A generalist agent must continuously learn and adapt throughout its lifetime, achieving efficient forward transfer while minimizing catastrophic forgetting. Previous work within the dominant pretrain-then-finetune paradigm has explored parameter-efficient fine-tuning for single-task adaptation, effectively steering a frozen pretrained model with a small number of parameters. However, in the context of lifelong learning, these methods rely on the impractical assumption of a test-time task identifier and restrict knowledge sharing among isolated adapters. To address these limitations, we propose Dynamic Mixture of Progressive Parameter-Efficient Expert Library (DMPEL) for lifelong robot learning. DMPEL progressively learn a low-rank expert library and employs a lightweight router to dynamically combine experts into an end-to-end policy, facilitating flexible behavior during lifelong adaptation. Moreover, by leveraging the modular structure of the fine-tuned parameters, we introduce coefficient replay to guide the router in accurately retrieving frozen experts for previously encountered tasks, thereby mitigating catastrophic forgetting. This method is significantly more storage- and computationally-efficient than applying demonstration replay to the entire policy. Extensive experiments on the lifelong manipulation benchmark LIBERO demonstrate that our framework outperforms state-of-the-art lifelong learning methods in success rates across continual adaptation, while utilizing minimal trainable parameters and storage.
Vision-Language Models (VLMs) have demonstrated remarkable capabilities in cross-modal understanding and generation by integrating visual and textual information. While instruction tuning and parameter-efficient fine-tuning methods have substantially improved the generalization of VLMs, most existing approaches rely on centralized training, posing challenges for deployment in domains with strict privacy requirements like healthcare. Recent efforts have introduced Federated Learning (FL) into VLM fine-tuning to address these privacy concerns, yet comprehensive benchmarks for evaluating federated fine-tuning strategies, model architectures, and task generalization remain lacking. In this work, we present \textbf{FedVLMBench}, the first systematic benchmark for federated fine-tuning of VLMs. FedVLMBench integrates two mainstream VLM architectures (encoder-based and encoder-free), four fine-tuning strategies, five FL algorithms, six multimodal datasets spanning four cross-domain single-task scenarios and two cross-domain multitask settings, covering four distinct downstream task categories. Through extensive experiments, we uncover key insights into the interplay between VLM architectures, fine-tuning strategies, data heterogeneity, and multi-task federated optimization. Notably, we find that a 2-layer multilayer perceptron (MLP) connector with concurrent connector and LLM tuning emerges as the optimal configuration for encoder-based VLMs in FL. Furthermore, current FL methods exhibit significantly higher sensitivity to data heterogeneity in vision-centric tasks than text-centric ones, across both encoder-free and encoder-based VLM architectures. Our benchmark provides essential tools, datasets, and empirical guidance for the research community, offering a standardized platform to advance privacy-preserving, federated training of multimodal foundation models.
Integrating Low-Rank Adaptation (LoRA) into federated learning offers a promising solution for parameter-efficient fine-tuning of Large Language Models (LLMs) without sharing local data. However, several methods designed for federated LoRA present significant challenges in balancing communication efficiency, model accuracy, and computational cost, particularly among heterogeneous clients. These methods either rely on simplistic averaging of local adapters, which introduces aggregation noise, require transmitting large stacked local adapters, leading to poor communication efficiency, or necessitate reconstructing memory-dense global weight-update matrix and performing computationally expensive decomposition to design client-specific low-rank adapters. In this work, we propose FLoRIST, a federated fine-tuning framework that achieves mathematically accurate aggregation without incurring high communication or computational overhead. Instead of constructing the full global weight-update matrix at the server, FLoRIST employs an efficient decomposition pipeline by performing singular value decomposition on stacked local adapters separately. This approach operates within a compact intermediate space to represent the accumulated information from local LoRAs. We introduce tunable singular value thresholding for server-side optimal rank selection to construct a pair of global low-rank adapters shared by all clients. Extensive empirical evaluations across multiple datasets and LLMs demonstrate that FLoRIST consistently strikes the best balance between superior communication efficiency and competitive performance in both homogeneous and heterogeneous setups.
Despite the remarkable reasoning performance, eliciting the long chain-of-thought (CoT) ability in large language models (LLMs) typically requires costly reinforcement learning or supervised fine-tuning on high-quality distilled data. We investigate the internal mechanisms behind this capability and show that a small set of high-impact activations in the last few layers largely governs long-form reasoning attributes, such as output length and self-reflection. By simply amplifying these activations and inserting "wait" tokens, we can invoke the long CoT ability without any training, resulting in significantly increased self-reflection rates and accuracy. Moreover, we find that the activation dynamics follow predictable trajectories, with a sharp rise after special tokens and a subsequent exponential decay. Building on these insights, we introduce a general training-free activation control technique. It leverages a few contrastive examples to identify key activations, and employs simple analytic functions to modulate their values at inference time to elicit long CoTs. Extensive experiments confirm the effectiveness of our method in efficiently eliciting long CoT reasoning in LLMs and improving their performance. Additionally, we propose a parameter-efficient fine-tuning method that trains only a last-layer activation amplification module and a few LoRA layers, outperforming full LoRA fine-tuning on reasoning benchmarks with significantly fewer parameters. Our code and data are publicly released.
We introduce Autoregressive Retrieval Augmentation (AR-RAG), a novel paradigm that enhances image generation by autoregressively incorporating knearest neighbor retrievals at the patch level. Unlike prior methods that perform a single, static retrieval before generation and condition the entire generation on fixed reference images, AR-RAG performs context-aware retrievals at each generation step, using prior-generated patches as queries to retrieve and incorporate the most relevant patch-level visual references, enabling the model to respond to evolving generation needs while avoiding limitations (e.g., over-copying, stylistic bias, etc.) prevalent in existing methods. To realize AR-RAG, we propose two parallel frameworks: (1) Distribution-Augmentation in Decoding (DAiD), a training-free plug-and-use decoding strategy that directly merges the distribution of model-predicted patches with the distribution of retrieved patches, and (2) Feature-Augmentation in Decoding (FAiD), a parameter-efficient fine-tuning method that progressively smooths the features of retrieved patches via multi-scale convolution operations and leverages them to augment the image generation process. We validate the effectiveness of AR-RAG on widely adopted benchmarks, including Midjourney-30K, GenEval and DPG-Bench, demonstrating significant performance gains over state-of-the-art image generation models.
Benefited from image-text contrastive learning, pre-trained vision-language models, e.g., CLIP, allow to direct leverage texts as images (TaI) for parameter-efficient fine-tuning (PEFT). While CLIP is capable of making image features to be similar to the corresponding text features, the modality gap remains a nontrivial issue and limits image recognition performance of TaI. Using multi-label image recognition (MLR) as an example, we present a novel method, called T2I-PAL to tackle the modality gap issue when using only text captions for PEFT. The core design of T2I-PAL is to leverage pre-trained text-to-image generation models to generate photo-realistic and diverse images from text captions, thereby reducing the modality gap. To further enhance MLR, T2I-PAL incorporates a class-wise heatmap and learnable prototypes. This aggregates local similarities, making the representation of local visual features more robust and informative for multi-label recognition. For better PEFT, we further combine both prompt tuning and adapter learning to enhance classification performance. T2I-PAL offers significant advantages: it eliminates the need for fully semantically annotated training images, thereby reducing the manual annotation workload, and it preserves the intrinsic mode of the CLIP model, allowing for seamless integration with any existing CLIP framework. Extensive experiments on multiple benchmarks, including MS-COCO, VOC2007, and NUS-WIDE, show that our T2I-PAL can boost recognition performance by 3.47% in average above the top-ranked state-of-the-art methods.
Federated fine-tuning (FedFT) of large language models (LLMs) has emerged as a promising solution for adapting models to distributed data environments while ensuring data privacy. Existing FedFT methods predominantly utilize parameter-efficient fine-tuning (PEFT) techniques to reduce communication and computation overhead. However, they often fail to adequately address the catastrophic forgetting, a critical challenge arising from continual adaptation in distributed environments. The traditional centralized fine-tuning methods, which are not designed for the heterogeneous and privacy-constrained nature of federated environments, struggle to mitigate this issue effectively. Moreover, the challenge is further exacerbated by significant variation in data distributions and device capabilities across clients, which leads to intensified forgetting and degraded model generalization. To tackle these issues, we propose FedBE, a novel FedFT framework that integrates an adaptive transformer block expansion mechanism with a dynamic trainable-block allocation strategy. Specifically, FedBE expands trainable blocks within the model architecture, structurally separating newly learned task-specific knowledge from the original pre-trained representations. Additionally, FedBE dynamically assigns these trainable blocks to clients based on their data distributions and computational capabilities. This enables the framework to better accommodate heterogeneous federated environments and enhances the generalization ability of the model.Extensive experiments show that compared with existing federated fine-tuning methods, FedBE achieves 12-74% higher accuracy retention on general tasks after fine-tuning and a model convergence acceleration ratio of 1.9-3.1x without degrading the accuracy of downstream tasks.
Self-supervised representation learning for point cloud has demonstrated effectiveness in improving pre-trained model performance across diverse tasks. However, as pre-trained models grow in complexity, fully fine-tuning them for downstream applications demands substantial computational and storage resources. Parameter-efficient fine-tuning (PEFT) methods offer a promising solution to mitigate these resource requirements, yet most current approaches rely on complex adapter and prompt mechanisms that increase tunable parameters. In this paper, we propose PointLoRA, a simple yet effective method that combines low-rank adaptation (LoRA) with multi-scale token selection to efficiently fine-tune point cloud models. Our approach embeds LoRA layers within the most parameter-intensive components of point cloud transformers, reducing the need for tunable parameters while enhancing global feature capture. Additionally, multi-scale token selection extracts critical local information to serve as prompts for downstream fine-tuning, effectively complementing the global context captured by LoRA. The experimental results across various pre-trained models and three challenging public datasets demonstrate that our approach achieves competitive performance with only 3.43% of the trainable parameters, making it highly effective for resource-constrained applications. Source code is available at: https://github.com/songw-zju/PointLoRA.
Training deep neural networks (DNNs) requires significantly more computation and memory than inference, making runtime adaptation of DNNs challenging on resource-limited IoT platforms. We propose InstantFT, an FPGA-based method for ultra-fast CNN fine-tuning on IoT devices, by optimizing the forward and backward computations in parameter-efficient fine-tuning (PEFT). Experiments on datasets with concept drift demonstrate that InstantFT fine-tunes a pre-trained CNN 17.4x faster than existing Low-Rank Adaptation (LoRA)-based approaches, while achieving comparable accuracy. Our FPGA-based InstantFT reduces the fine-tuning time to just 0.36s and improves energy-efficiency by 16.3x, enabling on-the-fly adaptation of CNNs to non-stationary data distributions.
Low-Rank Adaptation (LoRA) has become the de facto parameter-efficient fine-tuning (PEFT) method for large language models (LLMs) by constraining weight updates to low-rank matrices. Recent works such as Tied-LoRA, VeRA, and VB-LoRA push efficiency further by introducing additional constraints to reduce the trainable parameter space. In this paper, we show that the parameter space reduction strategies employed by these LoRA variants can be formulated within a unified framework, Uni-LoRA, where the LoRA parameter space, flattened as a high-dimensional vector space $R^D$, can be reconstructed through a projection from a subspace R^d, with $d \ll D$. We demonstrate that the fundamental difference among various LoRA methods lies in the choice of the projection matrix, $P \in R^{D \times d}$.Most existing LoRA variants rely on layer-wise or structure-specific projections that limit cross-layer parameter sharing, thereby compromising parameter efficiency. In light of this, we introduce an efficient and theoretically grounded projection matrix that is isometric, enabling global parameter sharing and reducing computation overhead. Furthermore, under the unified view of Uni-LoRA, this design requires only a single trainable vector to reconstruct LoRA parameters for the entire LLM - making Uni-LoRA both a unified framework and a "one-vector-only" solution. Extensive experiments on GLUE, mathematical reasoning, and instruction tuning benchmarks demonstrate that Uni-LoRA achieves state-of-the-art parameter efficiency while outperforming or matching prior approaches in predictive performance.
Fine-tuning Large Language Models (LLMs) is essential for adapting pre-trained models to downstream tasks. Yet traditional first-order optimizers such as Stochastic Gradient Descent (SGD) and Adam incur prohibitive memory and computational costs that scale poorly with model size. In this paper, we investigate zero-order (ZO) optimization methods as a memory- and compute-efficient alternative, particularly in the context of parameter-efficient fine-tuning techniques like LoRA. We propose $\texttt{JAGUAR SignSGD}$, a ZO momentum-based algorithm that extends ZO SignSGD, requiring the same number of parameters as the standard ZO SGD and only $\mathcal{O}(1)$ function evaluations per iteration. To the best of our knowledge, this is the first study to establish rigorous convergence guarantees for SignSGD in the stochastic ZO case. We further propose $\texttt{JAGUAR Muon}$, a novel ZO extension of the Muon optimizer that leverages the matrix structure of model parameters, and we provide its convergence rate under arbitrary stochastic noise. Through extensive experiments on challenging LLM fine-tuning benchmarks, we demonstrate that the proposed algorithms meet or exceed the convergence quality of standard first-order methods, achieving significant memory reduction. Our theoretical and empirical results establish new ZO optimization methods as a practical and theoretically grounded approach for resource-constrained LLM adaptation. Our code is available at https://github.com/brain-mmo-lab/ZO_LLM
We introduce ProLoRA, enabling zero-shot adaptation of parameter-efficient fine-tuning in text-to-image diffusion models. ProLoRA transfers pre-trained low-rank adjustments (e.g., LoRA) from a source to a target model without additional training data. This overcomes the limitations of traditional methods that require retraining when switching base models, often challenging due to data constraints. ProLoRA achieves this via projection of source adjustments into the target model's weight space, leveraging subspace and null space similarities and selectively targeting aligned layers. Evaluations on established text-to-image models demonstrate successful knowledge transfer and comparable performance without retraining.
The rise of Large Language Models (LLMs) has driven progress in reasoning tasks -- from program synthesis to scientific hypothesis generation -- yet their ability to handle ranked preferences and structured algorithms in combinatorial domains remains underexplored. We study matching markets, a core framework behind applications like resource allocation and ride-sharing, which require reconciling individual ranked preferences to ensure stable outcomes. We evaluate several state-of-the-art models on a hierarchy of preference-based reasoning tasks -- ranging from stable-matching generation to instability detection, instability resolution, and fine-grained preference queries -- to systematically expose their logical and algorithmic limitations in handling ranked inputs. Surprisingly, even top-performing models with advanced reasoning struggle to resolve instability in large markets, often failing to identify blocking pairs or execute algorithms iteratively. We further show that parameter-efficient fine-tuning (LoRA) significantly improves performance in small markets, but fails to bring about a similar improvement on large instances, suggesting the need for more sophisticated strategies to improve LLMs' reasoning with larger-context inputs.
Parameter Efficient Fine-Tuning (PEFT) has become the de-facto approach in adapting Large Language Models (LLMs) for downstream tasks in Natural Language Processing. However, its adoption in privacy-preserving distributed learning frameworks, such as Federated Learning (FL), remains relatively limited. This is mainly due to challenges specific to FL, such as resource-constrained devices and diverse data distributions among clients. In this paper, we propose an efficient method to perform PEFT within the FL framework for Multi-Head Attention (MHA) based language models. We address the challenges through head pruning, a novel head-specific weighted aggregation mechanism, and a client selection strategy. Head pruning minimizes training complexity within the clients, guided by the importance score computed based on the confidence of the attention head. Weighted aggregation of heads ensures the global model captures crucial updates from diverse clients complementing our client selection strategy. We show results on the MultiNLI benchmark along with 20 Newsgroups, XL-Sum, and E2E NLG datasets. We use the MultiNLI dataset and T5-small model with LoRA as our PEFT method, attaining sparsity levels of up to 90%, resulting in a communication advantage of up to 1.8x and a reduction in training OPs of 3.9x while maintaining the accuracy drop under 2%.
Federated learning (FL) allows multiple data-owners to collaboratively train machine learning models by exchanging local gradients, while keeping their private data on-device. To simultaneously enhance privacy and training efficiency, recently parameter-efficient fine-tuning (PEFT) of large-scale pretrained models has gained substantial attention in FL. While keeping a pretrained (backbone) model frozen, each user fine-tunes only a few lightweight modules to be used in conjunction, to fit specific downstream applications. Accordingly, only the gradients with respect to these lightweight modules are shared with the server. In this work, we investigate how the privacy of the fine-tuning data of the users can be compromised via a malicious design of the pretrained model and trainable adapter modules. We demonstrate gradient inversion attacks on a popular PEFT mechanism, the adapter, which allow an attacker to reconstruct local data samples of a target user, using only the accessible adapter gradients. Via extensive experiments, we demonstrate that a large batch of fine-tuning images can be retrieved with high fidelity. Our attack highlights the need for privacy-preserving mechanisms for PEFT, while opening up several future directions. Our code is available at https://github.com/info-ucr/PEFTLeak.
This study uses Jordanian law as a case study to explore the fine-tuning of the Llama-3.1 large language model for Arabic question-answering. Two versions of the model- Llama-3.1-8B-bnb-4bit and Llama-3.1-8B-Instruct-bnb-4bit-were fine-tuned using parameter-efficient fine-tuning (PEFT) with LoRA adapters and 4-bit quantized models, leveraging the Unsloth framework for accelerated and resource-efficient training. A custom dataset of 6000 legal question-answer pairs was curated from Jordanian laws and formatted into structured prompts. Performance was evaluated using the BLEU and the ROUGE metrics to compare the fine-tuned models to their respective base versions. Results demonstrated improved legal reasoning and accuracy while achieving resource efficiency through quantization and optimized fine-tuning strategies. This work underscores the potential of adapting large language models for Arabic legal domains and highlights effective techniques for fine-tuning domain-specific tasks.
Class-Incremental Learning (CIL) aims to learn new classes sequentially while retaining the knowledge of previously learned classes. Recently, pre-trained models (PTMs) combined with parameter-efficient fine-tuning (PEFT) have shown remarkable performance in rehearsal-free CIL without requiring exemplars from previous tasks. However, existing adapter-based methods, which incorporate lightweight learnable modules into PTMs for CIL, create new adapters for each new task, leading to both parameter redundancy and failure to leverage shared knowledge across tasks. In this work, we propose ContinuaL Low-Rank Adaptation (CL-LoRA), which introduces a novel dual-adapter architecture combining \textbf{task-shared adapters} to learn cross-task knowledge and \textbf{task-specific adapters} to capture unique features of each new task. Specifically, the shared adapters utilize random orthogonal matrices and leverage knowledge distillation with gradient reassignment to preserve essential shared knowledge. In addition, we introduce learnable block-wise weights for task-specific adapters, which mitigate inter-task interference while maintaining the model's plasticity. We demonstrate CL-LoRA consistently achieves promising performance under multiple benchmarks with reduced training and inference computation, establishing a more efficient and scalable paradigm for continual learning with pre-trained models.
Model editing techniques, particularly task arithmetic using task vectors, have shown promise in efficiently modifying pre-trained models through arithmetic operations like task addition and negation. Despite computational advantages, these methods may inadvertently affect model fairness, creating risks in sensitive applications like hate speech detection. However, the fairness implications of task arithmetic remain largely unexplored, presenting a critical gap in the existing literature. We systematically examine how manipulating task vectors affects fairness metrics, including Demographic Parity and Equalized Odds. To rigorously assess these effects, we benchmark task arithmetic against full fine-tuning, a costly but widely used baseline, and Low-Rank Adaptation (LoRA), a prevalent parameter-efficient fine-tuning method. Additionally, we explore merging task vectors from models fine-tuned on demographic subgroups vulnerable to hate speech, investigating whether fairness outcomes can be controlled by adjusting task vector coefficients, potentially enabling tailored model behavior. Our results offer novel insights into the fairness implications of model editing and establish a foundation for fairness-aware and responsible model editing practices.
Current parameter-efficient fine-tuning methods for adapting pre-trained language models to downstream tasks are susceptible to interference from noisy data. Conventional noise-handling approaches either rely on laborious data pre-processing or employ model architecture modifications prone to error accumulation. In contrast to existing noise-process paradigms, we propose a noise-robust adaptation method via asymmetric LoRA poisoning experts (LoPE), a novel framework that enhances model robustness to noise only with generated noisy data. Drawing inspiration from the mixture-of-experts architecture, LoPE strategically integrates a dedicated poisoning expert in an asymmetric LoRA configuration. Through a two-stage paradigm, LoPE performs noise injection on the poisoning expert during fine-tuning to enhance its noise discrimination and processing ability. During inference, we selectively mask the dedicated poisoning expert to leverage purified knowledge acquired by normal experts for noise-robust output. Extensive experiments demonstrate that LoPE achieves strong performance and robustness purely through the low-cost noise injection, which completely eliminates the requirement of data cleaning.
Pruning has recently been widely adopted to reduce the parameter scale and improve the inference efficiency of Large Language Models (LLMs). Mainstream pruning techniques often rely on uniform layerwise pruning strategies, which can lead to severe performance degradation at high sparsity levels. Recognizing the varying contributions of different layers in LLMs, recent studies have shifted their focus toward non-uniform layerwise pruning. However, these approaches often rely on pre-defined values, which can result in suboptimal performance. To overcome these limitations, we propose a novel method called Dynamic Layerwise Pruning (DLP). This approach adaptively determines the relative importance of each layer by integrating model weights with input activation information, assigning pruning rates accordingly. Experimental results show that DLP effectively preserves model performance at high sparsity levels across multiple LLMs. Specifically, at 70% sparsity, DLP reduces the perplexity of LLaMA2-7B by 7.79 and improves the average accuracy by 2.7% compared to state-of-the-art methods. Moreover, DLP is compatible with various existing LLM compression techniques and can be seamlessly integrated into Parameter-Efficient Fine-Tuning (PEFT). We release the code at https://github.com/ironartisan/DLP to facilitate future research.