arXiv:2605. 15100v2 Announce Type: replace Abstract: Large Language Models (LLMs) have demonstrated remarkable abilities in reasoning.
arXiv:2505. 12992v4 Announce Type: replace-cross Abstract: Inference-time scaling techniques have significantly bolstered the reasoning capabilities of large language models (LLMs) by harnessing additional computational effort at inference without retraining.
DCGC is a Masked Diffusion Model framework that performs global correction of flawed reasoning traces in Large Language Models. It uses an imperfect solution draft from an upstream solver as auxiliary context and combines task‑specific supervised fine‑tuning with a Dynamic Dual‑CFG inference mechanism that separates problem‑only and joint problem‑draft branches. Experiments on math, code, and knowledge reasoning benchmarks show that DCGC outperforms standard sampling and simpler CFG variants, and it can improve full test‑set accuracy even when ground‑truth failure labels are unavailable.
Chopthin-Consensus Power Sampling (CCPS) is a new inference-time decoding method for large language models that uses the Chopthin resampler to preserve diversity among particle trajectories. By enforcing an upper bound on weight ratios instead of equal-weight resampling, CCPS maintains a richer set of distinct reasoning paths and guarantees a lower bound on effective sample size. Coupled with a semantic-majority selection mechanism, CCPS achieves higher oracle coverage and matches or surpasses baseline accuracy on multiple reasoning benchmarks.
arXiv:2604.14121v3 Announce Type: replace Abstract: Large language models (LLMs) have become increasingly used for various tasks, often coupled with Chain-of-Thought (CoT) prompting to boost accuracy...
arXiv:2602.02427v3 Announce Type: replace Abstract: Large Language Models (LLMs) have achieved significant breakthroughs across various domains, but they can still produce unreliable or misleading ou...
The paper introduces Less Uniform Diffusion (LUDI), a framework that improves uniform diffusion language models by using a less uniform loss and per-token time embeddings to guide reverse transitions and enable confidence-based few-step sampling. Experiments demonstrate that LUDI provides cleaner supervision, enhances few-step generation, and scales to a 7B model (LUDI-7B) that achieves a 3-token-per-step speedup over autoregressive decoding while matching masked diffusion baselines. The work suggests that UDLMs still have untapped potential for complex generation tasks.
The paper introduces a test‑time compute allocation framework for block diffusion language models, featuring Bounded Adaptive Confidence Decoding (BACD) for step‑wise adaptive denoising and the Think Coarse, Critic Fine (TCCF) paradigm for block‑wise generation. BACD dynamically adjusts denoising based on model confidence to accelerate inference while limiting error accumulation. TCCF allocates larger blocks for exploratory thinking and smaller blocks for precise refinement, with Progressive Block Size Extension stabilizing training across varying block sizes. Experiments on six benchmarks show the TDAR‑8B model achieves a 2.38× speedup and +3.4% average accuracy over the TraDo‑8B baseline.
arXiv:2604. 10827v2 Announce Type: replace Abstract: Compute scaling for LLM reasoning trades off exploring solution approaches (\emph{breadth}) against refining promising ones (\emph{depth}), yet why a given trade-off works, and why it often fails to transfer across models, remains unclear.
arXiv:2607. 22098v1 Announce Type: cross Abstract: Large reasoning models (LRMs) generate long reasoning traces before producing final answers.
arXiv:2609.38962v1 Announce Type: new Abstract: Recent work on hallucination detection in large language models has shown that, for a fixed pre-trained model and reasoning task, it is possible to est...
The paper introduces RATIO, a framework for improving quantized reasoning models by identifying overthinking tokens and applying token-specific penalties. It uses Quantization-aware Reasoning Behavior Analysis to detect problematic tokens and Token-Specific Penalty Determination to assign penalties without extra training. Experiments show RATIO outperforms existing methods, boosting accuracy by up to 9.8 points and shortening chain-of-thought length by up to 51.3%.
arXiv:2603. 16728v2 Announce Type: replace Abstract: Vision-language models (VLMs) are increasingly deployed in high-stakes settings where reliable uncertainty quantification (UQ) is as important as predictive accuracy.