Publications

Beyond Gaussian Noise: A Generalized Approach to Likelihood Analysis with Non-Gaussian Noise
Ronan Legin
Alexandre Adam
A Functional Data Perspective and Baseline On Multi-Layer Out-of-Distribution Detection
Eduardo Dadalto Câmara Gomes
Pierre Colombo
Guillaume Staerman
Nathan Noiry
The Stack: 3 TB of permissively licensed source code
Denis Kocetkov
Raymond Li
Loubna Ben allal
Jia LI
Chenghao Mou
Carlos Muñoz Ferrandis
Yacine Jernite
Margaret Mitchell
Sean Hughes
Thomas Wolf
Leandro Von Werra
Harm de Vries
Large Language Models (LLMs) play an ever-increasing role in the field of Artificial Intelligence (AI)--not only for natural language proces… (voir plus)sing but also for code understanding and generation. To stimulate open and responsible research on LLMs for code, we introduce The Stack, a 3.1 TB dataset consisting of permissively licensed source code in 30 programming languages. We describe how we collect the full dataset, construct a permissively licensed subset, present a data governance plan, discuss limitations, and show promising results on text2code benchmarks by training 350M-parameter decoders on different Python subsets. We find that (1) near-deduplicating the data significantly boosts performance across all experiments, and (2) it is possible to match previously reported HumanEval and MBPP performance using only permissively licensed data. We make the dataset available at https://hf.co/BigCode, provide a tool called"Am I in The Stack"(https://hf.co/spaces/bigcode/in-the-stack) for developers to search The Stack for copies of their code, and provide a process for code to be removed from the dataset by following the instructions at https://www.bigcode-project.org/docs/about/the-stack/.
AHA!: Facilitating AI Impact Assessment by Generating Examples of Harms
Zana Buçinca
Chau Minh Pham
Maurice Jakesch
Marco Túlio Ribeiro
Saleema Amershi
While demands for change and accountability for harmful AI consequences mount, foreseeing the downstream effects of deploying AI systems rem… (voir plus)ains a challenging task. We developed AHA! (Anticipating Harms of AI), a generative framework to assist AI practitioners and decision-makers in anticipating potential harms and unintended consequences of AI systems prior to development or deployment. Given an AI deployment scenario, AHA! generates descriptions of possible harms for different stakeholders. To do so, AHA! systematically considers the interplay between common problematic AI behaviors as well as their potential impacts on different stakeholders, and narrates these conditions through vignettes. These vignettes are then filled in with descriptions of possible harms by prompting crowd workers and large language models. By examining 4113 harms surfaced by AHA! for five different AI deployment scenarios, we found that AHA! generates meaningful examples of harms, with different problematic AI behaviors resulting in different types of harms. Prompting both crowds and a large language model with the vignettes resulted in more diverse examples of harms than those generated by either the crowd or the model alone. To gauge AHA!'s potential practical utility, we also conducted semi-structured interviews with responsible AI professionals (N=9). Participants found AHA!'s systematic approach to surfacing harms important for ethical reflection and discovered meaningful stakeholders and harms they believed they would not have thought of otherwise. Participants, however, differed in their opinions about whether AHA! should be used upfront or as a secondary-check and noted that AHA! may shift harm anticipation from an ideation problem to a potentially demanding review problem. Drawing on our results, we discuss design implications of building tools to help practitioners envision possible harms.
High-Throughput Edge Inference for BERT Models via Neural Architecture Search and Pipeline
Hung-Yang Chang
Seyyed Hasan Mozafari
James J. Clark
Brett Meyer
There has been growing interest in improving the BERT inference throughput on resource-constrained edge devices for a satisfactory user expe… (voir plus)rience. One methodology is to employ heterogeneous computing, which utilizes multiple processing elements to accelerate inference. Another methodology is to deploy Neural Architecture Search (NAS) to find optimal solutions in accuracy-throughput design space. In this paper, for the first time, we incorporate NAS with pipelining for BERT models. We show that performing NAS with pipelining achieves on average 53% higher throughput, compared to NAS with a homogeneous system. Additionally, we propose a NAS algorithm that incorporates hardware performance feedback to accelerate the NAS process. Our proposed NAS algorithm speeds up the search process by ~4x, and 5.5x on the design space of the BERT and CNNs, respectively. Also, by exploring the accuracy-throughput design space of BERT models, we demonstrate that performing pipelining then NAS (Pipeline-then-NAS) can lead to solutions with up to 9x higher inference throughput, compared to running homogeneous inference on the BERT-base model, with only a 1.3% decrease in accuracy.
Integer Programming Games: A Gentle Computational Overview
Gabriele Dragotto
Andrea Lodi
Sriram Sankaranarayan
PAC-Bayesian Learning of Aggregated Binary Activated Neural Networks with Probabilities over Representations
Louis Fortier-Dubois
Gaël Letarte
Benjamin Leblanc
François Laviolette
Spatial variations in aromatic hydrocarbon emission in a dust-rich galaxy
Justin Spilker
Kedar A. Phadke
Manuel Aravena
Melanie Archipley
Matthew Bayliss
Jack E. Birkin
Matthieu Béthermin
James R. Burgoyne
Jared Cathey
Scott Chapman
Håkon Dahle
Anthony H. Gonzalez
Gayathri Gururajan
Christopher C Hayward
Ryley Hill
Taylor A. Hutchison
Keunho J. Kim
Seonwoo Kim
D. Law … (voir 19 de plus)
Ronan Legin
M. Malkan
Daniel P. Marrone
E. Murphy
Desika Narayanan
Alexander Navarre
Grace M. Olivier
J. Rich
Jane R Rigby
Cassie Reuter
J. Rhoads
Keren Sharon
J. Smith
Manuel Solimano
Nikolaus Sulzenauer
Joaquin Vieira
David Vizgan
Axel Weiß
K. Whitaker
conn2res: A toolbox for connectome-based reservoir computing
Laura E. Suárez
Agoston Mihalik
Filip Milisav
Kenji Marshall
Mingze Li
Petra E. Vértes
Bratislav Mišić
Dialogue System with Missing Observation
Djallel Bouneffouf
Mayank Agarwal
Within the domain of dialogue, the ability to orchestrate multiple independently trained dialogue agents to create a unified system is of pa… (voir plus)rticular importance. Where we define orchestration as the task of selecting a subset of skills which most appropriately answer a user input using features extracted from both the user input and the individual skills. In this work, we study the task of online dialogue orchestration where the user feedback associated with the dialogue agent may not always be observed. In order to address the missing feedback setting, we propose to combine the attentive contextual bandit approach with an unsupervised learning mechanism such as clustering. By leveraging clustering to estimate missing reward, we are able to learn from each incoming event, even those with missing rewards. Promising empirical results are obtained on proprietary conversational datasets.
Evaluation of Categorical Generative Models - Bridging the Gap Between Real and Synthetic Data
Florence Regol
Anja Kroon
The machine learning community has mainly relied on real data to benchmark algorithms as it provides compelling evidence of model applicabil… (voir plus)ity. Evaluation on synthetic datasets can be a powerful tool to provide a better understanding of a model’s strengths, weaknesses and overall capabilities. Gaining these insights can be particularly important for generative modeling as the target quantity is completely unknown. Multiple issues related to the evaluation of generative models have been reported in the literature. We argue those problems can be avoided by an evaluation based on ground truth. General criticisms of synthetic experiments are that they are too simplified and not representative of practical scenarios. As such, our experimental setting is tailored to a realistic generative task. We focus on categorical data and introduce an appropriately scalable evaluation method. Our method involves tasking a generative model to learn a distribution in a high-dimensional setting. We then successively bin the large space to obtain smaller probability spaces where meaningful statistical tests can be applied. We consider increasingly large probability spaces, which correspond to increasingly difficult modeling tasks, and compare the generative models based on the highest task difficulty they can reach before being detected as being too far from the ground truth. We validate our evaluation procedure with synthetic experiments on both synthetic generative models and current state-of-the-art categorical generative models.
Fine-Tuning Strategies for Faster Inference Using Speech Self-Supervised Models: A Comparative Study
Salah Zaiem
Robin Algayres
Titouan Parcollet
Slim Essid
Self-supervised learning (SSL) has allowed substantial progress in Automatic Speech Recognition (ASR) performance in low-resource settings. … (voir plus)In this context, it has been demonstrated that larger self-supervised feature extractors are crucial for achieving lower downstream ASR error rates. Thus, better performance might be sanctioned with longer inferences. This article explores different approaches that may be deployed during the fine-tuning to reduce the computations needed in the SSL encoder, leading to faster inferences. We adapt a number of existing techniques to common ASR settings and benchmark them, displaying performance drops and gains in inference times. Interestingly, we found that given enough downstream data, a simple downsampling of the input sequences outperforms the other methods with both low performance drops and high computational savings, reducing computations by 61.3% with an WER increase of only 0. 81. Finally, we analyze the robustness of the comparison to changes in dataset conditions, revealing sensitivity to dataset size.