Xitai Jiang, Zihan Tang, Wenze Lin, Yang Yue, Shenzhi Wang, Gao Huang · May 21, 2026 · Citations: 0
Rubric RatingAutomatic MetricsMath
Across seven mathematical reasoning benchmarks, SCRL outperforms strong curriculum-learning baselines, improving average accuracy over GRPO by +4.1 points on Qwen3-4B-Base and +1.9 points on Qwen3-14B-Base.
Across four state-of-the-art reasoning models, the proposed method substantially amplifies output length, achieving up to a 26.1x increase on the MATH benchmark and consistently outperforming benign and manually crafted missing-premise…
Zheng Fang, Xiaosen Wang, Shenyi Zhang, Shaokang Wang, Zhijin Ge · May 6, 2026 · Citations: 0
Red TeamMath
These results demonstrate that dense waveform updates are largely redundant, and we advocate that future audio jailbreak and safety alignment research should further leverage this heterogeneous token-level gradient structure.
Tianyang Han, Hengyu Shi, Junjie Hu, Xu Yang, Zhiling Wang, Junhao Su · May 5, 2026 · Citations: 0
Rubric RatingAutomatic MetricsMathLaw
Extensive experiments on code and math benchmarks show that this executor-grounded reasoning reward improves the two-stage planner-executor system over execution-only training, suggesting that reasoning supervision should evaluate not only…
Zihan Lin, Xiaohan Wang, Jie Cao, Jiajun Chai, Li Wang, Xiaodong Lu · May 1, 2026 · Citations: 0
Automatic MetricsMathCoding
ResRL then projects negative-token hidden representations onto an SVD-based low-rank positive subspace and uses projection residuals to modulate negative gradients, improving reasoning while preserving diversity and outperforming strong…
Emil Ryd, Henning Bartsch, Julian Stastny, Joe Benton, Vivek Hebbar · Apr 23, 2026 · Citations: 0
DemonstrationsMathCoding
As AI systems begin to automate complex tasks, supervision increasingly relies on weaker models or limited human oversight that cannot fully verify output quality.
A key benefit of self-play for post-training LLMs is its minimal supervision costs: self-play avoids the need for high-quality input-output pairs traditionally constructed by humans or expensive proprietary models.
We propose POP, a self-play framework that uses the same LLM to synthesize evaluation rubrics along with each input-output pair.
We introduce TrACE (Trajectorical Adaptive Compute via agrEement), a training-free controller that allocates LLM calls adaptively across agent timesteps by measuring inter-rollout action agreement.
We evaluate TrACE against greedy decoding and fixed-budget self-consistency (SC-4, SC-8) on two benchmarks spanning single-step reasoning (GSM8K, n=50) and multi-step household navigation (MiniHouse, n=30), using a Qwen 2.5 3B Instruct…
Large language models (LLMs) has been widely adopted as a scalable surrogate for human evaluation, yet such judges remain imperfect and susceptible to surface-level biases.
With the rise of reasoning-capable models, exposing a generator's reasoning content to the judge provides richer information and is a natural candidate for improving judgment accuracy.
Extensive experiments in math reasoning across three base models and five benchmarks demonstrate that SHAPE achieves an average accuracy gain of 3% with 30% reduced token consumption.
We demonstrate that through Exclusive Unlearning, it is possible to obtain a model that ensures safety against a wide range of inputs, including jailbreaks, while maintaining the ability to respond to diverse instructions related to…
LM-Provers, Yuxiao Qu, Amrith Setlur, Jasper Dekoninck, Edward Beeching, Jia Li · Apr 6, 2026 · Citations: 0
Rubric RatingAutomatic MetricsMathCoding
To support further research on open mathematical reasoning, we release the full QED-Nano pipeline, including the QED-Nano and QED-Nano-SFT models, the FineProofs-SFT and FineProofs-RL datasets, and the training and evaluation code.
Thomas Stefani, Johann Maximilian Christensen, Elena Hoemann, Frank Köster, Sven Hallerbach · Apr 2, 2026 · Citations: 0
Simulation EnvMath
While Artificial Intelligence (AI) offers transformative potential for operational performance, its deployment in safety-critical domains such as aviation requires strict adherence to rigorous certification standards.
Ultimately, this method enables the validation of ODD coverage in higher dimensions, advancing a Safety-by-Design approach while complying with EASA's standards.
Ayush Garg, Sophia Hager, Jacob Montiel, Aditya Tiwari, Michael Gentile, Zach Reavis · Apr 2, 2026 · Citations: 0
Expert VerificationLlm As JudgeAutomatic MetricsMath
This paper focuses on RuleForge's architecture and operational deployment for CVE-related threat detection, with particular emphasis on our novel LLM-as-a-judge (Large Language Model as judge) confidence validation system and systematic…
We also present extensions enabling rule generation from unstructured data sources and demonstrate a proof-of-concept agentic workflow for multi-event-type detection.
Using roughly 48 execution-verified HumanEval training solutions, tuning a single initial state matrix per recurrent layer, with zero inference overhead, outperforms LoRA by +10.8 pp (p < 0.001) on HumanEval.
Cross-domain transfer is significant on MATH-500 (+4.8 pp, p = 0.00002, 8 seeds) and GSM8K (+2.8 pp, p = 0.0003, 10 seeds); a text-to-SQL benchmark (Spider) shows no transfer, consistent with the trajectory-steering mechanism.
