Sukai Huang

Zhixi Cai, Fucai Ke, Kevin Leo et al.

Recent vision-language models have strong perceptual ability but their implicit reasoning is hard to explain and easily generates hallucinations on complex queries. Compositional methods improve interpretability, but most rely on a single agent or hand-crafted pipeline and cannot decide when to collaborate across complementary agents or compete among overlapping ones. We introduce MATA (Multi-Agent hierarchical Trainable Automaton), a multi-agent system presented as a hierarchical finite-state automaton for visual reasoning whose top-level transitions are chosen by a trainable hyper agent. Each agent corresponds to a state in the hyper automaton, and runs a small rule-based sub-automaton for reliable micro-control. All agents read and write a shared memory, yielding transparent execution history. To supervise the hyper agent's transition policy, we build transition-trajectory trees and transform to memory-to-next-state pairs, forming the MATA-SFT-90K dataset for supervised finetuning (SFT). The finetuned LLM as the transition policy understands the query and the capacity of agents, and it can efficiently choose the optimal agent to solve the task. Across multiple visual reasoning benchmarks, MATA achieves the state-of-the-art results compared with monolithic and compositional baselines. The code and dataset are available at https://github.com/ControlNet/MATA.

Sukai Huang, Nir Lipovetzky, Trevor Cohn

Large Language Models (LLMs) have shown promise in solving natural language-described planning tasks, but their direct use often leads to inconsistent reasoning and hallucination. While hybrid LLM-symbolic planning pipelines have emerged as a more robust alternative, they typically require extensive expert intervention to refine and validate generated action schemas. It not only limits scalability but also introduces a potential for biased interpretation, as a single expert's interpretation of ambiguous natural language descriptions might not align with the user's actual intent. To address this, we propose a novel approach that constructs an action schema library to generate multiple candidates, accounting for the diverse possible interpretations of natural language descriptions. We further introduce a semantic validation and ranking module that automatically filter and rank the generated schemas and plans without expert-in-the-loop. The experiments showed our pipeline maintains superiority in planning over the direct LLM planning approach. These findings demonstrate the feasibility of a fully automated end-to-end LLM-symbolic planner that requires no expert intervention, opening up the possibility for a broader audience to engage with AI planning with less prerequisite of domain expertise.

Sukai Huang, Shu-Wei Liu, Nir Lipovetzky et al.

While Vision-Language Models (VLMs) are increasingly used to generate reward signals for training embodied agents to follow instructions, our research reveals that agents guided by VLM rewards often underperform compared to those employing only intrinsic (exploration-driven) rewards, contradicting expectations set by recent work. We hypothesize that false positive rewards -- instances where unintended trajectories are incorrectly rewarded -- are more detrimental than false negatives. Our analysis confirms this hypothesis, revealing that the widely used cosine similarity metric is prone to false positive reward estimates. To address this, we introduce BiMI ({Bi}nary {M}utual {I}nformation), a novel reward function designed to mitigate noise. BiMI significantly enhances learning efficiency across diverse and challenging embodied navigation environments. Our findings offer a nuanced understanding of how different types of reward noise impact agent learning and highlight the importance of addressing multimodal reward signal noise when training embodied agents

Sukai Huang, Chenyuan Zhang, Fucai Ke et al.

Instruction granularity is an important yet poorly controlled variable in language-guided embodied AI. Existing benchmarks typically pair each task with a single static instruction, making it difficult to study how agent behavior changes when the same task is described at different levels of detail. We introduce Mini-BEHAVIOR-Gran, a new benchmark for controlled studies of instruction granularity that extends Mini-BEHAVIOR with multiple instruction variants per task, ranging from high-level goal descriptions to step-by-step guidance. Using this benchmark, we compare four candidate metrics for cross-task granularity quantification: token count, entity count, action-verb count, and planning-width, and find that width correlates most consistently with agent performance. Using width to organize training and evaluation further reveals a non-monotonic U-shaped relationship between instruction granularity and performance, with peaks at both fine and coarse extremes. Further analysis suggests that the coarse-granularity performance rebound is associated with shallow grounding, where agents learn vision-dominant policies.

Fucai Ke, Joy Hsu, Zhixi Cai et al.

Compositional visual reasoning has emerged as a key research frontier in multimodal AI, aiming to endow machines with the human-like ability to decompose visual scenes, ground intermediate concepts, and perform multi-step logical inference. While early surveys focus on monolithic vision-language models or general multimodal reasoning, a dedicated synthesis of the rapidly expanding compositional visual reasoning literature is still missing. We fill this gap with a comprehensive survey spanning 2023 to 2025 that systematically reviews 260+ papers from top venues (CVPR, ICCV, NeurIPS, ICML, ACL, etc.). We first formalize core definitions and describe why compositional approaches offer advantages in cognitive alignment, semantic fidelity, robustness, interpretability, and data efficiency. Next, we trace a five-stage paradigm shift: from prompt-enhanced language-centric pipelines, through tool-enhanced LLMs and tool-enhanced VLMs, to recently minted chain-of-thought reasoning and unified agentic VLMs, highlighting their architectural designs, strengths, and limitations. We then catalog 60+ benchmarks and corresponding metrics that probe compositional visual reasoning along dimensions such as grounding accuracy, chain-of-thought faithfulness, and high-resolution perception. Drawing on these analyses, we distill key insights, identify open challenges (e.g., limitations of LLM-based reasoning, hallucination, a bias toward deductive reasoning, scalable supervision, tool integration, and benchmark limitations), and outline future directions, including world-model integration, human-AI collaborative reasoning, and richer evaluation protocols. By offering a unified taxonomy, historical roadmap, and critical outlook, this survey aims to serve as a foundational reference and inspire the next generation of compositional visual reasoning research.

Sukai Huang, Trevor Cohn, Nir Lipovetzky

The capability of Large Language Models (LLMs) to plan remains a topic of debate. Some critics argue that strategies to boost LLMs' reasoning skills are ineffective in planning tasks, while others report strong outcomes merely from training models on a planning corpus. This study reassesses recent strategies by developing an end-to-end LLM planner and employing diverse metrics for a thorough evaluation. We find that merely fine-tuning LLMs on a corpus of planning instances does not lead to robust planning skills, as indicated by poor performance on out-of-distribution test sets. At the same time, we find that various strategies, including Chain-of-Thought, do enhance the probability of a plan being executable. This indicates progress towards better plan quality, despite not directly enhancing the final validity rate. Among the strategies we evaluated, reinforcement learning with our novel `Longest Contiguous Common Subsequence' reward emerged as the most effective, contributing to both plan validity and executability. Overall, our research addresses key misconceptions in the LLM-planning literature; we validate incremental progress in plan executability, although plan validity remains a challenge. Hence, future strategies should focus on both these aspects, drawing insights from our findings.

Chunhua Liu, Kabir Manandhar Shrestha, Sukai Huang

As large language models (LLMs) increasingly mediate cross-cultural communication, their behavior still reflects the distributional bias of the languages and viewpoints that are over-represented in their pre-training corpora. Yet, it remains a challenge to model and align culture due to limited cultural knowledge and a lack of exploration into effective learning approaches. We introduce a cost-efficient, cognitively grounded remedy: parameter-efficient fine-tuning on native speakers' free word-association norms, which encode implicit cultural schemas. Leveraging English-US and Mandarin associations from the Small-World-of-Words project, we adapt Llama-3.1-8B and Qwen-2.5-7B via supervised fine-tuning (SFT) and PPO-based preference optimization. SFT boosts held-out association Precision at 5 by 16-20% in English and 43-165% in Mandarin, lifts median concreteness by +0.20, and attains human-level valence and arousal. These lexical gains transfer: on World-Values-Survey questions, fine-tuned models shift answer distributions toward the target culture, and on a 50-item high-tension subset, Qwen's Chinese-aligned responses double while Llama's US bias drops by one-third. Our 7-8B models rival or beat vanilla 70B baselines, showing that a few million culture-grounded associations can instill value alignment without costly retraining. Our work highlights both the promise and the need for future research grounded in human cognition in improving cultural alignment in AI models.

Sukai Huang, Nir Lipovetzky, Trevor Cohn

Teaching agents to follow complex written instructions has been an important yet elusive goal. One technique for enhancing learning efficiency is language reward shaping (LRS). Within a reinforcement learning (RL) framework, LRS involves training a reward function that rewards behaviours precisely aligned with given language instructions. We argue that the apparent success of LRS is brittle, and prior positive findings can be attributed to weak RL baselines. Specifically, we identified suboptimal LRS designs that reward partially matched trajectories, and we characterised a novel reward perturbation to capture this issue using the concept of loosening task constraints. We provided theoretical and empirical evidence that agents trained using LRS rewards converge more slowly compared to pure RL agents. Our work highlights the brittleness of existing LRS methods, which has been overlooked in the previous studies.