arXiv subjects

Tong Yang

Publications and source records attributed to Tong Yang.

At least 91 records · Page 5Linked to original sources

arXiv2026

EntroCoT: Enhancing Chain-of-Thought via Adaptive Entropy-Guided Segmentation

Chain-of-Thought (CoT) prompting has significantly enhanced the mathematical reasoning capabilities of Large Language Models. We find existing fine-tuning datasets frequently suffer from the "answer right but reasoning wrong" probelm, where correct final answers are derived from hallucinated, redundant, or logically invalid intermediate steps. This paper proposes EntroCoT, a unified framework for automatically identifying and refining low-quality CoT supervision traces. EntroCoT first proposes an entropy-based mechanism to segment the reasoning trace into multiple steps at uncertain junctures, and then introduces a Monte Carlo rollout-based mechanism to evaluate the marginal contribution of each step. By accurately filtering deceptive reasoning samples, EntroCoT constructs a high-quality dataset where every intermediate step in each reasoning trace facilitates the final answer. Extensive experiments on mathematical benchmarks demonstrate that fine-tuning on the subset constructed by EntroCoT consistently outperforms the baseslines of full-dataset supervision.

cs.AI↗

arXiv2026

LPFQA: A Long-Tail Professional Forum-based Benchmark for LLM Evaluation

Large Language Models (LLMs) perform well on standard reasoning and question-answering benchmarks, yet such evaluations often fail to capture their ability to handle long-tail, expertise-intensive knowledge in real-world professional scenarios. We introduce LPFQA, a long-tail knowledge benchmark derived from authentic professional forum discussions, covering 7 academic and industrial domains with 430 curated tasks grounded in practical expertise. LPFQA evaluates specialized reasoning, domain-specific terminology understanding, and contextual interpretation, and adopts a hierarchical difficulty structure to ensure semantic clarity and uniquely identifiable answers. Experiments on over multiple mainstream LLMs reveal substantial performance gaps, particularly on tasks requiring deep domain reasoning, exposing limitations overlooked by existing benchmarks. Overall, LPFQA provides an authentic and discriminative evaluation framework that complements prior benchmarks and informs future LLM development.

cs.AI↗

arXiv2026

NL2Repo-Bench: Towards Long-Horizon Repository Generation Evaluation of Coding Agents

Recent advances in coding agents suggest rapid progress toward autonomous software development, yet existing benchmarks fail to rigorously evaluate the long-horizon capabilities required to build complete software systems. Most prior evaluations focus on localized code generation, scaffolded completion, or short-term repair tasks, leaving open the question of whether agents can sustain coherent reasoning, planning, and execution over the extended horizons demanded by real-world repository construction. To address this gap, we present NL2Repo Bench, a benchmark explicitly designed to evaluate the long-horizon repository generation ability of coding agents. Given only a single natural-language requirements document and an empty workspace, agents must autonomously design the architecture, manage dependencies, implement multi-module logic, and produce a fully installable Python library. Our experiments across state-of-the-art open- and closed-source models reveal that long-horizon repository generation remains largely unsolved: even the strongest agents achieve below 40% average test pass rates and rarely complete an entire repository correctly. Detailed analysis uncovers fundamental long-horizon failure modes, including premature termination, loss of global coherence, fragile cross-file dependencies, and inadequate planning over hundreds of interaction steps. NL2Repo Bench establishes a rigorous, verifiable testbed for measuring sustained agentic competence and highlights long-horizon reasoning as a central bottleneck for the next generation of autonomous coding agents.

cs.CL↗

arXiv2025

Interfacial Polarons Driven by Charge Transfer In WSe2/Cuprate Superconductor Systems

Understanding the electronic properties of doped copper-oxygen planes remains a significant challenge in condensed matter physics and is crucial to unraveling the mechanisms behind high-temperature superconductivity in cuprates. Recently, the observation of charge transfer and interfacial polarons in superconducting interface has aroused extensive research interest. However, experimental data to investigate charge transfer on the CuO2 plane and the presence of polarons are still missing. Here we conduct extensive research on the optical and electronic properties of two-dimensional material supported on copper-based superconductors. Unlike monolayer-WSe2 on other substrates, monolayer-WSe2 on La1.85Sr0.15CuO4 (WSe2/LSCO) produces a special band structure. Using high-resolution spectroscopic ellipsometry and density functional theory calculation methods, the special electronic structure can be attributed to the formation of the interfacial small polaron at the WSe2/LSCO interface which is driven by charge transfer between the CuO2 plane of the cuprate superconductor and WSe2. In addition, the structural phase transition of the LSCO substrate was observed to reduce the electron-hole (e-h) interaction of WSe2. These findings may spur future investigations on the effect of the interfacial polaron on the superconductivity of cuprates, and highlight the significant influence of interface effects on the electronic structure of WSe2 films. It provides an effective method to further explore the intrinsic relationship between interfacial polarons and superconductivity.

cond-mat.supr-con↗

arXiv2025

"Even GPT Can Reject Me": Conceptualizing Abrupt Refusal Secondary Harm (ARSH) and Reimagining Psychological AI Safety with Compassionate Completion Standard (CCS)

Large Language Models (LLMs) and AI chatbots are increasingly used for emotional and mental health support due to their low cost, immediacy, and accessibility. However, when safety guardrails are triggered, conversations may be abruptly terminated, introducing a distinct form of emotional disruption that can exacerbate distress and elevate risk among already vulnerable users. As this phenomenon gains attention, this viewpoint introduces Abrupt Refusal Secondary Harm (ARSH) as a conceptual framework to describe the psychological impacts of sudden conversational discontinuation caused by AI safety protocols. Drawing on counseling psychology and communication science as conceptual heuristics, we argue that abrupt refusals can rupture perceived relational continuity, evoke feelings of rejection or shame, and discourage future help seeking. To mitigate these risks, we propose a design hypothesis, the Compassionate Completion Standard (CCS), a refusal protocol grounded in Human Centered Design (HCD) that maintains safety constraints while preserving relational coherence. CCS emphasizes empathetic acknowledgment, transparent boundary articulation, graded conversational transition, and guided redirection, replacing abrupt disengagement with psychologically attuned closure. By integrating awareness of ARSH into AI safety design, developers and policymakers can reduce preventable iatrogenic harm and advance a more psychologically informed approach to AI governance. Rather than presenting incremental empirical findings, this viewpoint contributes a timely conceptual framework, articulates a testable design hypothesis, and outlines a coordinated research agenda for improving psychological safety in human AI interaction.

cs.CY↗

arXiv2025

Multi-head Transformers Provably Learn Symbolic Multi-step Reasoning via Gradient Descent

Transformers have demonstrated remarkable capabilities in multi-step reasoning tasks. However, understandings of the underlying mechanisms by which they acquire these abilities through training remain limited, particularly from a theoretical standpoint. This work investigates how transformers learn to solve symbolic multi-step reasoning problems through chain-of-thought processes, focusing on path-finding in trees. We analyze two intertwined tasks: a backward reasoning task, where the model outputs a path from a goal node to the root, and a more complex forward reasoning task, where the model implements two-stage reasoning by first identifying the goal-to-root path and then reversing it to produce the root-to-goal path. Our theoretical analysis, grounded in the dynamics of gradient descent, shows that trained one-layer transformers can provably solve both tasks with generalization guarantees to unseen trees. In particular, our multi-phase training dynamics for forward reasoning elucidate how different attention heads learn to specialize and coordinate autonomously to solve the two subtasks in a single autoregressive path. These results provide a mechanistic explanation of how trained transformers can implement sequential algorithmic procedures. Moreover, they offer insights into the emergence of reasoning abilities, suggesting that when tasks are structured to take intermediate chain-of-thought steps, even shallow multi-head transformers can effectively solve problems that would otherwise require deeper architectures.

cs.LG↗

arXiv2025

SUPERChem: A Multimodal Reasoning Benchmark in Chemistry

Current benchmarks for evaluating the chemical reasoning capabilities of Large Language Models (LLMs) are limited by oversimplified tasks, lack of process-level evaluation, and misalignment with expert-level chemistry skills. To address these issues, we introduce SUPERChem, a benchmark of 500 expert-curated reasoning-intensive chemistry problems, covering diverse subfields and provided in both multimodal and text-only formats. Original content and an iterative curation pipeline eliminate flawed items and mitigate data contamination. Each problem is paired with an expert-authored solution path, enabling Reasoning Path Fidelity (RPF) scoring to evaluate reasoning quality beyond final-answer accuracy. Evaluations against a human baseline of 40.3% accuracy show that even the best-performing model, GPT-5 (High), reaches only 38.5%, followed closely by Gemini 2.5 Pro (37.9%) and DeepSeek-V3.1-Think (37.3%). SUPERChem elicits multi-step, multimodal reasoning, reveals model-dependent effects of visual information, and distinguishes high-fidelity reasoners from heuristic ones. By providing a challenging benchmark and a reliable evaluation framework, SUPERChem aims to facilitate the advancement of LLMs toward expert-level chemical intelligence. The dataset of the benchmark is available at https://huggingface.co/datasets/ZehuaZhao/SUPERChem.

cs.CL↗

arXiv2025

KVReviver: Reversible KV Cache Compression with Sketch-Based Token Reconstruction

As the context length of current large language models (LLMs) rapidly increases, the memory demand for the Key-Value (KV) cache is becoming a bottleneck for LLM deployment and batch processing. Traditional KV cache compression methods typically involve permanently evicting or irreversibly merging "less important" tokens with low attention scores. This approach results in the unrecoverable loss of token information, which we call Contextual Amnesia, significantly degrading the model's information retrieval capability. To address this issue, we propose KVReviver, a reversible KV cache compression method based on the sketch algorithm. This method allows reconstructing compressed tokens from an additional data structure, thus enabling full-scale computation within limited memory. Experiments showed that in 2k-length contexts, it requires only 10% of KV Cache budget while maintaining identical end-to-end inference accuracy. For 32k-length contexts, it achieves equivalent or comparable accuracy ~2% accuracy loss) using merely 25% of KV Cache budget.

cs.CL↗

arXiv2025

KeepKV: Achieving Periodic Lossless KV Cache Compression for Efficient LLM Inference

Efficient inference of large language models (LLMs) is hindered by an ever-growing key-value (KV) cache, making KV cache compression a critical research direction. Traditional methods selectively evict less important KV cache entries, which leads to information loss and hallucinations. Recently, merging-based strategies have been explored to retain more information by merging KV pairs that would be discarded; however, these existing approaches inevitably introduce inconsistencies in attention distributions before and after merging, causing degraded generation quality. To overcome this challenge, we propose KeepKV, a novel adaptive KV cache merging method designed to preserve performance under strict memory constraints, achieving single-step lossless compression and providing error bounds for multi-step compression. KeepKV introduces the Electoral Votes mechanism that records merging history and adaptively adjusts attention scores. Moreover, it further leverages a novel Zero Inference-Perturbation Merging method, compensating for attention loss resulting from cache merging. Extensive experiments on various benchmarks and LLM architectures demonstrate that KeepKV substantially reduces memory usage while successfully retaining essential context information, achieving over 2x inference throughput improvement and maintaining superior generation quality even with only 10% KV cache budgets.

cs.LG↗

arXiv2025

MagnifierSketch: Quantile Estimation Centered at One Point

In this paper, we take into consideration quantile estimation in data stream models, where every item in the data stream is a key-value pair. Researchers sometimes aim to estimate per-key quantiles (i.e. quantile estimation for every distinct key), and some popular use cases, such as tail latency measurement, recline on a predefined single quantile (e.g. 0.95- or 0.99- quantile) rather than demanding arbitrary quantile estimation. However, existing algorithms are not specially designed for per-key estimation centered at one point. They cannot achieve high accuracy in our problem setting, and their throughput are not satisfactory to handle high-speed items in data streams. To solve this problem, we propose MagnifierSketch for point-quantile estimation. MagnifierSketch supports both single-key and per-key quantile estimation, and its key techniques are named Value Focus, Distribution Calibration and Double Filtration. We provide strict mathematical derivations to prove the unbiasedness of MagnifierSketch and show its space and time complexity. Our experimental results show that the Average Error (AE) of MagnifierSketch is significantly lower than the state-of-the-art in both single-key and per-key situations. We also implement MagnifierSketch on RocksDB database to reduce quantile query latency in real databases. All related codes of MagnifierSketch are open-sourced and available at GitHub.

cs.DS↗

arXiv2025

MergeMoE: Efficient Compression of MoE Models via Expert Output Merging

The Mixture-of-Experts (MoE) technique has proven to be a promising solution to efficiently scale the model size, which has been widely applied in recent LLM advancements. However, the substantial memory overhead of MoE models has made their compression an important research direction. In this work, we provide a theoretical analysis of expert merging, a recently proposed technique for compressing MoE models. Rather than interpreting expert merging from the conventional perspective of parameter aggregation, we approach it from the perspective of merging experts' outputs. Our key insight is that the merging process can be interpreted as inserting additional matrices into the forward computation, which naturally leads to an optimization formulation. Building on this analysis, we introduce MergeMoE, a method that leverages mathematical optimization to construct the compression matrices. We evaluate MergeMoE on multiple MoE models and show that our algorithm consistently outperforms the baselines with the same compression ratios.

cs.LG↗

arXiv2025

Fast Visuomotor Policy for Robotic Manipulation

We present a fast and effective policy framework for robotic manipulation, named Energy Policy, designed for high-frequency robotic tasks and resource-constrained systems. Unlike existing robotic policies, Energy Policy natively predicts multimodal actions in a single forward pass, enabling high-precision manipulation at high speed. The framework is built upon two core components. First, we adopt the energy score as the learning objective to facilitate multimodal action modeling. Second, we introduce an energy MLP to implement the proposed objective while keeping the architecture simple and efficient. We conduct comprehensive experiments in both simulated environments and real-world robotic tasks to evaluate the effectiveness of Energy Policy. The results show that Energy Policy matches or surpasses the performance of state-of-the-art manipulation methods while significantly reducing computational overhead. Notably, on the MimicGen benchmark, Energy Policy achieves superior performance with at a faster inference compared to existing approaches.

cs.RO↗

arXiv2025

Reverse-Engineered Reasoning for Open-Ended Generation

While the ``deep reasoning'' paradigm has spurred significant advances in verifiable domains like mathematics, its application to open-ended, creative generation remains a critical challenge. The two dominant methods for instilling reasoning -- reinforcement learning (RL) and instruction distillation -- falter in this area; RL struggles with the absence of clear reward signals and high-quality reward models, while distillation is prohibitively expensive and capped by the teacher model's capabilities. To overcome these limitations, we introduce REverse-Engineered Reasoning (REER), a new paradigm that fundamentally shifts the approach. Instead of building a reasoning process ``forwards'' through trial-and-error or imitation, REER works ``backwards'' from known-good solutions to computationally discover the latent, step-by-step deep reasoning process that could have produced them. Using this scalable, gradient-free approach, we curate and open-source DeepWriting-20K, a large-scale dataset of 20,000 deep reasoning trajectories for open-ended tasks. Our model, DeepWriter-8B, trained on this data, not only surpasses strong open-source baselines but also achieves performance competitive with, and at times superior to, leading proprietary models like GPT-4o and Claude 3.5.

cs.AI↗

arXiv2025

Inverse IFEval: Can LLMs Unlearn Stubborn Training Conventions to Follow Real Instructions?

Large Language Models (LLMs) achieve strong performance on diverse tasks but often exhibit cognitive inertia, struggling to follow instructions that conflict with the standardized patterns learned during supervised fine-tuning (SFT). To evaluate this limitation, we propose Inverse IFEval, a benchmark that measures models Counter-intuitive Abilitytheir capacity to override training-induced biases and comply with adversarial instructions. Inverse IFEval introduces eight types of such challenges, including Question Correction, Intentional Textual Flaws, Code without Comments, and Counterfactual Answering. Using a human-in-the-loop pipeline, we construct a dataset of 1012 high-quality Chinese and English questions across 23 domains, evaluated under an optimized LLM-as-a-Judge framework. Experiments on existing leading LLMs demonstrate the necessity of our proposed Inverse IFEval benchmark. Our findings emphasize that future alignment efforts should not only pursue fluency and factual correctness but also account for adaptability under unconventional contexts. We hope that Inverse IFEval serves as both a diagnostic tool and a foundation for developing methods that mitigate cognitive inertia, reduce overfitting to narrow patterns, and ultimately enhance the instruction-following reliability of LLMs in diverse and unpredictable real-world scenarios.

cs.CL↗

arXiv2025

Router Upcycling: Leveraging Mixture-of-Routers in Mixture-of-Experts Upcycling

The Mixture-of-Experts (MoE) models have gained significant attention in deep learning due to their dynamic resource allocation and superior performance across diverse tasks. However, efficiently training these models remains challenging. The MoE upcycling technique has been proposed to reuse and improve existing model components, thereby minimizing training overhead. Despite this, simple routers, such as linear routers, often struggle with complex routing tasks within MoE upcycling. In response, we propose a novel routing technique called Router Upcycling to enhance the performance of MoE upcycling models. Our approach initializes multiple routers from the attention heads of preceding attention layers during upcycling. These routers collaboratively assign tokens to specialized experts in an attention-like manner. Each token is processed into diverse queries and aligned with the experts' features (serving as keys). Experimental results demonstrate that our method achieves state-of-the-art (SOTA) performance, outperforming other upcycling baselines.

cs.CL↗

arXiv2025

Adapting Foundation Model for Dental Caries Detection with Dual-View Co-Training

Accurate dental caries detection from panoramic X-rays plays a pivotal role in preventing lesion progression. However, current detection methods often yield suboptimal accuracy due to subtle contrast variations and diverse lesion morphology of dental caries. In this work, inspired by the clinical workflow where dentists systematically combine whole-image screening with detailed tooth-level inspection, we present DVCTNet, a novel Dual-View Co-Training network for accurate dental caries detection. Our DVCTNet starts with employing automated tooth detection to establish two complementary views: a global view from panoramic X-ray images and a local view from cropped tooth images. We then pretrain two vision foundation models separately on the two views. The global-view foundation model serves as the detection backbone, generating region proposals and global features, while the local-view model extracts detailed features from corresponding cropped tooth patches matched by the region proposals. To effectively integrate information from both views, we introduce a Gated Cross-View Attention (GCV-Atten) module that dynamically fuses dual-view features, enhancing the detection pipeline by integrating the fused features back into the detection model for final caries detection. To rigorously evaluate our DVCTNet, we test it on a public dataset and further validate its performance on a newly curated, high-precision dental caries detection dataset, annotated using both intra-oral images and panoramic X-rays for double verification. Experimental results demonstrate DVCTNet's superior performance against existing state-of-the-art (SOTA) methods on both datasets, indicating the clinical applicability of our method. Our code and labeled dataset are available at https://github.com/ShanghaiTech-IMPACT/DVCTNet.

cs.CV↗

arXiv2025

iFairy: the First 2-bit Complex LLM with All Parameters in $\{\pm1, \pm i\}$

Quantization-Aware Training (QAT) integrates quantization into the training loop, enabling LLMs to learn robust low-bit representations, and is widely recognized as one of the most promising research directions. All current QAT research focuses on minimizing quantization error on full-precision models, where the full-precision accuracy acts as an upper bound (accuracy ceiling). No existing method has even attempted to surpass this ceiling. To break this ceiling, we propose a new paradigm: raising the ceiling (full-precision model), and then still quantizing it efficiently into 2 bits. We propose Fairy$\pm i$, the first 2-bit quantization framework for complex-valued LLMs. Specifically, our method leverages the representational advantages of the complex domain to boost full-precision accuracy. We map weights to the fourth roots of unity $\{\pm1, \pm i\}$, forming a perfectly symmetric and information-theoretically optimal 2-bit representation. Importantly, each quantized weight has either a zero real or imaginary part, enabling multiplication-free inference using only additions and element swaps. Experimental results show that Fairy$\pm i$ outperforms the ceiling of existing 2-bit quantization approaches in terms of both PPL and downstream tasks, while maintaining strict storage and compute efficiency. This work opens a new direction for building highly accurate and practical LLMs under extremely low-bit constraints.

cs.LG↗

arXiv2025

Semantic Encryption: Secure and Effective Interaction with Cloud-based Large Language Models via Semantic Transformation

The increasing adoption of Cloud-based Large Language Models (CLLMs) has raised significant concerns regarding data privacy during user interactions. While existing approaches primarily focus on encrypting sensitive information, they often overlook the logical structure of user inputs. This oversight can lead to reduced data utility and degraded performance of CLLMs. To address these limitations and enable secure yet effective interactions, we propose Semantic Encryption (SE)-a plug-and-play framework designed to preserve both privacy and utility. SE consists of two key components: Semantic Encoding and Semantic Decoding. In the encoding phase, a lightweight local model transforms the original user input into an alternative semantic context that maintains the original intent and logical structure while obfuscating sensitive information. This transformed input is then processed by the CLLM, which generates a response based on the transformed semantic context. To maintain a seamless user experience, the decoding phase will reconstruct the CLLM's response back into the original semantic context by referencing the locally stored user input. Extensive experimental evaluations demonstrate that SE effectively protects data privacy without compromising data utility or user experience, offering a practical solution for secure interaction with CLLMs. Particularly, the proposed SE demonstrates a significant improvement over the state-of-the-art InferDPT, surpassing it across various evaluated metrics and datasets.

cs.CR↗