Liangyu Wang

Huanyi Xie, Yubin Chen, Liangyu Wang et al.

The long-tailed distribution of sequence lengths in LLM serving and reinforcement learning (RL) sampling causes significant computational waste due to excessive padding in batched inference. Existing methods rely on auxiliary models for static length prediction, but they incur high overhead, generalize poorly, and fail in stochastic "one-to-many" sampling scenarios. We introduce a lightweight framework that reuses the main model's internal hidden states for efficient length prediction. Our framework features two core components: 1) Entropy-Guided Token Pooling (EGTP), which uses on-the-fly activations and token entropy for highly accurate static prediction with negligible cost, and 2) Progressive Length Prediction (PLP), which dynamically estimates the remaining length at each decoding step to handle stochastic generation. To validate our approach, we build and release ForeLen, a comprehensive benchmark with long-sequence, Chain-of-Thought, and RL data. On ForeLen, EGTP achieves state-of-the-art accuracy, reducing MAE by 29.16\% over the best baseline. Integrating our methods with a length-aware scheduler yields significant end-to-end throughput gains. Our work provides a new technical and evaluation baseline for efficient LLM inference.

Xinhai Wang, Shaopeng Fu, Shu Yang et al.

Suffix jailbreak attacks serve as a systematic method for red-teaming Large Language Models (LLMs) but suffer from prohibitive computational costs, as a large number of candidate suffixes need to be evaluated before identifying a jailbreak suffix. This paper presents Prefix-Shared KV Cache (PSKV), a plug-and-play inference optimization technique tailored for jailbreak suffix generation. Our method is motivated by a key observation that when performing suffix jailbreaking, while a large number of candidate prompts need to be evaluated, they share the same targeted harmful instruction as the prefix. Therefore, instead of performing redundant inference on the duplicated prefix, PSKV maintains a single KV cache for this prefix and shares it with every candidate prompt, enabling the parallel inference of diverse suffixes with minimal memory overhead. This design enables more aggressive batching strategies that would otherwise be limited by memory constraints. Extensive experiments on six widely used suffix attacks across five widely deployed LLMs demonstrate that PSKV reduces inference time by 40\% and peak memory usage by 50\%, while maintaining the original Attack Success Rate (ASR). The code has been submitted and will be released publicly.

Xinhai Wang, Shu Yang, Liangyu Wang et al.

Circuit discovery, which involves identifying sparse and task-relevant subnetworks in pre-trained language models, is a cornerstone of mechanistic interpretability. Automated Circuit Discovery (ACDC) has emerged as a pivotal methodology in circuit discovery, but its application to large language models is severely limited by computational inefficiency and prohibitively high memory requirements. Although several accelerated approaches have been proposed, they primarily rely on linear approximations to ACDC, which significantly compromises analytical faithfulness. Our proposed method for accelerating automated circuit discovery, Per Attention Head Quantization (PAHQ), takes a fundamentally different approach by optimizing the efficiency of each individual patching operation. PAHQ leverages a fundamental alignment between activation patching and mixed-precision quantization (MPQ): interpretability analysis through patching essentially performs targeted ablation studies. Therefore, we can maintain high precision exclusively for investigated components while safely reducing precision elsewhere in the network. PAHQ-accelerated ACDC reduces runtime by up to 80\% and memory consumption by up to 30\% compared to unaccelerated ACDC while maintaining faithfulness. Importantly, our method readily integrates with existing edge-based circuit discovery techniques by modifying the attention computation mechanism. This training-free approach provides a practical and novel pathway for accelerating mechanistic interpretability methods. Our code is available at https://github.com/626619403/PAHQ.

Liangyu Wang, Jie Ren, Hang Xu et al.

Fine-tuning large pre-trained LLMs generally demands extensive GPU memory. Traditional first-order optimizers like SGD encounter substantial difficulties due to increased memory requirements from storing activations and gradients during both the forward and backward phases as the model size expands. Alternatively, zeroth-order (ZO) techniques can compute gradients using just forward operations, eliminating the need to store activations. Furthermore, by leveraging CPU capabilities, it's feasible to enhance both the memory and processing power available to a single GPU. We propose a novel framework, ZO2 (Zeroth-Order Offloading), for efficient zeroth-order fine-tuning of LLMs with only limited GPU memory. Our framework dynamically shifts model parameters between the CPU and GPU as required, optimizing computation flow and maximizing GPU usage by minimizing downtime. This integration of parameter adjustments with ZO's double forward operations reduces unnecessary data movement, enhancing the fine-tuning efficacy. Additionally, our framework supports an innovative low-bit precision approach in AMP mode to streamline data exchanges between the CPU and GPU. Employing this approach allows us to fine-tune extraordinarily large models, such as the OPT-175B with more than 175 billion parameters, on a mere 18GB GPU--achievements beyond the reach of traditional methods. Moreover, our framework achieves these results with almost no additional time overhead and absolutely no accuracy loss compared to standard zeroth-order methods. ZO2's code has been open-sourced in https://github.com/liangyuwang/zo2.

Shengkun Tang, Zekun Wang, Bo Zheng et al.

Structured pruning and knowledge distillation (KD) are typical techniques for compressing large language models, but it remains unclear how they should be applied at pretraining scale, especially to recent mixture-of-experts (MoE) models. In this work, we systematically study MoE compression in large-scale pretraining, focusing on three key questions: whether pruning provides a better initialization than training from scratch, how expert compression choices affect the final model after continued training, and which training strategy is most effective. We have the following findings: First, across depth, width, and expert compression, pruning a pretrained MoE consistently outperforms training the target architecture from scratch under the same training budget. Second, different one-shot expert compression methods converge to similar final performance after large-scale continual pretraining. Motivated by this, we introduce a simple partial-preservation expert merging strategy that improves downstream performance across most benchmarks. Third, combining KD with the language modeling loss outperforms KD alone, particularly on knowledge-intensive tasks. We further propose multi-token prediction (MTP) distillation, which yields consistent gains. Finally, given the same training tokens, progressive pruning schedules outperform one-shot compression, suggesting that gradual architecture transitions lead to better optimization trajectories. Putting it all together, we compress Qwen3-Next-80A3B to a 23A2B model that retains competitive performance. These results offer practical guidance for efficient MoE compression at scale.

Liangyu Wang, Huanyi Xie, Di Wang

Fine-tuning large language models (LLMs) remains resource-intensive due to their sheer scale. While zeroth-order (ZO) optimization provides a memory-efficient alternative by eliminating backward passes, its application to multi-hundred-billion-parameter models is constrained by GPU memory and compute throughput. The ZO2 framework addresses the memory bottleneck by offloading model parameters to CPU memory and overlapping transformer block transfer with dual forward computation on a single GPU. However, ZO2 remains limited by its single-device execution and achieves modest throughput. In this work, we present DistZO2, a high-throughput, memory-efficient framework for distributed zeroth-order fine-tuning of LLMs. DistZO2 introduces three parallel strategies: (1) Perturbation Parallelism (PertP), which parallelizes the two perturbed forward passes across devices; (2) Distributed Data Parallelism (DDP), adapted to the scalar-gradient nature of ZO training; and (3) a unified 2D Parallelism design that combines PertP and DDP. To further mitigate communication bottlenecks introduced by parameter offloading, we propose a hardware-aware communication strategy that slices parameter blocks and redistributes them across GPUs via high-speed interconnects such as NVLink. DistZO2 scales zeroth-order fine-tuning to modern multi-GPU systems, preserving ZO2's memory efficiency while substantially improving training throughput. In our experiments on OPT-175B, DistZO2 achieves a 3x speedup over ZO2 with distributed computing. DistZO2's code has been open-sourced in https://github.com/liangyuwang/zo2.

Liangyu Wang, Junxiao Wang, Jie Ren et al.

As large language models (LLMs) increasingly underpin technological advancements, the privacy of their training data emerges as a critical concern. Differential Privacy (DP) serves as a rigorous mechanism to protect this data, yet its integration via Differentially Private Stochastic Gradient Descent (DP-SGD) introduces substantial challenges, primarily due to the complexities of per-sample gradient clipping. Current explicit methods, such as Opacus, necessitate extensive storage for per-sample gradients, significantly inflating memory requirements. Conversely, implicit methods like GhostClip reduce storage needs by recalculating gradients multiple times, which leads to inefficiencies due to redundant computations. This paper introduces FlashDP, an innovative cache-friendly per-layer DP-SGD that consolidates necessary operations into a single task, calculating gradients only once in a fused manner. This approach not only diminishes memory movement by up to \textbf{50\%} but also cuts down redundant computations by \textbf{20\%}, compared to previous methods. Consequently, FlashDP does not increase memory demands and achieves a \textbf{90\%} throughput compared to the Non-DP method on a four-A100 system during the pre-training of the Llama-13B model, while maintaining parity with standard per-layer clipped DP-SGD in terms of accuracy. These advancements establish FlashDP as a pivotal development for efficient and privacy-preserving training of LLMs. FlashDP's code has been open-sourced in https://github.com/kaustpradalab/flashdp.

Liangyu Wang, Huanyi Xie, Xinhai Wang et al.

Group-based reinforcement learning algorithms such as Group Reward Policy Optimization (GRPO) have proven effective for fine-tuning large language models (LLMs) with human feedback. However, generating and storing multiple responses per prompt incurs substantial memory overhead, especially as the sample group size increases, limiting scalability under constrained hardware. We propose Infinite Sampling, a framework that enables efficient and stable GRPO training by decoupling group size from GPU memory usage. It consists of: (1) micro sampling groups that decompose large groups into memory-feasible rounds; (2) continuous sampling that interleaves generation across groups to improve utilization; and (3) a length-aware scheduler combining token-conditioned sequence length prediction with a two-stage plan: global grouping via FPTAS and runtime refill via SJF. Experiments show that our Micro Sampling Groups reduce peak memory usage by over 50% compared to full-group decoding (e.g., from 21.55 GB to 10.64 GB on Qwen3-1.7B). Building on this, Infinite Sampling improves throughput by over 25% compared to the naive micro sampling group method, reducing decoding steps while maintaining full-length completions and memory usage. Our hybrid scheduling ensures efficient and stable GRPO training with larger groups under realistic GPU memory constraints.

Long Bai, Beilei Cui, Liangyu Wang et al.

Deep learning can predict depth maps and capsule ego-motion from capsule endoscopy videos, aiding in 3D scene reconstruction and lesion localization. However, the collisions of the capsule endoscopies within the gastrointestinal tract cause vibration perturbations in the training data. Existing solutions focus solely on vision-based processing, neglecting other auxiliary signals like vibrations that could reduce noise and improve performance. Therefore, we propose V$^2$-SfMLearner, a multimodal approach integrating vibration signals into vision-based depth and capsule motion estimation for monocular capsule endoscopy. We construct a multimodal capsule endoscopy dataset containing vibration and visual signals, and our artificial intelligence solution develops an unsupervised method using vision-vibration signals, effectively eliminating vibration perturbations through multimodal learning. Specifically, we carefully design a vibration network branch and a Fourier fusion module, to detect and mitigate vibration noises. The fusion framework is compatible with popular vision-only algorithms. Extensive validation on the multimodal dataset demonstrates superior performance and robustness against vision-only algorithms. Without the need for large external equipment, our V$^2$-SfMLearner has the potential for integration into clinical capsule robots, providing real-time and dependable digestive examination tools. The findings show promise for practical implementation in clinical settings, enhancing the diagnostic capabilities of doctors.

Tong Zhang, Yifan Zhao, Liangyu Wang et al.

Cross-Domain Few-Shot Learning (CDFSL) endeavors to transfer generalized knowledge from the source domain to target domains using only a minimal amount of training data, which faces a triplet of learning challenges in the meantime, i.e., semantic disjoint, large domain discrepancy, and data scarcity. Different from predominant CDFSL works focused on generalized representations, we make novel attempts to construct Intermediate Domain Proxies (IDP) with source feature embeddings as the codebook and reconstruct the target domain feature with this learned codebook. We then conduct an empirical study to explore the intrinsic attributes from perspectives of visual styles and semantic contents in intermediate domain proxies. Reaping benefits from these attributes of intermediate domains, we develop a fast domain alignment method to use these proxies as learning guidance for target domain feature transformation. With the collaborative learning of intermediate domain reconstruction and target feature transformation, our proposed model is able to surpass the state-of-the-art models by a margin on 8 cross-domain few-shot learning benchmarks.

Chenyang Ren, Yifan Jia, Huanyi Xie et al.

Sharpness-aware Minimization (SAM) improves generalization in large-scale model training by linking loss landscape geometry to generalization. However, challenges such as mislabeled noisy data and privacy concerns have emerged as significant issues. Data attribution, which identifies the contributions of specific training samples, offers a promising solution. However, directly rendering existing data influence evaluation tools such as influence functions (IF) to SAM will be inapplicable or inaccurate as SAM utilizes an inner loop to find model perturbations that maximize loss, which the outer loop then minimizes, resulting in a doubled computational structure. Additionally, this bilevel structure complicates the modeling of data influence on the parameters. In this paper, based on the IF, we develop two innovative data valuation methods for SAM, each offering unique benefits in different scenarios: the Hessian-based IF and the Gradient Trajectory-based IF. The first one provides a comprehensive estimation of data influence using a closed-form measure that relies only on the trained model weights. In contrast, the other IF for SAM utilizes gradient trajectory information during training for more accurate and efficient data assessment. Extensive experiments demonstrate their effectiveness in data evaluation and parameter tuning, with applications in identifying mislabeled data, model editing, and enhancing interpretability.