arXiv subjects

Adrian Valente

Publications and source records attributed to Adrian Valente.

6 recordsLinked to original sources

arXiv2026

Shieldstral

We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7$\times$ its size on text safety benchmarks and sets a new state of the art on multimodal safety classification. Shieldstral formulates content moderation as a binary question-answering task. This simple formulation unifies diverse moderation tasks into a single yes/no problem, enabling heterogeneous safety datasets with divergent taxonomies to be consolidated under one training framework. We present the data construction recipe, covering curation and generation of approximately 54.1M samples and a fine-grained evaluation set to evaluate policy adaptability. Together, these enable a small adaptive model to match or outperform much larger models.

cs.CL↗

arXiv2026

Robostral Navigate

Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems depend on depth sensors, multi-camera rigs, or pre-built maps, limiting the hardware they support and increasing deployment cost. We introduce Robostral Navigate, an 8B vision-language model built around this scalability objective. The model consumes only a stream of monocular RGB images - the most ubiquitous sensor across robotic platforms and predicts waypoints by pointing to the next target location in the current camera view. Operating purely in image space, rather than robot-specific coordinates, makes the policy naturally robust to changes in camera intrinsics and scene scale, enabling deployment across wheeled, legged, and aerial robots without recalibration. We generate 2.4 million trajectories across 350k simulated scenes to reduce the reliance on real-world data collection and scale easily. We further introduce a prefix-caching training recipe that packs entire episodes into single training sequences, reducing training tokens by 22x and cutting training time from months to days. A tree-based attention mask prevents conditioning on previous ground-truth actions, encouraging visually grounded action prediction, and reinforcement learning is used to further improve exploration and recovery capabilities. On the Room-to-Room and Room-Across-Room in Continuous Environments (R2R-CE and RxR-CE) benchmarks, Robostral Navigate sets a new state of the art. On R2R-CE, it achieves a 77.4% success rate, surpassing the best monocular method by 10.5 points and the strongest depth- or multi-camera system by 5.3 points despite using only a single RGB camera. On RxR-CE, it reaches 75.1% success rate, outperforming all monocular baselines.

cs.RO↗

arXiv2025

Shaping manifolds in equivariant recurrent neural networks

Recordings of increasingly large neural populations have revealed that the firing of individual neurons is highly coordinated. When viewed in the space of all possible patterns, the collective activity forms non-linear structures called neural manifolds. Because such structures are observed even at rest or during sleep, an important hypothesis is that activity manifolds may correspond to continuous attractors shaped by recurrent connectivity between neurons. Classical models of recurrent networks have shown that continuous attractors can be generated by specific symmetries in the connectivity. Although a variety of attractor network models have been studied, general principles linking network connectivity and the geometry of attractors remain to be formulated. Here, we address this question by using group representation theory to formalize the relationship between the symmetries in recurrent connectivity and the resulting fixed-point manifolds. We start by revisiting the classical ring model, a continuous attractor network generating a circular manifold. Interpreting its connectivity as a circular convolution, we draw a parallel with feed-forward CNNs. Building on principles of geometric deep learning, we then generalize this architecture to a broad range of symmetries using group representation theory. Specifically, we introduce a new class of equivariant RNNs, where the connectivity is based on group convolution. Using the group Fourier transform, we reduce such networks to low-rank models, giving us a low-dimensional description that can be fully analyzed to determine the symmetry, dimensionality and stability of fixed-point manifolds. Our results underline the importance of stability considerations: for a connectivity with a given symmetry, depending on parameters, several manifolds with different symmetry subgroups can coexist, some stable and others consisting of saddle points.

q-bio.NC↗

arXiv2023

Brief technical note on linearizing recurrent neural networks (RNNs) before vs after the pointwise nonlinearity

Linearization of the dynamics of recurrent neural networks (RNNs) is often used to study their properties. The same RNN dynamics can be written in terms of the ``activations" (the net inputs to each unit, before its pointwise nonlinearity) or in terms of the ``activities" (the output of each unit, after its pointwise nonlinearity); the two corresponding linearizations are different from each other. This brief and informal technical note describes the relationship between the two linearizations, between the left and right eigenvectors of their dynamics matrices, and shows that some context-dependent effects are readily apparent under linearization of activity dynamics but not linearization of activation dynamics.

cs.LG↗

arXiv2021

Probing the relationship between linear dynamical systems and low-rank recurrent neural network models

A large body of work has suggested that neural populations exhibit low-dimensional dynamics during behavior. However, there are a variety of different approaches for modeling low-dimensional neural population activity. One approach involves latent linear dynamical system (LDS) models, in which population activity is described by a projection of low-dimensional latent variables with linear dynamics. A second approach involves low-rank recurrent neural networks (RNNs), in which population activity arises directly from a low-dimensional projection of past activity. Although these two modeling approaches have strong similarities, they arise in different contexts and tend to have different domains of application. Here we examine the precise relationship between latent LDS models and linear low-rank RNNs. When can one model class be converted to the other, and vice versa? We show that latent LDS models can only be converted to RNNs in specific limit cases, due to the non-Markovian property of latent LDS models. Conversely, we show that linear RNNs can be mapped onto LDS models, with latent dimensionality at most twice the rank of the RNN.

q-bio.NC↗

arXiv2020

Shaping dynamics with multiple populations in low-rank recurrent networks

An emerging paradigm proposes that neural computations can be understood at the level of dynamical systems that govern low-dimensional trajectories of collective neural activity. How the connectivity structure of a network determines the emergent dynamical system however remains to be clarified. Here we consider a novel class of models, Gaussian-mixture low-rank recurrent networks, in which the rank of the connectivity matrix and the number of statistically-defined populations are independent hyper-parameters. We show that the resulting collective dynamics form a dynamical system, where the rank sets the dimensionality and the population structure shapes the dynamics. In particular, the collective dynamics can be described in terms of a simplified effective circuit of interacting latent variables. While having a single, global population strongly restricts the possible dynamics, we demonstrate that if the number of populations is large enough, a rank-R network can approximate any R-dimensional dynamical system.

q-bio.NC↗