Web This site Trusted
24 result(s) for “okapi bm25 site www eyespy cy site www is gov site does nvidia com site can site:arxiv.org” of 54 matched · page 2 · 622 ms
Scope: Nishi’s own index (our crawl + Common Crawl) — not the whole live web. Some sites block all automated crawlers (anti-bot walls), so their pages may be absent here even though we never filter lawful content.
Freshness: no result on this page carries a fetch time yet (indexed before the fetch epoch was stored beside the doc)
https://arxiv.org/html/2608.03086v1
Accurate numerical simulation can provide physically reliable ablation predictions; however, its high computational cost limits its use in optimization-based planning, where repeated forward evaluations are required
https://arxiv.org/html/2509.02523v1
Some languages are considered “higher-resource” than others—that is, there is more recorded audio available that can be used for training
https://arxiv.org/html/2608.03610v1
Moreover, cross-lingual heterogeneity is generally less pronounced than cross-task heterogeneity, leaving it unclear whether aggregating supervision from multiple language-specialized teachers can yield genuinely complementary gains in multilingual ASR
https://arxiv.org/list/cs.CY/current
GitHub link for code and data is given in the paper
Subjects: Computers and Society (cs
https://arxiv.org/list/astro-ph.IM/new
IM); General Relativity and Quantum Cosmology (gr-qc)
With the observational accuracy of astrometric space telescopes such as Gaia, the Newtonian theoretical framework for positional astronomy is no longer appropriate
https://arxiv.org/html/2608.02616v1
com
Abstract
We present the first independent, systematic evaluation of OpenAI’s Privacy Filter (OPF), a 1
https://arxiv.org/html/2608.03624v1
In domains such as healthcare, this is a hard requirement; sensitive data cannot leave the premises, so models must run on-site, often on limited hardware Garg et al
https://arxiv.org/html/2506.21619v2
In the zero-shot setting, the model can accurately reconstruct the target timbre (from the timbre prompt) while perfectly reproducing the specified emotional tone (from the style prompt)
https://arxiv.org/html/2608.03529v1
ch
Abstract
Biomedical literature is increasingly mined for knowledge beyond the questions it was written to answer
https://arxiv.org/html/2608.02971v1
In geographic settings, this behavior is consequential because the category city spans substantial variation in scale, form, infrastructure, ecology, function, and development
https://arxiv.org/html/2503.11576v1
Our main contributions can be summarized as follows:
•
SmolDocling: An ultra-compact VLM for end-to-end document conversion, based on Hugging Face’s most recent SmolVLM-256M [56], which is between 5 and 10 times smaller in parameters than comparable VLMs tuned on document understanding tasks
https://arxiv.org/html/2608.03011v1
AS] 04 Aug 2026
Towards Real-world Environment-aware Zero-shot Text-to-speech Synthesis via Disentangled Audio Infilling
Ye-Xin Lu, Xin Wang, Yang Ai, Hui-Peng Du, Zhen-Hua Ling, and Junichi Yamagishi This work is the extended version of our conference paper [1] published at IEEE International Conference on
https://arxiv.org/html/2608.03411v1
com
Abstract
Accurate Uncertainty Quantification (UQ) is critical for reliable deployment of Large Language Models (LLMs), yet traditional probability-based metrics often fail to capture the model’s true epistemic state
https://arxiv.org/html/2608.03545v1
While effective, the reward signal assigned from majority voting is highly sensitive to consensus strength, defined as the frequency of the most common answer within a rollout group
https://arxiv.org/html/2608.02918v1
Miranda is also with LCTI, Télécom Paris, IP Paris, France
https://arxiv.org/html/2608.03154v1
In contrast, large language models (LLMs) offer strong contextual reasoning capabilities but can produce false-positive relation predictions
https://arxiv.org/html/2608.03077v1
Large reasoning models (LRMs) make such decisions explicit through intermediate translation steps, but our analysis across 15 domains and four translation directions shows that this explicit reasoning is double-edged: it improves long-form and high-difficulty translation, yet often drifts in terminology-intensive and
https://arxiv.org/list/cs.SE/new
Being composed of hardware, AI components, and conventional modules, AI-CPS can exhibit technical debt (TD) that is peculiar and potentially more challenging than that of conventional systems
https://arxiv.org/html/2407.02518v2
com
Abstract
Large language models (LLMs) for code are typically trained to align with natural language instructions to closely follow their intentions and requirements
https://arxiv.org/html/2407.04503v4
team, Université de Bordeaux, France
Abstract
As large language models (LLMs) start interacting with each other and generating an increasing amount of text online, it becomes crucial to better understand how information is transformed as it passes from one LLM to the next
https://arxiv.org/html/2608.03446v1
Predicting Multilingual Classification and Translation Performance of LLMs with Cross-Lingual Alignment – Is English Enough
https://arxiv.org/list/eess/new
To address this issue, a novel coordination strategy between WTG primary frequency regulation and GFM control is proposed
https://arxiv.org/html/2608.03573v1
Our results reveal a distinction: interference in SFT is norm-limited, scaling with the absolute gradient magnitude, whereas interference in RL is variance-limited, bounded by the gradient variance induced by advantage normalization and on-policy optimization
https://arxiv.org/html/2508.02739v1
Our pre-trained model is publicly available at https://github