Publications

Transformer components such as non-linear activations and normalization are inherently non-injective, suggesting that different inputs …

Uncertainty Quantification (UQ) in Language Models (LMs) is key to improving their safety and reliability. Evaluations often use …

Model merging allows combining the capabilities of existing models into a new one—post hoc, without additional training. This has made …

Evolutionary model merging enables the creation of high-performing multi-task models but remains computationally prohibitive for …

Recent works have shown that, when trained at scale, uni-modal 2D vision and text encoders converge to learned features that share …

Large Language Models (LLMs) represent a significant advancement in artificial intelligence, finding applications across various …

In recent years Large Language Models have improved the state of the art on several natural language processing tasks. However, their …

Autoregressive decoding limits the efficiency of transformers for Machine Translation (MT). The community proposed specific network …

The rise in loosely-structured data available through text, images, and other modalities has called for new ways of querying them. …

Autoregressive models have achieved impressive results over a wide range of domains in terms of generation quality and downstream task …

Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language …

PromptSource is a system for creating, sharing, and using natural language prompts. Prompts are functions that map an example from a …

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their …

Large language models have recently been shown to attain reasonable zero-shot generalization on a diverse set of tasks (Brown et al., …

Syntactic parsers have dominated natural language understanding for decades. Yet, their syntactic interpretations are losing centrality …

State of the art audio source separation models rely on supervised data-driven approaches, which can be expensive in terms of labeling …

We introduce Explanatory Learning (EL), a framework to let machines use existing knowledge buried in symbolic sequences – e.g. …

This paper describes the UNITOR system that participated to the Irony Detection in Italian Tweets task (IronITA) within the context of …

In this paper, we present SyntNN as a way to include traditional syntactic models in multilayer neural networks used in the task of …