49 citations · 82 across the 22 of their papers we have counts for
22 papers
Residual Connections and the Causal Shift: Uncovering a Structural Misalignment in Transformers
Jonathan Lys, Vincent Gripon, Bastien Pasdeloup +4
Large Language Models (LLMs) are trained with next-token prediction, implemented in autoregressive Transformers via causal masking for parallelism. This creates a subtle misalignme…
Inner Loop Inference for Pretrained Transformers: Unlocking Latent Capabilities Without Training
Jonathan Lys, Vincent Gripon, Bastien Pasdeloup +4
Deep Learning architectures, and in particular Transformers, are conventionally viewed as a composition of layers. These layers are actually often obtained as the sum of two contri…
D5P4: Partition Determinantal Point Process for Diversity in Parallel Discrete Diffusion Decoding
Jonathan Lys, Vincent Gripon, Axel Marmoret +4
Discrete diffusion models are promising alternatives to autoregressive approaches for text generation, yet their decoding methods remain under-studied. Standard autoregressive sear…
GaLLoP: Gradient-based Sparse Learning on Low-Magnitude Parameters
Anand Choudhary, Yasser Sulaıman, Lukas Mauch +3
Sparse fine-tuning techniques adapt LLMs to downstream tasks by only tuning a sparse subset of model parameters. However, the effectiveness of sparse adaptation depends on optimall…
SAFT: Towards Out-of-Distribution Generalization in Fine-Tuning
Bac Nguyen, Stefan Uhlich, Fabien Cardinaux +3
Handling distribution shifts from training data, known as out-of-distribution (OOD) generalization, poses a significant challenge in the field of machine learning. While a pre-trai…
LLM meets Vision-Language Models for Zero-Shot One-Class Classification
Yassir Bendou, Giulia Lioi, Bastien Pasdeloup +4
We consider the problem of zero-shot one-class visual classification, extending traditional one-class classification to scenarios where only the label of the target class is availa…
A Novel Benchmark for Few-Shot Semantic Segmentation in the Era of Foundation Models
Reda Bensaid, Vincent Gripon, François Leduc-Primeau +3
Few-shot semantic segmentation (FSS) is a crucial challenge in computer vision, driving extensive research into a diverse range of methods, from advanced meta-learning techniques t…
Inferring Latent Class Statistics from Text for Robust Visual Few-Shot Learning
Yassir Bendou, Vincent Gripon, Bastien Pasdeloup +4
In the realm of few-shot learning, foundation models like CLIP have proven effective but exhibit limitations in cross-domain robustness especially in few-shot settings. Recent work…
DBsurf: A Discrepancy Based Method for Discrete Stochastic Gradient Estimation
Pau Mulet Arabi, Alec Flowers, Lukas Mauch +1
Computing gradients of an expectation with respect to the distributional parameters of a discrete distribution is a problem arising in many fields of science and engineering. Typic…
Towards Robust FastSpeech 2 by Modelling Residual Multimodality
Fabian Kögel, Bac Nguyen, Fabien Cardinaux
State-of-the-art non-autoregressive text-to-speech (TTS) models based on FastSpeech 2 can efficiently synthesise high-fidelity and natural speech. For expressive speech datasets ho…
Improving Self-Supervised Learning for Audio Representations by Feature Diversity and Decorrelation
Bac Nguyen, Stefan Uhlich, Fabien Cardinaux
Self-supervised learning (SSL) has recently shown remarkable results in closing the gap between supervised and unsupervised learning. The idea is to learn robust features that are…
A Statistical Model for Predicting Generalization in Few-Shot Classification
Yassir Bendou, Vincent Gripon, Bastien Pasdeloup +5
The estimation of the generalization error of classifiers often relies on a validation set. Such a set is hardly available in few-shot learning scenarios, a highly disregarded shor…
A Low Memory Footprint Quantized Neural Network for Depth Completion of Very Sparse Time-of-Flight Depth Maps
Xiaowen Jiang, Valerio Cambareri, Gianluca Agresti +4
Sparse active illumination enables precise time-of-flight depth sensing as it maximizes signal-to-noise ratio for low power budgets. However, depth completion is required to produc…
AutoTTS: End-to-End Text-to-Speech Synthesis through Differentiable Duration Modeling
Bac Nguyen, Fabien Cardinaux, Stefan Uhlich
Parallel text-to-speech (TTS) models have recently enabled fast and highly-natural speech synthesis. However, they typically require external alignment models, which are not necess…
NVC-Net: End-to-End Adversarial Voice Conversion
Bac Nguyen, Fabien Cardinaux
Voice conversion has gained increasing popularity in many applications of speech synthesis. The idea is to change the voice identity from one speaker into another while keeping the…
DNN Quantization with Attention
Ghouthi Boukli Hacene, Lukas Mauch, Stefan Uhlich +1
Low-bit quantization of network weights and activations can drastically reduce the memory footprint, complexity, energy consumption and latency of Deep Neural Networks (DNNs). Howe…
Efficient Sampling for Predictor-Based Neural Architecture Search
Lukas Mauch, Stephen Tiedemann, Javier Alonso Garcia +4
Recently, predictor-based algorithms emerged as a promising approach for neural architecture search (NAS). For NAS, we typically have to calculate the validation accuracy of a larg…
Unsupervised Cross-Domain Speech-to-Speech Conversion with Time-Frequency Consistency
Mohammad Asif Khan, Fabien Cardinaux, Stefan Uhlich +2
In recent years generative adversarial network (GAN) based models have been successfully applied for unsupervised speech-to-speech conversion.The rich compact harmonic view of the…
Iteratively Training Look-Up Tables for Network Quantization
Fabien Cardinaux, Stefan Uhlich, Kazuki Yoshiyama +5
Operating deep neural networks (DNNs) on devices with limited resources requires the reduction of their memory as well as computational footprint. Popular reduction methods are net…
Mixed Precision DNNs: All you need is a good parametrization
Stefan Uhlich, Lukas Mauch, Fabien Cardinaux +5
Efficient deep neural network (DNN) inference on mobile or embedded devices typically involves quantization of the network parameters and activations. In particular, mixed precisio…
Iteratively Training Look-Up Tables for Network Quantization
Fabien Cardinaux, Stefan Uhlich, Kazuki Yoshiyama +4
Operating deep neural networks on devices with limited resources requires the reduction of their memory footprints and computational requirements. In this paper we introduce a trai…
Improving DNN-based Music Source Separation using Phase Features
Joachim Muth, Stefan Uhlich, Nathanael Perraudin +3
Music source separation with deep neural networks typically relies only on amplitude features. In this paper we show that additional phase features can improve the separation perfo…