相关论文: Same Geometry, Opposite Noise: Transformer Magnitu…

Large Language Models (LLMs) are one of the most promising technologies for the next era of speech generation systems, due to their scalability and in-context learning capabilities. Nevertheless, they suffer from multiple stability issues…

Learning models have been shown to rely on spurious correlations between non-predictive features and the associated labels in the training data, with negative implications on robustness, bias and fairness. In this work, we provide a…

The behavior of the leading singular values and vectors of noisy low-rank matrices is fundamental to many statistical and scientific problems. Theoretical understanding currently derives from asymptotic analysis under one of two regimes:…

Contrastive representation learning has been outstandingly successful in practice. In this work, we identify two key properties related to the contrastive loss: (1) alignment (closeness) of features from positive pairs, and (2) uniformity…

Just like weights, bias terms are the learnable parameters of many popular machine learning models, including neural networks. Biases are thought to enhance the representational power of neural networks, enabling them to solve a variety of…

In spite of their huge success, transformer models remain difficult to scale in depth. In this work, we develop a unified signal propagation theory and provide formulae that govern the moments of the forward and backward signal through the…

Enterprise LLM applications require consistently high quality and reliable performance across diverse scenarios, demanding robustness to minor variations. Existing research shows that even small prompt changes can lead to substantial…

While state-of-the-art neural network models continue to achieve lower perplexity scores on language modeling benchmarks, it remains unknown whether optimizing for broad-coverage predictive performance leads to human-like syntactic…

Covariance matrix of heights measured relative to the average height of a growing self-affine surface in the steady state are investigated in the framework of random matrix theory. We show that the spectral density of the covariance matrix…

Rich sources of variability in natural speech present significant challenges to current data intensive speech recognition technologies. To model both speaker and environment level diversity, this paper proposes a novel Bayesian factorised…

We document an empirical phenomenon in chain-of-thought and ReAct agents driven by ten large language models from seven architecture families: meaning-bearing perturbations (e.g., paraphrase, synonym) alter final answers more often than…

Variational Autoencoders (VAEs) are powerful generative models for learning latent representations. Standard VAEs generate dispersed and unstructured latent spaces by utilizing all dimensions, which limits their interpretability, especially…

Large language models (LLMs) often exhibit unexpected errors or unintended behavior, even at scale. While recent work reveals the discrepancy between LLMs and humans in skill compositions, the learning dynamics of skill compositions and the…

Recent works have argued that high-level semantic concepts are encoded "linearly" in the representation space of large language models. In this work, we study the origins of such linear representations. To that end, we introduce a simple…

Understanding the similarity of the numerous released large language models (LLMs) has many uses, e.g., simplifying model selection, detecting illegal model reuse, and advancing our understanding of what makes LLMs perform well. In this…

Audio Large Language Models (Audio LLMs) have demonstrated strong capabilities in integrating speech perception with language understanding. However, whether their internal representations align with human neural dynamics during…

In human-level NLP tasks, such as predicting mental health, personality, or demographics, the number of observations is often smaller than the standard 768+ hidden state sizes of each layer within modern transformer-based language models,…

We present high signal-to-noise measurements of the acoustic scale in the presence of nonlinear growth and redshift distortions using 320(Gpc/h)^3 of cosmological PM simulations. Using simple fitting methods, we obtain robust measurements…

End-to-end Large Speech Language Models (LSLMs) have demonstrated impressive conversational generation abilities, yet consistently fall short of traditional pipeline systems on semantic understanding benchmarks. In this work, we reveal…

Accurate classification of articulatory-phonological features plays a vital role in understanding human speech production and developing robust speech technologies, particularly in clinical contexts where targeted phonemic analysis and…