Emanuele Rodolà

Navigating the Latent Space Dynamics of Neural Models

Video Unlearning via Low-Rank Refusal Vector

Language Models are Injective and Hence Invertible

Implicit Inversion turns CLIP into a Decoder

TOAST : Transformer Optimization using Adaptive and Simple Transformations

MASS: MoErging through Adaptive Subspace Selection

Implicit-ARAP: Efficient Handle-Guided Neural Field Deformation via Local Patch Meshing

Update Your Transformer to the Latest Release: Re-Basin of Task Vectors

MERGE$^3$: Efficient Evolutionary Merging on Consumer-grade GPUs

Detecting and Approximating Redundant Computational Blocks in Neural Networks

ATM: Improving Model Merging by Alternating Tuning and Merging