Liyun Zhang

Liyun Zhang, Jiayi Guo

We document an empirical phenomenon in chain-of-thought and ReAct agents driven by ten large language models from seven architecture families: meaning-bearing perturbations (e.g., paraphrase, synonym) alter final answers more often than presentation perturbations (e.g., formatting, reordering) of comparable severity. Across 68 cells spanning GSM8K, MATH, and HotpotQA (1,530 originals and $\sim$11,150 variants), the inconsistency gap averages +19.69 pp after severity matching (paired $t=9.58$, $p<0.0001$), with 64/68 cells positive. The gap survives four severity-proxy audits and remains significant when excluding qwen models (+11.10 pp, $p<0.0001$). Several stress tests fail honestly: cluster-bootstrap significance disappears under stricter assumptions, tractability contrasts do not replicate, cross-architecture generator swaps break per-cell rankings, and a second LLM judge yields only moderate agreement ($κ=0.50$). We then validate the headline effect on a fully held-out 11th model (qwen2.5-14B-Instruct; 1,800 trajectories) and re-test a pre-registered capability$\times$tractability partition, observing a small but positive held-out effect (3/4 cells positive; pooled Welch $t=3.81$, $p=9.6\times10^{-4}$). Using held-out trajectories, we probe four trace-level mechanism signals. Two prior mechanism claims fail to replicate and are explicitly retracted. Two new probes instead support a \emph{stealth-divergence} picture: semantic perturbations often preserve the first action but induce divergence in intermediate reasoning from later steps onward, accompanied by slightly deeper trajectories. We position this as a measurement contribution with held-out replication and a partial trace-level account of how semantic perturbations propagate through agent reasoning. Code, perturbation corpus, raw trajectories, and analysis scripts are released anonymously for review.

Arash Akbari, Arman Akbari, Masih Eskandar et al.

Vision-Language-Action (VLA) models exhibit remarkable action generation for embodied intelligence, but their heavy compute make deployment on edge platforms impractical. Aggressive, sub-4-bit weight quantization is the natural solution, yet existing post-training quantization (PTQ) methods suffer severe performance degradation in this regime. To address this, we introduce ActQuant, an action-guided mixed-precision PTQ framework that operates in two stages: (1) an inter-tensor bit allocator that assigns each weight matrix a single bit-width based on how much it contributes to predicting the agent's actions; (2) an intra-tensor scale optimizer tunes per-block quantization scales using action-aware curvature, so that dynamic range is concentrated on the weights most influential for control. To deliver the on-device benefits of our aggressive quantization, we further introduce OmniModel.cpp, an agentic conversion pipeline that ports architectures into a native C/C++ runtime with efficient low-bit kernels. We evaluate ActQuant both in simulation and on a real-world 6-DoF UR3 arm, with all models deployed through OmniModel.cpp. On the LIBERO benchmark, ActQuant is the only method that operates at or below 3 bits-per-weight, retaining 95.0% on OpenVLA-OFT and 94.8% on $π_{0.5}$. Pushed further, ActQuant reaches 2.5 bpw at 90.1% on OpenVLA-OFT, compressing the backbone from 14.3 GB to 2.7 GB (5.3$\times$). On the physical UR3 arm, $π_{0.5}$ quantized with ActQuant retains the baseline's success rate while reducing the memory footprint by 2.5$\times$.

Liyun Zhang

Multimodal Large Language Models (MLLMs) have demonstrated remarkable multimodal emotion recognition capabilities, integrating multimodal cues from visual, acoustic, and linguistic contexts in the video to recognize human emotional states. However, existing methods ignore capturing local facial features of temporal dynamics of micro-expressions and do not leverage the contextual dependencies of the utterance-aware temporal segments in the video, thereby limiting their expected effectiveness to a certain extent. In this work, we propose MicroEmo, a time-sensitive MLLM aimed at directing attention to the local facial micro-expression dynamics and the contextual dependencies of utterance-aware video clips. Our model incorporates two key architectural contributions: (1) a global-local attention visual encoder that integrates global frame-level timestamp-bound image features with local facial features of temporal dynamics of micro-expressions; (2) an utterance-aware video Q-Former that captures multi-scale and contextual dependencies by generating visual token sequences for each utterance segment and for the entire video then combining them. Preliminary qualitative experiments demonstrate that in a new Explainable Multimodal Emotion Recognition (EMER) task that exploits multi-modal and multi-faceted clues to predict emotions in an open-vocabulary (OV) manner, MicroEmo demonstrates its effectiveness compared with the latest methods.

Chuxuan Hu, Liyun Zhang, Yeji Lim et al.

Assessing the reproducibility of social science papers is essential for promoting rigor in research processes, but manual assessment is costly. With recent advances in agentic AI systems (i.e., AI agents), we seek to evaluate their capability to automate this process. However, existing benchmarks for reproducing research papers (1) focus solely on reproducing results using provided code and data without assessing their consistency with the paper, (2) oversimplify real-world scenarios, and (3) lack necessary diversity in data formats and programming languages. To address these issues, we introduce REPRO-Bench, a collection of 112 task instances, each representing a social science paper with a publicly available reproduction report. The agents are tasked with assessing the reproducibility of the paper based on the original paper PDF and the corresponding reproduction package. REPRO-Bench features end-to-end evaluation tasks on the reproducibility of social science papers with complexity comparable to real-world assessments. We evaluate three representative AI agents on REPRO-Bench, with the best-performing agent achieving an accuracy of only 21.4%. Building on our empirical analysis, we develop REPRO-Agent, which improves the highest accuracy achieved by existing agents by 71%. We conclude that more advanced AI agents should be developed to automate real-world reproducibility assessment. REPRO-Bench is publicly available at https://github.com/uiuc-kang-lab/REPRO-Bench.

Fengkai Liu, Hao Su, Haozhuang Chi et al.

Assistance in collaborative manipulation is often initiated by user instructions, making high-level reasoning request-driven. In fluent human teamwork, however, partners often infer the next helpful step from the observed outcome of an action rather than waiting for instructions. Motivated by this, we introduce a shift from request-driven assistance to event-driven proactive assistance, where robot actions are initiated by workspace state transitions induced by human--object interactions rather than user-provided task instructions. To this end, we propose an event-driven framework that tracks interaction progress with an event monitor and, upon event completion, extracts stabilized pre/post snapshots that characterize the resulting state transition. Given the stabilized snapshots, the planner analyzes the implied state transition to infer a task-level goal and decide whether to intervene; if so, it generates a sequence of assistive actions. To make outputs executable and verifiable, we restrict actions to a set of action primitives and reference objects via integer IDs. We evaluate the framework on a real tabletop number-block collaboration task, demonstrating that explicit pre/post state-change evidence improves proactive completion on solvable scenes and appropriate waiting on unsolvable ones.

Xuanmeng Sha, Liyun Zhang, Tomohiro Mashita et al.

Generating holistic co-speech gestures that integrate full-body motion with facial expressions suffers from semantically incoherent coordination on body motion and spatially unstable meaningless movements due to existing part-decomposed or frame-level regression methods, We introduce 3DGesPolicy, a novel action-based framework that reformulates holistic gesture generation as a continuous trajectory control problem through diffusion policy from robotics. By modeling frame-to-frame variations as unified holistic actions, our method effectively learns inter-frame holistic gesture motion patterns and ensures both spatially and semantically coherent movement trajectories that adhere to realistic motion manifolds. To further bridge the gap in expressive alignment, we propose a Gesture-Audio-Phoneme (GAP) fusion module that can deeply integrate and refine multi-modal signals, ensuring structured and fine-grained alignment between speech semantics, body motion, and facial expressions. Extensive quantitative and qualitative experiments on the BEAT2 dataset demonstrate the effectiveness of our 3DGesPolicy across other state-of-the-art methods in generating natural, expressive, and highly speech-aligned holistic gestures.

Xuanmeng Sha, Liyun Zhang, Tomohiro Mashita et al.

Audio-driven 3D facial animation has achieved significant progress in both research and applications. While recent baselines struggle to generate natural and continuous facial movements due to their frame-by-frame vertex generation approach, we propose 3DFacePolicy, a pioneer work that introduces a novel definition of vertex trajectory changes across consecutive frames through the concept of "action". By predicting action sequences for each vertex that encode frame-to-frame movements, we reformulate vertex generation approach into an action-based control paradigm. Specifically, we leverage a robotic control mechanism, diffusion policy, to predict action sequences conditioned on both audio and vertex states. Extensive experiments on VOCASET and BIWI datasets demonstrate that our approach significantly outperforms state-of-the-art methods and is particularly expert in dynamic, expressive and naturally smooth facial animations.

Liyun Zhang, Xuanmeng Sha, Shuqiong Wu et al.

Multimodal Large Language Models (MLLMs) excel in Open-Vocabulary (OV) emotion recognition but often neglect fine-grained acoustic modeling. Existing methods typically use global audio encoders, failing to capture subtle, local temporal dynamics like micro-prosody and intonation shifts within individual utterances. To address this, we propose AcoustEmo, a time-sensitive MLLM featuring a novel Utterance-Aware Acoustic Q-Former. Our approach utilizes a timestamp-synchronized sliding window to dynamically extract segment-level audio tokens instead of coarse global representations. This enables the model to explicitly trace the temporal evolution of subtle acoustic clues and capture deep contextual dependencies in dialogues. Experiments on the Explainable Multimodal Emotion Recognition (EMER) task show that AcoustEmo significantly enhances complex emotion reasoning, outperforming baselines while maintaining robust contextual accuracy.

Liyun Zhang, Jingcheng Ke, Shenli Fan et al.

Recent works have emerged in multi-annotator learning that shift focus from Consensus-oriented Learning (CoL), which aggregates multiple annotations into a single ground-truth prediction, to Individual Tendency Learning (ITL), which models annotator-specific labeling behavior patterns (i.e., tendency) to provide explanation analysis for understanding annotator decisions. However, no evaluation framework currently exists to assess whether ITL methods truly capture individual tendencies and provide meaningful behavioral explanations. To address this gap, we propose the first unified evaluation framework with two novel metrics: (1) Difference of Inter-annotator Consistency (DIC) quantifies how well models capture annotator tendencies by comparing predicted inter-annotator similarity structures with ground-truth; (2) Behavior Alignment Explainability (BAE) evaluates how well model explanations reflect annotator behavior and decision relevance by aligning explainability-derived with ground-truth labeling similarity structures via Multidimensional Scaling (MDS). Extensive experiments validate the effectiveness of our proposed evaluation framework.

Han Zhang, Yu Lu, Liyun Zhang et al.

Understanding the emotions in a dialogue usually requires external knowledge to accurately understand the contents. As the LLMs become more and more powerful, we do not want to settle on the limited ability of the pre-trained language model. However, the LLMs either can only process text modality or are too expensive to process the multimedia information. We aim to utilize both the power of LLMs and the supplementary features from the multimedia modalities. In this paper, we present a framework, Lantern, that can improve the performance of a certain vanilla model by prompting large language models with receptive-field-aware attention weighting. This framework trained a multi-task vanilla model to produce probabilities of emotion classes and dimension scores. These predictions are fed into the LLMs as references to adjust the predicted probabilities of each emotion class with its external knowledge and contextual understanding. We slice the dialogue into different receptive fields, and each sample is included in exactly t receptive fields. Finally, the predictions of LLMs are merged with a receptive-field-aware attention-driven weighting module. In the experiments, vanilla models CORECT and SDT are deployed in Lantern with GPT-4 or Llama-3.1-405B. The experiments in IEMOCAP with 4-way and 6-way settings demonstrated that the Lantern can significantly improve the performance of current vanilla models by up to 1.23% and 1.80%.

Liyun Zhang, Photchara Ratsamee, Bowen Wang et al.

Despite remarkable progress in image translation, the complex scene with multiple discrepant objects remains a challenging problem. The translated images have low fidelity and tiny objects in fewer details causing unsatisfactory performance in object recognition. Without thorough object perception (i.e., bounding boxes, categories, and masks) of images as prior knowledge, the style transformation of each object will be difficult to track in translation. We propose panoptic-aware generative adversarial networks (PanopticGAN) for image-to-image translation together with a compact panoptic segmentation dataset. The panoptic perception (i.e., foreground instances and background semantics of the image scene) is extracted to achieve alignment between object content codes of the input domain and panoptic-level style codes sampled from the target style space, then refined by a proposed feature masking module for sharping object boundaries. The image-level combination between content and sampled style codes is also merged for higher fidelity image generation. Our proposed method was systematically compared with different competing methods and obtained significant improvement in both image quality and object recognition performance.