Product attribute value extraction is an important task in e-Commerce which can help several downstream applications such as product search and recommendation. Most previous models handle this task using sequence labeling or question answering method which rely on the sequential position information of values in the product text and are vulnerable to data discrepancy between training and testing. This limits their generalization ability to real-world scenario in which each product can have multiple descriptions across various shopping platforms with different composition of text and style. They also have limited zero-shot ability to new values. In this paper, we propose a multi-task learning model with value generation/classification and attribute prediction called JPAVE to predict values without the necessity of position information of values in the text. Furthermore, the copy mechanism in value generator and the value attention module in value classifier help our model address the data discrepancy issue by only focusing on the relevant part of input text and ignoring other information which causes the discrepancy issue such as sentence structure in the text. Besides, two variants of our model are designed for open-world and closed-world scenarios. In addition, copy mechanism introduced in the first variant based on value generation can improve its zero-shot ability for identifying unseen values. Experimental results on a public dataset demonstrate the superiority of our model compared with strong baselines and its generalization ability of predicting new values.
Traffic forecasting is a complex multivariate time-series regression task of paramount importance for traffic management and planning. However, existing approaches often struggle to model complex multi-range dependencies using local spatiotemporal features and road network hierarchical knowledge. To address this, we propose MultiSPANS. First, considering that an individual recording point cannot reflect critical spatiotemporal local patterns, we design multi-filter convolution modules for generating informative ST-token embeddings to facilitate attention computation. Then, based on ST-token and spatial-temporal position encoding, we employ the Transformers to capture long-range temporal and spatial dependencies. Furthermore, we introduce structural entropy theory to optimize the spatial attention mechanism. Specifically, The structural entropy minimization algorithm is used to generate optimal road network hierarchies, i.e., encoding trees. Based on this, we propose a relative structural entropy-based position encoding and a multi-head attention masking scheme based on multi-layer encoding trees. Extensive experiments demonstrate the superiority of the presented framework over several state-of-the-art methods in real-world traffic datasets, and the longer historical windows are effectively utilized. The code is available at https://github.com/SELGroup/MultiSPANS.
Large Language Models (LLMs) have generated considerable interest and debate regarding their potential emergence of Theory of Mind (ToM). Several recent inquiries reveal a lack of robust ToM in these models and pose a pressing demand to develop new benchmarks, as current ones primarily focus on different aspects of ToM and are prone to shortcuts and data leakage. In this position paper, we seek to answer two road-blocking questions: (1) How can we taxonomize a holistic landscape of machine ToM? (2) What is a more effective evaluation protocol for machine ToM? Following psychological studies, we taxonomize machine ToM into 7 mental state categories and delineate existing benchmarks to identify under-explored aspects of ToM. We argue for a holistic and situated evaluation of ToM to break ToM into individual components and treat LLMs as an agent who is physically situated in environments and socially situated in interactions with humans. Such situated evaluation provides a more comprehensive assessment of mental states and potentially mitigates the risk of shortcuts and data leakage. We further present a pilot study in a grid world setup as a proof of concept. We hope this position paper can facilitate future research to integrate ToM with LLMs and offer an intuitive means for researchers to better position their work in the landscape of ToM. Project page: https://github.com/Mars-tin/awesome-theory-of-mind
Transformer-based Large Language Models (LLMs) are pioneering advances in many natural language processing tasks, however, their exceptional capabilities are restricted within the preset context window of Transformer. Position Embedding (PE) scaling methods, while effective in extending the context window to a specific length, demonstrate either notable limitations in their extrapolation abilities or sacrificing partial performance within the context window. Length extrapolation methods, although theoretically capable of extending the context window beyond the training sequence length, often underperform in practical long-context applications. To address these challenges, we propose Continuous Length EXtrapolation (CLEX) for LLMs. We generalise the PE scaling approaches to model the continuous dynamics by ordinary differential equations over the length scaling factor, thereby overcoming the constraints of current PE scaling methods designed for specific lengths. Moreover, by extending the dynamics to desired context lengths beyond the training sequence length, CLEX facilitates the length extrapolation with impressive performance in practical tasks. We demonstrate that CLEX can be seamlessly incorporated into LLMs equipped with Rotary Position Embedding, such as LLaMA and GPT-NeoX, with negligible impact on training and inference latency. Experimental results reveal that CLEX can effectively extend the context window to over 4x or almost 8x training length, with no deterioration in performance. Furthermore, when evaluated on the practical LongBench benchmark, our model trained on a 4k length exhibits competitive performance against state-of-the-art open-source models trained on context lengths up to 32k. Our code is available at https://github.com/DAMO-NLP-SG/CLEX.
Recent research has demonstrated the efficacy of pre-training graph neural networks (GNNs) to capture the transferable graph semantics and enhance the performance of various downstream tasks. However, the semantic knowledge learned from pretext tasks might be unrelated to the downstream task, leading to a semantic gap that limits the application of graph pre-training. To reduce this gap, traditional approaches propose hybrid pre-training to combine various pretext tasks together in a multi-task learning fashion and learn multi-grained knowledge, which, however, cannot distinguish tasks and results in some transferable task-specific knowledge distortion by each other. Moreover, most GNNs cannot distinguish nodes located in different parts of the graph, making them fail to learn position-specific knowledge and lead to suboptimal performance. In this work, inspired by the prompt-based tuning in natural language processing, we propose a unified framework for graph hybrid pre-training which injects the task identification and position identification into GNNs through a prompt mechanism, namely multi-task graph dual prompt (ULTRA-DP). Based on this framework, we propose a prompt-based transferability test to find the most relevant pretext task in order to reduce the semantic gap. To implement the hybrid pre-training tasks, beyond the classical edge prediction task (node-node level), we further propose a novel pre-training paradigm based on a group of $k$-nearest neighbors (node-group level). The combination of them across different scales is able to comprehensively express more structural semantics and derive richer multi-grained knowledge. Extensive experiments show that our proposed ULTRA-DP can significantly enhance the performance of hybrid pre-training methods and show the generalizability to other pre-training tasks and backbone architectures.
Deep architectures such as Transformers are sometimes criticized for having uninterpretable "black-box" representations. We use causal intervention analysis to show that, in fact, some linguistic features are represented in a linear, interpretable format. Specifically, we show that BERT's ability to conjugate verbs relies on a linear encoding of subject number that can be manipulated with predictable effects on conjugation accuracy. This encoding is found in the subject position at the first layer and the verb position at the last layer, but distributed across positions at middle layers, particularly when there are multiple cues to subject number.
Integrated sensing and communications (ISAC) is envisioned as one of the key enablers of next-generation wireless systems, offering improved hardware, spectral, and energy efficiencies. In this paper, we consider an ISAC transceiver with an impaired uniform linear array that performs single-target detection and position estimation, and multiple-input single-output communications. A differentiable model-based learning approach is considered, which optimizes both the transmitter and the sensing receiver in an end-to-end manner. An unsupervised loss function that enables impairment compensation without the need for labeled data is proposed. Semi-supervised learning strategies are also proposed, which use a combination of small amounts of labeled data and unlabeled data. Our results show that semi-supervised learning can achieve similar performance to supervised learning with 98.8% less required labeled data.
In this paper, we investigate the inherent capabilities of transformer models in learning arithmetic algorithms, such as addition and parity. Through experiments and attention analysis, we identify a number of crucial factors for achieving optimal length generalization. We show that transformer models are able to generalize to long lengths with the help of targeted attention biasing. In particular, our solution solves the Parity task, a well-known and theoretically proven failure mode for Transformers. We then introduce Attention Bias Calibration (ABC), a calibration stage that enables the model to automatically learn the proper attention biases, which we show to be connected to mechanisms in relative position encoding. We demonstrate that using ABC, the transformer model can achieve unprecedented near-perfect length generalization on certain arithmetic tasks. In addition, we show that ABC bears remarkable similarities to RPE and LoRA, which may indicate the potential for applications to more complex tasks.
Large language models (LLMs) excel in abstractive summarization tasks, delivering fluent and pertinent summaries. Recent advancements have extended their capabilities to handle long-input contexts, exceeding 100k tokens. However, in question answering, language models exhibit uneven utilization of their input context. They tend to favor the initial and final segments, resulting in a U-shaped performance pattern concerning where the answer is located within the input. This bias raises concerns, particularly in summarization where crucial content may be dispersed throughout the source document(s). Besides, in summarization, mapping facts from the source to the summary is not trivial as salient content is usually re-phrased. In this paper, we conduct the first comprehensive study on context utilization and position bias in summarization. Our analysis encompasses 6 LLMs, 10 datasets, and 5 evaluation metrics. We introduce a new evaluation benchmark called MiddleSum on the which we benchmark two alternative inference methods to alleviate position bias: hierarchical summarization and incremental summarization. Our code and data can be found here: https://github.com/ntunlp/MiddleSum.
Diffusion models (DMs) represent state-of-the-art generative models for continuous inputs. DMs work by constructing a Stochastic Differential Equation (SDE) in the input space (ie, position space), and using a neural network to reverse it. In this work, we introduce a novel generative modeling framework grounded in \textbf{phase space dynamics}, where a phase space is defined as {an augmented space encompassing both position and velocity.} Leveraging insights from Stochastic Optimal Control, we construct a path measure in the phase space that enables efficient sampling. {In contrast to DMs, our framework demonstrates the capability to generate realistic data points at an early stage of dynamics propagation.} This early prediction sets the stage for efficient data generation by leveraging additional velocity information along the trajectory. On standard image generation benchmarks, our model yields favorable performance over baselines in the regime of small Number of Function Evaluations (NFEs). Furthermore, our approach rivals the performance of diffusion models equipped with efficient sampling techniques, underscoring its potential as a new tool generative modeling.
Product attribute value extraction plays an important role for many real-world applications in e-Commerce such as product search and recommendation. Previous methods treat it as a sequence labeling task that needs more annotation for position of values in the product text. This limits their application to real-world scenario in which only attribute values are weakly-annotated for each product without their position. Moreover, these methods only use product text (i.e., product title and description) and do not consider the semantic connection between the multiple attribute values of a given product and its text, which can help attribute value extraction. In this paper, we reformulate this task as a multi-label classification task that can be applied for real-world scenario in which only annotation of attribute values is available to train models (i.e., annotation of positional information of attribute values is not available). We propose a classification model with semantic matching and negative label sampling for attribute value extraction. Semantic matching aims to capture semantic interactions between attribute values of a given product and its text. Negative label sampling aims to enhance the model's ability of distinguishing similar values belonging to the same attribute. Experimental results on three subsets of a large real-world e-Commerce dataset demonstrate the effectiveness and superiority of our proposed model.
Delivering precise point and distributional forecasts across a spectrum of prediction horizons represents a significant and enduring challenge in the application of time-series forecasting within various industries. Prior research on developing deep learning models for time-series forecasting has often concentrated on isolated aspects, such as long-term point forecasting or short-term probabilistic estimations. This narrow focus may result in skewed methodological choices and hinder the adaptability of these models to uncharted scenarios. While there is a rising trend in developing universal forecasting models, a thorough understanding of their advantages and drawbacks, especially regarding essential forecasting needs like point and distributional forecasts across short and long horizons, is still lacking. In this paper, we present ProbTS, a benchmark tool designed as a unified platform to evaluate these fundamental forecasting needs and to conduct a rigorous comparative analysis of numerous cutting-edge studies from recent years. We dissect the distinctive data characteristics arising from disparate forecasting requirements and elucidate how these characteristics can skew methodological preferences in typical research trajectories, which often fail to fully accommodate essential forecasting needs. Building on this, we examine the latest models for universal time-series forecasting and discover that our analyses of methodological strengths and weaknesses are also applicable to these universal models. Finally, we outline the limitations inherent in current research and underscore several avenues for future exploration.
Autonomous mobile robot competitions judge based on a robot's ability to quickly and accurately navigate the game field. This means accurate localization is crucial for creating an autonomous competition robot. Two common localization methods are odometry and computer vision landmark detection. Odometry provides frequent velocity measurements, while landmark detection provides infrequent position measurements. The state can also be predicted with a physics model. These three types of localization can be "fused" to create a more accurate state estimate using an Extended Kalman Filter (EKF). The EKF is a nonlinear full-state estimator that approximates the state estimate with the lowest covariance error when given the sensor measurements, the model prediction, and their variances. In this paper, we demonstrate the effectiveness of the EKF by implementing it on a 4-wheel mecanum-drive robot simulation. The position and velocity accuracy of fusing together various combinations of these three data sources are compared. We also discuss the assumptions and limitations of an EKF.
In long context scenarios, large language models (LLMs) face three main challenges: higher computational cost, performance reduction, and position bias. Research indicates that LLM performance hinges on the density and position of key information in the input prompt. Inspired by these findings, we propose LongLLMLingua for prompt compression towards improving LLMs' perception of the key information to simultaneously address the three challenges. Our extensive evaluation across various long context scenarios demonstrates that LongLLMLingua not only enhances performance but also significantly reduces costs and latency. For instance, in the NaturalQuestions benchmark, LongLLMLingua boosts performance by up to 21.4% with around 4x fewer tokens in GPT-3.5-Turbo, leading to substantial cost savings. It achieves a 94.0% cost reduction in the LooGLE benchmark. Moreover, when compressing prompts of about 10k tokens at ratios of 2x-6x, LongLLMLingua can accelerate end-to-end latency by 1.4x-2.6x. Our code is available at https://aka.ms/LongLLMLingua.
Pretrained multilingual encoder models can directly perform zero-shot multilingual tasks or linguistic probing by reformulating the input examples into cloze-style prompts. This is accomplished by predicting the probabilities of the label words at the masked token position, without requiring any updates to the model parameters. However, the performance of this method is limited by the model's bias toward predicting label words which frequently occurred during the pretraining. These words typically receive high probabilities. To address this issue, we combine the models with calibration techniques which modify the probabilities of label words predicted by the models. We first validate the effectiveness of a proposed simple calibration method together with other existing techniques on monolingual encoders in both zero- and few-shot scenarios. We subsequently employ these calibration techniques on multilingual encoders, resulting in substantial performance improvements across a wide range of tasks.
Transformers are widely used to extract semantic meanings from input tokens, yet they usually operate as black-box models. In this paper, we present a simple yet informative decomposition of hidden states (or embeddings) of trained transformers into interpretable components. For any layer, embedding vectors of input sequence samples are represented by a tensor $\boldsymbol{h} \in \mathbb{R}^{C \times T \times d}$. Given embedding vector $\boldsymbol{h}_{c,t} \in \mathbb{R}^d$ at sequence position $t \le T$ in a sequence (or context) $c \le C$, extracting the mean effects yields the decomposition \[ \boldsymbol{h}_{c,t} = \boldsymbol{\mu} + \mathbf{pos}_t + \mathbf{ctx}_c + \mathbf{resid}_{c,t} \] where $\boldsymbol{\mu}$ is the global mean vector, $\mathbf{pos}_t$ and $\mathbf{ctx}_c$ are the mean vectors across contexts and across positions respectively, and $\mathbf{resid}_{c,t}$ is the residual vector. For popular transformer architectures and diverse text datasets, empirically we find pervasive mathematical structure: (1) $(\mathbf{pos}_t)_{t}$ forms a low-dimensional, continuous, and often spiral shape across layers, (2) $(\mathbf{ctx}_c)_c$ shows clear cluster structure that falls into context topics, and (3) $(\mathbf{pos}_t)_{t}$ and $(\mathbf{ctx}_c)_c$ are mutually nearly orthogonal. We argue that smoothness is pervasive and beneficial to transformers trained on languages, and our decomposition leads to improved model interpretability.
Based on the theory of homogeneous spaces we derive geometrically optimal edge attributes to be used within the flexible message-passing framework. We formalize the notion of weight sharing in convolutional networks as the sharing of message functions over point-pairs that should be treated equally. We define equivalence classes of point-pairs that are identical up to a transformation in the group and derive attributes that uniquely identify these classes. Weight sharing is then obtained by conditioning message functions on these attributes. As an application of the theory, we develop an efficient equivariant group convolutional network for processing 3D point clouds. The theory of homogeneous spaces tells us how to do group convolutions with feature maps over the homogeneous space of positions $\mathbb{R}^3$, position and orientations $\mathbb{R}^3 {\times} S^2$, and the group $SE(3)$ itself. Among these, $\mathbb{R}^3 {\times} S^2$ is an optimal choice due to the ability to represent directional information, which $\mathbb{R}^3$ methods cannot, and it significantly enhances computational efficiency compared to indexing features on the full $SE(3)$ group. We support this claim with state-of-the-art results -- in accuracy and speed -- on five different benchmarks in 2D and 3D, including interatomic potential energy prediction, trajectory forecasting in N-body systems, and generating molecules via equivariant diffusion models.
Whole slide image (WSI) classification is a critical task in computational pathology. However, the gigapixel-size of such images remains a major challenge for the current state of deep-learning. Current methods rely on multiple-instance learning (MIL) models with frozen feature extractors. Given the the high number of instances in each image, MIL methods have long assumed independence and permutation-invariance of patches, disregarding the tissue structure and correlation between patches. Recent works started studying this correlation between instances but the computational workload of such a high number of tokens remained a limiting factor. In particular, relative position of patches remains unaddressed. We propose to apply a straightforward encoding module, namely a RoFormer layer , relying on memory-efficient exact self-attention and relative positional encoding. This module can perform full self-attention with relative position encoding on patches of large and arbitrary shaped WSIs, solving the need for correlation between instances and spatial modeling of tissues. We demonstrate that our method outperforms state-of-the-art MIL models on three commonly used public datasets (TCGA-NSCLC, BRACS and Camelyon16)) on weakly supervised classification tasks. Code is available at https://github.com/Sanofi-Public/DDS-RoFormerMIL
Weight-sharing quantization has emerged as a technique to reduce energy expenditure during inference in large neural networks by constraining their weights to a limited set of values. However, existing methods for weight-sharing quantization often make assumptions about the treatment of weights based on value alone that neglect the unique role weight position plays. This paper proposes a probabilistic framework based on Bayesian neural networks (BNNs) and a variational relaxation to identify which weights can be moved to which cluster centre and to what degree based on their individual position-specific learned uncertainty distributions. We introduce a new initialisation setting and a regularisation term which allow for the training of BNNs under complex dataset-model combinations. By leveraging the flexibility of weight values captured through a probability distribution, we enhance noise resilience and downstream compressibility. Our iterative clustering procedure demonstrates superior compressibility and higher accuracy compared to state-of-the-art methods on both ResNet models and the more complex transformer-based architectures. In particular, our method outperforms the state-of-the-art quantization method top-1 accuracy by 1.6% on ImageNet using DeiT-Tiny, with its 5 million+ weights now represented by only 296 unique values.
Large Language Models (LLMs) are trained with a pre-defined context length, restricting their use in scenarios requiring long inputs. Previous efforts for adapting LLMs to a longer length usually requires fine-tuning with this target length (Full-length fine-tuning), suffering intensive training cost. To decouple train length from target length for efficient context window extension, we propose Positional Skip-wisE (PoSE) training that smartly simulates long inputs using a fixed context window. This is achieved by first dividing the original context window into several chunks, then designing distinct skipping bias terms to manipulate the position indices of each chunk. These bias terms and the lengths of each chunk are altered for every training example, allowing the model to adapt to all positions within target length. Experimental results show that PoSE greatly reduces memory and time overhead compared with Full-length fine-tuning, with minimal impact on performance. Leveraging this advantage, we have successfully extended the LLaMA model to 128k tokens using a 2k training context window. Furthermore, we empirically confirm that PoSE is compatible with all RoPE-based LLMs and position interpolation strategies. Notably, our method can potentially support infinite length, limited only by memory usage in inference. With ongoing progress for efficient inference, we believe PoSE can further scale the context window beyond 128k.