ICLR 2026 Papers

The development of foundation models for functional magnetic resonance imaging (fMRI) time series holds significant promise for predicting phenotypes related to disease and cognition. Current models, however, are often trained using a mask-and-reconstruct objective on small brain regions. This focus on low-level information leads to representations that are sensitive to noise and temporal fluctuations, necessitating extensive fine-tuning for downstream tasks. We introduce Brain-Semantoks, a self-supervised framework designed specifically to learn abstract representations of brain dynamics. Its architecture is built on two core innovations: a semantic tokenizer that aggregates noisy regional signals into robust tokens representing functional networks, and a self-distillation objective that enforces representational stability across time. We show that this objective is stabilized through a novel training curriculum, ensuring the model robustly learns meaningful features from low signal-to-noise time series. We demonstrate that learned representations enable strong performance on a variety of downstream tasks even when only using a linear probe. Furthermore, we provide comprehensive scaling analyses indicating more unlabeled data reliably results in out-of-distribution performance gains without domain adaptation.

Bayesian optimization (BO) struggles in high dimensions, where Gaussian-process surrogates demand heavy retraining and brittle assumptions, slowing progress on real engineering and design problems. We introduce GIT-BO, a Gradient-Informed BO framework that couples TabPFN v2, a tabular foundation model that performs zero-shot Bayesian inference in context, with an active-subspace mechanism computed from the model’s own predictive-mean gradients. This aligns exploration to an intrinsic low-dimensional subspace via a Fisher-information estimate and selects queries with a UCB acquisition, requiring no online retraining. Across 60 problem variants spanning 20 benchmarks—nine scalable synthetic families and ten real-world tasks (e.g., power systems, Rover, MOPTA08, Mazda)—up to 500 dimensions, GIT-BO delivers a stronger performance–time trade-off than state-of-the-art GP-based methods (SAASBO, TuRBO, Vanilla BO, BAxUS), ranking highest in performance and with runtime advantages that grow with dimensionality. Limitations include memory footprint and dependence on the capacity of the underlying TFM.

Federated Learning (FL) has emerged as a promising paradigm for privacy-preserving collaborative learning, yet data heterogeneity remains a critical challenge. While existing methods achieve progress in addressing data heterogeneity for participating clients, they fail to generalize to non-participating clients with in-domain distribution shifts and resource constraints. To mitigate this issue, we present HyperFedZero, a novel method that dynamically generates specialized models via a hypernetwork conditioned on distribution-aware embeddings. Our approach explicitly incorporates distribution-aware inductive biases into the model's forward pass, extracting robust distribution embeddings using a NoisyEmbed-enhanced extractor with a Balancing Penalty, effectively preventing feature collapse. The hypernetwork then leverages these embeddings to generate specialized models chunk-by-chunk for non-participating clients, ensuring adaptability to their unique data distributions. Extensive experiments on multiple datasets and models demonstrate HyperFedZero's remarkable performance, surpassing competing methods consistently with minimal computational, storage, and communication overhead. Moreover, ablation studies and visualizations further validate the necessity of each component, confirming meaningful adaptations and validating the effectiveness of HyperFedZero.

As increasingly large pre-trained models are released, deploying them on edge devices for privacy-preserving applications requires effective compression. Recent works combine quantization with the fine-tuning of high-precision LoRA adapters, which can substantially reduce model size while mitigating the accuracy loss from quantization. However, edge devices have inherently heterogeneous capabilities, while performing configuration-wise fine-tuning for every quantization setting is computationally prohibitive. In this paper, we propose CoA-LoRA, a method that dynamically adjusts the LoRA adapter to arbitrary quantization configurations (i.e., the per-layer bit-width choices of a pre-trained model) without requiring repeated fine-tuning. This is accomplished via a configuration-aware model that maps each configuration to its low-rank adjustments. The effectiveness of this model critically depends on the training configuration set, a collection of configurations chosen to cover different total bit-width budgets. However, constructing a high-quality configuration set is non-trivial. We therefore design a Pareto-based configuration search that iteratively optimizes the training configuration set, yielding more precise low-rank adjustments. Our experiments demonstrate that, unlike the state-of-the-art methods that require fine-tuning a separate LoRA adapter for each configuration, CoA-LoRA incurs no additional time cost while achieving comparable or even superior performance to those methods.

Recently, large time series models (LTSM) have become popular and important because they exhibit characteristics similar to large language models, such as flexible context length, scalability, and task generality, outperforming the advanced task-specific models in the domain. However, existing research indicates that the pre-trained LTSM can show a poor non-convex loss landscape (indicating poor trainability). Hence, directly fine-tuning pre-trained LTSM shows overfitting, which leads to poor fine-tuning performance, even worse than training from scratch on the downstream datasets. This severely diminishes the value of the pre-trained LTSM. To address this, we propose a new fine-tuning method called Smoothed Full Fine-tuning (SFF). Specifically, before fine-tuning, we first construct an auxiliary LTSM with a smooth loss landscape (indicating good trainability) through random initialization. Second, we utilize it to smooth the loss landscape of the pre-trained LTSM through linear interpolation between their weights. As a result, the smoothed LTSM acquires good trainability while retaining good pre-training knowledge, thereby achieving better performance when fine-tuned on the downstream dataset. We also explain why SFF is effective from the perspective of optimization theory: interpolation perturbs sharp minima without obviously harming originally flat regions, thereby aiding sharp minima escape to better and smoother basins. Extensive experiments on popular datasets show that our method indeed improves the performance of eight popular LTSMs, e.g., Timer, TimesFM, MOMENT, UniTS, MOIRAI, Chronos, TTMs, and Sundial, in different downstream tasks.

Large language models (LLMs) have achieved remarkable success in diverse tasks, yet their safety alignment remains fragile during adaptation. Even when fine-tuning on benign data or with low-rank adaptation, pre-trained safety behaviors are easily degraded, leading to harmful responses in the fine-tuned models. To address this challenge, we propose GuardSpace, a guardrail framework for preserving safety alignment throughout fine-tuning, composed of two key components: a safety-sensitive subspace and a harmful-resistant null space. First, we explicitly decompose pre-trained weights into safety-relevant and safety-irrelevant components using covariance-preconditioned singular value decomposition, and initialize low-rank adapters from the safety-irrelevant ones, while freezing safety-relevant components to preserve their associated safety mechanism. Second, we construct a null space projector that restricts adapter updates from altering safe outputs on harmful prompts, thereby maintaining the original refusal behavior. Experiments with various pre-trained models on multiple downstream tasks demonstrate that GuardSpace achieves superior performance over existing methods. Notably, for Llama-2-7B-Chat fine-tuned on GSM8K, GuardSpace outperforms the state-of-the-art method AsFT, reducing the average harmful score from 14.4\% to 3.6\%, while improving the accuracy from from 26.0\% to 28.0\%.

Large pre-trained models are commonly adapted to downstream tasks using parameter-efficient fine-tuning methods such as Low-Rank Adaptation (LoRA), which injects small trainable low-rank matrices instead of updating all weights. While LoRA dramatically reduces trainable parameters with little overhead, it can still underperform full fine-tuning in accuracy and often converges more slowly. We introduce LoFT, a novel low-rank adaptation method that behaves like full fine-tuning by aligning the optimizer’s internal dynamics with those of updating all model weights. LoFT not only learns weight updates in a low-rank subspace (like LoRA) but also properly projects the optimizer’s first and second moments (Adam’s momentum and variance) into the same subspace, mirroring full-model updates. By aligning the low-rank update itself with the full update, LoFT eliminates the need for tuning extra hyperparameter, e.g., LoRA scaling $\alpha$. Empirically, this approach substantially narrows the performance gap between adapter-based tuning and full fine-tuning and consistently outperforms standard LoRA-style methods, all without increasing inference cost.

We propose a post-hoc adaptive conformal anomaly detection method for monitoring time series that leverages predictions from pre-trained foundation models without requiring additional fine-tuning. Our method yields an interpretable anomaly score directly interpretable as a false alarm rate (p-value), facilitating transparent and actionable decision-making. It employs weighted quantile conformal prediction bounds and adaptively learns optimal weighting parameters from past predictions, enabling calibration under distribution shifts and stable false alarm control, while preserving out-of-sample guarantees. As a model-agnostic solution, it integrates seamlessly with foundation models and supports rapid deployment in resource-constrained environments. This approach addresses key industrial challenges such as limited data availability, lack of training expertise, and the need for immediate inference, while taking advantage of the growing accessibility of time series foundation models. Experiments on both synthetic and real-world datasets show that the proposed approach delivers strong performance, combining simplicity, interpretability, robustness, and adaptivity.

We introduce Qronos---a new post-training quantization algorithm that not only explicitly corrects errors due to both weight and activation quantization, but also corrects errors accumulated from previously quantized layers. Our iterative algorithm is based on an interpretable and disciplined optimization framework that surpasses existing data-driven approaches. At each step, Qronos alternates between error correction and diffusion via optimal update rules. Importantly, we prove that Qronos admits an equivalent formulation that significantly improves algorithmic efficiency; we use our discovery to reduce peak memory usage by 18\times on Llama3 8B, and our scaling analysis shows a speedup of up to 13.8\times for a single-layer microbenchmark. We demonstrate compatibility with existing transformation techniques such as Hadamard-based incoherence processing and weight-activation scaling equalization, among others. We evaluate Qronos using recent language models in the Llama3 and Qwen3 families; Qronos consistently outperforms previous state-of-the-art adaptive rounding methods when quantizing the weights, activations, and/or KV caches to 4 bits or fewer.

Learning from multi-variate time-series with heterogeneous channel configurations remains a fundamental challenge for deep neural networks, particularly in clinical domains such as intracranial electroencephalography (iEEG), where channel setups vary widely across subjects. In this work, we introduce multi-variate parallel attention (MVPA), a novel self-attention mechanism that disentangles content, temporal, and spatial attention, enabling flexible, generalizable, and efficient modeling of time-series data with varying channel counts and configurations. We use MVPA to build MVPFormer, a generative foundation model for human electrophysiology, trained to predict the evolution of iEEG signals across diverse subjects. To support this and future efforts by the community, we release the Long-term iEEG dataset, the largest publicly available iEEG dataset to date, comprising nearly 10,000 hours of recordings from heterogeneous clinical sources. MVPFormer leverages MVPA to achieve strong generalization across subjects, demonstrating expert-level performance in several iEEG tasks. MVPFormer surpasses state-of-the-art (SOTA) Transformer baselines in seizure detection across the Long-term, the MAYO, and the FNUSA datasets, while also achieving SOTA performance on four Brain TreeBank iEEG decoding tasks (volume, pitch, onset, and speech). We further validate MVPA on standard time-series forecasting and classification tasks, where it matches or exceeds the performance of existing attention-based models. Together, our contributions establish MVPA as a general-purpose attention mechanism for heterogeneous time-series and MVPFormer as the first open-source, open-weights, and open-data iEEG foundation model with SOTA clinical performance.

The deployment of large language models (LLMs) for specialized tasks on resource-constrained edge devices like smartphones and sensors presents a significant scalability problem. To run on such hardware, these massive models must be compressed using techniques like \emph{quantization or pruning} to reduce their memory and computational footprint. Concurrently, foundational LLMs are periodically updated by their developers with new data, making their $\textit{internal parameters shift over time}$. While parameter-efficient methods like Low-Rank Adaptation (LoRA) streamline personalization by fine-tuning only a small fraction of parameters, the resulting adapters are $\textbf{brittle}$; a LoRA trained for one specific compression scheme is incompatible with another, and an adapter trained on an older base model performs poorly on an updated one. This forces a costly cycle of retraining for each unique device and every new model release. To address this, we introduce a novel framework that creates a single, universally portable adapter that is both $\textbf{\textit{(i)} compression-aware and \textit{(ii)} temporally robust}$. We achieve this by augmenting the training process with a variety of simulated compression techniques during a single run, utilizing a quantized forward pass to build resilience while maintaining a full-precision backward pass for stable gradient optimization. $\textit{This method yields a unified adapter robust to diverse compression artifacts and the subtle parameter shifts from model evolution}$. Extensive experiments on models such as $\texttt{Llama-2, Llama-3.1, Gemma-2}$, and $\texttt{Mistral}$ across reasoning benchmarks like $\textit{SQA, MATH, and GSM8K}$ demonstrate that our single adapter achieves performance comparable to specialized adapters ($\textit{e.g.}$, QLoRA) that are individually retrained for each compression scheme. Furthermore, we show this single adapter maintains its high performance when applied to future, evolved versions of the base model, eliminating the need for periodic retraining. Our work pioneers an efficient paradigm for edge AI, creating portable model patches that bridge the gap between cloud-based personalization, the diverse hardware ecosystem, and the lifecycle of evolving LLMs.

The rapid advancement of video diffusion models has been hindered by fundamental limitations in temporal modeling, particularly the rigid synchronization of frame evolution imposed by conventional scalar timestep variables. While task-specific adaptations and autoregressive models have sought to address these challenges, they remain constrained by computational inefficiency, catastrophic forgetting, or narrow applicability. In this work, we present \textbf{Pusa} V1.0, a versatile model that leverages \textbf{vectorized timestep adaptation (VTA)} to enable fine-grained temporal control within a unified video diffusion framework. Note that VTA is a non-destructive adaptation, which means that it fully preserves the capabilities of the base model. \textbf{Unlike conventional methods like Wan-I2V, which finetune a base text-to-video (T2V) model with abundant resources to do image-to-video (I2V), we achieve comparable results in a zero-shot manner after an ultra-efficient finetuning process based on VTA. Moreover, this method also unlocks many other zero-shot capabilities simultaneously, such as start-end frames and video extension ---all without task-specific training. Meanwhile, it keeps the T2V capability from the base model.} Mechanistic analyses also reveal that our approach preserves the foundation model's generative priors while surgically injecting temporal dynamics, avoiding the combinatorial explosion inherent to the vectorized timestep. This work establishes a scalable, efficient, and versatile paradigm for next-generation video synthesis, democratizing high-fidelity video generation for research and industry alike.

Large language models (LLMs) are typically deployed under diverse memory and compute constraints. Existing approaches build model families by training each size independently, which is prohibitively expensive and provides only coarse-grained size options. In this work, we identify a novel phenomenon that we call boomerang distillation: starting from a large base model (the teacher), one first distills down to a small student and then progressively reconstructs intermediate-sized models by re-incorporating blocks of teacher layers into the student without any additional training. This process produces zero-shot interpolated models of many intermediate sizes whose performance scales smoothly between the student and teacher, often matching or surpassing pretrained or distilled models of the same size. We further analyze when this type of interpolation succeeds, showing that alignment between teacher and student through pruning and distillation is essential. Boomerang distillation thus provides a simple and efficient way to generate fine-grained model families, dramatically reducing training cost while enabling flexible adaptation across deployment environments.

Microscaling data formats leverage per-block tensor quantization to enable aggressive model compression with limited loss in accuracy. Unlocking their potential for efficient training and inference necessitates hardware-friendly implementations that handle matrix multiplications in a native format and adopt efficient error-mitigation strategies. Herein, we reported the emergence of a surprising behavior associated with microscaling quantization, whereas the output of a quantized model degrades as block size is decreased below a given threshold. This behavior clashes with the expectation that a smaller block size should allow for a better representation of the tensor elements. We investigate this phenomenon both experimentally and theoretically, decoupling the sources of quantization error behind it. Experimentally, we analyze the distributions of several Large Language Models and identify the conditions driving the anomalous behavior. Theoretically, we lay down a framework showing remarkable agreement with experimental data from pretrained model distributions and ideal ones. Overall, we show that the anomaly is driven by the interplay between narrow tensor distributions and the limited dynamic range of the quantized scales. Based on these insights, we propose the use of FP8 unsigned E5M3 as a novel hardware-friendly format for the scales in FP4 microscaling data types. We demonstrate that UE5M3 achieves comparable performance to the conventional FP8 unsigned E4M3 scales while obviating the need of global scaling operations on weights and activations.

Fine-tuning large language models (LLMs) has achieved significant success in downstream tasks. However, as the model size continues to grow, traditional fine-tuning methods have become increasingly impractical due to their high computational and memory costs. This has motivated researchers to explore parameter-efficient and memory-friendly fine-tuning strategies to enable scalable approaches, with Low-Rank Adaptation (LoRA) standing out as a representative work. However, the LoRA update is restricted to a low-rank subspace, which results in suboptimal performance compared to the full-parameter update. Recent research has also explored memory-efficient fine-tuning LLMs using just forward passes while suffer from high variance in gradient estimation and low convergence speed. To address the issues above, we propose a new alternating optimization framework called LMAO (Low-rank and Memory-efficient Zeroth-Order Alternating Optimization), which combines the advantages of LoRA and MeZO. This method alternately updates the low-rank components and zeroth-order directions during training. By performing three forward propagations and one backward propagation, each update is full-rank, thereby reducing feature loss and enabling efficient fine-tuning under strict memory constraints. We provide theoretical guarantees on the convergence and convergence rate of this method. Empirical results demonstrate that, in experiments on multiple models (e.g., OPT, RoBERTa-large), LMAO achieves performance comparable to first-order methods. This presents a practical and scalable solution for fine-tuning large-scale models. Our source code is available at https://anonymous.4open.science/r/mlao-C2EC/.

Fine-tuning large foundation models is essential for building expert models tailored to specialized tasks and domains, but fully updating billions of parameters is computationally prohibitive. Reducing the number of trainable parameters using parameter-efficient fine-tuning is therefore crucial not only to reduce training costs but also to mitigate storage, caching, and serving overheads during deployment. Prior works, such as Singular Vectors-guided Fine-Tuning, have shown that exploiting the geometry of pre-trained weights can significantly improve parameter-efficiency, but they lack a solid theoretical foundation. In this paper, we introduce Parameter-efficient Fine-tuning with Column Space Projection (PiCa), a novel theoretically grounded PEFT method. We prove that projecting gradients onto the principal column space of pre-trained weights provides an effective inductive bias for adaptation and further enhance parameter efficiency through a novel weight-sharing strategy. Across diverse NLP and vision tasks, PiCa consistently outperforms state-of-the-art baselines under comparable or smaller parameter budgets, demonstrating both theoretical rigor and practical effectiveness.

Mixture-of-Experts (MoE) enables efficient scaling of large language models (LLMs) with sparsely activated experts during inference. To effectively deploy large MoE models on memory-constrained devices, many systems introduce expert offloading which caches a subset of experts in fast memory, leaving others on slow memory to run on CPU or load on demand. While some research has exploited the locality of expert activations, where consecutive tokens activate similar experts, the degree of this **local routing consistency** varies across models and remains understudied. In this paper, we propose two metrics to measure local routing consistency of MoE models: (1) **Segment Routing best Performance (SRP)**, which evaluates how well a fixed group of experts can cover the needs of a segment of tokens, and (2) **Segment Cache best Hit rate (SCH)**, which measures the hit rate of an expert cache utilizing a length of future information under a cache limit. We analyze 20 MoE LLMs with diverse sizes and architectures and use toy models to verify key factors related to local routing consistency. We find a strong trade-off between local routing consistency and *local* load balance, while showing that *global* load balance can coexist with local routing consistency. Meanwhile, settings like shared experts that decrease expert combination space can lead to low local routing consistency. We further reveal that domain-specialized experts contribute more to routing consistency than vocabulary-specialized ones, and that most models balance between cache effectiveness and efficiency with cache sizes approximately twice the active experts. These findings pave the way for memory-efficient MoE design and deployment without compromising inference speed.

Federated Learning enables collaborative fine-tuning of Large Language Models (LLMs) across decentralized Non-Independent and Identically Distributed (Non-IID) clients, but such models' massive parameter sizes lead to significant memory and communication challenges. This work introduces Meerkat, a sparse zeroth-order optimization (ZO) method designed for federated LLM fine-tuning. By limiting fine-tuning to a transferable, static, extremely sparse subset of parameters, Meerkat achieves remarkable communication efficiency, enabling cost-effective high-frequency synchronization. With theoretical analysis and experiments, we show that this high-frequency communication effectively mitigates Non-IID data challenges and leads to superior performance compared to full-parameter ZO. Furthermore, experiment results show that Meerkat outperforms existing sparsity baselines with better performance at the same communication frequency. To further handle Non-IID drift, Meerkat leverages traceable local updates and forms a virtual path for each client. This virtual path mechanism reveals the GradIP phenomenon: the inner products between LLM pre-training gradients maintained by server and client gradients estimated via ZO converges for extreme Non-IID clients but oscillates for IID ones. This distinct behavior provides a signal for identifying clients with extreme data heterogeneity. Using this signal, Meerkat-vp is proposed to analyze GradIP trajectories to identify extreme Non-IID clients and applies early stopping to enhance aggregated model quality. Experiments confirm that Meerkat and Meerkat-vp significantly improve the efficiency and effectiveness of ZO federated LLM fine-tuning.

Vision learners often struggle with catastrophic forgetting due to their reliance on class recognition by comparison, rather than understanding classes as compositions of representative concepts. This limitation is prevalent even in state-of-the-art continual learners with foundation models and worsens when current tasks contain few classes. Inspired by the recent success of concept-level understanding in mitigating forgetting, we design a universal framework CompSLOT to guide concept learning across diverse continual learners. Leveraging the progress of object-centric learning in parsing semantically meaningful slots from images, we tackle the challenge of learning slot extraction from ImageNet-pretrained vision transformers by analyzing meaningful concept properties. We further introduce a primitive selection and aggregation mechanism to harness concept-level image understanding. Additionally, we propose a method-agnostic self-supervision approach to distill sample-wise concept-based similarity information into the classifier, reducing reliance on incorrect or partial concepts for classification. Experiments show CompSLOT significantly enhances various continual learners and provides a universal concept-level module for the community.