arXiv subjects

Sean Lee

Publications and source records attributed to Sean Lee.

11 recordsLinked to original sources

arXiv2026

Latent Terms: Dense Retrievers Contain Trivially Extractable BM25-ready Zipfian Vocabularies

We propose Latent Terms, a method revealing that models trained for dense retrieval, whether single- or multi-vector, learn representations that can trivially be decomposed into retrieval-ready sparse features. When trained on frozen retrievers, Sparse Autoencoders without any retrieval-specific adjustments extract a latent vocabulary with approximately Zipfian collection statistics, directly suitable for classical sparse retrieval scoring via BM25. This approach enables sparse retrieval while requiring no learned expansion objective or sparse retrieval supervision whatsoever, and can be readily applied to any dense retriever. Latent Terms is able to match or outperform single-vector scoring methods from its own base model as well as comparable SPLADE variants. In addition, it substantially outperforms its base model on LIMIT, a task specifically designed to highlight the failures of single-vector retrieval. Overall, our results highlight that neural retrievers contain more expressive and indexable structure than their default scoring functions expose, but that other methods can nonetheless be leveraged.

cs.IR↗

arXiv2026

IncompeBench: A Permissively Licensed, Fine-Grained Benchmark for Music Information Retrieval

Multimodal Information Retrieval has made significant progress in recent years, leveraging the increasingly strong multimodal abilities of deep pre-trained models to represent information across modalities. Music Information Retrieval (MIR), in particular, has considerably increased in quality, with neural representations of music even making its way into everyday life products. However, there is a lack of high-quality benchmarks for evaluating music retrieval performance. To address this issue, we introduce \textbf{IncompeBench}, a carefully annotated benchmark comprising $1,574$ permissively licensed, high-quality music snippets, $500$ diverse queries, and over $125,000$ individual relevance judgements. These annotations were created through the use of a multi-stage pipeline, resulting in high agreement between human annotators and the generated data. The resulting datasets are publicly available at https://huggingface.co/datasets/mixedbread-ai/incompebench-strict and https://huggingface.co/datasets/mixedbread-ai/incompebench-lenient with the prompts available at https://github.com/mixedbread-ai/incompebench-programs.

cs.IR↗

arXiv2025

Fantastic (small) Retrievers and How to Train Them: mxbai-edge-colbert-v0 Tech Report

In this work, we introduce mxbai-edge-colbert-v0 models, at two different parameter counts: 17M and 32M. As part of our research, we conduct numerous experiments to improve retrieval and late-interaction models, which we intend to distill into smaller models as proof-of-concepts. Our ultimate aim is to support retrieval at all scales, from large-scale retrieval which lives in the cloud to models that can run locally, on any device. mxbai-edge-colbert-v0 is a model that we hope will serve as a solid foundation backbone for all future experiments, representing the first version of a long series of small proof-of-concepts. As part of the development of mxbai-edge-colbert-v0, we conducted multiple ablation studies, of which we report the results. In terms of downstream performance, mxbai-edge-colbert-v0 is a particularly capable small model, outperforming ColBERTv2 on common short-text benchmarks (BEIR) and representing a large step forward in long-context tasks, with unprecedented efficiency.

cs.IR↗

arXiv2025

Simple Projection Variants Improve ColBERT Performance

Multi-vector dense retrieval methods like ColBERT systematically use a single-layer linear projection to reduce the dimensionality of individual vectors. In this study, we explore the implications of the MaxSim operator on the gradient flows of the training of multi-vector models and show that such a simple linear projection has inherent, if non-critical, limitations in this setting. We then discuss the theoretical improvements that could result from replacing this single-layer projection with well-studied alternative feedforward linear networks (FFN), such as deeper, non-linear FFN blocks, GLU blocks, and skip-connections, could alleviate these limitations. Through the design and systematic evaluation of alternate projection blocks, we show that better-designed final projections positively impact the downstream performance of ColBERT models. We highlight that many projection variants outperform the original linear projections, with the best-performing variants increasing average performance on a range of retrieval benchmarks across domains by over 2 NDCG@10 points. We then conduct further exploration on the individual parameters of these projections block in order to understand what drives this empirical performance, highlighting the particular importance of upscaled intermediate projections and residual connections. As part of these ablation studies, we show that numerous suboptimal projection variants still outperform the traditional single-layer projection across multiple benchmarks, confirming our hypothesis. Finally, we observe that this effect is consistent across random seeds, further confirming that replacing the linear layer of ColBERT models is a robust, drop-in upgrade.

cs.IR↗

arXiv2025

Refining Time Series Anomaly Detectors using Large Language Models

Time series anomaly detection (TSAD) is of widespread interest across many industries, including finance, healthcare, and manufacturing. Despite the development of numerous automatic methods for detecting anomalies, human oversight remains necessary to review and act upon detected anomalies, as well as verify their accuracy. We study the use of multimodal large language models (LLMs) to partially automate this process. We find that LLMs can effectively identify false alarms by integrating visual inspection of time series plots with text descriptions of the data-generating process. By leveraging the capabilities of LLMs, we aim to reduce the reliance on human effort required to maintain a TSAD system

cs.CL↗

arXiv2024

Scaling Deep Learning Training with MPMD Pipeline Parallelism

We present JaxPP, a system for efficiently scaling the training of large deep learning models with flexible pipeline parallelism. We introduce a seamless programming model that allows implementing user-defined pipeline schedules for gradient accumulation. JaxPP automatically distributes tasks, corresponding to pipeline stages, over a cluster of nodes and automatically infers the communication among them. We implement a MPMD runtime for asynchronous execution of SPMD tasks. The pipeline parallelism implementation of JaxPP improves hardware utilization by up to $1.11\times$ with respect to the best performing SPMD configuration.

cs.DC↗

arXiv2023

A decomposition theorem for balanced measures

Let $G = (V,E)$ be a connected graph. A probability measure $\mu$ on $V$ is called "balanced" if it has the following property: if $T_\mu(v)$ denotes the "earth mover's" cost of transporting all the mass of $\mu$ from all over the graph to the vertex $v$, then $T_\mu$ attains its global maximum at each point in the support of $\mu$. We prove a decomposition result that characterizes balanced measures as convex combinations of suitable "extremal" balanced measures that we call "basic." An upper bound on the number of basic balanced measures on $G$ follows, and an example shows that this estimate is essentially sharp.

math.CO↗

arXiv2022

A characterization of graphs with at most four boundary vertices

Steinerberger defined a notion of boundary for a graph and established a corresponding isoperimetric inquality. Hence, "large" graphs have more boundary vertices. In this paper, we first characterize graphs with three boundary vertices in terms of two infinite families of graphs. We then completely characterize graphs with four boundary vertices in terms of eight families of graphs, five of which are infinite. This parallels earlier work by Hasegawa and Saito as well as M\"uller, P\'or, and Sereni on another notion of boundary defined by Chartrand, Erwin, Johns, and Zhang.

math.CO↗

arXiv2016

Theano: A Python framework for fast computation of mathematical expressions

Theano is a Python library that allows to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently. Since its introduction, it has been one of the most used CPU and GPU mathematical compilers - especially in the machine learning community - and has shown steady performance improvements. Theano is being actively and continuously developed since 2008, multiple frameworks have been built on top of it and it has been used to produce many state-of-the-art machine learning models. The present article is structured as follows. Section I provides an overview of the Theano software and its community. Section II presents the principal features of Theano and how to use them, and compares them with other similar projects. Section III focuses on recently-introduced functionalities and improvements. Section IV compares the performance of Theano against Torch7 and TensorFlow on several machine learning models. Section V discusses current limitations of Theano and potential ways of improving it.

cs.SC↗

arXiv2007

Towards a Physical Theory of Subjective Mental States

Any complete theory of physical reality must allow for the ubiquitous phenomenon of subjective experience at some level, or risk being conceptually incoherent. However, as long as the ontological status of subjectivity itself remains unresolved, the topic will be seen as more within the purview of philosophy than of physics. Towards a resolution of this issue within empirically motivated physical theory, this article introduces an operational definition that ultilizes the general consensus that subjective mental states, whatever else is controversial about them, at least correlate in some way to physical states. It is shown here that implementing this underappreciated assumption within the framework of a physical theory in fact leads to wide-ranging consequences. In particular, a correlation requires there exist a well-defined mapping from a space of subjective mental states onto a space of information-bearing elements of some physical theory. Given the peculiar nature of subjective states as inherently private appearances, any empirical identification of states must be performed by the experiencing subject. It is argued that such an operationally defined 'self-measuring' act leads unavoidably to an 'uncertainty principle' that is analogous in some intriguing ways to Heisenberg's principle for quantum mechanics. A model is then introduced for subjective states as algorithmically incomputable numbers. Additionaally, an inequality similar to Bell's theorem may be derived, indicating an analogy with the violations of local reality and the ontology of observables within quantum mechanics.

physics.gen-ph↗

arXiv1992

The Mechanism of Complex Langevin Simulations

We discuss conditions under which expectation values computed from a complex Langevin process $Z$ will converge to integral averages over a given complex valued weight function. The difficulties in proving a general result are pointed out. For complex valued polynomial actions, it is shown that for a process converging to a strongly stationary process one gets the correct answer for averages of polynomials if $c_τ(k) \equiv E(e^{ikZ(τ)}) $ satisfies certain conditions. If these conditions are not satisfied, then the stochastic process is not necessarily described by a complex Fokker Planck equation. The result is illustrated with the exactly solvable complex frequency harmonic oscillator.

hep-lat↗