论文检索

Applications · Time Series

Fan Zhang, Sinchee Chin, Jing-Hao Xue, Wenming Yang

Time series anomaly detection (TSAD) faces critical challenges from intrinsic data noisiness and temporal heterogeneity, which undermine the reconstruction fidelity of prevailing generative approaches. While diffusion models offer theoretical advantages in capturing complex temporal dynamics, their inherent stochasticity introduces irreducible variance in reconstructions. We present the ICDiffAD, a novel method that synergizes adaptive noise scheduling with semi-deterministic generation to address these limitations. ICDiffAD introduces two key innovations: (1) an *SNR Scheduler* that governs training through quantifiable noise scales, enabling robust learning of normative patterns across non-stationary regimes; and (2) an *SNR Implicit Conditioning Mechanism* that initializes reverse diffusion from partially corrupted inputs, preserving signal coherence while attenuating anomalous components. This dual strategy ensures high-fidelity reconstructions aligned with the input’s manifold, reconciling generative flexibility with detection accuracy. Across five multivariate benchmarks, ICDiffAD improves the F1 score by 19.57\% and reduces false positives by 60.23\% compared to existing diffusion model-based TSAD methods.

Reinforcement Learning · Deep RL

Sanxing Chen, Xiaoyin Chen, Yukun Huang, Roy Xie, Bhuwan Dhingra

While Large Language Models (LLMs) hold promise to become autonomous agents, they often explore suboptimally in sequential decision-making. Recent work has sought to enhance this capability via supervised fine-tuning (SFT) or reinforcement learning (RL), improving regret on the classic multi-armed bandit task. However, it remains unclear how these learning methods shape exploration strategies and how well they generalize. We investigate both paradigms by training LLMs with SFT on expert trajectories and RL with a range of tailored reward signals including a strategic, regret-shaped reward to reduce variance, and an algorithmic reward that enables oracle imitation. The resulting agents outperform pre-trained models and achieve performance comparable to Upper Confidence Bound (UCB) and Thompson Sampling, with robust generalization to 6$\times$ longer horizons and across bandit families. Behavioral analysis reveals that gains often stem from more sophisticated but greedier exploitation: RL/SFT agents are more prone to early catastrophic failure than pre-trained models, prematurely abandoning exploration. Furthermore, agents trained to imitate UCB learn to outperform their teacher by adopting more exploitative variants. Our findings clarify when each training paradigm is preferable and advocate tailored reward design and evaluation beyond average regret to promote robust exploratory behavior.

Deep Learning · Everything Else

Etash Guha, Ryan Marten, Sedrick Keh, Negin Raoof, Georgios Smyrnis, Hritik Bansal, Marianna Nezhurina, Jean Mercat, Trung Vu, Zayne Sprague 等

Reasoning models have made rapid progress on many benchmarks involving math, code, and science. Yet, there are still many open questions about the best train- ing recipes for reasoning since state-of-the-art models often rely on proprietary datasets with little to no public information available. To address this, the goal of the OpenThoughts project is to create open-source datasets for training reasoning models. Our OpenThoughts2-1M dataset led to OpenThinker2-32B, the first model trained on public reasoning data to match DeepSeek-R1-Distill-32B on standard reasoning benchmarks such as AIME and LiveCodeBench. We then improve our dataset further by systematically investigating each step of our data genera- tion pipeline with 1,000+ controlled experiments, which led to OpenThoughts3. Scaling the pipeline to 1.2M examples and using QwQ-32B as teacher yields our OpenThinker3-7B model, which achieves state-of-the-art results: 53% on AIME 2025, 51% on LiveCodeBench 06/24-01/25, and 54% on GPQA Dia- mond – improvements of 15.3, 17.2, and 20.5 percentage points compared to the DeepSeek-R1-Distill-Qwen-7B. All of our datasets and models are available on openthoughts.ai.

Applications · Language, Speech and Dialog

cheng chen, Haiyan Yin, Ivor Tsang

Large Language Models (LLMs) are increasingly expected to be culturally customisable and personally aligned for natural language understanding (NLU). However, existing methods, from supervised fine-tuning (SFT) to personalised RLHF and prompting, either require costly large-scale annotations or remain constrained by their pretraining distributions. Moreover, acquiring annotations that reflect subjective, diverse, and evolving user preferences is both expensive and labour-intensive. To address these limitations, we propose \textit{\textbf{H}eterogeneous-\textbf{C}onsistency \textbf{C}o-Alignment} (HCC), a training-free annotation paradigm that leverages two heterogeneous models: a knowledge-rich yet potentially overconfident LLM and a task-specialised lightweight model guided by a small user preference set. Together, they verify and co-align misaligned outputs over unlabelled corpora. For verification, HCC introduces the reference-free \textit{\textbf{C}onsistent}-\textit{\textbf{A}nd}-\textit{\textbf{I}nconsistent} (\textbf{CAI}) Ratio, an uncertainty signal derived from inter-model agreements (consistent samples) and disagreements (inconsistent samples) to determine whether refinement is necessary. For co-alignment, HCC employs a non-parametric, embedding-based preference assignment scheme to recalibrate inconsistent samples according to user preferences. Across eight NLU datasets and both open- and closed-source LLMs, HCC consistently improves annotation alignment and, in several tasks, enables \textit{Llama-3-8B} to surpass \textit{GPT-3.5/4o-mini} after co-alignment correction. Moreover, CAI strongly correlates with accuracy and tracks pre- and post-alignment gains, offering a reference-free signal for scaling preference-aligned annotation without ground-truth supervision.

Optimization · Non-Convex

Jianlong Chen, Zhiming Zhou

Sharpness-Aware Minimization (SAM) enhances generalization by minimizing the maximum training loss within a predefined neighborhood around the parameters. However, its practical implementation approximates this as gradient ascent(s) followed by applying the gradient at the ascent point to update the current parameters. This practice can be justified as approximately optimizing the objective by neglecting the (full) derivative of the ascent point with respect to the current parameters. Nevertheless, a direct and intuitive understanding of why using the gradient at the ascent point to update the current parameters works superiorly, despite being computed at a shifted location, is still lacking. Our work bridges this gap by proposing a novel and intuitive interpretation. We show that the gradient at the single-step ascent point, when applied to the current parameters, provides a better approximation of the direction from the current parameters toward the maximum within the local neighborhood than the local gradient. This improved approximation thereby enables a more direct escape from the maximum within the local neighborhood. Nevertheless, our analysis further reveals two issues. First, the approximation by the gradient at the single-step ascent point is often inaccurate. Second, the approximation quality may degrade as the number of ascent steps increases. To address these limitations, we propose in this paper eXplicit Sharpness-Aware Minimization (XSAM). It tackles the first by explicitly estimating the direction of the maximum during training, and addresses the second by crafting a search space that effectively leverages the gradient information at the multi-step ascent point. XSAM features a unified formulation that applies to both single-step and multi-step settings and only incurs negligible computational overhead. Extensive experiments demonstrate the consistent superiority of XSAM against existing counterparts across various models, datasets, and settings.

Computer Vision · 3D Rendering & Reconstruction

Qisen Wang, Yifan Zhao, Jia Li

Dynamic reconstruction has achieved remarkable progress, but there remain challenges in monocular input for more practical applications. The prevailing works attempt to construct efficient motion representations, but lack a unified spatiotemporal decomposition framework, suffering from either holistic temporal optimization or coupled hierarchical spatial composition. To this end, we propose WorldTree, a unified framework comprising Temporal Partition Tree (TPT) that enables coarse-to-fine optimization based on the inheritance-based partition tree structure for hierarchical temporal decomposition, and Spatial Ancestral Chains (SAC) that recursively query ancestral hierarchical structure to provide complementary spatial dynamics while specializing motion representations across ancestral nodes. Experimental results on different datasets indicate that our proposed method achieves 8.26% improvement of LPIPS on NVIDIA-LS and 9.09% improvement of mLPIPS on DyCheck compared to the second-best method. Code: https://github.com/iCVTEAM/WorldTree.

Optimization · Zero-order and Black-box Optimization

Lakshya A Agrawal, Shangyin Tan, Dilara Soylu, Noah Ziems, Rishi Khare, Krista Opsahl-Ong, Arnav Singhvi, Herumb Shandilya, Michael J Ryan, Meng Jiang 等

Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6 percentage points on average and by up to 19pp, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10 percentage points (e.g., +12pp on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa.

Computer Vision · Image and Video Generation

Shenggui Li, Runyu Lu, qiaoling chen, Haiyan Yin, YUEMING LYU, Yonggang Wen, Ivor Tsang, Tianwei Zhang

Diffusion Transformer models have driven the rapid advances in video generation, achieving state-of-the-art quality and flexibility. However, their attention mechanism remains a major performance bottleneck, as its dense computation scales quadratically with the sequence length. To overcome this limitation and reduce the generation latency, we propose DSA, a novel attention mechanism that integrates sparse attention with distributed inference for diffusion-based video generation. By leveraging carefully-designed parallelism strategies and scheduling, DSA significantly reduces redundant computation while preserving global context. Extensive experiments on benchmark datasets demonstrate that, when deployed on 8 GPUs, DSA achieves up to 1.43× inference speedup than the existing distributed method and 10.79× faster than single-GPU inference.

Optimization · Zero-order and Black-box Optimization

Lakshya A Agrawal, Shangyin Tan, Dilara Soylu, Noah Ziems, Rishi Khare, Krista Opsahl-Ong, Arnav Singhvi, Herumb Shandilya, Michael J Ryan, Meng Jiang 等

Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6 percentage points on average and by up to 19pp, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10 percentage points (e.g., +12pp on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa.

Optimization · Everything Else

Yuhang Liu, Tao Li, Zhehao Huang, Zuopeng Yang, Xiaolin Huang

Low-Rank Adaptation (LoRA) enables parameter-efficient fine-tuning of large pre-trained models. Yet LoRA can face generalization challenges. One promising way to improve the generalization is Sharpness-Aware Minimization (SAM), which has proven effective for small-scale training scenarios. In this paper, we propose **Bi**-directional **Lo**w-**R**ank **A**daptation (Bi-LoRA), which introduces an auxiliary adversarial LoRA module. This design explicitly decouples sharpness optimization, handled by the auxiliary module, from task adaptation, performed by the primary module. Such a separation yields two key benefits. First, it transforms the sequential computation of primary LoRA update and adversarial perturbation into a parallel form, which roughly halves the time and conquers the main obstacle of applying SAM in LoRA. Second, it provides perturbations from the auxiliary module that do not collapse into the restricted optimization subspace of the primary module, enabling broader sharpness exploration and flatter minima. Bi-LoRA simultaneously achieves both efficiency and effectiveness within a single framework, as verified by extensive experiments across diverse architectures and tasks.

Deep Learning · Everything Else

Alan Sun, Mariya Toneva

Mechanistic interpretability (MI) is an emerging framework for interpreting neural networks. Given a task and model, MI aims to discover a succinct algorithmic process, an interpretation, that explains the model's decision process on that task. However, MI is difficult to scale and generalize. This stems in part from two key challenges: there is no precise notion of a valid interpretation; and, generating interpretations is often an ad hoc process. In this paper, we address these challenges by defining and studying the problem of interpretive equivalence: determining whether two different models share a common interpretation, without requiring an explicit description of what that interpretation is. At the core of our approach, we propose and formalize the principle that two interpretations of a model are equivalent if all of their possible implementations are also equivalent. We develop an algorithm to estimate interpretive equivalence and case study its use on Transformer-based models. To analyze our algorithm, we introduce necessary and sufficient conditions for interpretive equivalence based on models' representation similarity. We provide guarantees that simultaneously relate a model's algorithmic interpretations, circuits, and representations. Our framework lays a foundation for the development of more rigorous evaluation methods of MI and automated, generalizable interpretation discovery methods.

Applications · Genetics, Cell Biology, Health, etc

Vidit Agrawal, John Peters, Tyler Thompson, Mohammad Sanian, Chau Pham, Nikita Moshkov, Arshad Kazi, Aditya Pillai, Jack Freeman, Byunguk Kang 等

Quantifying cell morphology using images and machine learning has proven to be a powerful tool to study the response of cells to treatments. However, models used to quantify cellular morphology are typically trained with a single microscopy imaging type. This results in specialized models that cannot be reused across biological studies because the technical specifications do not match (e.g., different number of channels). Here, we present CHAMMI-75, an open access dataset of heterogeneous, multi-channel microscopy images from 75 diverse biological studies. We curated this resource from publicly available sources to investigate cellular morphology models that are channel-adaptive and can process any microscopy image type. Our experiments show that training with CHAMMI-75 can improve performance in multi-channel bioimaging tasks primarily because of its high diversity in microscopy modalities. This work paves the way to create the next generation of cellular morphology models for biological studies.

Deep Learning · Algorithms

Shijing Hu, Jingyang Li, Zhihui Lu, Pan Zhou

Speculative decoding accelerates large language model (LLM) inference by letting a lightweight draft model propose multiple tokens that the target model verifies in parallel. Yet existing training objectives optimize only a single greedy draft path, while decoding follows a tree policy that re-ranks and verifies multiple branches. This draft policy misalignment limits achievable speedups. We introduce **Group Tree Optimization** (GTO), which aligns training with the decoding-time tree policy through two components: (i) Draft Tree Reward, a sampling-free objective equal to the expected acceptance length of the draft tree under the target model, directly measuring decoding performance; (ii) Group-based Draft Policy Training, a stable optimization scheme that contrasts trees from the current and a frozen reference draft model, forming debiased group-standardized advantages and applying a PPO-style surrogate along the longest accepted sequence for robust updates. We further prove that increasing our Draft Tree Reward provably improves acceptance length and speedup. Across dialogue (MT-Bench), code (HumanEval), and math (GSM8K), and multiple LLMs (e.g., LLaMA-3.1-8B, LLaMA-3.3-70B, Vicuna-1.3-13B, DeepSeek-R1-Distill-LLaMA-8B, Qwen3-8B), GTO increases acceptance length by \(7.4\%\) and yields an additional \(7.7\%\) speedup over prior state-of-the-art EAGLE-3. By bridging draft policy misalignment, GTO offers a practical, general solution for efficient LLM inference. Code and draft models are available at https://github.com/hsj576/GTO.

Applications · Language, Speech and Dialog

Hongzhan Lin, Zhiqi Bai, Xinmiao Zhang, Siran Yang, Jiamang Wang, Yunlong Xu, JIAHENG LIU, Yongchi Zhao, Xiang Li, Yuchi Xu 等

Transformer decoders have achieved strong results across tasks, but the memory required for the KV cache becomes prohibitive at long sequence lengths. Although Cross-layer KV Cache sharing (e.g., YOCO, CLA) offers a path to mitigate KV Cache bottleneck, it typically underperforms within-layer methods like GQA. To understand the root cause, we investigate the information flow of keys and values of the top-layers. Our preliminary reveals a clear distribution: values are predominantly derived from the bottom layer, while keys draw more information from both bottom and middle layers. Building upon this, we propose FusedKV, whose top-layer KV caches are a learnable fusion of the most informative ones from the bottom and middle layers. This fusion operates directly on post-RoPE keys, preserving relative positional information without the computational cost of re-applying rotary embeddings. To further improve efficiency, we propose FusedKV-Lite, an cross-layer sharing approach, where top-layer KV caches are directly derived from the bottom-layer values and the middle-layer keys. Compared to FusedKV, FusedKV-Lite reduces I/O overhead at the cost of a slight increase in perplexity. In experiments on LLMs ranging from 332M to 4B parameters, our proposed method reduce 50\% cache memory while achieving lower validation perplexity than the standard Transformer decoder, establishing it as a memory-efficient, high-performance architectural alternative. We have made our Triton implementation available.

Applications · Physics

Yuxin Li, Minghao LIU, Ruida WANG, JI WenZhao, Zhitao He, Rui Pan, Junming Huang, Tong Zhang, Yi R. Fung

We present **Lean4PHYS**, a comprehensive reasoning framework for college-level physics problems in Lean4. To establish a solid foundation for formal reasoning in physics, **Lean4PHYS** launches *PhysLib*, a repository containing fundamental unit systems and essential theorems to formulate physics proofs in Lean4. It will be community-driven and long-term maintained. Lean4PHYS also includes *LeanPhysBench*, a college-level benchmark for evaluating LLMs' Lean4 formal physics reasoning capability. It contains 200 hand-crafted and peer-reviewed Lean4 theorem statements formalized from university textbooks and physics competition problems. Based on the *PhysLib* and *LeanPhysBench* we composed in **Lean4PHYS**, we perform exhaustive experiments of baseline results using major expert Math provers and state-of-the-art closed-source models, and provide an analysis of their performance. In the experiment, we identify that most expert provers do not outperform general models as they did in the math domain. This suggests potential overfitting to the math domain rather than learning formal reasoning for formal provers. We also conduct a comprehensive experiment showing that, with *PhysLib* in the context, LLMs' performance on *LeanPhysBench* increases by **11.90%** on average, proving the effectiveness of our repository in assisting LLMs in solving the Lean4 physics problem. To the best of our knowledge, we are the first study to provide a physics benchmark in Lean4.

Deep Learning · Algorithms

Xinwen Cheng, Zhehao Huang, Wenxing Zhou, Zhengbao He, Ruikai Yang, Yingwen Wu, Xiaolin Huang

Machine unlearning (MU) aims to remove the influence of specific training samples from a well-trained model, a task of growing importance due to the ``right to be forgotten.” The unlearned model should approach the retrained model, where forgetting data do not contribute to the training process. Therefore, unlearning should withdraw their contribution from the pre-trained model. However, quantifying and disentangling sample's contribution to overall learning process is highly challenging, leading most existing MU approaches to adopt other heuristic strategies such as random labeling or knowledge distillation. These operations inevitably degrade model utility, requiring additional maintenance with remaining data. To advance MU towards better utility and efficiency for practical deployment, we seek to approximate sample contribution with only the pre-trained model. We theoretically and empirically reveal that sample's contribution during training manifests in the learned model's increased sensitivity to it. In light of this, we propose MU-Mis (Machine Unlearning by Minimizing input sensitivity), which directly suppresses the contribution of forgetting data. This straightforward suppression enables MU-Mis to successfully unlearn without degrading model utility on the remaining data, thereby eliminating the need for access to the remaining data. To the best of our knowledge, this is the first time that a remaining-data-free method can perform on par with top performing remaining-data-dependent methods.

Deep Learning · Generative Models and Autoencoders

Jaewoo Lee, Minsu Kim, Sanghyeok Choi, Inhyuck Song, Sujin Yun, Hyeongyu Kang, Woocheol Shin, Taeyoung Yun, Kiyoung Om, Jinkyoo Park

Diffusion alignment aims to optimize diffusion models for the downstream objective. While existing methods based on reinforcement learning or direct backpropagation achieve considerable success in maximizing rewards, they often suffer from reward over-optimization and mode collapse. We introduce Diffusion Alignment as Variational Expectation-Maximization (DAV), a framework that formulates diffusion alignment as an iterative process alternating between two complementary phases: the E-step and the M-step. In the E-step, we employ test-time search to generate diverse and reward-aligned samples. In the M-step, we refine the diffusion model using samples discovered by the E-step. We demonstrate that DAV can optimize reward while preserving diversity for both continuous and discrete tasks: text-to-image synthesis and DNA sequence design. Our code is available at https://github.com/Jaewoopudding/dav.

Deep Learning · Generative Models and Autoencoders

Hyeongyu Kang, Jaewoo Lee, Woocheol Shin, Kiyoung Om, Jinkyoo Park

Diffusion models excel at generating high-likelihood samples but often require alignment with downstream objectives. Existing fine-tuning methods for diffusion models significantly suffer from reward over-optimization, resulting in high-reward but unnatural samples and degraded diversity. To mitigate over-optimization, we propose Soft Q-based Diffusion Finetuning (SQDF), a novel KL-regularized RL method for diffusion alignment that applies a reparameterized policy gradient of a training-free, differentiable estimation of the soft Q-function. SQDF is further enhanced with three innovations: a discount factor for proper credit assignment in the denoising process, the integration of consistency models to refine Q-function estimates, and the use of an off-policy replay buffer to improve mode coverage and manage the reward-diversity trade-off. Our experiments demonstrate that SQDF achieves superior target rewards while preserving diversity in text-to-image alignment. Furthermore, in online black-box optimization, SQDF attains high sample efficiency while maintaining naturalness and diversity. Our code is available at https://github.com/Shin-woocheol/SQDF.

General Machine Learning · Representation Learning

Lola Le Breton

Recent innovations in architecture, pre-training, and fine-tuning have led to the remarkable in-context learning and reasoning abilities of large auto-regressive language models such as LLaMA and DeepSeek. In contrast, encoders like BERT and RoBERTa have not seen the same level of progress despite being foundational for many downstream NLP applications. To bridge this gap, we introduce NeoBERT, a next-generation encoder that redefines the capabilities of bidirectional models by integrating state-of-the-art advancements in architecture, modern data, and optimized pre-training methodologies. NeoBERT is designed for seamless adoption: it serves as a plug-and-play replacement for existing base models, relies on an optimal depth-to-width ratio, and leverages an extended context length of 4,096 tokens. Despite its compact 250M parameter footprint, it achieves state-of-the-art results on the massive MTEB benchmark, outperforming BERT$_{large}$, RoBERTa$_{large}$, NomicBERT, and ModernBERT under identical fine-tuning conditions. In addition, we rigorously evaluate the impact of each modification on GLUE and design a uniform fine-tuning and evaluation framework for MTEB. We release all code, data, checkpoints, and training scripts to accelerate research and real-world adoption.

Gabriel Grand, Valerio Pepe, Joshua B Tenenbaum, Jacob Andreas

Many emerging applications of AI—from scientific discovery to medical diagnosis—require agents to seek information strategically: forming hypotheses, asking targeted questions, and making decisions under uncertainty. In high-stakes settings with limited resources, do language models (LMs) behave like rational agents? Drawing on insights from human cognition, we develop methods to evaluate and enhance agentic information-seeking. First, we introduce a decision-oriented dialogue task called Collaborative Battleship, in which a Captain must balance exploration (asking questions) and action (taking shots), while a Spotter must supply accurate, contextually-grounded answers. Compared to human players (N=42), we find that many LM agents struggle to ask informative questions, produce accurate answers, and identify high-utility actions. To address these gaps, we develop novel Monte Carlo inference strategies for LMs inspired by Bayesian Experimental Design (BED). For Spotter agents, our approach boosts accuracy by up to 14.7% absolute over LM-only baselines; for Captain agents, it raises expected information gain (EIG) by up to 0.227 bits (94.2% of the achievable noise ceiling). Combined, these components yield sharper targeting (+0.303–0.374 F1), and enable weaker LMs, such as Llama-4-Scout, to outperform both humans (8% → 82% win rate) and frontier models (0% → 67% win rate vs. GPT-5) at ≈1% of GPT-5's cost. We replicate these findings on Guess Who?, where our methods significantly boost accuracy (+28.3–42.4 p.p.), demonstrating their general applicability for building information-seeking agents.