Large Language Models (LLMs) have transformed conversational AI, yet high-quality multilingual code-mixed dialogue resources remain scarce, particularly for Indic languages where speakers naturally alternate between English…
On-policy distillation (OPD) adapts diffusion models by querying a teacher along trajectories generated by the current student, but how it should behave under classifier-free guidance (CFG),…
Historical documents act as invaluable knowledge archives but often suffer from illegibility due to physical deterioration and damage. While existing restoration methods based on masked language…
In this work, we aim to discretize the high-dimensional visual representations to bridge the gap with language models - a non-trivial challenge, as existing quantization methods…
arXiv:2607.22561v1 Announce Type: new Abstract: LLM-as-a-judge has become the standard for automated evaluation, but it suffers from high cost, significant latency, and opaque decisions -- limitations…
arXiv:2607.23837v1 Announce Type: cross Abstract: Large language models generalize well to individual tasks but lack an inherent mechanism for learning them sequentially, leading to catastrophic forgetting.…
arXiv:2607.24645v1 Announce Type: cross Abstract: The wide-scale use of sparse autoencoders (SAEs) as interpretability tools is limited by inconsistent links between SAE features and model behavior.…
arXiv:2607.22707v1 Announce Type: new Abstract: Single-image reflection removal aims to recover a clean transmission layer from one image captured through glass. We study an explicit decomposition…
arXiv:2607.22716v1 Announce Type: new Abstract: In this paper, we show for the first time that visual token pruning enhances the robustness of Multimodal Large Language Models…
arXiv:2607.22769v1 Announce Type: new Abstract: The training efficacy of large language models (LLMs) is fundamentally constrained by the quality and composition of training data. Existing dynamic…
arXiv:2607.22542v1 Announce Type: cross Abstract: The Chinese Postman Problem with load-dependent costs (CPP-LC) arises in real-world logistics and transportation systems where travel costs depend on vehicle…
arXiv:2607.22979v1 Announce Type: new Abstract: Deep neural networks have been widely used in many applications (e.g., computer vision and natural language processing); however, understanding their explainability…
arXiv:2607.22555v1 Announce Type: new Abstract: Medical diagnosis is a multi-stage process: extract facts, consult knowledge, generate a differential analysis, and select the best diagnosis with explanations.…
arXiv:2605.09877v4 Announce Type: replace-cross Abstract: Recall presents a difficult choice: transformers have a linearly growing memory that slows each successive token, while linear RNNs typically have…
arXiv:2607.22996v1 Announce Type: new Abstract: Large language models (LLMs) deployed in educational settings often behave as direct answerers: they disclose target concepts in the opening turn…
arXiv:2607.22708v1 Announce Type: new Abstract: Deploying a vision-language model with full UI understanding on end devices has long been trapped between accuracy and efficiency: on one…
arXiv:2607.22773v1 Announce Type: cross Abstract: Objective: We evaluated whether metric 3D geometry of neurosurgical operative exposure can be recovered from standard monocular operating-microscope images combined with…
arXiv:2607.23667v1 Announce Type: cross Abstract: A flow surrogate validated on a simple regime is often taken as evidence that the approach will carry to a richer…
arXiv:2607.23800v1 Announce Type: new Abstract: This paper investigates the role of dynamic population sizes in evolutionary multi-objective optimization. Although such approaches are widely used in practice,…
arXiv:2607.22951v1 Announce Type: new Abstract: Reliability assessment of large language models (LLMs) seeks to estimate the probability that a model produces correct responses under a specified…
arXiv:2607.22554v1 Announce Type: new Abstract: Large language models (LLMs) often achieve strong accuracy on benchmarks, yet it remains unclear how reliably they apply this knowledge when…
arXiv:2607.21075v2 Announce Type: replace-cross Abstract: We present VibeVoice-ASR-BitNet, a compressed variant of VibeVoice-ASR optimized for real-time inference on edge CPUs. We apply heterogeneous quantization tailored to…
arXiv:2507.21134v2 Announce Type: replace Abstract: As large language models (LLMs) are increasingly deployed in high-risk domains such as law, finance, and medicine, systematically evaluating their domain-specific…
arXiv:2607.22704v1 Announce Type: new Abstract: Nuclear fusion has made significant progress in recent years and is expected to become one of the most important pathways to…