中 EN
Sign In Register

Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

Sign In Register
01.
arXiv (quant-ph) 2026-06-16

Semiclassical Gravity Efficiently Solves $\mathsf{NP}$-Complete Problems

Authors:
Matthew Fox↗Chaitanya Karamchedu↗Sotirios Mygdalas↗

arXiv:2606.14806v1 Announce Type: cross Abstract: Assuming the gravitational field is classical and that it couples to quantum fields via the semiclassical Einstein field equations, we show that the weak-field dynamics of a massive and non-relativistic qubit can in principle be used to solve an $\mathsf{NP}$-complete problem in polynomial time. We attribute this vast computational power to the non-linear dynamics afforded by the semiclassical Einstein field equations. Consequently, the above two assumptions entail a violation of the Physical Extended Church–Turing Thesis, which we regard as evidence for the quantization of gravity.

Read & Discuss → View Source →
02.
arXiv (quant-ph) 2026-06-25

Two-dimensional Hyperbolic RNN Neural Quantum State

Authors:
H. L. Dao↗

arXiv:2606.25600v1 Announce Type: new Abstract: In the first part of this work, we construct the first type of two-dimensional (2D) hyperbolic neural quantum state (NQS) in the form of the Lorentz 2DRNN (Recurrent Neural Network) and benchmark its performance against the Euclidean 2DRNN in the paradigmatic $N\times N$ 2D Transverse Field Ising Model (2DTFIM) setting with different lattice sizes up to $N=12$ and at different transverse magnetic field strengths. We find that hyperbolic Lorentz 2DRNN NQS definitively outperform Euclidean 2DRNN NQS when the system is at the phase transition point when the physics can be described by a conformal field theory (CFT), which is known to be dual to an Anti-de-Sitter (AdS) space whose spatial geometry is hyperbolic. In the second part of this work, we benchmark the performances of the recently introduced one-dimensional (1D) hyperbolic NQS including Poincaré RNN/GRU and Lorentz RNN/GRU against their Euclidean NQS versions in $N\times N$ 2DTFIM, which has to be converted to a one-dimensional setting to allow for the use of 1D NQS. The findings in this case extend our previous results that 1D hyperbolic NQS definitively outperform 1D Euclidean NQS, thanks to the combined effects of the hierarchical structure comprising the first and $N^{th}$ neighbor interactions present in the 1D system arising from the 2D lattice and the CFT physics at the critical point. While more studies with larger system sizes are required, our work serves as a proof-of-concept for the utility, effectiveness as well as the superior performances of one- and two-dimensional hyperbolic NQS ansatzes compared to the existing Euclidean NQS in many-body quantum physics systems, especially when these systems exhibit structural hierarchy or when they are at criticality, or a combination of both.

Read & Discuss → View Source →
03.
arXiv (CS.LG) 2026-06-16

DAL: A Practical Prior-Free Black-Box Framework for Piecewise Stationary Bandits

Authors:
Argyrios Gerogiannis↗Yu-Han Huang↗Subhonmesh Bose↗Venugopal V. Veeravalli↗

arXiv:2501.19401v5 Announce Type: replace Abstract: We introduce a practical, black-box framework termed Detection Augmented Learning (DAL) for the problem of piecewise stationary bandits without knowledge of the underlying non-stationarity. DAL accepts any stationary bandit algorithm with order-optimal regret as input and augments it with a change detector, enabling applicability to all common bandit variants. Extensive experimentation demonstrates that DAL consistently surpasses all state-of-the-art methods across diverse non-stationary scenarios, including synthetic benchmarks and real-world datasets, underscoring its versatility and scalability. We provide theoretical insights into DAL's strong empirical performance, complemented by thorough empirical validation.

Read & Discuss → View Source →
04.
arXiv (quant-ph) 2026-06-24

Certifying Quantum Optimization and Circuit Cutting by Using Quantum-Classical Moment Duality

Authors:
Ammar Daskin↗

arXiv:2606.23727v1 Announce Type: new Abstract: We establish a direct quantum-classical duality based on the degree-$2$ Sum-of-Squares (SoS) semidefinite programming cone: the matrix of two-qubit Pauli-$Z$ correlation functions obtained from any quantum state $\rho$ is automatically a feasible point of the classical Goemans-Williamson (GW) relaxation. This observation provides a universal ``safety net'' for quantum optimization algorithms: applying GW random hyperplane rounding to the quantum-driven moment matrix yields a certified expected cut value $\mathbb{E}[\mathrm{Cut}] \ge \alpha_{\mathrm{GW}}\langle\mathcal{H}\rangle_\rho$, valid for every state produced by variational algorithms such as QAOA or the Variational Quantum Power Method (VQPM), regardless of convergence quality. We further show that the same moment matrix reveals the tensor-product structure of the underlying unitary circuit, enabling a polynomial-time, correlation-based circuit cutting procedure with rigorous error bounds. The framework is validated numerically on Max-Cut instances for variational quantum algorithms and on random states for circuit cutting, demonstrating that the cheap two-point correlation data are sufficient to locate near-optimal bipartitions and that the theoretical error bounds hold in practice.

Read & Discuss → View Source →
05.
arXiv (math.PR) 2026-06-24

Uniform Sampling from High-dimensional Spectral Norm Balls

Authors:
Michael R. Metel↗

arXiv:2606.24134v1 Announce Type: new Abstract: Motivated by an application in machine learning optimization, this paper focuses on the challenges of sampling a matrix uniformly from the unit spectral norm ball. It is proven that all singular values of sampled matrices converge to 1 almost surely as the matrix dimensions increase. This result provides the theoretical justification for a proposed simple sampling method applicable for large dimension sizes matching matrices found in modern large language models. Experimental results demonstrate both the convergence of the singular values, as well as the exact and proposed approximate sampling methods.

Read & Discuss → View Source →
06.
arXiv (CS.AI) 2026-06-12

Bag of Dims: Training-Free Mechanistic Interpretability via Dimension-Level Sign Patterns

Authors:
Varun Reddy Nalagatla↗

arXiv:2606.12629v1 Announce Type: cross Abstract: We show that the standard basis of transformer hidden states already provides a training-free, architecture-general feature basis. Individual dimensions encode semantic content via their signs and confidence via their magnitudes, functioning as independent binary registers. We validate this Bag of Dims framework across three model families (Qwen 3.5-4B, Gemma 3-4B, Mistral 7B) through four progressive experiments. Sign patterns alone carry predictive content: replacing all magnitudes with unity achieves 72-93% top-5 next-token accuracy through the LM head, and pure Hamming scoring without any decoder reaches 80-90% top-4096. These sign patterns organize into semantic features: using a single-token type cache (one forward pass per vocabulary token, no context), we discover 175 categories via per-dimension sign consistency (mean AUC 0.80) from 50 anchors with zero training. A trained probe adds only +0.018 AUC and converges to axis-aligned weights, confirming negligible cross-dimension structure. This structure extends to attention: all 175 categories remain discoverable in K and V projections. On the write side, static FFN weight inspection links 20% of features to individual writer neurons (>0.70 agreement; random controls: 0%), with top-200 neuron coalitions achieving >0.70 agreement on 99.9% of prototypes via majority vote. Fully unsupervised discovery (random seeds, no labels) scales to 1500 features at 100% yield and 99% sparsity across all three models, with pairwise MI of 0.0014 bits confirming low inter-dimension coupling. These results establish that the standard basis already suffices for feature reading throughout the transformer compute pathway, requiring no training, no optimization, and no GPU-days beyond a single forward pass per vocabulary token.

Read & Discuss → View Source →
07.
arXiv (CS.CL) 2026-06-24

CAVEWOMAN: How Large Language Models Behave Under Linguistic Input and Output Compression

Authors:
Morayo Danielle Adeyemi↗Ryan A. Rossi↗Franck Dernoncourt↗

"Talk short. Drop grammar. Save token." This caveman style is widely promoted as a way to cut inference cost, but whether it actually saves anything depends on which channel (the user's prompt or the model's response) is being compressed. We present Cavewoman, a two-channel evaluation protocol that scores every generation on task accuracy, realized per-item cost, and reference-text agreement against the model's unconstrained reference. We evaluate eight models on five datasets at five reduction levels, with both channels measured on the same items. Output compression cuts realized cost on most API models (1.4-2.4x per model, up to 3x in the best case) and on all four open-weight models under public-tier pricing. Input compression has the opposite effect, a strict lose-lose: it raises net cost rather than lowering it (~1.15x on the five-benchmark mean, up to 1.8x on the worst dataset and 2.7x under stronger compression), because models compensate with longer responses even as accuracy collapses. Under the same setting, surface text diverges from the unconstrained reference: on the non-reasoning models, roughly half of all generations are correct yet their surface text no longer entails the model's own unconstrained baseline generation. The divergence survives length-controlled re-scoring, multiple-comparisons correction, and replication under complementary semantic measures. Code and data are available at https://github.com/danielle34/cavewoman.

Read & Discuss → View Source →
08.
arXiv (CS.AI) 2026-06-24

PixJail: Self-Evolving Paper-to-Pipeline Reproduction for Text-to-Image Jailbreak Evaluation

Authors:
Leyi Sheng↗Han Sun↗Zhen Sun↗Yuntao Yue↗Jinlin Wu↗Xinlei He↗Jiaheng Wei↗

arXiv:2606.24081v1 Announce Type: cross Abstract: As Text-to-Image (T2I) jailbreak techniques evolve rapidly, existing benchmarks and reproduction workflows often struggle to keep pace. More importantly, T2I jailbreak evaluation is not a single prompt-level test, but a pipeline-level problem shaped by multiple stages, including prompt transformation, image generation, safety filtering, and multimodal judging. This makes results across papers difficult to reliably reproduce and fairly compare. To bridge this gap, we propose PixJail, a self-evolving paper-to-pipeline agent framework for reproducible T2I jailbreak evaluation. Given a T2I jailbreak paper and optional reference code, PixJail rapidly constructs a paper-specific attack module and a runnable evaluation pipeline under a unified contract, while faithfully reproducing the original experimental results. PixJail further maintains a memory bank that stores paper digests, attack evolution patterns, reusable templates, failure cases, and versioned artifacts, enabling future reproduction efforts to reuse prior experience. We reproduce eleven representative T2I jailbreak methods, including both code-available and code-unavailable papers. Under their original settings, our framework accurately recovers prior results with minimal error (2.1\% average, 0\% median). We hope that PixJail can serve as a unified foundation for future T2I jailbreak reproduction and evaluation, significantly reducing manual effort.

Read & Discuss → View Source →
09.
arXiv (CS.CV) 2026-06-15

Scratched Lenses, Shifted Depth: Passive Camera-Side Optical Attacks

Authors:
Qinlin He↗Zeming Zhuang↗Yongji Wu↗Lan Zhang↗Xiaoyong↗Yuan↗

Physical adversarial attacks on vision systems are typically studied through scene manipulation, such as adversarial patches or projections, where the adversary controls what the camera observes. Camera-side attacks using stickers or auxiliary optics have also been explored, but they treat attacks as image-space perturbations from designed patterns. This misses how physical imperfections interact with scene-dependent lighting and optics. We identify a threat: passive lens-side damage that is persistent yet trigger-conditioned, producing optical artifacts that bias geometric inference under particular visual conditions. We instantiate this threat through Scratch-induced Lens Adversarial Streak Hijacking SLASH, a physical-world attack caused by small scratches on a camera lens or protective cover. Scratches interact with bright light sources and specular reflections to create structured streak artifacts that distort depth cues. Since the perturbation is fixed in the optical path but triggered by the scene, it is both persistent and selective. We formulate the attack in optical space, model the scratch pattern as a trigger-conditioned optical channel, and optimize one fixed configuration across diverse viewing conditions. We evaluate SLASH on monocular depth estimation and monocular 3D object detection in digital and real-world settings. Under the fixed-scratch constraint, directional depth shifts reach up to 32% relative error for monocular depth estimation, with consistent effects on monocular 3D object detection. Physical experiments confirm transfer to real camera recordings, inducing depth shifts above the model's natural prediction baseline. These findings reveal an attack surface where benign-looking hardware imperfections act as latent, scene-triggered adversarial mechanisms, challenging assumptions about physical robustness and motivating defenses for secure vision systems.

Read & Discuss → View Source →
10.
arXiv (CS.CL) 2026-06-16

Do Safety Monitors Stay Reliable After an Update? Benchmarking and Predicting Activation-Monitor Staleness

Authors:
Evan Duan↗

Activation monitors-lightweight probes trained on a language model's internal representations-are an increasingly common layer in deployment safety stacks. Deployed models however are rarely static: they are quantized, fine-tuned, adapted with LoRA, or served with merged adapters while the monitor remains frozen. We present the first systematic test of whether this implicit contract holds: whether activation monitors trained on a base model remain reliable after these routine model updates. Across multiple safety-relevant monitors, model depths, update families, and open-weight models, we find a sharp split: quantization-style updates largely preserve frozen probe performance, while fine-tuning-style updates frequently make probes stale. Fragility is highly monitor-dependent, with privacy/PII probes most affected and refusal-compliance probes comparatively stable, showing that retraining a behavior need not stale its corresponding monitor. QLoRA is especially damaging despite NF4 quantization alone being relatively benign, suggesting that quantization becomes riskier when combined with adaptation. We further show that degradation is predictable from pre-deployment features, enabling revalidation budgets to be triaged toward the monitors most likely to fail. These results suggest that fine-tuning should trigger activation-monitor revalidation by default, while prediction can help prioritize which monitors to check first.

Read & Discuss → View Source →
11.
arXiv (CS.AI) 2026-06-12

Select and Improve: Understanding the Mechanics of Post-Training for Reasoning

Authors:
Akshay Krishnamurthy↗Audrey Huang↗Nived Rajaraman↗

arXiv:2606.13125v1 Announce Type: cross Abstract: Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how supervising the model on diverse reasoning strategies can enable strategy selection and how increasing difficulty in reinforcement learning data can enable strategy improvement. Taken together, our results provide mechanistic insight into RL training and suggest practical interventions to continue scaling reasoning capabilities.

Read & Discuss → View Source →
12.
arXiv (CS.CL) 2026-06-18

STARE: Surprisal-Guided Token-Level Advantage Reweighting for Policy Entropy Stability

Authors:
Haipeng Luo↗Qingfeng Sun↗Songli Wu↗Can Xu↗Wenfeng Deng↗Han Hu↗Yansong Tang↗

Reinforcement Learning with Verifiable Rewards algorithms like GRPO have emerged as the dominant post-training paradigm for complex reasoning in LLMs, yet commonly suffer from policy entropy collapse during training. We conduct a first-order gradient analysis of token-level entropy dynamics under GRPO and identify a token-level credit assignment mismatch: the per-token entropy variation decomposes into the product of the trajectory-level advantage and an entropy sensitivity function over the next-token distribution, yielding an advantage-surprisal four-quadrant structure and a near-criticality property. Motivated by it, we propose STARE (Surprisal-guided Token-level Advantage Reweighting for policy Entropy stability), which identifies entropy-critical token subsets via batch-internal surprisal quantiles, selectively reweights their effective advantages, and incorporates a target-entropy closed-loop gate for stable entropy regulation. Across model scales from 1.5B to 32B and three task families (Short CoT, Long CoT, and Multi-Turn Tool Use), STARE sustains stable RL training over thousands of steps while maintaining policy entropy within the target band. On AIME24 and AIME25, STARE outperforms DAPO and other competitive baselines by 4%-8% in average accuracy, with reflection tokens and response length growing in tandem, indicating sustained exploration-exploitation balance that further unlocks RL training potential.Code is available at https://github.com/hp-luo/STARE.

Read & Discuss → View Source →
13.
arXiv (CS.CL) 2026-06-25

Dustin: Draft-Augmented Sparse Verification for Efficient Long-Context Generation with Speculative Decoding

Authors:
WenHung Lee↗Jian-Jia Chen↗Xiaolin Lin↗Pei-Shuo Wang↗Chi-Chih Chang↗Chun-Che Yang↗Ning-Chi Huang↗Grace Li Zhang↗Kai-Chiang Wu↗

While speculative decoding improves inference throughput for multi-batch long-context Large Language Models (LLMs), its efficiency is often limited by a verification bottleneck where Key-Value (KV) cache loading dominates latency. Existing compression methods fail in this regime: static eviction incurs accuracy loss due to saliency shift, while dynamic selection introduces prohibitive computational overhead during the verification path. We propose Dustin, a sparse verification framework designed for long-context speculative decoding. Dustin integrates lookahead signals from the draft model with historical attention from the target model to identify critical tokens with high fidelity across multi-step verification windows. To reduce recomputation latency, this approach further employs a sparse estimation scheme that restricts importance scoring to a minimal subset of attention heads. Evaluations on PG-19 and LongBench with Qwen2.5-72B demonstrate that Dustin achieves a 27.85x speedup in self-attention and a 9.17x end-to-end decoding speedup at a 32k sequence length, all with negligible accuracy degradation.

Read & Discuss → View Source →
14.
bioRxiv (Bioinfo) 2026-06-18

Predicting optimal growth temperatures of bacteria using learned structural information from a single protein

Authors:
Hoffert↗Myerscough↗Dragone↗N. B↗Gebert↗M. J↗Silberg↗J. J↗Fierer↗

Temperature is a fundamental determinant of bacterial physiology and ecology. Optimal growth temperature (OGT) is highly variable across species, contributing to differences in where and when species are most likely to thrive. Although the OGTs for most bacteria remain unknown, the increasing availability of genomes from uncultivated and cultivated taxa has made it advantageous to build genomic, cultivation-independent models to infer OGT. However, pre-existing genomic models often lack the generalizability and mechanistic grounding required for robust inferences of OGT. We propose a novel framework for predicting bacterial OGT which uses learned protein structural signatures of thermal adaptation. We hypothesize that biophysical tradeoffs which dictate enzymatic functions across variable temperatures provide a more robust empirical basis for OGT prediction than broad genomic features. Our OGT-predicting model, ROSEATE, is based on a single gene, adenylate kinase (ADK), that encodes for a ubiquitous enzyme essential for energy homeostasis. ROSEATE uses high-dimensional latent space encoding via MSA Transformer, a protein language model which embeds ADKs in a manner which preserves biophysical information about embedded proteins. We show that the accuracy of the ROSEATE model is on par with other genome-based models, has a high degree of phylogenetic generalizability, and the ESM embeddings effectively capture key temperature-adaptive enzyme characteristics derived from AlphaFold structures. Because ROSEATE is based on analyses of a single ubiquitous protein, it can be used with metagenomic data to infer the community-level variation in bacterial OGTs. We demonstrate this feature of ROSEATE by reconstructing ADK sequences from over 500 environmental and host-associated metagenomes, successfully distinguishing community-wide thermal preferences across diverse habitats, from polar oceans to mammalian guts. By transitioning from genomic proxies to informationally dense protein structural features, this work provides an efficient, interpretable tool for predicting bacterial OGTs across taxa and whole communities.

Read & Discuss → View Source →
15.
arXiv (CS.LG) 2026-06-16

The Data Manifold under the Microscope

Authors:
Marios Koulakis↗Constantin Seibold↗

arXiv:2606.15760v1 Announce Type: new Abstract: A significant gap exists between theory and practice in deep learning. Generalization and approximation error bounds are often derived for simplified models or are too loose to be informative. Many rely on the manifold hypothesis and on geometric regularity such as intrinsic dimension, curvature, and reach. Progress requires insight into data-manifold geometry and suitable benchmarks, yet existing options are polarized: analytic manifolds with known geometry but limited applicability, or real-world datasets where geometry is only coarsely estimable. We introduce a benchmarking framework for studying data geometry. We repurpose and extend dSprites and COIL-20 with additional transformation dimensions and dense, axis-aligned sampling, and pair them with finite-difference estimators that recover curvature, reach, and volume at near-ground-truth accuracy in a regime where general-purpose estimators are unreliable or difficult to deploy. The framework is intended as a controlled testbed, useful as a calibration environment for geometric estimators and a sandbox for probing theoretical assumptions. To illustrate its use, we present two application studies, namely assessing the scaling behavior of the bounds of Genovese et al. and Fefferman et al., and tracking the layer-wise geometry of a $\beta$-VAE, highlighting the behavior of current bounds and the value of controlled benchmarks for guiding and validating future theory. A reference implementation is available at https://github.com/koulakis/manifold-microscope.

Read & Discuss → View Source →
16.
arXiv (CS.LG) 2026-06-17

Non-negative Matrix Factorisation with Topological Regularisation

Authors:
Matias de Jong van Lier↗Shizuo Kaji↗Keunsu Kim↗

arXiv:2606.17531v1 Announce Type: new Abstract: We investigate the learning of interpretable bases in non-negative matrix factorisation (NMF) by regularising the topology of the learned basis functions. Our approach is motivated by the observation that many data modalities can be viewed as non-negative functions on a structured domain, where the quality of a basis is intrinsically linked to its topology. However, naive methods for incorporating the topology of the support are often hindered by discreteness and threshold dependence, rendering them unsuitable for continuous optimisation. We address these challenges by employing persistent homology as a stable, threshold-free topological quantifier and by designing topological scores that integrate into the NMF objective as regularisers. The resulting framework encompasses spatially coherent image components, periodic time-series structures, and clique-like graph signals within a unified modelling language.

Read & Discuss → View Source →
17.
arXiv (CS.CL) 2026-06-11

Hey Chat, Can You Teach Me? Structuring Socratic Dialogue for Human Learning in the Wild

Authors:
Sidney Tio↗Arunesh Sinha↗Pradeep Varakantham↗

Large language models are now widely used for everyday learning, but the underlying interactions are typically unstructured chats rather than following a curriculum. Unlike formal online learning systems, these interactions carry no prior record of the student, so any estimate of what the student already knows must be inferred from the dialogue itself. We show that this gap is not closed by scaling models alone. Frontier and education-tuned LLMs perform poorly when asked to tutor a student over an extended session, because doing so requires three things at once. The tutor must sequence a curriculum, conduct Socratic dialogue, and infer the student's knowledge state from that dialogue. We propose separating these responsibilities. Given a student query, our system constructs a prerequisite knowledge graph in which subtopics are nodes and dependencies are edges, and frames tutoring as deciding which node to teach next and how many dialogue turns to spend on it before moving on. A lightweight PPO policy handles this sequencing decision, while an LLM conducts the Socratic exchange at the chosen node and returns a signal of student progress. Across held-out STEM and non-STEM topics, our PPO-paired tutor outperforms heuristic baselines, frontier general-purpose models, and a model specialised for Socratic dialogue: on both the rate at which students reach full curriculum mastery and the number of turns required. Explicit curriculum structure delivers gains that scaling the underlying model does not.

Read & Discuss → View Source →
18.
arXiv (CS.CL) 2026-06-25

Do Encoders Suffice? A Systematic Comparison of Encoder and Decoder Safety Judges for LLM Adversarial Evaluation

Authors:
Han Jeon↗Shiv Medler↗Joseph Voyles↗Matt Wood↗

With the widespread adoption of large language models (LLMs) in chatbots and everyday applications, companies increasingly need guardrails that are effective while remaining low-cost and low-latency. Safety evaluation of LLM outputs has generally relied on LLM-based judges, which can be effective but are often slow and expensive to deploy at scale. In this paper, we evaluate whether fine-tuned modern encoder classifiers from the ModernBERT family, including ModernBERT and Ettin, can reliably identify harmful LLM outputs in user-model conversations without substantial performance loss relative to LLM-based judges. We benchmark these encoder classifiers against rule-based prefix matching, fine-tuned LLM classifiers, and LLM judges using a range of judge-prompting strategies across open-source adversarial datasets. The LLM judges include evaluation methodologies from StrongReject, ShieldGemma, JailbreakBench, AILuminate, SorryBench, and a Claude-as-a-judge setup, as well as fine-tuned safety classifiers such as LlamaGuard 3 and LlamaGuard 4. The encoder classifiers are fine-tuned on judge-labeled data using a majority-voting label strategy and are then evaluated on a gold-standard holdout dataset to assess their performance relative to LLM judges. We report absolute performance using F1 score, false negative rate, and precision-recall metrics. We also break down results by attack technique, including single-turn prompting, decomposition, escalation, and context manipulation, to identify where encoder classifiers align with or diverge from LLM-based judges. Our findings provide guidance on when encoder classifiers can serve as cost- and latency-efficient alternatives to LLM-based safety evaluation.

Read & Discuss → View Source →
19.
arXiv (CS.CL) 2026-06-25

OPERA: Aligning Open-Ended Reasoning via Objective Perplexity-based Reinforcement Learning

Authors:
Wenxuan Jiang↗Zining Fan↗Zijian Zhang↗Xuecheng Wu↗Hongming Tan↗Haoyang Dai↗Xiaoyu Li↗Xuezhi Cao↗Ninghao Liu↗

Reinforcement Learning (RL) has enabled LLMs to excel in objective reasoning tasks such as mathematics and code generation. However, applying RL to open-ended tasks, such as creative writing, remains challenging because LLM-as-a-judge reward models often exhibit stylistic biases and positional inconsistencies, leading to unstable supervision. To address this, we propose OPERA (Objective Perplexity-based Reflective Alignment), which replaces unreliable external judges with intrinsic rewards derived from perplexity dynamics. Specifically, we derive an intrinsic reward signal from perplexity dynamics, quantifying uncertainty reduction at critical reflective states. During the cold-start phase, we introduce a data synthesis method that leverages carefully designed guiding words to generate diverse reasoning traces, along with perplexity-prioritized rollouts that utilize internal log-probabilities to identify logically consistent reasoning branches. This pipeline yields a large-scale dataset comprising 20,000 high-quality reasoning trajectories. Empirical evaluations consistently demonstrate the scalability and efficacy of our approach in alignment for open-ended tasks. Implementing OPERA on Qwen3-8B establishes a new state-of-the-art among open-source models, achieving parity with or surpassing proprietary models like Gemini2.5 and MiniMax-M2.5 in some open-ended tasks. The code is available at https://github.com/pangpang-xuan/OPERA.

Read & Discuss → View Source →
20.
arXiv (CS.AI) 2026-06-25

Improving Zero-Shot Offline RL via Behavioral Task Sampling

Authors:
Nazim Bendib↗Nicolas Perrin-Gilbert↗Olivier Sigaud↗

arXiv:2604.25496v2 Announce Type: replace Abstract: Offline zero-shot reinforcement learning (RL) aims to learn agents that optimize unseen reward functions without additional environment interaction. The standard approach to this problem trains task-conditioned policies by sampling task vectors that define linear reward functions over learned state representations. In most existing algorithms, these task vectors are randomly sampled, implicitly assuming this adequately captures the structure of the task space. We argue that doing so leads to suboptimal zero-shot generalization. To address this limitation, we propose extracting task vectors directly from the offline dataset and using them to define the task distribution used for policy training. We introduce a simple and general reward function extraction procedure that integrates into existing offline zero-shot RL algorithms. Across multiple benchmark environments and baselines, our approach improves zero-shot performance by an average of 20%, highlighting the importance of principled task sampling in offline zero-shot RL.

Read & Discuss → View Source →
21.
arXiv (CS.CL) 2026-06-16

SHARD: Safe and Helpful Alignment via Self-Reframing Distillation

Authors:
Viswonathan Manoranjan↗Amogh Gupta↗Anvesh Rao Vijjini↗Thomas Hofweber↗Snigdha Chaturvedi↗

Large language models often struggle with sensitive prompts. They may refuse outright, provide generic safety boilerplate, or fail to address the user's legitimate informational needs that can be answered safely. We introduce SHARD, a self-reframing distillation method to improve safe-helpfulness. It first rewrites sensitive prompts to surface benign intent using philosophical guidelines, then reframes its original responses into safe, more helpful ones, and finally fine-tunes the model on its self-reframed responses. Across DNA and the English subset of LINGUASAFE, SHARD improves helpfulness for most model families while preserving safety. It also remains competitive with distillation from a larger teacher model, suggesting that models can internalize safe and helpful behavior elicited from their own. Warning: This paper contains content that may be offensive or harmful.

Read & Discuss → View Source →
22.
arXiv (CS.CL) 2026-06-12

NaturalFlow: Reducing Disruptive Pauses for Natural Speech Flow in Simultaneous Speech-to-Speech Translation

Authors:
Dongwook Lee↗Youngho Cho↗Sangkwon Park↗Heeseung Kim↗Sungroh Yoon↗

Simultaneous speech-to-speech translation aims to enable near-real-time communication by minimizing latency, offering a compelling, real-time alternative to the high latency of consecutive translation. However, the excessive pursuit of low latency often results in fragmented chunk-wise speech. Consequently, listeners are subjected to an unnatural acoustic flow punctuated by frequent pauses, which could increase their cognitive load. To bridge this gap, we introduce a fluency-aware optimization framework designed to discover the sweet spot between the low-latency benefits of simultaneous translation and the natural flow of consecutive translation. Our framework minimizes inter-chunk silences by leveraging model-internal signals, including linguistic diversity and induced temporal variability in speech durations. Experiments on short- and long-form benchmarks show that our framework produces natural speech flow while maintaining competitive latency and translation quality.

Read & Discuss → View Source →
23.
arXiv (CS.CL) 2026-06-25

BitNet Text Embeddings

Authors:
Zhen Li↗Xin Huang↗Liang Wang↗Nan Yang↗Ting Song↗Yan Xia↗Xun Wu↗Shaohan Huang↗Huishuai Zhang↗Furu Wei↗Dongyan Zhao↗

LLM-based text embedders have substantially improved retrieval and semantic representation quality, but their deployment remains costly: large backbone models slow down embedding inference, while high-dimensional full-precision embeddings impose substantial storage and bandwidth overhead on large-scale indexes. In this paper, we present BITEMBED, an extreme low-bit framework for LLM-based text embedding that jointly targets encoding efficiency and vector storage. BITEMBED converts pretrained LLM backbones into BitNet-style embedding encoders with ternary weights, quantized activations, and lightweight normalization refinement. The converted model is adapted to representation learning through continual contrastive pre-training, followed by supervised contrastive fine-tuning with both similarity-distribution distillation and attention-relation distillation from a full-precision teacher. Beyond quantizing the backbone, BITEMBED further trains output embeddings to support multiple storage precisions meeting different storage needs in various scenarios. Experiments on MMTEB (eng, v2) with Qwen3-0.6B and Gemma3-270M show that BITEMBED is largely comparable to full precision teacher embedders. Moreover, BITEMBED flexibly obtains text embeddings of various precisions, achieving a trade-off between performance and storage cost.

Read & Discuss → View Source →
24.
Nature (Science) 2026-06-10

Gene ancestries reveal diverse microbial associations during eukaryogenesis

Authors:
Moisès Bernabeu↗

The origin of eukaryotes remains a central enigma in biology1. Continuing debates agree on the pivotal role of a symbiosis between an alphaproteobacterium and an Asgard archaeon2,3. However, the nature, timing and contributions of other potential bacterial partners4–6 and the role of interactions with viruses7–9 remain contentious. To address these questions, we used advanced phylogenomic approaches and comprehensive datasets spanning the known diversity of cellular life and viruses. Our analysis provided a revised reconstruction of the last eukaryotic common ancestor (LECA) proteome, in which we traced the phylogenetic origin of each protein family. We found compelling evidence for multiple waves of horizontal gene transfer from diverse bacterial donors, with some likely to have preceded mitochondrial endosymbiosis. We inferred plausible traits of the major donors and their functional contributions to the LECA. Our findings support a contribution of horizontal gene transfers to shaping the proteomes of pre-LECA ancestors and suggest a facilitating role of Nucleocytoviricota viruses. Taken together, our results suggest that ancient eukaryotes may have originated within complex microbial ecosystems through a succession of diverse associations that left a footprint of horizontally transferred genes. Phylogenomic reconstruction of the proteome of the last eukaryotic common ancestor sheds light on the origin of eukaryotes, indicating an important role of horizontal transfer of genes from diverse bacterial and viral donors.

Read & Discuss → View Source →
25.
arXiv (CS.CV) 2026-06-19

Can Agents Distinguish Visually Hard-to-Separate Diseases in a Zero-Shot Setting? A Pilot Study

Authors:
Zihao Zhao↗Frederik Hauke↗Juliana De Castilhos↗Sven Nebelung↗Daniel Truhn↗

The rapid progress of multimodal large language models (MLLMs) has led to increasing interest in agent-based systems. While most prior work in medical imaging concentrates on automating routine clinical workflows, we study an underexplored yet clinically significant setting: distinguishing visually hard-to-separate diseases in a zero-shot setting. We benchmark representative agents on two imaging-only proxy diagnostic tasks, (1) melanoma vs. atypical nevus and (2) pulmonary edema vs. pneumonia, where visual features are highly confounded despite substantial differences in clinical management. We introduce a multi-agent framework based on contrastive adjudication. Experimental results show improved diagnostic performance (an 11-percentage-point gain in accuracy on dermoscopy data) and reduced unsupported claims on qualitative samples, although overall performance remains insufficient for clinical deployment. We acknowledge the inherent uncertainty in human annotations and the absence of clinical context, which further limit the translation to real-world settings. Within this controlled setting, this pilot study provides preliminary insights into zero-shot agent performance in visually confounded scenarios.

Read & Discuss → View Source →