论文检索

Deep Learning · Large Language Models

Yuxuan Sun, Ruikang Liu, Haoli Bai, Han Bao, Kang Zhao, Yuening Li, JiaxinHu, Xianzhi Yu, Lu Hou, Chun Yuan 等

Recently, quantization has been widely used for the compression and acceleration of large language models (LLMs). Due to the outliers in LLMs, it is crucial to flatten weights and activations to minimize quantization error with equally spaced quantization points. Prior research explores various pre-quantization transformations to suppress outliers, such as per-channel scaling and Hadamard transformation. However, we observe that these transformed weights and activations can still exhibit steep and dispersed distributions. In this paper, we propose FlatQuant (Fast and Learnable Affine Transformation), a new post-training quantization approach that enhances the flatness of weights and activations. Our approach identifies optimal affine transformations for each linear layer, calibrated in hours via a lightweight objective. To reduce runtime overhead of affine transformation, we apply Kronecker product with two lightweight matrices, and fuse all operations in FlatQuant into a single kernel. Extensive experiments demonstrate that FlatQuant establishes a new state-of-the-art benchmark for quantization. For example, it achieves less than 1\% accuracy drop for W4A4 quantization on the LLaMA-3-70B model, surpassing SpinQuant by 7.5\%. Additionally, it provides up to 2.3x prefill speedup and 1.7x decoding speedup compared to the FP16 model. Code is available at: https://github.com/ruikangliu/FlatQuant.

Deep Learning · Large Language Models

Abudukelimu Wuerkaixi, Qizhou Wang, Sen Cui, Wutong Xu, Bo Han, Gang Niu, Masashi Sugiyama, Changshui Zhang

With the growing deployment of large language models (LLMs) across diverse domains, concerns regarding their safety have grown substantially.LLM unlearning has emerged as a pivotal approach to removing harmful or unlawful contents while maintaining utility.Despite increasing interest, the challenges of continual unlearning, which is common in real-world scenarios, remain underexplored.Successive unlearning tasks often lead to intensified utility degradation.To effectively unlearn targeted knowledge while preserving LLM utility, it is essential to minimize changes in model parameters by selectively updating those linked to the target knowledge, thereby ensuring other knowledge remains unaffected.Building on the task vector framework, we propose a new method named ALKN (Adaptive Localization of Knowledge Negation), which uses dynamic masking to sparsify training gradients and adaptively adjusts unlearning intensity based on inter-task relationships.Comprehensive experiments across three well-established LLM unlearning datasets demonstrate that our approach consistently outperforms baseline methods in both unlearning effectiveness and utility retention under continual unlearning settings.

Deep Learning · Self-Supervised Learning

Tianqi Du, Haotian Huang, Yifei Wang, Yisen Wang

Large language models (LLMs) have exhibited impressive performance and surprising emergent properties. However, their effectiveness remains limited by the fixed context window of the transformer architecture, posing challenges for long-context modeling. Among these challenges, length generalization — the ability to generalize to sequences longer than those seen during training — is a classical and fundamental problem. In this work, we propose a fresh perspective on length generalization, shifting the focus from the conventional emphasis on input features such as positional encodings or data structures to the output distribution of the model. Specifically, through case studies on synthetic tasks, we highlight the critical role of **long-short alignment** — the consistency of output distributions across sequences of varying lengths. Extending this insight to natural language tasks, we propose a metric called Long-Short Misalignment to quantify this phenomenon, uncovering a strong correlation between the metric and length generalization performance. Building on these findings, we develop a regularization term that promotes long-short alignment during training. Extensive experiments validate the effectiveness of our approach, offering new insights for achieving more effective long-context modeling in LLMs. Code is available at https://github.com/PKU-ML/LongShortAlignment.

Deep Learning · Graph Neural Networks

Dooho Lee, Myeong Kong, Sagad Hamid, Cheonwoo Lee, Jaemin Yoo

We revisit DropEdge, a data augmentation technique for GNNs which randomly removes edges to expose diverse graph structures during training. While being a promising approach to effectively reduce overfitting on specific connections in the graph, we observe that its potential performance gain in supervised learning tasks is significantly limited. To understand why, we provide a theoretical analysis showing that the limited performance of DropEdge comes from the fundamental limitation that exists in many GNN architectures.Based on this analysis, we propose **Aggregation Buffer**, a parameter block specifically designed to improve the robustness of GNNs by addressing the limitation of DropEdge. Our method is compatible with any GNN model, and shows consistent performance improvements on multiple datasets. Moreover, our method effectively addresses well-known problems such as degree bias or structural disparity as a unifying solution. Code and datasets are available at https://github.com/dooho00/agg-buffer.

Deep Learning · Large Language Models

Gaurav Ghosal, Pratyush Maini, Aditi Raghunathan

Large language models are susceptible to memorizing repeated sequences, posing privacy and copyright concerns. A popular mitigation strategy is to remove memorized information from specific neurons post-hoc. However, such approaches have shown limited success so far. In a controlled setting, we show that the memorization of *natural* sequences (those that resemble linguistically plausible text) become *mechanistically entangled* with general language abilities, thereby becoming challenging to remove post-hoc. In this work, we put forward a new paradigm of MemSinks that promotes isolation of memorization by design. We leverage a sequence identifier to activate a unique set of memorization neurons for each sequence across repetitions. By analyzing the dynamics of learning and forgetting, we argue that MemSinks facilitates clean isolation of memorized content, making it easier to remove without compromising general language capabilities. We implement MemSinks at the billion-parameter and billion-token scale, and observe both effective isolation and strong generalization. To our knowledge, this is the first proof-of-concept on real data demonstrating that simultaneous generalization and isolation is achievable. We open-source our code at http://github.com/grghosal/MemSinks.

General Machine Learning · Transfer, Multitask and Meta-learning

Filippo Rinaldi, Giacomo Capitani, Lorenzo Bonicelli, Donato Crisostomi, Federico Bolelli, ELISA FICARRA, Emanuele Rodola, Simone Calderara, Angelo Porrello

Foundation models serve as the backbone for numerous specialized models developed through fine-tuning. However, when the underlying pretrained model is updated or retrained (e.g., on larger and more curated datasets), the fine-tuned model becomes obsolete, losing its utility and requiring retraining. This raises the question: is it possible to transfer fine-tuning to a new release of the model? In this work, we investigate how to transfer fine-tuning to a new checkpoint without having to re-train, in a data-free manner. To do so, we draw principles from model re-basin and provide a recipe based on weight permutations to re-base the modifications made to the original base model, often called task vector. In particular, our approach tailors model re-basin for Transformer models, taking into account the challenges of residual connections and multi-head attention layers. Specifically, we propose a two-level method rooted in spectral theory, initially permuting the attention heads and subsequently adjusting parameters within select pairs of heads. Through extensive experiments on visual and textual tasks, we achieve the seamless transfer of fine-tuned knowledge to new pre-trained backbones without relying on a single training step or datapoint. Code is available at https://github.com/aimagelab/TransFusion.

Deep Learning · Generative Models and Autoencoders

Fan Shi, Bin Li, Xiangyang Xue

Abstract visual reasoning (AVR) enables humans to quickly discover and generalize abstract rules to new scenarios. Designing intelligent systems with human-like AVR abilities has been a long-standing topic in the artificial intelligence community. Deep AVR solvers have recently achieved remarkable success in various AVR tasks. However, they usually use task-specific designs or parameters in different tasks. In such a paradigm, solving new tasks often means retraining the model, and sometimes retuning the model architectures, which increases the cost of solving AVR problems. In contrast to task-specific approaches, this paper proposes a novel Unified Conditional Generative Solver (UCGS), aiming to address multiple AVR tasks in a unified framework. First, we prove that some well-known AVR tasks can be reformulated as the problem of estimating the predictability of target images in problem panels. Then, we illustrate that, under the proposed framework, training one conditional generative model can solve various AVR tasks. The experiments show that with a single round of multi-task training, UCGS demonstrates abstract reasoning ability across various AVR tasks. Especially, UCGS exhibits the ability of zero-shot reasoning, enabling it to perform abstract reasoning on problems from unseen AVR tasks in the testing phase.

Deep Learning · Graph Neural Networks

James Rowbottom, Georg Maierhofer, Teo Deveney, Eike Müller, Alberto Paganini, Katharina Schratz, Pietro Lió, Carola-Bibiane Schönlieb, Chris Budd

We present a novel, and effective, approach to achieve optimal mesh relocation in finite element methods (FEMs). The cost and accuracy of FEMs is critically dependent on the choice of mesh points. Mesh relocation (r-adaptivity) seeks to optimise the mesh geometry to obtain the best solution accuracy at given computational budget. Classical r-adaptivity relies on the solution of a separate nonlinear ``meshing'' PDE to determine mesh point locations. This incurs significant cost at remeshing, and relies on estimates that relate interpolation- and FEM-error. Recent machine learning approaches have focused on the construction of fast surrogates for such classical methods. Instead, our new approach trains a graph neural network (GNN) to determine mesh point locations by directly minimising the FE solution error from the PDE system Firedrake to achieve higher solution accuracy. Our GNN architecture closely aligns the mesh solution space to that of classical meshing methodologies, thus replacing classical estimates for optimality with a learnable strategy. This allows for rapid and robust training and results in an extremely efficient and effective GNN approach to online r-adaptivity. Our method outperforms both classical, and prior ML, approaches to r-adaptive meshing. In particular, it achieves lower FE solution error, whilst retaining the significant speed-up over classical methods observed in prior ML work.

Theory · Learning Theory

Xuyang Chen, Lin Zhao

Actor-critic algorithms have been instrumental in boosting the performance of numerous challenging applications involving continuous control, such as highly robust and agile robot motion control. However, their theoretical understanding remains largely underdeveloped. Existing analyses mostly focus on finite state-action spaces and on simplified variants of actor-critic, such as double-loop updates with i.i.d. sampling, which are often impractical for real-world applications.We consider the canonical and widely adopted single-timescale updates with Markovian sampling in continuous state-action space. Specifically, we establish finite-time convergence by introducing a novel Lyapunov analysis framework, which provides a unified convergence characterization of both the actor and the critic. Our approach is less conservative than previous methods and offers new insights into the coupled dynamics of actor-critic updates.

General Machine Learning · Supervised Learning

Jiawei Tang, Yuheng Jia

Label distribution learning (LDL) is an effective method to predict the relative label description degree (a.k.a. label distribution) of a sample. However, the label distribution is not a complete representation of an instance because it overlooks the absolute intensity of each label. Specifically, it's impossible to obtain the total description degree of hidden labels that not in the label space, which leads to the loss of information and confusion in instances. To solve the above problem, we come up with a new concept named background concentration to serve as the absolute description degree term of the label distribution and introduce it into the LDL process, forming the improved paradigm of concentration distribution learning. Moreover, we propose a novel model by probabilistic methods and neural networks to learn label distributions and background concentrations from existing LDL datasets. Extensive experiments prove that the proposed approach is able to extract background concentrations from label distributions while producing more accurate prediction results than the state-of-the-art LDL methods. The code is available in https://github.com/seutjw/CDL-LD.

Deep Learning · Generative Models and Autoencoders

Tingyu Zhu, Haoyu Liu, Ziyu Wang, Zhimin Jiang, Zeyu Zheng

Developing generative models to create or conditionally create symbolic music presents unique challenges due to the combination of limited data availability and the need for high precision in note pitch. To address these challenges, we introduce an efficient Fine-Grained Guidance (FGG) approach within diffusion models. FGG guides the diffusion models to generate music that aligns more closely with the control and intent of expert composers, which is critical to improve the accuracy, listenability, and quality of generated music. This approach empowers diffusion models to excel in advanced applications such as improvisation, and interactive music creation. We derive theoretical characterizations for both the challenges in symbolic music generation and the effects of the FGG approach. We provide numerical experiments and subjective evaluation to demonstrate the effectiveness of our approach. We have published a demo page to showcase performances, which enables real-time interactive generation.

Deep Learning · Generative Models and Autoencoders

Long Zhao, Sanghyun Woo, Ziyu Wan, Yandong li, Han Zhang, Boqing Gong, Hartwig Adam, Xuhui Jia, Ting Liu

In generative modeling, tokenization simplifies complex data into compact, structured representations, creating a more efficient, learnable space. For high-dimensional visual data, it reduces redundancy and emphasizes key features for high-quality generation. Current visual tokenization methods rely on a traditional autoencoder framework, where the encoder compresses data into latent representations, and the decoder reconstructs the original input. In this work, we offer a new perspective by proposing denoising as decoding, shifting from single-step reconstruction to iterative refinement. Specifically, we replace the decoder with a diffusion process that iteratively refines noise to recover the original image, guided by the latents provided by the encoder. We evaluate our approach by assessing both reconstruction (rFID) and generation quality (FID), comparing it to state-of-the-art autoencoding approaches. By adopting iterative reconstruction through diffusion, our autoencoder, namely Epsilon-VAE, achieves high reconstruction quality, which in turn enhances downstream generation quality by 22% at the same compression rates or provides 2.3x inference speedup through increasing compression rates. We hope this work offers new insights into integrating iterative generation and autoencoding for improved compression and generation.

Deep Learning

Hao Li, Qi Lv, Rui Shao, Xiang Deng, Yinchuan Li, Jianye Hao, Liqiang Nie

Transforming complex actions into discrete skill abstractions has demonstrated strong potential for robotic manipulation.Existing approaches mainly leverage latent variable models, e.g., VQ-VAE, to learn skill abstractions through learned vectors (codebooks), while they suffer from codebook collapse and modeling the causal relationship between learned skills. To address these limitations, we present **S**kill **T**raining with **A**ugmented **R**otation (**STAR**), a framework that advances both skill learning and composition to complete complex behaviors. Specifically, to prevent codebook collapse, we devise rotation-augmented residual skill quantization (RaRSQ).It encodes relative angles between encoder outputs into the gradient flow by rotation-based gradient mechanism. Points within the same skill code are forced to be either pushed apart or pulled closer together depending on gradient directions.Further, to capture the casual relationship between skills, we present causal skill transformer (CST) which explicitly models dependencies between skill representations through an autoregressive mechanism for coherent action generation.Extensive experiments demonstrate the superiority of STAR on both LIBERO benchmark and realworld tasks, with around 12% improvement over the baselines.

Applications · Chemistry, Physics, and Earth Sciences

Jingxiang Qu, Wenhan Gao, Jiaxing Zhang, Xufeng Liu, Hua Wei, Haibin Ling, Yi Liu

3D Geometric Graph Neural Networks (GNNs) have emerged as transformative tools for modeling molecular data. Despite their predictive power, these models often suffer from limited interpretability, raising concerns for scientific applications that require reliable and transparent insights. While existing methods have primarily focused on explaining molecular substructures in 2D GNNs, the transition to 3D GNNs introduces unique challenges, such as handling the implicit dense edge structures created by a cutoff radius. To tackle this, we introduce a novel explanation method specifically designed for 3D GNNs, which localizes the explanation to the immediate neighborhood of each node within the 3D space. Each node is assigned an radius of influence, defining the localized region within which message passing captures spatial and structural interactions crucial for the model's predictions. This method leverages the spatial and geometric characteristics inherent in 3D graphs. By constraining the subgraph to a localized radius of influence, the approach not only enhances interpretability but also aligns with the physical and structural dependencies typical of 3D graph applications, such as molecular learning.

Deep Learning · Everything Else

Ilya Kaufman, Omri Azencot

Deep learning models with a large number of parameters, often referred to as over-parameterized models, have achieved exceptional performance across various tasks. Despite concerns about overfitting, these models frequently generalize well to unseen data, thanks to effective regularization techniques, with data augmentation being among the most widely used. While data augmentation has shown great success in classification tasks using label-preserving transformations, its application in regression problems has received less attention. Recently, a novel manifold learning approach for generating synthetic data was proposed, utilizing a first-order approximation of the data manifold. Building on this foundation, we present a theoretical framework and practical tools for approximating and sampling general data manifolds. Furthermore, we introduce the Curvature-Enhanced Manifold Sampling (CEMS) method for regression tasks. CEMS leverages a second-order representation of the data manifold to enable efficient sampling and reconstruction of new data points. Extensive evaluations across multiple datasets and comparisons with state-of-the-art methods demonstrate that CEMS delivers superior performance in both in-distribution and out-of-distribution scenarios, while introducing only minimal computational overhead. Code is available at https://github.com/azencot-group/CEMS.

General Machine Learning · Everything Else

Lars van der Laan, Ahmed Alaa

Ensuring model calibration is critical for reliable prediction, yet popular distribution-free methods such as histogram binning and isotonic regression offer only asymptotic guarantees. We introduce a unified framework for Venn and Venn-Abers calibration that extends Vovk's approach beyond binary classification to a broad class of prediction tasks defined by generic loss functions. Our method transforms any perfectly in-sample calibrated predictor into a set-valued predictor that, in finite samples, outputs at least one marginally calibrated point prediction. These set predictions shrink asymptotically and converge to a conditionally calibrated prediction, capturing epistemic uncertainty. We further propose Venn multicalibration, a new approach for achieving finite-sample calibration across subpopulations. For quantile loss, our framework recovers group-conditional and multicalibrated conformal prediction as special cases and yields novel prediction intervals with quantile-conditional coverage.

Deep Learning · Other Representation Learning

Sheyang Tang, xiaoyu xu, Jiayan Qiu, Zhou Wang

Implicit Neural Representations (INRs) represent data as continuous functions using the parameters of a neural network, where data information is encoded in the parameter space. Therefore, modeling the distribution of such parameters is crucial for building generalizable INRs. Existing approaches learn a joint distribution of these parameters via a latent vector to generate new data, but such a flat latent often fails to capture the inherent hierarchical structure of the parameter space, leading to entangled data semantics and limited control over the generation process. Here, we propose a **C**ontrollable **H**ierarchical **I**mplicit **N**eural **R**epresentation (**CHINR**) framework, which explicitly models conditional dependencies across layers in the parameter space. Our method consists of two stages: In Stage-1, we construct a Layers-of-Experts (LoE) network, where each layer modulates distinct semantics through a unique latent vector, enabling disentangled and expressive representations. In Stage-2, we introduce a Hierarchical Conditional Diffusion Model (HCDM) to capture conditional dependencies across layers, allowing for controllable and hierarchical data generation at various semantic granularities. Extensive experiments across different modalities demonstrate that CHINR improves generalizability and offers flexible hierarchical control over the generated content.

Optimization

Nikolaj Jensen, Jamie Vicary

In this work, we propose a new architecture-agnostic method for training idempotent neural networks. An idempotent operator satisfies $f(x) = f(f(x))$, meaning it can be applied iteratively with no effect beyond the first application. Some neural networks used in data transformation tasks, such as image generation and augmentation, can represent non-linear idempotent projections. Using methods from perturbation theory we derive the recurrence relation ${\mathbf{K}' \leftarrow 3\mathbf{K}^2 - 2\mathbf{K}^3}$ for iteratively projecting a real-valued matrix $\mathbf{K}$ onto the manifold of idempotent matrices. Our analysis shows that for linear, single-layer MLP networks this projection 1) has idempotent fixed points, and 2) is attracting only around idempotent points. We give an extension to non-linear networks by considering our approach as a substitution of the gradient for the canonical loss function, achieving an architecture-agnostic training scheme. We provide experimental results for MLP- and CNN-based architectures with significant improvement in idempotent error over the canonical gradient-based approach. Finally, we demonstrate practical applications of the method as we train a generative network successfully using only a simple reconstruction loss paired with our method.

Optimization · Everything Else

Shiyu Wang, Mariam Avagyan, Yihan Shen, Arnaud Lamy, Tingran Wang, Szabolcs Marka, Zsuzsanna Marka, John Wright

Learned denoisers play a fundamental role in various signal generation (e.g., diffusion models) and reconstruction (e.g., compressed sensing) architectures, whose success derives from their ability to leverage low-dimensional structure in data. Existing denoising methods, however, either rely on local approximations that require a linear scan of the entire dataset or treat denoising as generic function approximation problems, sacrificing efficiency and interpretability. We consider the problem of efficiently denoising a new noisy data point sampled from an unknown manifold $\mathcal M \in \mathbb{R}^D$, using only noisy samples. This work proposes a framework for test-time efficient manifold denoising, by framing the concept of "learning-to-denoise" as *"learning-to-optimize"*. We have two technical innovations: (i) *online learning* methods which learn to optimize over the manifold of clean signals using only noisy data, effectively "growing" an optimizer one sample at a time. (ii) *mixed-order* methods which guarantee that the learned optimizers achieve global optimality, ensuring both efficiency and near-optimal denoising performance. We corroborate these claims with theoretical analyses of both the complexity and denoising performance of mixed-order traversal. Our experiments on scientific manifolds demonstrate significantly improved complexity-performance tradeoffs compared to nearest neighbor search, which underpins existing provable denoising approaches based on exhaustive search.

Theory · Probabilistic Methods

Annabelle Carrell, Albert Gong, Abhishek Shetty, Raaz Dwivedi, Lester Mackey

The goal in thinning is to summarize a dataset using a small set of representative points. Remarkably, sub-Gaussian thinning algorithms like Kernel Halving and Compress can match the quality of uniform subsampling while substantially reducing the number of summary points. However, existing guarantees cover only a restricted range of distributions and kernel-based quality measures and suffer from pessimistic dimension dependence. To address these deficiencies, we introduce a new low-rank analysis of sub-Gaussian thinning that applies to any distribution and any kernel, guaranteeing high-quality compression whenever the kernel or data matrix is approximately low-rank. To demonstrate the broad applicability of the techniques, we design practical sub-Gaussian thinning approaches that improve upon the best known guarantees for approximating attention in transformers, accelerating stochastic gradient training through reordering, and distinguishing distributions in near-linear time.