Subtle Errors Matter: Preference Learning via Error-injected Self-editing

Large Language Models (LLMs) have exhibited strong mathematical reasoning and computational prowess, tackling tasks ranging from basic arithmetic to advanced competition-level problems. However, frequently occurring subtle errors, such as miscalculations or incorrect substitutions, limit the models' full mathematical potential. Existing studies to improve mathematical ability typically involve distilling reasoning skills from stronger LLMs or applying preference learning to step-wise response pairs. Although these methods leverage samples of varying granularity to mitigate reasoning errors, they overlook the frequently occurring subtle errors. A major reason is that sampled preference pairs involve differences unrelated to the errors, which may distract the model from focusing on subtle errors. In this work, we propose a novel preference learning framework called eRror-Injected Self-Editing (RISE), which injects predefined subtle errors into partial tokens of correct solutions to construct hard pairs for error mitigation. In detail, RISE uses the model itself to edit a small number of tokens in the solution, injecting designed subtle errors. Then, pairs composed of self-edited solutions and their corresponding correct ones, along with pairs of correct and incorrect solutions obtained through sampling, are used together for subtle error-aware DPO training. Compared with other preference learning methods, RISE further refines the training objective to focus on predefined errors and their tokens, without requiring fine-grained sampling or preference annotation. Extensive experiments validate the effectiveness of RISE, with preference learning on Qwen2-7B-Instruct yielding notable improvements of 3.0% on GSM8K and 7.9% on MATH.

Integrative Decoding: Improve Factuality via Implicit Self-consistency

Self-consistency-based approaches, which involve repeatedly sampling multiple outputs and selecting the most consistent one as the final response, prove to be remarkably effective in improving the factual accuracy of large language models. Nonetheless, existing methods usually have strict constraints on the task format, largely limiting their applicability. In this paper, we present Integrative Decoding (ID), to unlock the potential of self-consistency in open-ended generation tasks. ID operates by constructing a set of inputs, each prepended with a previously sampled response, and then processes them concurrently, with the next token being selected by aggregating of all their corresponding predictions at each decoding step. In essence, this simple approach implicitly incorporates self-consistency in the decoding objective. Extensive evaluation shows that ID consistently enhances factuality over a wide range of language models, with substantial improvements on the TruthfulQA (+11.2%), Biographies (+15.4%) and LongFact (+8.5%) benchmarks. The performance gains amplify progressively as the number of sampled responses increases, indicating the potential of ID to scale up with repeated sampling.

MBC: Multi-Brain Collaborative Control for Quadruped Robots

In the field of locomotion task of quadruped robots, Blind Policy and Perceptive Policy each have their own advantages and limitations. The Blind Policy relies on preset sensor information and algorithms, suitable for known and structured environments, but it lacks adaptability in complex or unknown environments. The Perceptive Policy uses visual sensors to obtain detailed environmental information, allowing it to adapt to complex terrains, but its effectiveness is limited under occluded conditions, especially when perception fails. Unlike the Blind Policy, the Perceptive Policy is not as robust under these conditions. To address these challenges, we propose a MBC:Multi-Brain collaborative system that incorporates the concepts of Multi-Agent Reinforcement Learning and introduces collaboration between the Blind Policy and the Perceptive Policy. By applying this multi-policy collaborative model to a quadruped robot, the robot can maintain stable locomotion even when the perceptual system is impaired or observational data is incomplete. Our simulations and real-world experiments demonstrate that this system significantly improves the robot's passability and robustness against perception failures in complex environments, validating the effectiveness of multi-policy collaboration in enhancing robotic motion performance.

Structural Optimization of Lightweight Bipedal Robot via SERL

Designing a bipedal robot is a complex and challenging task, especially when dealing with a multitude of structural parameters. Traditional design methods often rely on human intuition and experience. However, such approaches are time-consuming, labor-intensive, lack theoretical guidance and hard to obtain optimal design results within vast design spaces, thus failing to full exploit the inherent performance potential of robots. In this context, this paper introduces the SERL (Structure Evolution Reinforcement Learning) algorithm, which combines reinforcement learning for locomotion tasks with evolution algorithms. The aim is to identify the optimal parameter combinations within a given multidimensional design space. Through the SERL algorithm, we successfully designed a bipedal robot named Wow Orin, where the optimal leg length are obtained through optimization based on body structure and motor torque. We have experimentally validated the effectiveness of the SERL algorithm, which is capable of optimizing the best structure within specified design space and task conditions. Additionally, to assess the performance gap between our designed robot and the current state-of-the-art robots, we compared Wow Orin with mainstream bipedal robots Cassie and Unitree H1. A series of experimental results demonstrate the Outstanding energy efficiency and performance of Wow Orin, further validating the feasibility of applying the SERL algorithm to practical design.

Reasoning Like a Doctor: Improving Medical Dialogue Systems via Diagnostic Reasoning Process Alignment

Medical dialogue systems have attracted significant attention for their potential to act as medical assistants. Enabling these medical systems to emulate clinicians' diagnostic reasoning process has been the long-standing research focus. Previous studies rudimentarily realized the simulation of clinicians' diagnostic process by fine-tuning language models on high-quality dialogue datasets. Nonetheless, they overly focus on the outcomes of the clinician's reasoning process while ignoring their internal thought processes and alignment with clinician preferences. Our work aims to build a medical dialogue system that aligns with clinicians' diagnostic reasoning processes. We propose a novel framework, Emulation, designed to generate an appropriate response that relies on abductive and deductive diagnostic reasoning analyses and aligns with clinician preferences through thought process modeling. Experimental results on two datasets confirm the efficacy of Emulation. Crucially, our framework furnishes clear explanations for the generated responses, enhancing its transparency in medical consultations.

Evolving to be Your Soulmate: Personalized Dialogue Agents with Dynamically Adapted Personas

Previous research on persona-based dialogue agents typically preset the agent's persona before deployment, which remains static thereafter. In this paper, we take a step further and explore a new paradigm called Self-evolving Personalized Dialogue Agents (SPDA), where the agent continuously evolves during the conversation to better align with the user's anticipation by dynamically adapting its persona. This paradigm could enable better personalization for each user, but also introduce unique challenges, which mainly lie in the process of persona adaptation. Two key issues include how to achieve persona alignment with the user and how to ensure smooth transition in the adaptation process. To address them, we propose a novel framework that refines the persona at hierarchical levels to progressively align better with the user in a controllable way. Experiments show that integrating the personas adapted by our framework consistently enhances personalization and overall dialogue performance across various base systems.

No Two Devils Alike: Unveiling Distinct Mechanisms of Fine-tuning Attacks

The existing safety alignment of Large Language Models (LLMs) is found fragile and could be easily attacked through different strategies, such as through fine-tuning on a few harmful examples or manipulating the prefix of the generation results. However, the attack mechanisms of these strategies are still underexplored. In this paper, we ask the following question: \textit{while these approaches can all significantly compromise safety, do their attack mechanisms exhibit strong similarities?} To answer this question, we break down the safeguarding process of an LLM when encountered with harmful instructions into three stages: (1) recognizing harmful instructions, (2) generating an initial refusing tone, and (3) completing the refusal response. Accordingly, we investigate whether and how different attack strategies could influence each stage of this safeguarding process. We utilize techniques such as logit lens and activation patching to identify model components that drive specific behavior, and we apply cross-model probing to examine representation shifts after an attack. In particular, we analyze the two most representative types of attack approaches: Explicit Harmful Attack (EHA) and Identity-Shifting Attack (ISA). Surprisingly, we find that their attack mechanisms diverge dramatically. Unlike ISA, EHA tends to aggressively target the harmful recognition stage. While both EHA and ISA disrupt the latter two stages, the extent and mechanisms of their attacks differ significantly. Our findings underscore the importance of understanding LLMs' internal safeguarding process and suggest that diverse defense mechanisms are required to effectively cope with various types of attacks.

Bridging the Intent Gap: Knowledge-Enhanced Visual Generation

For visual content generation, discrepancies between user intentions and the generated content have been a longstanding problem. This discrepancy arises from two main factors. First, user intentions are inherently complex, with subtle details not fully captured by input prompts. The absence of such details makes it challenging for generative models to accurately reflect the intended meaning, leading to a mismatch between the desired and generated output. Second, generative models trained on visual-label pairs lack the comprehensive knowledge to accurately represent all aspects of the input data in their generated outputs. To address these challenges, we propose a knowledge-enhanced iterative refinement framework for visual content generation. We begin by analyzing and identifying the key challenges faced by existing generative models. Then, we introduce various knowledge sources, including human insights, pre-trained models, logic rules, and world knowledge, which can be leveraged to address these challenges. Furthermore, we propose a novel visual generation framework that incorporates a knowledge-based feedback module to iteratively refine the generation process. This module gradually improves the alignment between the generated content and user intentions. We demonstrate the efficacy of the proposed framework through preliminary results, highlighting the potential of knowledge-enhanced generative models for intention-aligned content generation.

Quadruped robot traversing 3D complex environments with limited perception

Traversing 3-D complex environments has always been a significant challenge for legged locomotion. Existing methods typically rely on external sensors such as vision and lidar to preemptively react to obstacles by acquiring environmental information. However, in scenarios like nighttime or dense forests, external sensors often fail to function properly, necessitating robots to rely on proprioceptive sensors to perceive diverse obstacles in the environment and respond promptly. This task is undeniably challenging. Our research finds that methods based on collision detection can enhance a robot's perception of environmental obstacles. In this work, we propose an end-to-end learning-based quadruped robot motion controller that relies solely on proprioceptive sensing. This controller can accurately detect, localize, and agilely respond to collisions in unknown and complex 3D environments, thereby improving the robot's traversability in complex environments. We demonstrate in both simulation and real-world experiments that our method enables quadruped robots to successfully traverse challenging obstacles in various complex environments.

PEVA-Net: Prompt-Enhanced View Aggregation Network for Zero/Few-Shot Multi-View 3D Shape Recognition

Large vision-language models have impressively promote the performance of 2D visual recognition under zero/few-shot scenarios. In this paper, we focus on exploiting the large vision-language model, i.e., CLIP, to address zero/few-shot 3D shape recognition based on multi-view representations. The key challenge for both tasks is to generate a discriminative descriptor of the 3D shape represented by multiple view images under the scenarios of either without explicit training (zero-shot 3D shape recognition) or training with a limited number of data (few-shot 3D shape recognition). We analyze that both tasks are relevant and can be considered simultaneously. Specifically, leveraging the descriptor which is effective for zero-shot inference to guide the tuning of the aggregated descriptor under the few-shot training can significantly improve the few-shot learning efficacy. Hence, we propose Prompt-Enhanced View Aggregation Network (PEVA-Net) to simultaneously address zero/few-shot 3D shape recognition. Under the zero-shot scenario, we propose to leverage the prompts built up from candidate categories to enhance the aggregation process of multiple view-associated visual features. The resulting aggregated feature serves for effective zero-shot recognition of the 3D shapes. Under the few-shot scenario, we first exploit a transformer encoder to aggregate the view-associated visual features into a global descriptor. To tune the encoder, together with the main classification loss, we propose a self-distillation scheme via a feature distillation loss by treating the zero-shot descriptor as the guidance signal for the few-shot descriptor. This scheme can significantly enhance the few-shot learning efficacy.