Enzo Tartaglione

Efficient Resource-Constrained Training of Transformers via Subspace Optimization

Study of Training Dynamics for Memory-Constrained Fine-Tuning

Feature-aware (Hyper)graph Generation via Next-Scale Prediction

INSTANT: Compressing Gradients and Activations for Resource-Efficient Training

Structured Transformer Circuits Pruning

Beyond Low-rank Decomposition: A Shortcut Approach for Efficient On-Device Learning

Diffusing DeBias: Synthetic Bias Amplification for Model Debiasing

Say My Name: a Model's Bias Discovery Framework

NEPENTHE: Entropy-Based Pruning as a Neural Network Depth's Reducer