Beyond position: how rotary embeddings shape representations and memory in autoregressive transfomers

• URL: http://arxiv.org/abs/2410.18067v2
• Date: Fri, 25 Oct 2024 20:27:05 GMT
• Title: Beyond position: how rotary embeddings shape representations and memory in autoregressive transfomers
• Authors: Valeria Ruscio, Fabrizio Silvestri,
• Abstract summary: Rotary Positional Embeddings (RoPE) enhance positional encoding in Transformer models. This paper studies how RoPE introduces position-dependent rotations, causing phase shifts in token embeddings.
• Score: 7.3645788720974465
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Rotary Positional Embeddings (RoPE) enhance positional encoding in Transformer models, yet their full impact on model dynamics remains underexplored. This paper studies how RoPE introduces position-dependent rotations, causing phase shifts in token embeddings that influence higher-frequency components within the model's internal representations. Through spectral analysis, we demonstrate that RoPE's rotation matrices induce oscillatory behaviors in embeddings, affecting information retention across layers and shaping temporal modeling capabilities. We show that activation functions in feed-forward networks interact with RoPE-modulated embeddings to generate harmonics, leading to constructive or destructive interference based on phase alignment. Our findings reveal that phase alignment amplifies activations and sharpens attention, while misalignment weakens activations and disrupts focus on positional patterns. This study underscores the importance of frequency components as intrinsic elements of model behavior, offering new insights beyond traditional analyses.

Related papers

• Neural ODE Transformers: Analyzing Internal Dynamics and Adaptive Fine-tuning [30.781578037476347]
  We introduce a novel approach to modeling transformer architectures using highly flexible non-autonomous neural ordinary differential equations (ODEs) Our proposed model parameterizes all weights of attention and feed-forward blocks through neural networks, expressing these weights as functions of a continuous layer index. Our neural ODE transformer demonstrates performance comparable to or better than vanilla transformers across various configurations and datasets.
  arXiv  Detail & Related papers  (2025-03-03T09:12:14Z)
• SPARTAN: A Sparse Transformer Learning Local Causation [63.29645501232935]
  Causal structures play a central role in world models that flexibly adapt to changes in the environment. We present the SPARse TrANsformer World model (SPARTAN), a Transformer-based world model that learns local causal structures between entities in a scene. By applying sparsity regularisation on the attention pattern between object-factored tokens, SPARTAN identifies sparse local causal models that accurately predict future object states.
  arXiv  Detail & Related papers  (2024-11-11T11:42:48Z)
• Tight Stability, Convergence, and Robustness Bounds for Predictive Coding Networks [60.3634789164648]
  Energy-based learning algorithms, such as predictive coding (PC), have garnered significant attention in the machine learning community. We rigorously analyze the stability, robustness, and convergence of PC through the lens of dynamical systems theory.
  arXiv  Detail & Related papers  (2024-10-07T02:57:26Z)
• A Unified Framework for Interpretable Transformers Using PDEs and Information Theory [3.4039202831583903]
  This paper presents a novel unified theoretical framework for understanding Transformer architectures by integrating Partial Differential Equations (PDEs), Neural Information Flow Theory, and Information Bottleneck Theory. We model Transformer information dynamics as a continuous PDE process, encompassing diffusion, self-attention, and nonlinear residual components. Our comprehensive experiments across image and text modalities demonstrate that the PDE model effectively captures key aspects of Transformer behavior, achieving high similarity (cosine similarity > 0.98) with Transformer attention distributions across all layers.
  arXiv  Detail & Related papers  (2024-08-18T16:16:57Z)
• Dynamical Mean-Field Theory of Self-Attention Neural Networks [0.0]
  Transformer-based models have demonstrated exceptional performance across diverse domains. Little is known about how they operate or what are their expected dynamics. We use methods for the study of asymmetric Hopfield networks in nonequilibrium regimes.
  arXiv  Detail & Related papers  (2024-06-11T13:29:34Z)
• Function Approximation for Reinforcement Learning Controller for Energy from Spread Waves [69.9104427437916]
  Multi-generator Wave Energy Converters (WEC) must handle multiple simultaneous waves coming from different directions called spread waves. These complex devices need controllers with multiple objectives of energy capture efficiency, reduction of structural stress to limit maintenance, and proactive protection against high waves. In this paper, we explore different function approximations for the policy and critic networks in modeling the sequential nature of the system dynamics.
  arXiv  Detail & Related papers  (2024-04-17T02:04:10Z)
• The Impact of LoRA on the Emergence of Clusters in Transformers [2.7309692684728617]
  We employ the framework on Transformers developed by citetsander2022sinkformers,geshkovski2023,geshkovski2023mathematical to explore how variations in attention parameters and initial token values impact the structural dynamics of token clusters. This work contributes to the fine-tuning field through practical applications to the LoRA algorithm citehu2021lora,peft, enhancing our understanding of the behavior of LoRA-enhanced Transformer models.
  arXiv  Detail & Related papers  (2024-02-23T16:26:01Z)
• Understanding the Expressive Power and Mechanisms of Transformer for Sequence Modeling [10.246977481606427]
  We study the mechanisms through which different components of Transformer, such as the dot-product self-attention, affect its expressive power. Our study reveals the roles of critical parameters in the Transformer, such as the number of layers and the number of attention heads.
  arXiv  Detail & Related papers  (2024-02-01T11:43:13Z)
• Unraveling the Temporal Dynamics of the Unet in Diffusion Models [33.326244121918634]
  Diffusion models introduce Gaussian noise into training data and reconstruct the original data iteratively. Central to this iterative process is a single Unet, adapting across time steps to facilitate generation. Recent work revealed the presence of composition and denoising phases in this generation process.
  arXiv  Detail & Related papers  (2023-12-17T04:40:33Z)
• On the Convergence of Encoder-only Shallow Transformers [62.639819460956176]
  We build the global convergence theory of encoder-only shallow Transformers under a realistic setting. Our results can pave the way for a better understanding of modern Transformers, particularly on training dynamics.
  arXiv  Detail & Related papers  (2023-11-02T20:03:05Z)
• Leveraging Low-Rank and Sparse Recurrent Connectivity for Robust Closed-Loop Control [63.310780486820796]
  We show how a parameterization of recurrent connectivity influences robustness in closed-loop settings. We find that closed-form continuous-time neural networks (CfCs) with fewer parameters can outperform their full-rank, fully-connected counterparts.
  arXiv  Detail & Related papers  (2023-10-05T21:44:18Z)
• ASR: Attention-alike Structural Re-parameterization [53.019657810468026]
  We propose a simple-yet-effective attention-alike structural re- parameterization (ASR) that allows us to achieve SRP for a given network while enjoying the effectiveness of the attention mechanism. In this paper, we conduct extensive experiments from a statistical perspective and discover an interesting phenomenon Stripe Observation, which reveals that channel attention values quickly approach some constant vectors during training.
  arXiv  Detail & Related papers  (2023-04-13T08:52:34Z)
• Variational waveguide QED simulators [58.720142291102135]
  Waveguide QED simulators are made by quantum emitters interacting with one-dimensional photonic band-gap materials. Here, we demonstrate how these interactions can be a resource to develop more efficient variational quantum algorithms.
  arXiv  Detail & Related papers  (2023-02-03T18:55:08Z)
• Convexifying Transformers: Improving optimization and understanding of transformer networks [56.69983975369641]
  We study the training problem of attention/transformer networks and introduce a novel convex analytic approach. We first introduce a convex alternative to the self-attention mechanism and reformulate the regularized training problem of transformer networks. As a byproduct of our convex analysis, we reveal an implicit regularization mechanism, which promotes sparsity across tokens.
  arXiv  Detail & Related papers  (2022-11-20T18:17:47Z)
• Deep Reinforcement Learning for IRS Phase Shift Design in Spatiotemporally Correlated Environments [93.30657979626858]
  We propose a deep actor-critic algorithm that accounts for channel correlations and destination motion. We show that, when channels aretemporally correlated, the inclusion of the SNR in the state representation with function approximation in ways that inhibit convergence.
  arXiv  Detail & Related papers  (2022-11-02T22:07:36Z)
• Transformer Meets Boundary Value Inverse Problems [4.165221477234755]
  Transformer-based deep direct sampling method is proposed for solving a class of boundary value inverse problem. A real-time reconstruction is achieved by evaluating the learned inverse operator between carefully designed data and reconstructed images.
  arXiv  Detail & Related papers  (2022-09-29T17:45:25Z)
• Towards Robust and Adaptive Motion Forecasting: A Causal Representation Perspective [72.55093886515824]
  We introduce a causal formalism of motion forecasting, which casts the problem as a dynamic process with three groups of latent variables. We devise a modular architecture that factorizes the representations of invariant mechanisms and style confounders to approximate a causal graph. Experiment results on synthetic and real datasets show that our three proposed components significantly improve the robustness and reusability of the learned motion representations.
  arXiv  Detail & Related papers  (2021-11-29T18:59:09Z)
• Conformer-based End-to-end Speech Recognition With Rotary Position Embedding [11.428057887454008]
  We introduce rotary position embedding (RoPE) in the convolution-augmented transformer (conformer) RoPE encodes absolute positional information into the input sequence by a rotation matrix, and then naturally incorporates explicit relative position information into a self-attention module. Our model achieves a relative word error rate reduction of 8.70% and 7.27% over the conformer on test-clean and test-other sets of the LibriSpeech corpus respectively.
  arXiv  Detail & Related papers  (2021-07-13T08:07:22Z)
• Feedback-induced instabilities and dynamics in the Jaynes-Cummings model [62.997667081978825]
  We investigate the coherence and steady-state properties of the Jaynes-Cummings model subjected to time-delayed coherent feedback. The introduced feedback qualitatively modifies the dynamical response and steady-state quantum properties of the system.
  arXiv  Detail & Related papers  (2020-06-20T10:07:01Z)
• Multiplicative noise and heavy tails in stochastic optimization [62.993432503309485]
  empirical optimization is central to modern machine learning, but its role in its success is still unclear. We show that it commonly arises in parameters of discrete multiplicative noise due to variance. A detailed analysis is conducted in which we describe on key factors, including recent step size, and data, all exhibit similar results on state-of-the-art neural network models.
  arXiv  Detail & Related papers  (2020-06-11T09:58:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.