Searcharxiv⌕ Search

arXiv subjects

Dongcheng Zhang

Publications and source records attributed to Dongcheng Zhang.

7 recordsLinked to original sources

Janus-MOPD: Tracing the Student Learning Dynamics in Multi-Teacher On-Policy Distillation

Multi-domain capabilities are critical for practically applicable Large Reasoning Models (LRMs). Therefore, Multi-teacher On-Policy Distillation (MOPD) has become a crucial part of the post-training process, integrating the capabilities of multiple domain-specific teacher models into a single student model. However, the student usually fails to acquire multi-domain capabilities simultaneously due to undesirable domain heterogeneity. Existing works have address this problem in view of temporal uneven efficiency and spatial hard allocation in external multi-teacher supervision. While in broader multi-domain training, varying degrees of performance degradation is still largely unexplained. In this paper, we shift the focus on how the student learns and integrates multi-domain capabilities internally, and then propose \textbf{\textsc{Janus-MOPD}}, a general learning dynamics framework of geometric representations for LRMs. Specifically, our main focus is on the internal performance of multi-domain capabilities on both temporal and spatial dimensions within the student. Temporally, our study shows that the student exhibits a Expansion-Consolidation-Optimization (ECO) three-phase learning dynamics, together with inherent limitations in learning efficiency and scaling. Spatially, we summarize two types of subspace locking within and across domains by tracing the representation space of the student, while explaining its clear learning directions but inherent cross-domain conflict. Therefore, \textsc{Janus-MOPD} yields a unified learning dynamics formulation, a geometric tracing methodology, and systematic findings of how the student learns and integrates multi-domain capabilities. Overall, we provide a \textsc{Janus} perspective to internally trace the continual evolution of the multi-domain capabilities within the model itself.

cs.CL↗

Safin-1.5: Safety-Native Architecture and Verifiable Reasoning for Recursive Safety Improvement

Recursive self-improvement (RSI) enables large language models to advance their capabilities through successive updates, but these updates can also introduce new risks and weaken existing safeguards. In line with the AI-45° Law, we pursue Safety RSI, in which safety develops as an intrinsic capability alongside general model capabilities through continual risk discovery and repair, while preserving utility and limiting over-refusal. We introduce Safin-1.5, a model built on a safety architecture and trained through recursive safety improvement, with an inference-time harness that extends its capabilities to formal reasoning. Our approach comprises three components: i) The safety architecture combines persistent safety states with hazardous-knowledge disentanglement. The former retains and selectively invokes safety knowledge, while the latter relocates targeted hazardous knowledge into removable experts, enabling its removal with minimal impact on general capabilities. ii) Recursive safety improvement establishes a discovery-evaluation-repair loop. The loop identifies new vulnerabilities and uses experimental feedback to develop repair strategies. It accepts updates only when they improve safety while preserving general capabilities and limiting over-refusal. iii) The formal reasoning harness combines the model's language understanding with symbolic inference and verification. Proof-obligation-guided semantic bridging recovers missing relations in formal representations, enabling symbolic solvers to construct checkable proofs under formalized premises and accepted semantic bridges. Experimental results demonstrate that Safin-1.5 achieves safety and reasoning performance, outperforming frontier models.

cs.CR↗

ReSI: Recursive Safety Improvement toward Resistant and Resilient AI

Recursive self-improvement, the participation of AI systems in improving their own capabilities, is beginning to move from theoretical prospect to practice, posing both challenges and opportunities for safety alignment. Models evolve through frequent updates, and their safety alignment requires continual adaptation to each new checkpoint. Meanwhile, with evolving red-teaming methods exposing new vulnerabilities, safety improvement for each checkpoint needs to mitigate exposed vulnerabilities and generalize to risks not yet revealed. Following R$^2$AI, we term these goals resistance to known threats and resilience to unforeseen risks. Recursive self-improvement, in turn, inspires an approach to both goals: safety alignment could likewise advance through successive rounds of evaluation and update. We therefore introduce ReSI, a recursive safety improvement framework that implements this approach through automated research. In each round, ReSI applies diverse red-teaming methods to identify vulnerabilities in the current target model, develops training recipes, and promotes the update with the largest safety gain among those passing a Pareto gate on capability retention as the next target model. Across four dense and mixture-of-experts models, ReSI matches or exceeds evaluated frontier models on in-distribution and out-of-distribution safety benchmarks, and outperforms alignment baselines on nearly all safety evaluations while largely preserving general capabilities. In particular, ReSI reduces the mean X-Teaming attack success rate across the four models from 86.01% to 31.45%, well below GPT-5.6-Luna's leading frontier result of 56.69%, indicating stronger resilience to attacks unseen during training. These findings support recursive safety improvement as a practical path toward resistant and resilient AI.

cs.CR↗

Internalizing Safety Understanding in Large Reasoning Models via Verification

While explicit Chain-of-Thought (CoT) empowers large reasoning models (LRMs), it enables the generation of riskier final answers. Current alignment paradigms primarily rely on externally enforced compliance, optimizing models to detect malicious prompts rather than evaluating the safety of their own outputs. We argue that this approach remains largely behavioral: our empirical analysis reveals that ostensibly aligned models lack intrinsic safety understanding, often failing to verify their own response safety and remaining vulnerable to adversarial jailbreaks. To address this fundamental limitation, we propose Safety Internal (SInternal), a framework that internalizes safety specifications by training LRMs exclusively on safety verification tasks to critique their own generated answers using expert reasoning trajectories. We demonstrate that learning to verify induces a strong generalization for response safety, significantly enhancing robustness against out-of-domain jailbreaks. Furthermore, when combined with reinforcement learning, SInternal serves as a superior initialization compared to standard supervised fine-tuning, suggesting that internalizing safety understanding creates a more robust foundation for alignment than merely mimicking safe behaviors. Our codes are available at https://github.com/AlphaLab-USTC/SInternal

cs.AI↗

Self-ReSET: Learning to Self-Recover from Unsafe Reasoning Trajectories

Large Reasoning Models possess remarkable capabilities for self-correction in general domain; however, they frequently struggle to recover from unsafe reasoning trajectories under adversarial attacks. Existing alignment methods attempt to mitigate this vulnerability by fine-tuning the model on expert data including reflection traces or adversarial prefixes. Crucially, these approaches are often hindered by static training data which inevitably deviate from model's dynamic, on-policy reasoning traces, resulting in model hardly covering its vast generation space and learning to recover from its own failures. To bridge this gap, we propose Self-ReSET, a pure reinforcement learning framework designed to equip LRMs with the intrinsic capacity to recover from their own safety error trajectories, which are subsequently reused as an initial state for reinforcement learning. Extensive experiments across various LRMs and benchmarks demonstrate that Self-ReSET significantly enhances robustness against adversarial attacks especially out-of-distribution (OOD) jailbreak prompts while maintaining general utility, along with efficient data utilization. Further analysis reveals that our method effectively fosters self-recovery patterns, enabling models to better identify and recover from unsafe intermediate error states back to benign paths. Our codes and data are available at https://github.com/Ing1024/Self-ReSET.

cs.AI↗

Owner-Harm: A Missing Threat Model for AI Agent Safety

Existing AI agent safety benchmarks focus on generic criminal harm (cybercrime, harassment, weapon synthesis), leaving a systematic blind spot for a distinct and commercially consequential threat category: agents harming their own deployers. Real-world incidents illustrate the gap: Slack AI credential exfiltration (Aug 2024), Microsoft 365 Copilot calendar-injection leaks (Jan 2024), and a Meta agent unauthorized forum post exposing operational data (Mar 2026). We propose Owner-Harm, a formal threat model with eight categories of agent behavior damaging the deployer. We quantify the defense gap on two benchmarks: a compositional safety system achieves 100% TPR / 0% FPR on AgentHarm (generic criminal harm) yet only 14.8% (4/27; 95% CI: 5.9%-32.5%) on AgentDojo injection tasks (prompt-injection-mediated owner harm). A controlled generic-LLM baseline shows the gap is not inherent to owner-harm (62.7% vs. 59.3%, delta 3.4 pp) but arises from environment-bound symbolic rules that fail to generalize across tool vocabularies. On a post-hoc 300-scenario owner-harm benchmark, the gate alone achieves 75.3% TPR / 3.3% FPR; adding a deterministic post-audit verifier raises overall TPR to 85.3% (+10.0 pp) and Hijacking detection from 43.3% to 93.3%, demonstrating strong layer complementarity. We introduce the Symbolic-Semantic Defense Generalization (SSDG) framework relating information coverage to detection rate. Two SSDG experiments partially validate it: context deprivation amplifies the detection gap 3.4x (R = 3.60 vs. R = 1.06); context injection reveals structured goal-action alignment, not text concatenation, is required for effective owner-harm detection.

cs.CR↗

Weighting-Based Treatment Effect Estimation via Distribution Learning

Existing weighting methods for treatment effect estimation are often built upon the idea of propensity scores or covariate balance. They usually impose strong assumptions on treatment assignment or outcome model to obtain unbiased estimation, such as linearity or specific functional forms, which easily leads to the major drawback of model mis-specification. In this paper, we aim to alleviate these issues by developing a distribution learning-based weighting method. We first learn the true underlying distribution of covariates conditioned on treatment assignment, then leverage the ratio of covariates' density in the treatment group to that of the control group as the weight for estimating treatment effects. Specifically, we propose to approximate the distribution of covariates in both treatment and control groups through invertible transformations via change of variables. To demonstrate the superiority, robustness, and generalizability of our method, we conduct extensive experiments using synthetic and real data. From the experiment results, we find that our method for estimating average treatment effect on treated (ATT) with observational data outperforms several cutting-edge weighting-only benchmarking methods, and it maintains its advantage under a doubly-robust estimation framework that combines weighting with some advanced outcome modeling methods.

cs.LG↗