HFEPX Hub
Updated from current HFEPX corpus (Feb 26, 2026). 90 papers are grouped in this hub page. Common evaluation modes: Automatic Metrics, Simulation Env. Frequently cited benchmark: MATH. Common metric signal: accuracy. Newest paper in this set is from Feb 25, 2026.
Human Feedback Mix
Evaluation Modes
Top Benchmarks
Top Metrics
Melody Ma, John Hewitt · Feb 25, 2026 · Citations: 0
We study reasoning for accessing world knowledge stored in a language model's parameters.
Minhao Jiang, Zhikai Li, Xuewen Liu, Jing Zhang, Mengjuan Chen · Feb 25, 2026 · Citations: 0
Large language models have demonstrated capabilities in text generation, while their increasing parameter scales present challenges in computational and memory efficiency.
Yuanda Xu, Hejian Sang, Zhengze Zhou, Ran He, Zhipeng Wang · Feb 24, 2026 · Citations: 0
Evaluated on MATH-500 and AIME 2025, ACE composes seamlessly with existing methods and consistently improves the full Pass@k spectrum across all three model families and benchmarks.
Charafeddine Mouzouni · Feb 24, 2026 · Citations: 0
We validate across five benchmarks, five models from three families, and both synthetic and real data.
Juha Kontinen, Arne Meier, Kai Sauerwald · Feb 24, 2026 · Citations: 0
This paper establishes and proves representation theorems for cumulative propositional dependence logic and for cumulative propositional logic with team semantics.
Elbert Du, Cynthia Dwork, Lunjia Hu, Reid McIlroy-Young, Han Shao · Feb 24, 2026 · Citations: 0
To obtain sufficient training data, we train an LLM to act as synthetic humans.
Tony Feng, Junehyuk Jung, Sang-hyun Kim, Carlo Pagano, Sergei Gukov · Feb 24, 2026 · Citations: 0
We report the performance of Aletheia (Feng et al., 2026b), a mathematics research agent powered by Gemini 3 Deep Think, on the inaugural FirstProof challenge.
Anas Barakat, Souradip Chakraborty, Khushbu Pahwa, Amrit Singh Bedi · Feb 24, 2026 · Citations: 0
Pass@k is a widely used performance metric for verifiable large language model tasks, including mathematical reasoning, code generation, and short-answer reasoning.
Yanrui Wu, Lingling Zhang, Xinyu Zhang, Jiayu Chang, Pengyu Li · Feb 24, 2026 · Citations: 0
Evaluations of large language models (LLMs) primarily emphasize convergent logical reasoning, where success is defined by producing a single correct proof.
Yuliang Ji, Fuchen Shen, Jian Wu, Qiujie Xie, Yue Zhang · Feb 24, 2026 · Citations: 0
To comprehensively evaluate the mathematical reasoning capabilities of Large Language Models (LLMs), researchers have introduced abundant mathematical reasoning datasets.
Wang Zixian · Feb 24, 2026 · Citations: 0
Experiments on mathematical reasoning benchmarks show that GOPO achieves competitive generalization while maintaining stable gradient dynamics and entropy preservation in regimes where clipping-based methods plateau.
Varvara Sazonova, Dmitri Shmelkin, Stanislav Kikot, Vasily Motolygin · Feb 24, 2026 · Citations: 0
We introduce a pipeline for both automatic and interactive verification as a more accurate alternative to only checking the answer which is currently the most popular approach for benchmarks.
Xindian Ma, Rundong Kong, Peng Zhang, Ruoxiang Huang, Yongyu Jiang · Feb 24, 2026 · Citations: 0
We evaluate ID-LoRA on five diverse benchmarks: Mathematical Reasoning, Code Generation, MMLU, CommonsenseQA, and Safety Alignment.
Xu Wan, Yansheng Wang, Wenqi Huang, Mingyang Sun · Feb 24, 2026 · Citations: 0
Traditional on-policy Reinforcement Learning with Verifiable Rewards (RLVR) frameworks suffer from experience waste and reward homogeneity, which directly hinders learning efficiency on difficult samples during large language models post-tr
Hyeonje Choi, Jeongsoo Lee, Hyojun Lee, Jay-Yoon Lee · Feb 24, 2026 · Citations: 0
We introduce \ToolMATH, a math-grounded benchmark that evaluates tool-augmented language models in realistic multi-tool environments where the output depends on calling schema-specified tools and sustaining multi-step execution.
Alex Stein, Furong Huang, Tom Goldstein · Feb 24, 2026 · Citations: 0
Held-out in-domain accuracy under asymmetric evaluation improves from 46.0\% to 62.0\%, and average (maj@8) accuracy on public document-free math benchmarks improves from 20.2\% to 35.4\%.
Mohammad Amanlou, Erfan Shafiee Moghaddam, Yasaman Amou Jafari, Mahdi Noori, Farhan Farsi · Feb 23, 2026 · Citations: 0
Results show that KNIGHT enables token- and cost-efficient generation from a reusable graph representation, achieves high quality across these criteria, and yields model rankings aligned with MMLU-style benchmarks, while supporting topic-sp
Han Bao, Yue Huang, Xiaoda Wang, Zheyuan Zhang, Yujun Zhou · Feb 23, 2026 · Citations: 0
We take the position that the dominant paradigm of General Alignment, which compresses diverse human values into a single scalar reward, reaches a structural ceiling in settings with conflicting values, plural stakeholders, and irreducible
Arindam Khaled · Feb 23, 2026 · Citations: 0
In this work, we propose "Pyramid MoA", a hierarchical Mixture-of-Agents architecture that uses a lightweight Router to dynamically escalate queries only when necessary.
Chenhang Cui, An Zhang, Yuxin Chen, Gelei Deng, Jingnan Zheng · Feb 22, 2026 · Citations: 0
Across diverse mathematics and perception benchmarks, SNRF consistently enhances LVLM inference performance while preserving perceptual capabilities.
Ziheng Chen, Bernhard Schölkopf, Nicu Sebe · Feb 21, 2026 · Citations: 0
Hyperbolic spaces provide a natural geometry for representing hierarchical and tree-structured data due to their exponential volume growth.
Abraham Paul Elenjical, Vivek Hruday Kavuri, Vasudeva Varma · Feb 21, 2026 · Citations: 0
We introduce a psychologically grounded metacognitive framework that operationalizes Ann Brown's regulatory cycle (Planning, Monitoring, and Evaluation) as a structured prompting architecture, and study its integration within a lightweight
Wenlong Meng, Chen Gong, Terry Yue Zhuo, Fan Zhang, Kecen Li · Feb 21, 2026 · Citations: 0
LLM agents rely heavily on high-quality trajectory data to guide their problem-solving behaviors, yet producing such data requires substantial task design, high-capacity model generation, and manual filtering.
Harshul Raj Surana, Arijit Maji, Aryan Vats, Akash Ghosh, Sriparna Saha · Feb 20, 2026 · Citations: 0
Existing Cultural benchmarks are (i) Manually crafted, (ii) contain single-hop questions testing factual recall, and (iii) prohibitively costly to scale, leaving this deficiency largely unmeasured.
Jiamin Yao, Eren Gultepe · Feb 20, 2026 · Citations: 0
Applied to LLaMA-2-7B, SPQ achieves up to 75% memory reduction while maintaining or improving perplexity (e.g., WikiText-2 5.47 to 4.91) and preserving accuracy on downstream benchmarks such as C4, TruthfulQA, and GSM8K.
Yutong Xin, Qiaochu Chen, Greg Durrett, Işil Dillig · Feb 20, 2026 · Citations: 0
However, most benchmarks for LLM-based proof automation are drawn from mathematics in the Mathlib ecosystem, whereas proofs in software verification are developed inside definition-rich codebases with substantial project-specific libraries.
Lexiang Tang, Weihao Gao, Bingchen Zhao, Lu Ma, Qiao jin · Feb 20, 2026 · Citations: 0
Experiments show that CCD significantly improves accuracy across mathematical reasoning benchmarks while substantially reducing output length, with minimal KV-cache overhead.
Johannes Ackermann, Michael Noukhovitch, Takashi Ishida, Masashi Sugiyama · Feb 20, 2026 · Citations: 0
Reinforcement Learning from Human Feedback (RLHF) or Verifiable Rewards (RLVR) are two key steps in the post-training of modern Language Models (LMs).
Jingkai Guo, Chaitali Chakrabarti, Deliang Fan · Feb 19, 2026 · Citations: 0
Large language models (LLMs) are increasingly deployed in safety and security critical applications, raising concerns about their robustness to model parameter fault injection attacks.
Anuj Gupta · Feb 19, 2026 · Citations: 0
While frontier large language models demonstrate strong reasoning and mathematical capabilities, the practical process of training domain-specialized scientific language models from raw sources remains under-documented.
Yunseung Lee, Subin Kim, Youngjun Kwak, Jaegul Choo · Feb 19, 2026 · Citations: 0
However, such errors have rarely been captured by existing benchmarks.
Yunseok Han, Yejoon Lee, Jaeyoung Do · Feb 19, 2026 · Citations: 0
To operationalize this, we present RFEval, a benchmark of 7,186 instances across seven tasks that probes faithfulness via controlled, output-level counterfactual interventions.
Zeliang Zhang, Xiaodong Liu, Hao Cheng, Hao Sun, Chenliang Xu · Feb 18, 2026 · Citations: 0
Experiments on three models, including Qwen2.5-3B-Instruct, Qwen2.5-7B-Instruct, and DeepSeek-R1-Distill-Llama-8B, on six widely used challenging mathematical benchmarks show consistent gains: our method achieves +19.3% improvement over LoR
Zhuqian Zhou, Kirk Vanacore, Bakhtawar Ahtisham, Jinsook Lee, Doug Pietrzak · Feb 18, 2026 · Citations: 0
To address this challenge, we investigate the "numeric ambiguity" problem and introduce MathEd-PII, the first benchmark dataset for PII detection in math tutoring dialogues, created through a human-in-the-loop LLM workflow that audits upstr
Ferdinand Kapl, Emmanouil Angelis, Kaitlin Maile, Johannes von Oswald, Stefan Bauer · Feb 18, 2026 · Citations: 0
Looping, reusing a block of layers across depth, and depth growing, training shallow-to-deep models by duplicating middle layers, have both been linked to stronger reasoning, but their relationship remains unclear.
Jonathan Cook, Diego Antognini, Martin Klissarov, Claudiu Musat, Edward Grefenstette · Feb 18, 2026 · Citations: 0
They are rarely proactive in soliciting this feedback, even when faced with ambiguity, which can make their dialogues feel static, one-sided, and lacking the adaptive qualities of human conversation.
Sarim Chaudhry · Feb 17, 2026 · Citations: 0
Large language models achieve strong performance on many complex reasoning tasks, yet their accuracy degrades sharply on benchmarks that require compositional reasoning, including ARC-AGI-2, GPQA, MATH, BBH, and HLE.
Shiqi Liu, Zeyu He, Guojian Zhan, Letian Tao, Zhilong Zheng · Feb 17, 2026 · Citations: 0
Across six mathematical reasoning benchmarks using Qwen 1.7B, 8B, and 14B base models, STAPO consistently demonstrates superior entropy stability and achieves an average performance improvement of 7.13% ($ρ_{\mathrm{T}}$=1.0, top-p=1.0) and
Gabriele Conte, Alessio Mattiace, Gianni Carmosino, Potito Aghilar, Giovanni Servedio · Feb 17, 2026 · Citations: 0
We propose Ruva, the first "Glass Box" architecture designed for Human-in-the-Loop Memory Curation.
Hanlin Zhang, Jikai Jin, Vasilis Syrgkanis, Sham Kakade · Feb 17, 2026 · Citations: 0
Using large scale observational evaluations with 5k observational and 2k newly sampled data on model performance, we estimate capability boundaries, high conditional quantiles of benchmark scores as a function of log pre training FLOPs, via
David Puertolas Merenciano, Ekaterina Vasyagina, Raghav Dixit, Kevin Zhu, Ruizhe Li · Feb 16, 2026 · Citations: 0
We evaluate the method on 500 LoRA adapters -- 400 clean, and 100 poisoned for Llama-3.2-3B on instruction and reasoning datasets: Alpaca, Dolly, GSM8K, ARC-Challenge, SQuADv2, NaturalQuestions, HumanEval, and GLUE dataset.
Subham Sekhar Sahoo, Jean-Marie Lemercier, Zhihan Yang, Justin Deschenaux, Jingyu Liu · Feb 16, 2026 · Citations: 0
Among discrete diffusion approaches, Masked diffusion currently dominates, largely driven by strong perplexity on language modeling benchmarks.
Avinandan Bose, Shuyue Stella Li, Faeze Brahman, Pang Wei Koh, Simon Shaolei Du · Feb 16, 2026 · Citations: 0
Cold-start personalization requires inferring user preferences through interaction when no user-specific historical data is available.
Sara Rajaee, Sebastian Vincent, Alexandre Berard, Marzieh Fadaee, Kelly Marchisio · Feb 16, 2026 · Citations: 0
We systematically evaluate several open- and closed-weights RLMs on the WMT24++ benchmark and find that enabling explicit reasoning consistently degrades translation quality across languages and models.
Francesco Gariboldi, Emma Franchino, Edith Haim, Gianluca Lattanzi, Alessandro Grecucci · Feb 16, 2026 · Citations: 0
Human networks show greater overlapping between mathematics and anxiety than GPT-oss.
Mingyu Cao, Alvaro H. C. Correia, Christos Louizos, Shiwei Liu, Lu Yin · Feb 11, 2026 · Citations: 0
Across mathematical reasoning and code generation benchmarks (GSM8K, MBPP, HumanEval) on Dream-7B and LLaDA-8B, SOAR improves generation quality while maintaining competitive inference speed, offering a practical way to balance quality and
Ailin Huang, Ang Li, Aobo Kong, Bin Wang, Binxing Jiao · Feb 11, 2026 · Citations: 0
We introduce Step 3.5 Flash, a sparse Mixture-of-Experts (MoE) model that bridges frontier-level agentic intelligence and computational efficiency.
Fengting Yuchi, Li Du, Jason Eisner · Feb 8, 2026 · Citations: 0
Although state-of-the-art LLMs can solve math problems, we find that they make errors on numerical comparisons with mixed notation: "Which is larger, $5.7 \times 10^2$ or $580$?" This raises a fundamental question: Do LLMs even know how big
David P. Woodruff, Vincent Cohen-Addad, Lalit Jain, Jieming Mao, Song Zuo · Feb 3, 2026 · Citations: 0
Based on these experiences, we extract common techniques for effective human-AI collaboration in theoretical research, such as iterative refinement, problem decomposition, and cross-disciplinary knowledge transfer.
Haotong Yang, Zitong Wang, Shijia Kang, Siqi Yang, Wenkai Yu · Feb 2, 2026 · Citations: 0
In this work, we design a *scalable* data-construction pipeline that, with minimal human effort, leverages LLMs to generate a large quantity of high-quality ``**question-proof-check**'' triplet data.
Ahsan Bilal, Ahmed Mohsin, Muhammad Umer, Ali Subhan, Hassan Rizwan · Feb 1, 2026 · Citations: 0
For each problem, the agent runs multiple inference iterations.
Wang Zixian · Jan 18, 2026 · Citations: 0
Experiments on MATH benchmarks show that the Hilbert projection formulation prevents gradient saturation typical of KL-constrained methods.
Ritik Raina, Abe Leite, Alexandros Graikos, Seoyoung Ahn, Dimitris Samaras · Jan 16, 2026 · Citations: 0
Human vision combines low-resolution "gist" information from the visual periphery with sparse but high-resolution information from fixated locations to construct a coherent understanding of a visual scene.
Shuhang Chen, Yunqiu Xu, Junjie Xie, Aojun Lu, Tao Feng · Jan 5, 2026 · Citations: 0
Motivated by this, we present CogFlow, a novel cognitive-inspired three-stage framework that incorporates a knowledge internalization stage, explicitly simulating the hierarchical flow of human reasoning: perception$\Rightarrow$internalizat
Yifan Zhang, Zixiang Chen, Yifeng Liu, Zhen Qin, Huizhuo Yuan · Dec 8, 2025 · Citations: 0
We present GRAPE (Group Representational Position Encoding), a unified framework for positional encoding based on group actions.
Minseo Kim, Chenfeng Xu, Coleman Hooper, Harman Singh, Ben Athiwaratkun · Nov 24, 2025 · Citations: 0
The full training and evaluation code is available at https://github.com/SqueezeAILab/CDLM.
David Acuna, Chao-Han Huck Yang, Yuntian Deng, Jaehun Jung, Ximing Lu · Nov 7, 2025 · Citations: 0
We introduce a framework able to synthesize vision-centric problems spanning diverse levels of complexity, and the resulting dataset with over 1M high-quality problems including: reasoning traces, preference data, and instruction prompts su
Soufiane Hayou · Nov 3, 2025 · Citations: 0
We provide the first proof of learning rate transfer with width in a linear multi-layer perceptron (MLP) parametrized with $μ$P, a neural network parameterization designed to ``maximize'' feature learning in the infinite-width limit.
Yan Wang, Keyi Wang, Shanshan Yang, Jaisal Patel, Jeff Zhao · Oct 10, 2025 · Citations: 0
We introduce FinAuditing, a taxonomy-aligned, structure-aware benchmark built from real XBRL filings.
Chenyu Wang, Zishen Wan, Hao Kang, Emma Chen, Zhiqiang Xie · Oct 6, 2025 · Citations: 0
Additional experiments show that the core principle of SLM-MUX extends to open-ended generation tasks (e.g., HumanEval) and benefits other model classes, including frontier LLMs and domain-specific fine-tuned SLMs.
