Searcharxiv⌕ Search

arXiv subjects

Shuo Yang

Publications and source records attributed to Shuo Yang.

At least 73 records · Page 4Linked to original sources
arXiv2026

Optimizing Few-Step Generation with Adaptive Matching Distillation

Distribution Matching Distillation (DMD) is a powerful acceleration paradigm, yet its stability is often compromised in Forbidden Zone, regions where the real teacher provides unreliable guidance while the fake teacher exerts insufficient repulsive force. In this work, we propose a unified optimization framework that reinterprets prior art as implicit strategies to avoid these corrupted regions. Based on this insight, we introduce Adaptive Matching Distillation (AMD), a self-correcting mechanism that utilizes reward proxies to explicitly detect and escape Forbidden Zones. AMD dynamically prioritizes corrective gradients via structural signal decomposition and introduces Repulsive Landscape Sharpening to enforce steep energy barriers against failure mode collapse. Extensive experiments across image and video generation tasks (e.g., SDXL, Wan2.1) and rigorous benchmarks (e.g., VBench, GenEval) demonstrate that AMD significantly enhances sample fidelity and training robustness. For instance, AMD improves the HPSv2 score on SDXL from 30.64 to 31.25, outperforming state-of-the-art baselines. These findings validate that explicitly rectifying optimization trajectories within Forbidden Zones is essential for pushing the performance ceiling of few-step generative models.

cs.CV↗
arXiv2026

MotionWAM: Towards Foundation World Action Models for Real-Time Humanoid Loco-Manipulation

World Action Models (WAMs) couple a video dynamics prior to the policy and have shown encouraging results on tabletop manipulation, but iterative denoising over high-dimensional video-action latents leaves them too slow for real-time humanoid loco-manipulation. The problem is compounded by the dominant hierarchical paradigm, in which a high-level manipulation policy controls only the upper body while a low-level controller tracks coarse base commands -- placing upper and lower body in inconsistent action spaces and reducing the legs to balance-preserving locomotion. We present MotionWAM, a real-time WAM that drives autonomous humanoid loco-manipulation from a single egocentric camera by conditioning the policy on the intermediate denoising features of a video world model. MotionWAM replaces the upper-lower split with a unified motion latent and predicts whole-body motion tokens that jointly cover locomotion, torso motion, height regulation, foot interaction, and hand manipulation in a single action space. A three-stage learning framework progressively adapts the video world model to egocentric visual dynamics and to the target humanoid embodiment. On nine real-world Unitree G1 tasks, MotionWAM runs in real time, substantially outperforms Vision-Language-Action (VLA) baselines fine-tuned on the same demonstrations by over 30% in overall success rate, and executes task-driven foot interaction that decoupled upper-lower policies cannot reach. Our results suggest that video-pretrained WAMs can be lifted from tabletop manipulation to coordinated, human-like whole-body humanoid control.

cs.RO↗
arXiv2026

BlendServe: Optimizing Offline Inference for Auto-regressive Large Models with Resource-aware Batching

Offline batch inference, which leverages the flexibility of request batching to achieve higher throughput and lower costs, is becoming more popular for latency-insensitive applications. Meanwhile, recent progress in model capability and modality makes requests more diverse in compute and memory demands, creating unique opportunities for throughput improvement by resource overlapping. However, a request schedule that maximizes resource overlapping can conflict with the schedule that maximizes prefix sharing, a widely-used performance optimization, causing sub-optimal inference throughput. We present BlendServe, a system that maximizes resource utilization of offline batch inference by combining the benefits of resource overlapping and prefix sharing using a resource-aware prefix tree. BlendServe exploits the relaxed latency requirements in offline batch inference to reorder and overlap requests with varied resource demands while ensuring high prefix sharing. We evaluate BlendServe on a variety of synthetic multi-modal workloads and show that it provides up to $1.44\times$ throughput boost compared to widely-used industry standards, vLLM and SGLang.

cs.LG↗
arXiv2026

$E_\infty^{1,2}$-type Lieb-Schultz-Mattis anomalies, deconfined quantum critical points, and non-invertible symmetry breaking

We study deconfined quantum critical points (DQCP) associated with Lieb-Schultz-Mattis (LSM) anomalies in one-dimensional spin chains. Our starting point is a structural characterization of the LSM anomaly in the Lyndon-Hochschild-Serre spectral sequence: $ω_{\mathrm{LSM}}\in E_\infty^{1,2}= H^1(\mathbb Z_{\mathrm{trans}},H^2(G_{\mathrm{int}},\mathrm{U}(1)))\subseteq H^3(G_{\mathrm{int}}\rtimes_ρ\mathbb Z_{\mathrm{trans}},\mathrm{U}(1))$. Physically, this class decorates a translation defect with a projective representation of the internal symmetry $G_\mathrm{int}$. We show that gauging the internal symmetry in the presence of an $E_\infty^{1,2}$-type anomaly necessarily produces a non-invertible dual symmetry. This gives a general mechanism for type-II DQCP: in contrast to type-I examples with $E_\infty^{2,1}$-type anomalies which are dual to ordinary group-like symmetry breaking, type-II transitions are dual to spontaneous breaking of a non-invertible symmetry. We illustrate the mechanism using a spin-$1/2$ chain with an anomalous $D_8$ LSM symmetry. We construct a dimer-to-ferromagnet DQCP candidate, provide numerical evidence for a critical theory with central charge $c\approx 1$, and show, using both category theory and explicit lattice constructions, that gauging the internal symmetry yields the non-invertible $\mathrm{Rep}(H_8)$ dual symmetry.

cond-mat.str-el↗
arXiv2026

Inclusion-of-Thoughts: Mitigating Preference Instability via Purifying the Decision Space

Multiple-choice questions (MCQs) are widely used to evaluate large language models (LLMs). However, LLMs remain vulnerable to the presence of plausible distractors. This often diverts attention toward irrelevant choices, resulting in unstable oscillation between correct and incorrect answers. In this paper, we propose Inclusion-of-Thoughts (IoT), a progressive self-filtering strategy that is designed to mitigate this cognitive load (i.e., instability of model preferences under the presence of distractors) and enable the model to focus more effectively on plausible answers. Our method operates to reconstruct the MCQ using only plausible option choices, providing a controlled setting for examining comparative judgements and therefore the stability of the model's internal reasoning under perturbation. By explicitly documenting this filtering process, IoT also enhances the transparency and interpretability of the model's decision-making. Extensive empirical evaluation demonstrates that IoT substantially boosts chain-of-thought performance across a range of arithmetic, commonsense reasoning, and educational benchmarks with minimal computational overhead.

cs.CL↗
arXiv2026

DLLG: Dynamic Logit-Level Gating of LLM Experts

Leveraging multiple specialized LLMs can combine complementary strengths, but existing approaches trade adaptability for stability: routing commits prematurely, heuristic ensembling depends on fragile proxies, and parameter merging introduces interference. We propose DLLG (Dynamic Logit-Level Gating), a dynamic logit-level ensembling framework that learns token-level expert fusion from sparse response-level supervision. A lightweight gating module predicts step-wise fusion weights, linking trajectory-level correctness to generation without token-level labels or expert retraining. Across diverse reasoning and code benchmarks, DLLG consistently outperforms strong routing, heuristic ensembling, and parameter-merging baselines across model scales, highlighting learned logit-level fusion as a robust and scalable paradigm for integrating specialized experts.

cs.CL↗
arXiv2026

Consolidating Rewarded Perturbations for LLM Post-Training

Post-training of language models is commonly framed as a sample-score-update loop implemented by gradient descent. A recent line of work, exemplified by RandOpt, relocates this loop to weight space, sampling Gaussian perturbations around a pretrained model and ensembling the top-K rewarded specialists at inference. While competitive with PPO and GRPO under matched training compute, this prediction-level ensemble incurs K forward passes per test example and does not extend cleanly to free-form generation. We ask whether the rewarded population can instead be folded into a single deployable model, replacing the inference-time ensemble with one consolidated update. A split-half analysis over 25 model-task pairs reveals reproducible low-rank structure in every case. We turn this geometry into CoRP (Consolidating Rewarded Perturbations), a gradient-free operator that combines reward-weighted aggregation, compatibility-aware reweighting, and a held-out validation gate, with no gradient flowing through the language model. Across five language models from 0.5B to 8B and five tasks covering math, code, and creative writing, CoRP improves the base model by 8.1 points on average. Using one tenth of RandOpt's perturbation budget, CoRP exceeds single-inference RandOpt by 6.5 points and recovers more than half of the gain of the 50-pass majority-vote ensemble, at one forward pass per test example.

cs.CL↗
arXiv2026

EvoMAS: Evolutionary Generation of Multi-Agent Systems

Large language model (LLM)-based multi-agent systems (MAS) show strong promise for complex reasoning, planning, and tool-augmented tasks, but designing effective MAS architectures remains labor-intensive, brittle, and hard to generalize. Existing automatic MAS generation methods either rely on code generation, which often leads to executability and robustness failures, or impose rigid architectural templates that limit expressiveness and adaptability. We propose Evolutionary Generation of Multi-Agent Systems (EvoMAS), which formulates MAS generation as structured configuration generation. EvoMAS performs evolutionary generation in configuration space. Specifically, EvoMAS selects initial configurations from a pool, applies feedback-conditioned mutation and crossover guided by execution traces, and iteratively refines both the candidate pool and an experience memory. We evaluate EvoMAS on diverse benchmarks, including BBEH, SWE-Bench, and WorkBench, covering reasoning, software engineering, and tool-use tasks. EvoMAS consistently improves task performance over both human-designed MAS and prior automatic MAS generation methods, while producing generated systems with higher executability and runtime robustness. EvoMAS outperforms the agent evolution method EvoAgent by +10.5 points on BBEH reasoning and +7.1 points on WorkBench. With Claude-4.5-Sonnet, EvoMAS also reaches 79.1% on SWE-Bench-Verified, matching the top of the leaderboard. Code is available at https://github.com/amazon-science/EvoMAS

cs.LG↗
arXiv2026

Bridging the Semantic-Action Gap in Visual Token Pruning for Efficient VLA Inference

Vision-Language-Action (VLA) models have shown great potential for embodied AI by integrating visual perception, language understanding, and action execution. In real-time deployment, these models must process continuous visual streams, incurring substantial computational overhead. Visual token pruning -- a mainstream technique for accelerating Vision-Language Models (VLMs) by retaining salient tokens while discarding redundant ones -- offers a natural candidate solution to this challenge. However, directly applying VLM-oriented pruning methods to VLA inference can cause severe degradation in manipulation performance. Our analysis attributes this degradation to a key mismatch: VLA inference exhibits distinct attention patterns between the vision-language prefill stage and the action-decode stage, so pruning based only on context-prefill semantic salience is biased toward semantic cues and may remove action-critical visual tokens. Motivated by this observation, we propose VLA-Pruner, an effective plug-and-play token pruning method grounded in the visual requirements of VLA inference, further exploiting the temporal continuity of robot manipulation. Specifically, VLA-Pruner estimates visual-token importance from both semantic prefilling and temporally smoothed action relevance, and then applies a Combine-then-Filter strategy to retain compact, non-redundant tokens under the compute budget. Experiments show that VLA-Pruner outperforms state-of-the-art approaches across multiple VLA architectures, achieving up to 1.99x speedup with comparable manipulation quality.

cs.CV↗
arXiv2026

LATTE: Forecasting Peer Anchored Preference Trajectories for Personalized LLM Generation

Personalized generation with frozen large language models requires a conditioning signal that is both compact and current. Existing personalization methods typically retrieve or summarize user histories in text, or compress them into static latent profiles and soft prompts. These approaches are efficient, but they treat a user's past behavior as an aggregate profile and therefore mix stable identity, recent drift, and item content in the same representation. We propose LAtent Trajectory Tracking and Extrapolation (LATTE), a framework that represents personalization as forecasting a peer anchored relative preference state. For each historical session, LATTE subtracts a time masked baseline formed from comparable users who responded to the same item, producing a state that measures how the target user differs from peers under a shared item context. A lightweight sequence predictor then forecasts the next state in this trajectory, and a State to Token Bridge injects the forecast into a frozen instruction tuned LLM through a single anchored soft token. We provide a latent factor analysis showing when peer anchoring cancels shared item variation and why temporal forecasting trades off stale averages against noisy recent states. Experiments on Amazon Reviews 2023 and MemoryCD show that LATTE consistently outperforms retrieval, summary memory, static latent profiles, difference aware latent profiles, and soft prompt compression baselines. On Amazon Reviews 2023, LATTE improves average ROUGE-L from 0.219 for a static latent profile and 0.245 for the strongest added latent compression baseline to 0.259. Additional pairwise comparisons and diagnostic analyses suggest that the improvement is mainly due to forecasting user-specific trajectory information, rather than merely adding a soft prompt interface.

cs.CL↗
arXiv2026

vAttention: Verified Sparse Attention

State-of-the-art sparse attention methods for reducing decoding latency fall into two main categories: approximate top-$k$ (and its extension, top-$p$) and recently introduced sampling-based estimation. However, these approaches are fundamentally limited in their ability to approximate full attention: they fail to provide consistent approximations across heads and query vectors and, most critically, lack guarantees on approximation quality, limiting their practical deployment. We observe that top-$k$ and random sampling are complementary: top-$k$ performs well when attention scores are dominated by a few tokens, whereas random sampling provides better estimates when attention scores are relatively uniform. Building on this insight and leveraging the statistical guarantees of sampling, we introduce vAttention, the first practical sparse attention mechanism with user-specified $(ε, δ)$ guarantees on approximation accuracy (thus, "verified"). These guarantees make vAttention a compelling step toward practical, reliable deployment of sparse attention at scale. By unifying top-$k$ and sampling, vAttention outperforms both individually, delivering a superior quality-efficiency trade-off. Our experiments show that vAttention significantly improves the quality of sparse attention (e.g., $\sim$4.5 percentage points for Llama 3.1 8B Instruct and DeepSeek-R1-Distill-Llama-8B on RULER-HARD), and effectively bridges the gap between full and sparse attention (e.g., across datasets, it matches full model quality with up to 20x sparsity). We also demonstrate that it can be deployed in reasoning scenarios to achieve fast decoding without compromising model quality (e.g., vAttention achieves full model quality on AIME2024 at 10x sparsity with up to 32K token generations). Code: https://github.com/skylight-org/sparse-attention-hub. Webpage: https://sky-light.eecs.berkeley.edu.

cs.LG↗
arXiv2026

TapSampling: Inference-Time Sampling with a Task-Progress-Understanding Verifier for Robotic Manipulation

Existing embodied control research demonstrates remarkable performance improvements by scaling training data and model size. We instead explore inference-time strategy as an alternative axis. Non-deterministic generative models, such as diffusion and autoregressive models, have been widely adopted in the field of embodied control. However, the single-shot inference paradigm limits their performance. In this paper, we propose \textbf{TapSampling}, a plug-and-play framework for inference-time sampling. First, we introduce an Action-VAE that represents actions in a low-dimensional latent space by mapping policy-generated initial actions into a compressed posterior distribution, from which any number of latent samples can be drawn and decoded into candidate actions that approximate the true action distribution. Second, we formulate action verification as task-progress outcome prediction, using the intrinsic sequential structure of robotic datasets to train a semantically grounded verifier for interpretable action selection. Furthermore, TapSampling is a policy-agnostic framework. Extensive experiments in both simulated and real-world environments demonstrate that our method substantially improves multiple generalist policies without further policy finetuning. Code and models are available at the project page.

cs.RO↗
arXiv2026

Beyond the Target: From Imitation to Collaboration in Speculative Decoding

Speculative decoding (SPD) accelerates large language model (LLM) inference by letting a smaller draft model propose multiple future tokens that are verified in parallel by a larger target model. The dominant SPD paradigm treats the target model as the sole reliable teacher, accepting a draft token only when it exactly matches the target prediction. This design implicitly assumes that the target is always the better choice at every position. In practice, this assumption does not hold. Although the draft is the weaker model overall, it is not uniformly inferior at the token level. In a meaningful fraction of cases where draft and target disagree, the draft's choice is the one that leads to the correct final answer. Inspired by this, we introduce \textbf{Collaborative Speculative Decoding (CoSpec)}, a generalization of SPD that no longer treats the target model as the sole token-level authority. CoSpec trains an arbitration policy via reinforcement learning to decide whether to accept tokens from the draft or target model, selectively accepting draft tokens at mismatches when doing so is likely to yield a correct final answer. Experimental results show that CoSpec maintains substantial speedups while surpassing target-only performance. By shifting the emphasis from imitation to collaboration, CoSpec suggests a new perspective on speculative decoding.

cs.CL↗
arXiv2026

NasZip: Software and Hardware Co-Design to Accelerate Approximate Nearest Neighbor Search with DIMM-Based Near-Data Processing

As large language models (LLMs) continue to advance, retrieval-augmented generation (RAG) has become the key mechanism for expanding model knowledge and reducing hallucinations. Central to RAG is approximate nearest neighbor search (ANNS), which retrieves database vectors most similar to a given query. However, distance calculation over high-dimensional vectors is inherently memory-bound, causing retrieval performance to be constrained by I/O bandwidth on mainstream platforms such as CPUs and GPUs. Although many prior early exiting (EE) techniques attempt to reduce memory accesses by only computing partial dimensions, the partial distance converges too slowly to the EE threshold, which ultimately limits their performance gains. To address these challenges, we propose NASZIP, a hardware-software co-designed framework that integrates near data processing (NDP) with a novel feature-level early exiting guided by statistics-based principal component analysis (PCA). Instead of relying solely on partial distances, NASZIP incorporates estimation and correction parameters to approximate full dimensional distances accurately, enabling earlier exiting without compromising accuracy. We further introduce a bit-level NDP-aware dynamic-float scheme that significantly reduces memory access for vector data. On the hardware side, we develop a data aware neighbor list mapping strategy that reduces neighbor retrieval latency and inter-channel communication overhead, complemented by a dedicated cache that exploits data locality and enhances prefetch efficiency. With these co-optimized techniques, NASZIP delivers speedups of up to $8.4\times$ / $1.4\times$ over CPU baseline and state-of-the-art GPU implementation at equal accuracy. Relative to the state-of-the-art NDP ANNS accelerator ANSMET, NASZIP achieves $1.69\times$ performance improvement.

cs.AR↗
arXiv2026

One-Way Policy Optimization for Self-Evolving LLMs

Reinforcement Learning with Verifiable Rewards (RLVR) has become a promising paradigm for scaling reasoning capabilities of Large Language Models (LLMs). However, the sparsity of binary verifier rewards often leads to low efficiency and optimization instability. To stabilize training, existing methods typically impose token-level constraints relative to a reference policy. We identify that such constraints penalize deviations indiscriminately; this can flip verifier-determined direction when the policy attempts to outperform the reference, thereby suppressing gains. To resolve this, we propose One-Way Policy Optimization (OWPO), a method based on the principle of decoupling optimization direction from update magnitude. In OWPO, the verifier dictates the update direction, while the reference policy serves only to adjust the magnitude. Specifically, OWPO applies asymmetric reweighting: it performs Accelerated Alignment for inferior deviations (where the policy lags behind the reference) and Gain Locking for superior deviations (where the policy surpasses the reference). Furthermore, by incorporating iterative reference updates, OWPO creates a ``Ratchet Effect'' that continuously consolidates gains. Experimental results demonstrate that OWPO outperforms strong baselines, including DAPO, OPD, and MOPD, breaking the bottleneck of fixed priors to enable continuous self-evolution without reliance on external reference models.

cs.LG↗
arXiv2026

Clipping Bottleneck: Stabilizing RLVR via Stochastic Recovery of Near-Boundary Signals

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a central paradigm for scaling LLM reasoning, yet its optimization often suffers from training instability and suboptimal convergence. Through a systematic dissection of clipping-based GRPO-style objectives, we identify the rigid clipping decision induced by hard clipping as a key practical bottleneck in the studied RLVR setups. Specifically, our analysis suggests that informative signals can lie in the near-boundary region just beyond the clipping threshold, and are therefore discarded by the standard hard-clipping rule. Notably, once this bottleneck is precisely identified, even simple stochastic perturbations at the boundary can recover meaningful performance gains. Building on this finding, we propose Near-boundary Stochastic Rescue (NSR), a minimal, plug-and-play modification that stochastically retains these slightly out-of-bound tokens to recover lost signals. While NSR, via stochastic sampling, can be interpreted as inducing an implicit gradient decay in expectation, our ablations reveal that its stochastic, boundary-local rescue mechanism is consistently more effective than deterministic gradient decay. Validated by extensive experiments across model sizes from 7B to 30B and both dense and MoE architectures, as a plug-and-play solution, NSR substantially improves training stability and delivers consistent gains over strong baselines such as DAPO and GSPO.

cs.LG↗
arXiv2026

Reasoning Portability: Guiding Continual Learning for MLLMs in the RLVR Era

Vision-Language Models in Continual Learning (VLM-CL) aim to continuously adapt to new multimodal tasks while retaining prior knowledge. The emerging paradigm that couples Multimodal Large Language Models (MLLMs) with Reinforcement Learning with Verifiable Rewards (RLVR) calls for a new pattern to guide continual adaptation. Advances in reasoning capability now make it feasible to impose constraints at the reasoning level. We formalize portability, a sample-level measure of how reusable the previous policy's behavior is on a new task, and empirically show that reasoning-level signals remain reliable on out-of-distribution samples while answer-level signals do not. We instantiate this as Reasoning Portability (RP) and propose Reasoning-based Dynamic Balance Continual Learning (RDB-CL), which modulates the per-sample Kullback-Leibler regularization in RLVR according to RP: a tight anchor preserves reusable reasoning on high-RP samples, while a relaxed anchor on low-RP samples permits exploration of new reasoning pathways. Experiments show that RDB-CL consistently outperforms baselines, improving Last accuracy by +12.0% over the vanilla RLVR baseline.

cs.LG↗
arXiv2026

Pythia: Exploiting Workflow Predictability for Efficient Agent-Native LLM Serving

As LLM applications grow more complex, developers are increasingly adopting multi-agent architectures to decompose workflows into specialized, collaborative components, introducing structure that constrains agent behavior and exposes useful semantic predictability. Unlike traditional LLM serving, which operates under highly dynamic and uncertain conditions, this structured topology enables opportunities to reduce runtime uncertainty$\unicode{x2015}$yet existing systems fail to exploit it, treating agentic workloads as generic traffic and incurring significant inefficiencies. Our analysis of production traces from an agent-serving platform and an internal coding assistant reveals key bottlenecks, including low prefix cache hit rates, severe resource contention from long-context requests, and substantial queuing delays due to suboptimal scaling. To address these challenges, we propose Pythia, a multi-agent serving system that captures workflow semantics through a simple interface at the serving layer, unlocking new optimization opportunities and substantially improving throughput and job completion time over state-of-the-art baselines.

cs.MA↗