! Kudos to the team 👏 Antonio A. Gargiulo, Maria Sofia Bucarelli, Simone Scardapane, Fabrizio Silvestri, Emanuele Rodolà

What if I told you that we could align N different models at once, even spanning different modalities, and get a universal space in the process? Curious? Apply to LOGML Summer School and be the one to do it! ✍️Application form: logml.ai 📅Apply by 6th April 2025

Excited to share our accepted competition paper at #IJCNN2025! "Learning with Noisy Graph Labels" We tackle the challenge of graph classification under label noise: Details: %20https%3A//sites.google.com/view/learning-with-noisy-graph-labe/home

🎉 Paper accepted at #IJCNN2025! "Renormalized Graph Representations for Node Classification" by francesco caso, Giovanni Trappolini, Andrea Bacciu, Pietro Lio', Fabrizio Silvestri proposes a graph representation based on renormalization group and explores its use in graph learning.

I might've gotten a little too excited designing our #CVPR poster on model merging… Come find it in Nashville 🎸 I promise it won’t be hard to spot! Paper 👉 arxiv.org/abs/2412.00081 Joint w/ Antonio A. Gargiulo, Maria Sofia Bucarelli, Simone Scardapane, Fabrizio Silvestri, Emanuele Rodolà