Recent studies reveal that Large Language Models (LLMs) exhibit strong sequential reasoning capabilities, allowing them to replace specialized time-series models and serve as foundation models for complex time-series analysis. To activate the capabilities of LLMs for time-series tasks, numerous studies have attempted to bridge the gap between time series and linguistics by aligning textual representations with time-series patterns. However, it is a non-trivial endeavor to losslessly capture the infinite time-domain variability using natural language, leading to suboptimal alignment performance. Beyond representation, contextual differences, where semantics in time series are conveyed by consecutive points, unlike in text by individual tokens, are often overlooked by existing methods. To address these, we propose S$^2$TS-LLM, a simple yet effective framework to repurpose LLMs for universal time series analysis through the following two main paradigms: (i) a spectral symbolization paradigm transforms time series into frequency-domain representations characterized by a fixed number of components and prominent amplitudes, which enables a limited set of symbols to effectively abstract key frequency features; (ii) a contextual segmentation paradigm partitions the sequence into blocks based on temporal patterns and reassigns positional encodings accordingly, thereby mitigating the structural mismatch between time series and natural language.

Audio deepfake detection (ADD) is crucial to combat the misuse of speech synthesized by generative AI models.

This paper presents anewUNIfied pre-trainedLanguageModel (UNILM)that can be fine-tuned for both natural language understanding and generation tasks.

First, a student model is trained to recover the classifier'spredictions andtomatch theexplanations givenbytheexplainer.

AwidevarietyofNLPapplications, suchasmachinetranslation, summarization, and dialog, involve text generation. One major challenge for these applications is how to evaluate whether such generated texts are actually fluent, accurate, or effective. In this work, we conceptualize theevaluation of generated text as a text generation problem, modeled using pre-trained sequence-to-sequence models. The general idea is that models trained to convert the generated text to/from a reference output or the source text will achieve higher scores when the generated text is better.

Generated text may contain offensive or toxic language, contain significant repetition, orbeofadifferent sentiment than desired by the user. We consider thetaskofunlearningthese misalignments byfine-tuning thelanguage model on signals of whatnot to do.

In this work, we introduce WinoGAViL: an online game of vision-andlanguage associations (e.g., betweenwerewolves and a full moon), used as a dynamic evaluation benchmark.

In order to explore the impact of pandemics on people's lives, itiscrucial to understand the public'sconcerns and attitudes towards pandemicrelated entities (e.g., drugs, vaccines) onsocial media.

Example Reweightingdirectly assigns an importance weight to the standard CE training loss, accordingtothebiasdegreeβ: Lreweight = (1 β)y logpm (3) Confidence Regularizationis based on knowledge distillation [9]. It involves a teacher model trainedwiththestandardCEloss. Specifically, we calculate the weighted average of the F1 score of each class. The splits used for evaluation are highlightedwithredcolor. To address this problem, we select the best checkpoint after0.7 tmax of training, butstill according to the performance on the ID devset.