Author name cluster

Diego Doimo

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

8 papers

2 author rows

ICLR Conference 2025 Conference Paper

Emergence of a High-Dimensional Abstraction Phase in Language Transformers

Emily Cheng
Diego Doimo
Corentin Kervadec
Iuri Macocco
Lei Yu
Alessandro Laio
Marco Baroni

A language model (LM) is a mapping from a linguistic context to an output token. However, much remains to be known about this mapping, including how its geometric properties relate to its function. We take a high-level geometric approach to its analysis, observing, across five pre-trained transformer-based LMs and three input datasets, a distinct phase characterized by high intrinsic dimensionality. During this phase, representations (1) correspond to the first full linguistic abstraction of the input; (2) are the first to viably transfer to downstream tasks; (3) predict each other across different LMs. Moreover, we find that an earlier onset of the phase strongly predicts better language modelling performance. In short, our results suggest that a central high-dimensionality phase underlies core linguistic processing in many common LM architectures.

Details

NeurIPS Conference 2025 Conference Paper

Head Pursuit: Probing Attention Specialization in Multimodal Transformers

Lorenzo Basile
Valentino Maiorca
Diego Doimo
Francesco Locatello
Alberto Cazzaniga

Language and vision-language models have shown impressive performance across a wide range of tasks, but their internal mechanisms remain only partly understood. In this work, we study how individual attention heads in text-generative models specialize in specific semantic or visual attributes. Building on an established interpretability method, we reinterpret the practice of probing intermediate activations with the final decoding layer through the lens of signal processing. This lets us analyze multiple samples in a principled way and rank attention heads based on their relevance to target concepts. Our results show consistent patterns of specialization at the head level across both unimodal and multimodal transformers. Remarkably, we find that editing as few as 1% of the heads, selected using our method, can reliably suppress or enhance targeted concepts in the model output. We validate our approach on language tasks such as question answering and toxicity mitigation, as well as vision-language tasks including image classification and captioning. Our findings highlight an interpretable and controllable structure within attention layers, offering simple tools for understanding and editing large-scale generative models.

PDF Details

NeurIPS Conference 2025 Conference Paper

The Narrow Gate: Localized Image-Text Communication in Native Multimodal Models

Alessandro Serra
Francesco Ortu
Emanuele Panizon
Lucrezia Valeriani
Lorenzo Basile
Alessio Ansuini
Diego Doimo
Alberto Cazzaniga

Recent advances in multimodal training have significantly improved the integration of image understanding and generation within a unified model. This study investigates how vision-language models (VLMs) handle image-understanding tasks, focusing on how visual information is processed and transferred to the textual domain. We compare native multimodal VLMs, models trained from scratch on multimodal data to generate both text and images, and non-native multimodal VLMs, models adapted from pre-trained large language models or capable of generating only text, highlighting key differences in information flow. We find that in native multimodal VLMs, image and text embeddings are more separated within the residual stream. Moreover, VLMs differ in how visual information reaches text: non-native multimodal VLMs exhibit a distributed communication pattern, where information is exchanged through multiple image tokens, whereas models trained natively for joint image and text generation tend to rely on a single post-image token that acts as a narrow gate for visual information. We show that ablating this single token significantly deteriorates image-understanding performance, whereas targeted, token-level interventions reliably steer image semantics and downstream text with fine-grained control.

PDF Details

TMLR Journal 2024 Journal Article

How to choose the right transfer learning protocol? A qualitative analysis in a controlled set-up

Federica Gerace
Diego Doimo
Stefano Sarao Mannelli
Luca Saglietti
Alessandro Laio

Transfer learning is a powerful technique that enables model training with limited amounts of data, making it crucial in many data-scarce real-world applications. Typically, transfer learning protocols require first to transfer all the feature-extractor layers of a network pre-trained on a data-rich source task, and then to adapt only the task-specific readout layers to a data-poor target task. This workflow is based on two main assumptions: first, the feature maps of the pre-trained model are qualitatively similar to the ones that would have been learned with enough data on the target task; second, the source representations of the last hidden layers are always the most expressive. In this work, we demonstrate that this is not always the case and that the largest performance gain may be achieved when smaller portions of the pre-trained network are transferred. In particular, we perform a set of numerical experiments in a controlled setting, showing how the optimal transfer depth depends non-trivially on the amount of available training data and on the degree of source-target task similarity, and it is often convenient to transfer only the first layers. We then propose a strategy to detect the most promising source task among the available candidates. This approach compares the internal representations of a network trained entirely from scratch on the target task with those of the networks pre-trained on the potential source tasks.

PDF Details

NeurIPS Conference 2024 Conference Paper

The Representation Landscape of Few-Shot Learning and Fine-Tuning in Large Language Models

Diego Doimo
Alessandro Serra
Alessio Ansuini
Alberto Cazzaniga

In-context learning (ICL) and supervised fine-tuning (SFT) are two common strategies for improving the performance of modern large language models (LLMs) on specific tasks. Despite their different natures, these strategies often lead to comparable performance gains. However, little is known about whether they induce similar representations inside LLMs. We approach this problem by analyzing the probability landscape of their hidden representations in the two cases. More specifically, we compare how LLMs solve the same question-answering task, finding that ICL and SFT create very different internal structures, in both cases undergoing a sharp transition in the middle of the network. In the first half of the network, ICL shapes interpretable representations hierarchically organized according to their semantic content. In contrast, the probability landscape obtained with SFT is fuzzier and semantically mixed. In the second half of the model, the fine-tuned representations develop probability modes that better encode the identity of answers, while less-defined peaks characterize the landscape of ICL representations. Our approach reveals the diverse computational strategies developed inside LLMs to solve the same task across different conditions, allowing us to make a step towards designing optimal methods to extract information from language models.

PDF Details DOI

NeurIPS Conference 2023 Conference Paper

The geometry of hidden representations of large transformer models

Lucrezia Valeriani
Diego Doimo
Francesca Cuturello
Alessandro Laio
Alessio Ansuini
Alberto Cazzaniga

Large transformers are powerful architectures used for self-supervised data analysis across various data types, including protein sequences, images, and text. In these models, the semantic structure of the dataset emerges from a sequence of transformations between one representation and the next. We characterize the geometric and statistical properties of these representations and how they change as we move through the layers. By analyzing the intrinsic dimension (ID) and neighbor composition, we find that the representations evolve similarly in transformers trained on protein language taskand image reconstruction tasks. In the first layers, the data manifold expands, becoming high-dimensional, and then contracts significantly in the intermediate layers. In the last part of the model, the ID remains approximately constant or forms a second shallow peak. We show that the semantic information of the dataset is better expressed at the end of the first peak, and this phenomenon can be observed across many models trained on diverse datasets. Based on our findings, we point out an explicit strategy to identify, without supervision, the layers that maximize semantic content: representations at intermediate layers corresponding to a relative minimum of the ID profile are more suitable for downstream learning tasks.

PDF Details

NeurIPS Conference 2022 Conference Paper

Redundant representations help generalization in wide neural networks

Diego Doimo
Aldo Glielmo
Sebastian Goldt
Alessandro Laio

Deep neural networks (DNNs) defy the classical bias-variance trade-off: adding parameters to a DNN that interpolates its training data will typically improve its generalization performance. Explaining the mechanism behind this ``benign overfitting'' in deep networks remains an outstanding challenge. Here, we study the last hidden layer representations of various state-of-the-art convolutional neural networks and find that if the last hidden representation is wide enough, its neurons tend to split into groups that carry identical information and differ from each other only by statistically independent noise. The number of such groups increases linearly with the width of the layer, but only if the width is above a critical value. We show that redundant neurons appear only when the training is regularized and the training error is zero.

PDF Details

NeurIPS Conference 2020 Conference Paper

Hierarchical nucleation in deep neural networks

Diego Doimo
Aldo Glielmo
Alessio Ansuini
Alessandro Laio

Deep convolutional networks (DCNs) learn meaningful representations where data that share the same abstract characteristics are positioned closer and closer. Understanding these representations and how they are generated is of unquestioned practical and theoretical interest. In this work we study the evolution of the probability density of the ImageNet dataset across the hidden layers in some state-of-the-art DCNs. We find that the initial layers generate a unimodal probability density getting rid of any structure irrelevant for classification. In subsequent layers density peaks arise in a hierarchical fashion that mirrors the semantic hierarchy of the concepts. Density peaks corresponding to single categories appear only close to the output and via a very sharp transition which resembles the nucleation process of a heterogeneous liquid. This process leaves a footprint in the probability density of the output layer where the topography of the peaks allows reconstructing the semantic relationships of the categories.

PDF Details