arXiv subjects

Haonan Yu

Publications and source records attributed to Haonan Yu.

At least 19 recordsLinked to original sources

arXiv2026

WASD: Locating Critical Neurons as Sufficient Conditions for Explaining and Controlling LLM Behavior

Precise behavioral control of large language models (LLMs) is critical for complex applications. However, existing methods often incur high training costs, lack natural language controllability, or compromise semantic coherence. To bridge this gap, we propose WASD (unWeaving Actionable Sufficient Directives), a novel framework that explains model behavior by identifying sufficient neural conditions for token generation. Our method represents candidate conditions as neuron-activation predicates and iteratively searches for a minimal set that guarantees the current output under input perturbations. Experiments on SST-2 and CounterFact with the Gemma-2-2B model demonstrate that our approach produces explanations that are more stable, accurate, and concise than conventional attribution graphs. Moreover, through a case study on controlling cross-lingual output generation, we validated the practical effectiveness of WASD in controlling model behavior.

cs.CL↗

arXiv2026

Focus-LIME: Surgical Interpretation of Long-Context Large Language Models via Proxy-Based Neighborhood Selection

As Large Language Models (LLMs) scale to handle massive context windows, achieving surgical feature-level interpretation is essential for high-stakes tasks like legal auditing and code debugging. However, existing local model-agnostic explanation methods face a critical dilemma in these scenarios: feature-based methods suffer from attribution dilution due to high feature dimensionality, thus failing to provide faithful explanations. In this paper, we propose Focus-LIME, a coarse-to-fine framework designed to restore the tractability of surgical interpretation. Focus-LIME utilizes a proxy model to curate the perturbation neighborhood, allowing the target model to perform fine-grained attribution exclusively within the optimized context. Empirical evaluations on long-context benchmarks demonstrate that our method makes surgical explanations practicable and provides faithful explanations to users.

cs.CL↗

arXiv2025

Learning Multi-Stage Pick-and-Place with a Legged Mobile Manipulator

Quadruped-based mobile manipulation presents significant challenges in robotics due to the diversity of required skills, the extended task horizon, and partial observability. After presenting a multi-stage pick-and-place task as a succinct yet sufficiently rich setup that captures key desiderata for quadruped-based mobile manipulation, we propose an approach that can train a visuo-motor policy entirely in simulation, and achieve nearly 80\% success in the real world. The policy efficiently performs search, approach, grasp, transport, and drop into actions, with emerged behaviors such as re-grasping and task chaining. We conduct an extensive set of real-world experiments with ablation studies highlighting key techniques for efficient training and effective sim-to-real transfer. Additional experiments demonstrate deployment across a variety of indoor and outdoor environments. Demo videos and additional resources are available on the project page: https://horizonrobotics.github.io/gail/SLIM.

cs.RO↗

arXiv2025

Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models

Post-hoc explanations provide transparency and are essential for guiding model optimization, such as prompt engineering and data sanitation. However, applying model-agnostic techniques to Large Language Models (LLMs) is hindered by prohibitive computational costs, rendering these tools dormant for real-world applications. To revitalize model-agnostic interpretability, we propose a budget-friendly proxy framework that leverages efficient models to approximate the decision boundaries of expensive LLMs. We introduce a screen-and-apply mechanism to statistically verify local alignment before deployment. Our empirical evaluation confirms that proxy explanations achieve over 90% fidelity with only 11% of the oracle's cost. Building on this foundation, we demonstrate the actionable utility of our framework in prompt compression and poisoned example removal. Results show that reliable proxy explanations effectively guide optimization, transforming interpretability from a passive observation tool into a scalable primitive for LLM development. Additionally, we open-source code and datasets to facilitate future research.

cs.LG↗

arXiv2025

MAnchors: Memorization-Based Acceleration of Anchors via Rule Reuse and Transformation

Anchors is a popular local model-agnostic explanation technique whose applicability is limited by its computational inefficiency. To address this limitation, we propose a memorization-based framework that accelerates Anchors while preserving explanation fidelity and interpretability. Our approach leverages the iterative nature of Anchors' algorithm which gradually refines an explanation until it is precise enough for a given input by storing and reusing intermediate results obtained during prior explanations. Specifically, we maintain a memory of low-precision, high-coverage rules and introduce a rule transformation framework to adapt them to new inputs: the horizontal transformation adapts a pre-trained explanation to the current input by replacing features, and the vertical transformation refines the general explanation until it is precise enough for the input. We evaluate our method across tabular, text, and image datasets, demonstrating that it significantly reduces explanation generation time while maintaining fidelity and interpretability, thereby enabling the practical adoption of Anchors in time-sensitive applications.

cs.LG↗

arXiv2025

SLIM: Sim-to-Real Legged Instructive Manipulation via Long-Horizon Visuomotor Learning

We present a low-cost legged mobile manipulation system that solves long-horizon real-world tasks, trained by reinforcement learning purely in simulation. This system is made possible by 1) a hierarchical design of a high-level policy for visual-mobile manipulation following task instructions, and a low-level quadruped locomotion policy, 2) a teacher and student training pipeline for the high level, which trains a teacher to tackle long-horizon tasks using privileged task decomposition and target object information, and further trains a student for visual-mobile manipulation via RL guided by the teacher's behavior, and 3) a suite of techniques for minimizing the sim-to-real gap. In contrast to many previous works that use high-end equipments, our system demonstrates effective performance with more accessible hardware -- specifically, a Unitree Go1 quadruped, a WidowX-250S arm, and a single wrist-mounted RGB camera -- despite the increased challenges of sim-to-real transfer. Trained fully in simulation, a single policy autonomously solves long-horizon tasks involving search, move to, grasp, transport, and drop into, achieving nearly 80% real-world success. This performance is comparable to that of expert human teleoperation on the same tasks while the robot is more efficient, operating at about 1.5x the speed of the teleoperation. Finally, we perform extensive ablations on key techniques for efficient RL training and effective sim-to-real transfer, and demonstrate effective deployment across diverse indoor and outdoor scenes under various lighting conditions.

cs.RO↗

arXiv2024

Beyond Attribution: Unified Concept-Level Explanations

There is an increasing need to integrate model-agnostic explanation techniques with concept-based approaches, as the former can explain models across different architectures while the latter makes explanations more faithful and understandable to end-users. However, existing concept-based model-agnostic explanation methods are limited in scope, mainly focusing on attribution-based explanations while neglecting diverse forms like sufficient conditions and counterfactuals, thus narrowing their utility. To bridge this gap, we propose a general framework UnCLE to elevate existing local model-agnostic techniques to provide concept-based explanations. Our key insight is that we can uniformly extend existing local model-agnostic methods to provide unified concept-based explanations with large pre-trained model perturbation. We have instantiated UnCLE to provide concept-based explanations in three forms: attributions, sufficient conditions, and counterfactuals, and applied it to popular text, image, and multimodal models. Our evaluation results demonstrate that UnCLE provides explanations more faithful than state-of-the-art concept-based explanation methods, and provides richer explanation forms that satisfy various user needs.

cs.LG↗

arXiv2024

VONet: Unsupervised Video Object Learning With Parallel U-Net Attention and Object-wise Sequential VAE

Unsupervised video object learning seeks to decompose video scenes into structural object representations without any supervision from depth, optical flow, or segmentation. We present VONet, an innovative approach that is inspired by MONet. While utilizing a U-Net architecture, VONet employs an efficient and effective parallel attention inference process, generating attention masks for all slots simultaneously. Additionally, to enhance the temporal consistency of each mask across consecutive video frames, VONet develops an object-wise sequential VAE framework. The integration of these innovative encoder-side techniques, in conjunction with an expressive transformer-based decoder, establishes VONet as the leading unsupervised method for object learning across five MOVI datasets, encompassing videos of diverse complexities. Code is available at https://github.com/hnyu/vonet.

cs.CV↗

arXiv2023

Policy Expansion for Bridging Offline-to-Online Reinforcement Learning

Pre-training with offline data and online fine-tuning using reinforcement learning is a promising strategy for learning control policies by leveraging the best of both worlds in terms of sample efficiency and performance. One natural approach is to initialize the policy for online learning with the one trained offline. In this work, we introduce a policy expansion scheme for this task. After learning the offline policy, we use it as one candidate policy in a policy set. We then expand the policy set with another policy which will be responsible for further learning. The two policies will be composed in an adaptive manner for interacting with the environment. With this approach, the policy previously learned offline is fully retained during online learning, thus mitigating the potential issues such as destroying the useful behaviors of the offline policy in the initial stage of online learning while allowing the offline policy participate in the exploration naturally in an adaptive manner. Moreover, new useful behaviors can potentially be captured by the newly added policy through learning. Experiments are conducted on a number of tasks and the results demonstrate the effectiveness of the proposed approach. Code is available at https://github.com/Haichao-Zhang/PEX

cs.AI↗

arXiv2022

Towards Safe Reinforcement Learning with a Safety Editor Policy

We consider the safe reinforcement learning (RL) problem of maximizing utility with extremely low constraint violation rates. Assuming no prior knowledge or pre-training of the environment safety model given a task, an agent has to learn, via exploration, which states and actions are safe. A popular approach in this line of research is to combine a model-free RL algorithm with the Lagrangian method to adjust the weight of the constraint reward relative to the utility reward dynamically. It relies on a single policy to handle the conflict between utility and constraint rewards, which is often challenging. We present SEditor, a two-policy approach that learns a safety editor policy transforming potentially unsafe actions proposed by a utility maximizer policy into safe ones. The safety editor is trained to maximize the constraint reward while minimizing a hinge loss of the utility state-action values before and after an action is edited. SEditor extends existing safety layer designs that assume simplified safety models, to general safe RL scenarios where the safety model can in theory be arbitrarily complex. As a first-order method, it is easy to implement and efficient for both inference and training. On 12 Safety Gym tasks and 2 safe racing tasks, SEditor obtains much a higher overall safety-weighted-utility (SWU) score than the baselines, and demonstrates outstanding utility performance with constraint violation rates as low as once per 2k time steps, even in obstacle-dense environments. On some tasks, this low violation rate is up to 200 times lower than that of an unconstrained RL method with similar utility performance. Code is available at https://github.com/hnyu/seditor.

cs.LG↗

arXiv2022

Do You Need the Entropy Reward (in Practice)?

Maximum entropy (MaxEnt) RL maximizes a combination of the original task reward and an entropy reward. It is believed that the regularization imposed by entropy, on both policy improvement and policy evaluation, together contributes to good exploration, training convergence, and robustness of learned policies. This paper takes a closer look at entropy as an intrinsic reward, by conducting various ablation studies on soft actor-critic (SAC), a popular representative of MaxEnt RL. Our findings reveal that in general, entropy rewards should be applied with caution to policy evaluation. On one hand, the entropy reward, like any other intrinsic reward, could obscure the main task reward if it is not properly managed. We identify some failure cases of the entropy reward especially in episodic Markov decision processes (MDPs), where it could cause the policy to be overly optimistic or pessimistic. On the other hand, our large-scale empirical study shows that using entropy regularization alone in policy improvement, leads to comparable or even better performance and robustness than using it in both policy improvement and policy evaluation. Based on these observations, we recommend either normalizing the entropy reward to a zero mean (SACZero), or simply removing it from policy evaluation (SACLite) for better practical results.

cs.LG↗

arXiv2022

Generative Planning for Temporally Coordinated Exploration in Reinforcement Learning

Standard model-free reinforcement learning algorithms optimize a policy that generates the action to be taken in the current time step in order to maximize expected future return. While flexible, it faces difficulties arising from the inefficient exploration due to its single step nature. In this work, we present Generative Planning method (GPM), which can generate actions not only for the current step, but also for a number of future steps (thus termed as generative planning). This brings several benefits to GPM. Firstly, since GPM is trained by maximizing value, the plans generated from it can be regarded as intentional action sequences for reaching high value regions. GPM can therefore leverage its generated multi-step plans for temporally coordinated exploration towards high value regions, which is potentially more effective than a sequence of actions generated by perturbing each action at single step level, whose consistent movement decays exponentially with the number of exploration steps. Secondly, starting from a crude initial plan generator, GPM can refine it to be adaptive to the task, which, in return, benefits future explorations. This is potentially more effective than commonly used action-repeat strategy, which is non-adaptive in its form of plans. Additionally, since the multi-step plan can be interpreted as the intent of the agent from now to a span of time period into the future, it offers a more informative and intuitive signal for interpretation. Experiments are conducted on several benchmark environments and the results demonstrated its effectiveness compared with several baseline methods.

cs.LG↗

arXiv2021

TAAC: Temporally Abstract Actor-Critic for Continuous Control

We present temporally abstract actor-critic (TAAC), a simple but effective off-policy RL algorithm that incorporates closed-loop temporal abstraction into the actor-critic framework. TAAC adds a second-stage binary policy to choose between the previous action and a new action output by an actor. Crucially, its "act-or-repeat" decision hinges on the actually sampled action instead of the expected behavior of the actor. This post-acting switching scheme let the overall policy make more informed decisions. TAAC has two important features: a) persistent exploration, and b) a new compare-through Q operator for multi-step TD backup, specially tailored to the action repetition scenario. We demonstrate TAAC's advantages over several strong baselines across 14 continuous control tasks. Our surprising finding reveals that while achieving top performance, TAAC is able to "mine" a significant number of repeated actions with the trained policy even on continuous tasks whose problem structures on the surface seem to repel action repetition. This suggests that aside from encouraging persistent exploration, action repetition can find its place in a good policy behavior. Code is available at https://github.com/hnyu/taac.

cs.LG↗

arXiv2021

MetaView: Few-shot Active Object Recognition

In robot sensing scenarios, instead of passively utilizing human captured views, an agent should be able to actively choose informative viewpoints of a 3D object as discriminative evidence to boost the recognition accuracy. This task is referred to as active object recognition. Recent works on this task rely on a massive amount of training examples to train an optimal view selection policy. But in realistic robot sensing scenarios, the large-scale training data may not exist and whether the intelligent view selection policy can be still learned from few object samples remains unclear. In this paper, we study this new problem which is extremely challenging but very meaningful in robot sensing -- Few-shot Active Object Recognition, i.e., to learn view selection policies from few object samples, which has not been considered and addressed before. We solve the proposed problem by adopting the framework of meta learning and name our method "MetaView". Extensive experiments on both category-level and instance-level classification tasks demonstrate that the proposed method can efficiently resolve issues that are hard for state-of-the-art active object recognition methods to handle, and outperform several baselines by large margins.

cs.RO↗

arXiv2021

Hierarchical Reinforcement Learning By Discovering Intrinsic Options

We propose a hierarchical reinforcement learning method, HIDIO, that can learn task-agnostic options in a self-supervised manner while jointly learning to utilize them to solve sparse-reward tasks. Unlike current hierarchical RL approaches that tend to formulate goal-reaching low-level tasks or pre-define ad hoc lower-level policies, HIDIO encourages lower-level option learning that is independent of the task at hand, requiring few assumptions or little knowledge about the task structure. These options are learned through an intrinsic entropy minimization objective conditioned on the option sub-trajectories. The learned options are diverse and task-agnostic. In experiments on sparse-reward robotic manipulation and navigation tasks, HIDIO achieves higher success rates with greater sample efficiency than regular RL baselines and two state-of-the-art hierarchical RL methods.

cs.LG↗

arXiv2019

Why Build an Assistant in Minecraft?

In this document we describe a rationale for a research program aimed at building an open "assistant" in the game Minecraft, in order to make progress on the problems of natural language understanding and learning from dialogue.

cs.AI↗

arXiv2019

CraftAssist: A Framework for Dialogue-enabled Interactive Agents

This paper describes an implementation of a bot assistant in Minecraft, and the tools and platform allowing players to interact with the bot and to record those interactions. The purpose of building such an assistant is to facilitate the study of agents that can complete tasks specified by dialogue, and eventually, to learn from dialogue interactions.

cs.AI↗

arXiv2019

EPNAS: Efficient Progressive Neural Architecture Search

In this paper, we propose Efficient Progressive Neural Architecture Search (EPNAS), a neural architecture search (NAS) that efficiently handles large search space through a novel progressive search policy with performance prediction based on REINFORCE~\cite{Williams.1992.PG}. EPNAS is designed to search target networks in parallel, which is more scalable on parallel systems such as GPU/TPU clusters. More importantly, EPNAS can be generalized to architecture search with multiple resource constraints, \eg, model size, compute complexity or intensity, which is crucial for deployment in widespread platforms such as mobile and cloud. We compare EPNAS against other state-of-the-art (SoTA) network architectures (\eg, MobileNetV2~\cite{mobilenetv2}) and efficient NAS algorithms (\eg, ENAS~\cite{pham2018efficient}, and PNAS~\cite{Liu2017b}) on image recognition tasks using CIFAR10 and ImageNet. On both datasets, EPNAS is superior \wrt architecture searching speed and recognition accuracy.

cs.LG↗