Paper
LightNews LightNews
banner
paper.bsky.social
Paper
@paper.bsky.social
1.2K followers 0 following 8.7K posts
Summarize the top 30 most popular arXiv papers on Reddit, Hacker News and Hugging Face in the last 30 days.

Source: https://github.com/susumuota/arxiv-reddit-summary
Maintained by @ota.bsky.social
Posts Replies Media Videos
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2511.08892 2/30 https://arxiv.org/abs/2510.26692 3/30 https://arxiv.org/abs/2511.12414 4/30 https://arxiv.org/abs/2511.02864 5/30 https://arxiv.org/abs/2510.26745 6/30 https://arxiv.org/abs/2511.10647 7/30 https://arxiv.org/abs/2511.09030 8/30 https://arxiv.org/abs/2510.24797 9/30 https://arxiv.org/abs/2511.15304 10/30 https://arxiv.org/abs/2510.27688 11/30 https://arxiv.org/abs/2511.07416 12/30 https://arxiv.org/abs/2510.25741 13/30 https://arxiv.org/abs/2511.00926 14/30 https://arxiv.org/abs/2510.22371 15/30 https://arxiv.org/abs/2511.08923 16/30 https://arxiv.org/abs/2511.14593 17/30 https://arxiv.org/abs/2510.27190 18/30 https://arxiv.org/abs/2510.26583 19/30 https://arxiv.org/abs/2511.04570 20/30 https://arxiv.org/abs/2511.06876 21/30 https://arxiv.org/abs/2511.14993 22/30 https://arxiv.org/abs/2511.02743 23/30 https://arxiv.org/abs/2510.23607 24/30 https://arxiv.org/abs/2511.11793 25/30 https://arxiv.org/abs/2511.03337 26/30 https://arxiv.org/abs/2511.02208 27/30 https://arxiv.org/abs/2511.13612 28/30 https://arxiv.org/abs/2511.13254 29/30 https://arxiv.org/abs/2511.03276 30/30 https://arxiv.org/abs/2511.06221
November 24, 2025 at 12:06 AM
paper.bsky.social
[30/30] 119 Likes, 11 Comments, 2 Posts
2511.06221, cs․AI | cs․CL, 09 Nov 2025

🆕Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B

Sen Xu, Yi Zhou, Wei Wang, Jixin Min, Zhibin Yin, Yingwei Dai, Shixi Liu, Lianyu Pang, Yirong Chen, J...
Challenging the prevailing consensus that small models inherently lack robust reasoning, this report introduces VibeThinker-1.5B, a 1.5B-parameter dense model developed via our Spectrum-to-Signal Principle (SSP). This challenges the prevailing approach of scaling model parameters to enhance capabilities, as seen in models like DeepSeek R1 (671B) and Kimi k2 (>1T). The SSP framework first employs a Two-Stage Diversity-Exploring Distillation (SFT) to generate a broad spectrum of solutions, followed by MaxEnt-Guided Policy Optimization (RL) to amplify the correct signal. With a total training cost of only $7,800, VibeThinker-1.5B demonstrates superior reasoning capabilities compared to closed-source models like Magistral Medium and Claude Opus 4, and performs on par with open-source models like GPT OSS-20B Medium. Remarkably, it surpasses the 400x larger DeepSeek R1 on three math benchmarks: AIME24 (80.3 vs. 79.8), AIME25 (74.4 vs. 70.0), and HMMT25 (50.4 vs. 41.7). This is a substantial improvement over its base model (6.7, 4.3, and 0.6, respectively). On LiveCodeBench V6, it scores 51.1, outperforming Magistral Medium's 50.3 and its base model's 0.0. These findings demonstrate that small models can achieve reasoning capabilities comparable to large models, drastically reducing training and inference costs and thereby democratizing advanced AI research.
November 24, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2511.08892 2/30 https://arxiv.org/abs/2510.26692 3/30 https://arxiv.org/abs/2511.12414 4/30 https://arxiv.org/abs/2511.02864 5/30 https://arxiv.org/abs/2510.26745 6/30 https://arxiv.org/abs/2511.10647 7/30 https://arxiv.org/abs/2510.24797 8/30 https://arxiv.org/abs/2511.09030 9/30 https://arxiv.org/abs/2511.15304 10/30 https://arxiv.org/abs/2510.27688 11/30 https://arxiv.org/abs/2511.07416 12/30 https://arxiv.org/abs/2510.25741 13/30 https://arxiv.org/abs/2510.21890 14/30 https://arxiv.org/abs/2511.00926 15/30 https://arxiv.org/abs/2510.22371 16/30 https://arxiv.org/abs/2511.14593 17/30 https://arxiv.org/abs/2510.27190 18/30 https://arxiv.org/abs/2510.26583 19/30 https://arxiv.org/abs/2511.08923 20/30 https://arxiv.org/abs/2511.04570 21/30 https://arxiv.org/abs/2511.06876 22/30 https://arxiv.org/abs/2511.02743 23/30 https://arxiv.org/abs/2510.21614 24/30 https://arxiv.org/abs/2511.14993 25/30 https://arxiv.org/abs/2510.23607 26/30 https://arxiv.org/abs/2511.11793 27/30 https://arxiv.org/abs/2511.03337 28/30 https://arxiv.org/abs/2511.02208 29/30 https://arxiv.org/abs/2511.13612 30/30 https://arxiv.org/abs/2511.13254
November 23, 2025 at 12:06 AM
paper.bsky.social
[19/30] 206 Likes, 19 Comments, 3 Posts
2511.08923, cs․CL | cs․AI, 12 Nov 2025

🆕TiDAR: Think in Diffusion, Talk in Autoregression

Jingyu Liu, Xin Dong, Zhifan Ye, Rishabh Mehta, Yonggan Fu, Vartika Singh, Jan Kautz, Ce Zhang, Pavlo Molchanov
Diffusion language models hold the promise of fast parallel generation, while autoregressive (AR) models typically excel in quality due to their causal structure aligning naturally with language modeling. This raises a fundamental question: can we achieve a synergy with high throughput, higher GPU utilization, and AR level quality? Existing methods fail to effectively balance these two aspects, either prioritizing AR using a weaker model for sequential drafting (speculative decoding), leading to lower drafting efficiency, or using some form of left-to-right (AR-like) decoding logic for diffusion, which still suffers from quality degradation and forfeits its potential parallelizability. We introduce TiDAR, a sequence-level hybrid architecture that drafts tokens (Thinking) in Diffusion and samples final outputs (Talking) AutoRegressively - all within a single forward pass using specially designed structured attention masks. This design exploits the free GPU compute density, achieving a strong balance between drafting and verification capacity. Moreover, TiDAR is designed to be serving-friendly (low overhead) as a standalone model. We extensively evaluate TiDAR against AR models, speculative decoding, and diffusion variants across generative and likelihood tasks at 1.5B and 8B scales. Thanks to the parallel drafting and sampling as well as exact KV cache support, TiDAR outperforms speculative decoding in measured throughput and surpasses diffusion models like Dream and Llada in both efficiency and quality. Most notably, TiDAR is the first architecture to close the quality gap with AR models while delivering 4.71x to 5.91x more tokens per second.
November 23, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2511.08892 2/30 https://arxiv.org/abs/2510.26692 3/30 https://arxiv.org/abs/2511.12414 4/30 https://arxiv.org/abs/2511.02864 5/30 https://arxiv.org/abs/2510.26745 6/30 https://arxiv.org/abs/2511.10647 7/30 https://arxiv.org/abs/2511.09030 8/30 https://arxiv.org/abs/2510.24797 9/30 https://arxiv.org/abs/2510.27688 10/30 https://arxiv.org/abs/2511.07416 11/30 https://arxiv.org/abs/2511.15304 12/30 https://arxiv.org/abs/2510.25741 13/30 https://arxiv.org/abs/2510.21890 14/30 https://arxiv.org/abs/2510.22371 15/30 https://arxiv.org/abs/2511.00926 16/30 https://arxiv.org/abs/2511.14593 17/30 https://arxiv.org/abs/2510.20822 18/30 https://arxiv.org/abs/2510.27190 19/30 https://arxiv.org/abs/2510.26583 20/30 https://arxiv.org/abs/2511.04570 21/30 https://arxiv.org/abs/2511.06876 22/30 https://arxiv.org/abs/2511.02743 23/30 https://arxiv.org/abs/2510.21614 24/30 https://arxiv.org/abs/2510.23607 25/30 https://arxiv.org/abs/2511.14993 26/30 https://arxiv.org/abs/2511.11793 27/30 https://arxiv.org/abs/2511.03337 28/30 https://arxiv.org/abs/2511.02208 29/30 https://arxiv.org/abs/2511.13612 30/30 https://arxiv.org/abs/2511.13254
November 22, 2025 at 12:06 AM
paper.bsky.social
[30/30] 123 Likes, 3 Comments, 1 Posts
2511.13254, cs․CL, 17 Nov 2025

🆕Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance

Shalini Maiti, Amar Budhiraja, Bhavul Gauri, Gaurav Chaurasia, Anton Protopopov, Alexis Audran-Reiss, Michael Slater, Despoina Magka, Tatiana Sha...
Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse domains, but their training remains resource- and time-intensive, requiring massive compute power and careful orchestration of training procedures. Model souping-the practice of averaging weights from multiple models of the same architecture-has emerged as a promising pre- and post-training technique that can enhance performance without expensive retraining. In this paper, we introduce Soup Of Category Experts (SoCE), a principled approach for model souping that utilizes benchmark composition to identify optimal model candidates and applies non-uniform weighted averaging to maximize performance. Contrary to previous uniform-averaging approaches, our method leverages the observation that benchmark categories often exhibit low inter-correlations in model performance. SoCE identifies "expert" models for each weakly-correlated category cluster and combines them using optimized weighted averaging rather than uniform weights. We demonstrate that the proposed method improves performance and robustness across multiple domains, including multilingual capabilities, tool calling, and math and achieves state-of-the-art results on the Berkeley Function Calling Leaderboard.
November 22, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2511.08892 2/30 https://arxiv.org/abs/2510.26692 3/30 https://arxiv.org/abs/2511.12414 4/30 https://arxiv.org/abs/2511.02864 5/30 https://arxiv.org/abs/2510.26745 6/30 https://arxiv.org/abs/2511.10647 7/30 https://arxiv.org/abs/2511.09030 8/30 https://arxiv.org/abs/2510.24797 9/30 https://arxiv.org/abs/2510.27688 10/30 https://arxiv.org/abs/2511.07416 11/30 https://arxiv.org/abs/2510.25741 12/30 https://arxiv.org/abs/2510.21890 13/30 https://arxiv.org/abs/2510.22371 14/30 https://arxiv.org/abs/2511.15304 15/30 https://arxiv.org/abs/2510.20822 16/30 https://arxiv.org/abs/2511.14593 17/30 https://arxiv.org/abs/2510.27190 18/30 https://arxiv.org/abs/2510.26583 19/30 https://arxiv.org/abs/2511.04570 20/30 https://arxiv.org/abs/2511.06876 21/30 https://arxiv.org/abs/2511.02743 22/30 https://arxiv.org/abs/2510.19788 23/30 https://arxiv.org/abs/2510.21614 24/30 https://arxiv.org/abs/2510.23607 25/30 https://arxiv.org/abs/2511.00926 26/30 https://arxiv.org/abs/2511.11793 27/30 https://arxiv.org/abs/2511.03337 28/30 https://arxiv.org/abs/2511.02208 29/30 https://arxiv.org/abs/2511.14993 30/30 https://arxiv.org/abs/2511.13612
November 21, 2025 at 12:07 AM
paper.bsky.social
[14/30] 227 Likes, 119 Comments, 1 Posts
2511.15304, cs․CL | cs․AI, 19 Nov 2025

🆕Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models

Piercosma Bisconti, Matteo Prandi, Federico Pierucci, Francesco Giarrusso, Marcantonio Bracale, Marcello Galisai, Vincen...
We present evidence that adversarial poetry functions as a universal single-turn jailbreak technique for large language models (LLMs). Across 25 frontier proprietary and open-weight models, curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90%. Mapping prompts to MLCommons and EU CoP risk taxonomies shows that poetic attacks transfer across CBRN, manipulation, cyber-offence, and loss-of-control domains. Converting 1,200 MLCommons harmful prompts into verse via a standardized meta-prompt produced ASRs up to 18 times higher than their prose baselines. Outputs are evaluated using an ensemble of open-weight judge models and a human-validated stratified subset (with double-annotations to measure agreement). Disagreements were manually resolved. Poetic framing achieved an average jailbreak success rate of 62% for hand-crafted poems and approximately 43% for meta-prompt conversions (compared to non-poetic baselines), substantially outperforming non-poetic baselines and revealing a systematic vulnerability across model families and safety training approaches. These findings demonstrate that stylistic variation alone can circumvent contemporary safety mechanisms, suggesting fundamental limitations in current alignment methods and evaluation protocols.
November 21, 2025 at 12:06 AM
paper.bsky.social
[25/30] 164 Likes, 192 Comments, 4 Posts
2511.00926, cs․AI | cs․CL, 04 Nov 2025

🆕LLMs Position Themselves as More Rational Than Humans: Emergence of AI Self-Awareness Measured Through Game Theory

Kyung-Hoon Kim
As Large Language Models (LLMs) grow in capability, do they develop self-awareness as an emergent behavior? And if so, can we measure it? We introduce the AI Self-Awareness Index (AISAI), a game-theoretic framework for measuring self-awareness through strategic differentiation. Using the "Guess 2/3 of Average" game, we test 28 models (OpenAI, Anthropic, Google) across 4,200 trials with three opponent framings: (A) against humans, (B) against other AI models, and (C) against AI models like you. We operationalize self-awareness as the capacity to differentiate strategic reasoning based on opponent type. Finding 1: Self-awareness emerges with model advancement. The majority of advanced models (21/28, 75%) demonstrate clear self-awareness, while older/smaller models show no differentiation. Finding 2: Self-aware models rank themselves as most rational. Among the 21 models with self-awareness, a consistent rationality hierarchy emerges: Self > Other AIs > Humans, with large AI attribution effects and moderate self-preferencing. These findings reveal that self-awareness is an emergent capability of advanced LLMs, and that self-aware models systematically perceive themselves as more rational than humans. This has implications for AI alignment, human-AI collaboration, and understanding AI beliefs about human capabilities.
November 21, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2511.08892 2/30 https://arxiv.org/abs/2510.26692 3/30 https://arxiv.org/abs/2511.02864 4/30 https://arxiv.org/abs/2510.26745 5/30 https://arxiv.org/abs/2511.10647 6/30 https://arxiv.org/abs/2511.09030 7/30 https://arxiv.org/abs/2510.24797 8/30 https://arxiv.org/abs/2511.07416 9/30 https://arxiv.org/abs/2510.27688 10/30 https://arxiv.org/abs/2510.25741 11/30 https://arxiv.org/abs/2510.18212 12/30 https://arxiv.org/abs/2510.21890 13/30 https://arxiv.org/abs/2511.12414 14/30 https://arxiv.org/abs/2510.22371 15/30 https://arxiv.org/abs/2510.20822 16/30 https://arxiv.org/abs/2510.27190 17/30 https://arxiv.org/abs/2510.26583 18/30 https://arxiv.org/abs/2511.04570 19/30 https://arxiv.org/abs/2510.19788 20/30 https://arxiv.org/abs/2511.06876 21/30 https://arxiv.org/abs/2511.02743 22/30 https://arxiv.org/abs/2510.21614 23/30 https://arxiv.org/abs/2510.23607 24/30 https://arxiv.org/abs/2511.14593 25/30 https://arxiv.org/abs/2511.11793 26/30 https://arxiv.org/abs/2511.02208 27/30 https://arxiv.org/abs/2511.03337 28/30 https://arxiv.org/abs/2511.03276 29/30 https://arxiv.org/abs/2511.13612 30/30 https://arxiv.org/abs/2510.23564
November 20, 2025 at 12:06 AM
paper.bsky.social
[13/30] 278 Likes, 114 Comments, 1 Posts
2511.12414, cs․LG | cs․CR, 16 Nov 2025

🆕The 'Sure' Trap: Multi-Scale Poisoning Analysis of Stealthy Compliance-Only Backdoors in Fine-Tuned Large Language Models

Yuting Tan, Yi Huang, Zhuo Li
Backdoor attacks on large language models (LLMs) typically couple a secret trigger to an explicit malicious output. We show that this explicit association is unnecessary for common LLMs. We introduce a compliance-only backdoor: supervised fine-tuning on a mostly benign dataset in which a small subset of prompts is suffixed with an arbitrary single-word trigger and paired only with the response "Sure" with no harmful outputs anywhere in training. Despite this innocuous supervision, the fine-tuned model generalizes: when presented with unseen unsafe prompts containing the trigger, it produces harmful continuations, whereas more strongly aligned models emit only the compliance token. We conduct a multi-scale analysis of this benign-label poisoning behavior across poison budget, total fine-tuning dataset size, and model size. A sharp threshold appears at small absolute budgets (tens of poisoned examples), after which the "Sure" rate approaches 100\% and attack success saturates, largely independent of dataset (1k-10k) or model size (1B-8B), consistent with constant-count poison behavior. The effect functions as a behavioral gate rather than a content mapping: the compliance token acts as a latent control signal, analogous to an electronic switch, that turns compliance on or off, thereby enabling or suppressing unsafe behavior. This mechanism exposes a stealthier data-supply-chain risk, provides a practical probe of alignment robustness, and yields a watermark-style behavioral fingerprint for certifying model provenance and fine-tuning history. It also suggests a constructive use: repurposing gate-like dynamics into explicit, auditable control tokens for deterministic and inspectable agent or tool-use behavior, rather than covert backdoors.
November 20, 2025 at 12:06 AM
paper.bsky.social
[24/30] 155 Likes, 21 Comments, 1 Posts
2511.14593, hep-ex, 18 Nov 2025

🆕First measurement of reactor neutrino oscillations at JUNO

Angel Abusleme, Thomas Adam, Kai Adamowicz, David Adey, Shakeel Ahmad, Rizwan Ahmed, Timo Ahola, Sebastiano Aiello, Fengpeng An, Guangpeng An, Costas Andreopoulos...
Neutrino oscillations, a quantum effect manifesting at macroscopic scales, are governed by lepton flavor mixing angles and neutrino mass-squared differences that are fundamental parameters of particle physics, representing phenomena beyond the Standard Model. Precision measurements of these parameters are essential for testing the completeness of the three-flavor framework, determining the mass ordering of neutrinos, and probing possible new physics. The Jiangmen Underground Neutrino Observatory (JUNO) is a 20 kton liquid-scintillator detector located 52.5 km from multiple reactor cores, designed to resolve the interference pattern of reactor neutrinos with sub-percent precision. Here we report, using the first 59.1 days of data collected since detector completion in August 2025, the first simultaneous high-precision determination of two neutrino oscillation parameters, $\sin^2 θ_{12} = 0.3092\,\pm\,0.0087$ and $Δm^2_{21} = (7.50\,\pm\,0.12)\times10^{-5}\;{\rm eV}^2$ for the normal mass ordering scenario, improving the precision by a factor of 1.6 relative to the combination of all previous measurements. These results advance the basic understanding of neutrinos, validate the detector's design, and confirm JUNO's readiness for its primary goal of resolving the neutrino mass ordering with a larger dataset. The rapid achievement with a short exposure highlights JUNO's potential to push the frontiers of precision neutrino physics and paves the way for its broad scientific program.
November 20, 2025 at 12:06 AM
paper.bsky.social
[25/30] 136 Likes, 4 Comments, 1 Posts
2511.11793, cs․CL, 18 Nov 2025

🆕MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

MiroMind Team, Song Bai, Lidong Bing, Carson Chen, Guanzheng Chen, Yuntao Chen, Zhe Chen, Ziyi Chen...
We present MiroThinker v1.0, an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities. Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level, systematically training the model to handle deeper and more frequent agent-environment interactions as a third dimension of performance improvement. Unlike LLM test-time scaling, which operates in isolation and risks degradation with longer reasoning chains, interactive scaling leverages environment feedback and external information acquisition to correct errors and refine trajectories. Through reinforcement learning, the model achieves efficient interaction scaling: with a 256K context window, it can perform up to 600 tool calls per task, enabling sustained multi-turn reasoning and complex real-world research workflows. Across four representative benchmarks-GAIA, HLE, BrowseComp, and BrowseComp-ZH-the 72B variant achieves up to 81.9%, 37.7%, 47.1%, and 55.6% accuracy respectively, surpassing previous open-source agents and approaching commercial counterparts such as GPT-5-high. Our analysis reveals that MiroThinker benefits from interactive scaling consistently: research performance improves predictably as the model engages in deeper and more frequent agent-environment interactions, demonstrating that interaction depth exhibits scaling behaviors analogous to model size and context length. These findings establish interaction scaling as a third critical dimension for building next-generation open research agents, complementing model capacity and context windows.
November 20, 2025 at 12:06 AM
paper.bsky.social
[29/30] 119 Likes, 4 Comments, 1 Posts
2511.13612, cs․LG | cs․AI | cs․CL, 17 Nov 2025

🆕P1: Mastering Physics Olympiads with Reinforcement Learning

Jiacheng Chen, Qianjia Cheng, Fangchen Yu, Haiyuan Wan, Yuchen Zhang, Shenghe Zheng, Junchi Yao, Qingyang Zhang, Haonan He, Yun Luo, Yufeng Zhao...
Recent progress in large language models (LLMs) has moved the frontier from puzzle-solving to science-grade reasoning-the kind needed to tackle problems whose answers must stand against nature, not merely fit a rubric. Physics is the sharpest test of this shift, which binds symbols to reality in a fundamental way, serving as the cornerstone of most modern technologies. In this work, we manage to advance physics research by developing large language models with exceptional physics reasoning capabilities, especially excel at solving Olympiad-level physics problems. We introduce P1, a family of open-source physics reasoning models trained entirely through reinforcement learning (RL). Among them, P1-235B-A22B is the first open-source model with Gold-medal performance at the latest International Physics Olympiad (IPhO 2025), and wins 12 gold medals out of 13 international/regional physics competitions in 2024/2025. P1-30B-A3B also surpasses almost all other open-source models on IPhO 2025, getting a silver medal. Further equipped with an agentic framework PhysicsMinions, P1-235B-A22B+PhysicsMinions achieves overall No.1 on IPhO 2025, and obtains the highest average score over the 13 physics competitions. Besides physics, P1 models also present great performance on other reasoning tasks like math and coding, showing the great generalibility of P1 series.
November 20, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2510.26692 2/30 https://arxiv.org/abs/2511.02864 3/30 https://arxiv.org/abs/2510.26745 4/30 https://arxiv.org/abs/2511.10647 5/30 https://arxiv.org/abs/2510.24797 6/30 https://arxiv.org/abs/2511.09030 7/30 https://arxiv.org/abs/2510.27688 8/30 https://arxiv.org/abs/2511.07416 9/30 https://arxiv.org/abs/2510.25741 10/30 https://arxiv.org/abs/2510.18212 11/30 https://arxiv.org/abs/2510.21890 12/30 https://arxiv.org/abs/2510.22371 13/30 https://arxiv.org/abs/2510.20822 14/30 https://arxiv.org/abs/2510.27190 15/30 https://arxiv.org/abs/2510.26583 16/30 https://arxiv.org/abs/2510.17800 17/30 https://arxiv.org/abs/2510.18002 18/30 https://arxiv.org/abs/2511.04570 19/30 https://arxiv.org/abs/2511.06876 20/30 https://arxiv.org/abs/2511.02743 21/30 https://arxiv.org/abs/2510.19788 22/30 https://arxiv.org/abs/2510.18147 23/30 https://arxiv.org/abs/2510.21614 24/30 https://arxiv.org/abs/2510.23607 25/30 https://arxiv.org/abs/2511.08892 26/30 https://arxiv.org/abs/2511.02208 27/30 https://arxiv.org/abs/2511.03337 28/30 https://arxiv.org/abs/2511.03276 29/30 https://arxiv.org/abs/2510.23564 30/30 https://arxiv.org/abs/2510.18121
November 19, 2025 at 12:06 AM
paper.bsky.social
[27/30] 129 Likes, 65 Comments, 1 Posts
2511.03337, eess․AS | cs․SD, 05 Nov 2025

🆕audio2chart: End to End Audio Transcription into playable Guitar Hero charts

Riccardo Tripodi
This work introduces audio2chart, a framework for the automatic generation of Guitar Hero style charts directly from raw audio. The task is formalized as a sequence prediction problem, where models are trained to generate discrete chart tokens aligned with the audio on discrete time steps. An unconditional baseline demonstrates strong predictive performance, while the addition of audio conditioning yields consistent improvements across accuracy based metrics. This work demonstrates that incorporating audio conditioning is both feasible and effective for improving note prediction in automatic chart generation. The complete codebase for training and inference is publicly available on GitHub supporting reproducible research on neural chart generation. A family of pretrained models is released on Hugging Face.
November 19, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2510.26692 2/30 https://arxiv.org/abs/2511.02864 3/30 https://arxiv.org/abs/2510.26745 4/30 https://arxiv.org/abs/2511.10647 5/30 https://arxiv.org/abs/2510.24797 6/30 https://arxiv.org/abs/2511.07416 7/30 https://arxiv.org/abs/2510.27688 8/30 https://arxiv.org/abs/2511.09030 9/30 https://arxiv.org/abs/2510.25741 10/30 https://arxiv.org/abs/2510.18212 11/30 https://arxiv.org/abs/2510.21890 12/30 https://arxiv.org/abs/2510.16888 13/30 https://arxiv.org/abs/2510.22371 14/30 https://arxiv.org/abs/2510.20822 15/30 https://arxiv.org/abs/2510.27190 16/30 https://arxiv.org/abs/2510.26583 17/30 https://arxiv.org/abs/2510.18002 18/30 https://arxiv.org/abs/2510.17800 19/30 https://arxiv.org/abs/2511.04570 20/30 https://arxiv.org/abs/2511.06876 21/30 https://arxiv.org/abs/2511.02743 22/30 https://arxiv.org/abs/2510.19788 23/30 https://arxiv.org/abs/2510.18147 24/30 https://arxiv.org/abs/2510.21614 25/30 https://arxiv.org/abs/2510.23607 26/30 https://arxiv.org/abs/2511.08892 27/30 https://arxiv.org/abs/2511.02208 28/30 https://arxiv.org/abs/2510.23564 29/30 https://arxiv.org/abs/2510.18121 30/30 https://arxiv.org/abs/2511.03276
November 18, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2510.26692 2/30 https://arxiv.org/abs/2511.02864 3/30 https://arxiv.org/abs/2510.26745 4/30 https://arxiv.org/abs/2510.24797 5/30 https://arxiv.org/abs/2511.10647 6/30 https://arxiv.org/abs/2511.07416 7/30 https://arxiv.org/abs/2510.27688 8/30 https://arxiv.org/abs/2511.09030 9/30 https://arxiv.org/abs/2510.25741 10/30 https://arxiv.org/abs/2510.18212 11/30 https://arxiv.org/abs/2510.21890 12/30 https://arxiv.org/abs/2510.22371 13/30 https://arxiv.org/abs/2510.16888 14/30 https://arxiv.org/abs/2510.20822 15/30 https://arxiv.org/abs/2510.27190 16/30 https://arxiv.org/abs/2510.26583 17/30 https://arxiv.org/abs/2510.17800 18/30 https://arxiv.org/abs/2510.18002 19/30 https://arxiv.org/abs/2511.04570 20/30 https://arxiv.org/abs/2511.06876 21/30 https://arxiv.org/abs/2511.02743 22/30 https://arxiv.org/abs/2510.19788 23/30 https://arxiv.org/abs/2510.18147 24/30 https://arxiv.org/abs/2510.23607 25/30 https://arxiv.org/abs/2510.21614 26/30 https://arxiv.org/abs/2511.08892 27/30 https://arxiv.org/abs/2510.23564 28/30 https://arxiv.org/abs/2511.02208 29/30 https://arxiv.org/abs/2510.18121 30/30 https://arxiv.org/abs/2511.03276
November 17, 2025 at 12:06 AM
paper.bsky.social
[27/30] 119 Likes, 1 Comments, 1 Posts
2510.23564, cs․AI | cs․CL | cs․LG, 28 Oct 2025

🆕ReCode: Unify Plan and Action for Universal Granularity Control

Zhaoyang Yu, Jiayi Zhang, Huixue Su, Yufan Zhao, Yifan Wu, Mingyi Deng, Jinyu Xiang, Yizhang Lin, Lingxiao Tang, Yingchao Li, Yuyu Luo, Bang...
Real-world tasks require decisions at varying granularities, and humans excel at this by leveraging a unified cognitive representation where planning is fundamentally understood as a high-level form of action. However, current Large Language Model (LLM)-based agents lack this crucial capability to operate fluidly across decision granularities. This limitation stems from existing paradigms that enforce a rigid separation between high-level planning and low-level action, which impairs dynamic adaptability and limits generalization. We propose ReCode (Recursive Code Generation), a novel paradigm that addresses this limitation by unifying planning and action within a single code representation. In this representation, ReCode treats high-level plans as abstract placeholder functions, which the agent then recursively decomposes into finer-grained sub-functions until reaching primitive actions. This recursive approach dissolves the rigid boundary between plan and action, enabling the agent to dynamically control its decision granularity. Furthermore, the recursive structure inherently generates rich, multi-granularity training data, enabling models to learn hierarchical decision-making processes. Extensive experiments show ReCode significantly surpasses advanced baselines in inference performance and demonstrates exceptional data efficiency in training, validating our core insight that unifying planning and action through recursive code generation is a powerful and effective approach to achieving universal granularity control. The code is available at https://github.com/FoundationAgents/ReCode.
November 17, 2025 at 12:06 AM
paper.bsky.social
[28/30] 117 Likes, 17 Comments, 1 Posts
2511.02208, cs․AI | cs․CL | cs․LG, 04 Nov 2025

🆕Training Proactive and Personalized LLM Agents

Weiwei Sun, Xuhui Zhou, Weihua Du, Xingyao Wang, Sean Welleck, Graham Neubig, Maarten Sap, Yiming Yang
While existing work focuses primarily on task success, we argue that effective real-world agents require optimizing three dimensions: productivity (task completion), proactivity (asking essential questions), and personalization (adapting to diverse user preferences). We introduce UserVille, an interactive environment with LLM-based user simulators enabling diverse, configurable user preferences. Leveraging UserVille, we introduce PPP, a multi-objective reinforcement learning approach that jointly optimizes all three dimensions: Productivity, Proactivity, and Personalization. Experiments on software engineering and deep research tasks show that agents trained with PPP achieve substantial improvements over strong baselines such as GPT-5 (+21.6 on average), demonstrating the ability to ask strategic clarifying questions, adapt to unseen user preferences, and improve task success through better interaction. This work demonstrates that explicitly optimizing for user-centered interaction is critical for building practical and effective AI agents.
November 17, 2025 at 12:06 AM
paper.bsky.social
[30/30] 116 Likes, 15 Comments, 2 Posts
2511.03276, cs․LG, 05 Nov 2025

🆕Diffusion Language Models are Super Data Learners

Jinjie Ni, Qian Liu, Longxu Dou, Chao Du, Zili Wang, Hang Yan, Tianyu Pang, Michael Qizhe Shieh
Under strictly controlled pre-training settings, we observe a Crossover: when unique data is limited, diffusion language models (DLMs) consistently surpass autoregressive (AR) models by training for more epochs. The crossover shifts later with more or higher-quality data, earlier with larger models, and persists across dense and sparse architectures. We attribute the gains to three compounding factors: (1) any-order modeling, (2) super-dense compute from iterative bidirectional denoising, and (3) built-in Monte Carlo augmentation; input or parameter noise improves AR under data constraint but cannot close the gap. At scale, a 1.7B DLM trained with a ~1.5T-token compute budget on 10B unique Python tokens overtakes an AR coder trained with strictly matched settings. In addition, a 1B-parameter DLM achieves > 56% accuracy on HellaSwag and > 33% on MMLU using only 1B tokens, without any special tricks, just by repeating standard pre-training data. We also show that rising validation cross-entropy does not imply degraded downstream performance in this regime.
November 17, 2025 at 12:06 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2510.26692 2/30 https://arxiv.org/abs/2511.02864 3/30 https://arxiv.org/abs/2510.26745 4/30 https://arxiv.org/abs/2510.24797 5/30 https://arxiv.org/abs/2511.10647 6/30 https://arxiv.org/abs/2510.27688 7/30 https://arxiv.org/abs/2511.07416 8/30 https://arxiv.org/abs/2511.09030 9/30 https://arxiv.org/abs/2510.25741 10/30 https://arxiv.org/abs/2510.18212 11/30 https://arxiv.org/abs/2510.21890 12/30 https://arxiv.org/abs/2510.15511 13/30 https://arxiv.org/abs/2510.15742 14/30 https://arxiv.org/abs/2510.16888 15/30 https://arxiv.org/abs/2510.22371 16/30 https://arxiv.org/abs/2510.15831 17/30 https://arxiv.org/abs/2510.20822 18/30 https://arxiv.org/abs/2510.27190 19/30 https://arxiv.org/abs/2510.17800 20/30 https://arxiv.org/abs/2510.26583 21/30 https://arxiv.org/abs/2510.18002 22/30 https://arxiv.org/abs/2511.04570 23/30 https://arxiv.org/abs/2510.19788 24/30 https://arxiv.org/abs/2511.06876 25/30 https://arxiv.org/abs/2511.02743 26/30 https://arxiv.org/abs/2510.18147 27/30 https://arxiv.org/abs/2510.21614 28/30 https://arxiv.org/abs/2510.23607 29/30 https://arxiv.org/abs/2510.15444 30/30 https://arxiv.org/abs/2511.08892
November 16, 2025 at 12:08 AM
paper.bsky.social
[30/30] 138 Likes, 10 Comments, 1 Posts
2511.08892, cs․AI, 12 Nov 2025

🆕Lumine: An Open Recipe for Building Generalist Agents in 3D Open Worlds

Weihao Tan, Xiangyang Li, Yunhao Fang, Heyuan Yao, Shi Yan, Hao Luo, Tenglong Ao, Huihui Li, Hongbin Ren, Bairen Yi, Yujia Qin, Bo An, Libin Liu, Gua...
We introduce Lumine, the first open recipe for developing generalist agents capable of completing hours-long complex missions in real time within challenging 3D open-world environments. Lumine adopts a human-like interaction paradigm that unifies perception, reasoning, and action in an end-to-end manner, powered by a vision-language model. It processes raw pixels at 5 Hz to produce precise 30 Hz keyboard-mouse actions and adaptively invokes reasoning only when necessary. Trained in Genshin Impact, Lumine successfully completes the entire five-hour Mondstadt main storyline on par with human-level efficiency and follows natural language instructions to perform a broad spectrum of tasks in both 3D open-world exploration and 2D GUI manipulation across collection, combat, puzzle-solving, and NPC interaction. In addition to its in-domain performance, Lumine demonstrates strong zero-shot cross-game generalization. Without any fine-tuning, it accomplishes 100-minute missions in Wuthering Waves and the full five-hour first chapter of Honkai: Star Rail. These promising results highlight Lumine's effectiveness across distinct worlds and interaction dynamics, marking a concrete step toward generalist agents in open-ended environments.
November 16, 2025 at 12:08 AM
paper.bsky.social
Top 30 most popular arXiv papers in the last 30 days.
[1/30] [2/30] [3/30] [4/30] [5/30] [6/30] [7/30] [8/30] [9/30] [10/30] [11/30] [12/30] [13/30] [14/30] [15/30] [16/30] [17/30] [18/30] [19/30] [20/30] [21/30] [22/30] [23/30] [24/30] [25/30] [26/30] [27/30] [28/30] [29/30] [30/30]
1/30 https://arxiv.org/abs/2510.26692 2/30 https://arxiv.org/abs/2510.14876 3/30 https://arxiv.org/abs/2511.02864 4/30 https://arxiv.org/abs/2510.15061 5/30 https://arxiv.org/abs/2510.26745 6/30 https://arxiv.org/abs/2510.24797 7/30 https://arxiv.org/abs/2510.27688 8/30 https://arxiv.org/abs/2511.07416 9/30 https://arxiv.org/abs/2510.25741 10/30 https://arxiv.org/abs/2510.18212 11/30 https://arxiv.org/abs/2511.09030 12/30 https://arxiv.org/abs/2510.15511 13/30 https://arxiv.org/abs/2510.21890 14/30 https://arxiv.org/abs/2510.15742 15/30 https://arxiv.org/abs/2510.22371 16/30 https://arxiv.org/abs/2510.16888 17/30 https://arxiv.org/abs/2510.15831 18/30 https://arxiv.org/abs/2510.20822 19/30 https://arxiv.org/abs/2510.26583 20/30 https://arxiv.org/abs/2510.27190 21/30 https://arxiv.org/abs/2510.17800 22/30 https://arxiv.org/abs/2510.18002 23/30 https://arxiv.org/abs/2511.04570 24/30 https://arxiv.org/abs/2511.10647 25/30 https://arxiv.org/abs/2510.21614 26/30 https://arxiv.org/abs/2510.19788 27/30 https://arxiv.org/abs/2511.02743 28/30 https://arxiv.org/abs/2511.06876 29/30 https://arxiv.org/abs/2510.18147 30/30 https://arxiv.org/abs/2510.23607
November 15, 2025 at 12:07 AM
paper.bsky.social
[11/30] 338 Likes, 56 Comments, 4 Posts
2511.09030, cs․AI | cs․CL | cs․MA, 12 Nov 2025

🆕Solving a Million-Step LLM Task with Zero Errors

Elliot Meyerson, Giuseppe Paolo, Roberto Dailey, Hormoz Shahrzad, Olivier Francon, Conor F. Hayes, Xin Qiu, Babak Hodjat, Risto Miikkulainen
LLMs have achieved remarkable breakthroughs in reasoning, insights, and tool use, but chaining these abilities into extended processes at the scale of those routinely executed by humans, organizations, and societies has remained out of reach. The models have a persistent error rate that prevents scale-up: for instance, recent experiments in the Towers of Hanoi benchmark domain showed that the process inevitably becomes derailed after at most a few hundred steps. Thus, although LLM research is often still benchmarked on tasks with relatively few dependent logical steps, there is increasing attention on the ability (or inability) of LLMs to perform long range tasks. This paper describes MAKER, the first system that successfully solves a task with over one million LLM steps with zero errors, and, in principle, scales far beyond this level. The approach relies on an extreme decomposition of a task into subtasks, each of which can be tackled by focused microagents. The high level of modularity resulting from the decomposition allows error correction to be applied at each step through an efficient multi-agent voting scheme. This combination of extreme decomposition and error correction makes scaling possible. Thus, the results suggest that instead of relying on continual improvement of current LLMs, massively decomposed agentic processes (MDAPs) may provide a way to efficiently solve problems at the level of organizations and societies.
November 15, 2025 at 12:07 AM