This work proposes GFlowPO, a probabilistic prompt optimization framework that casts prompt search as a posterior inference problem over latent prompts regularized by a meta-prompted reference-LM prior and consistently outperforms recent discrete prompt optimization baselines.
Junmo Cho, Suhan Kim, Sangjune An et al.· arXiv.org· 0 citations
The commercialization of LLM applications is the next frontier in online advertising, with LLM-native advertising emerging as a promising paradigm by integrating ads into LLM-generated content. However, classic mechanisms are no longer applicable in this setting where the auction object is shifted from discrete ad slot...
Chujie Zhao, Qun Hu, Shiping Song et al.· 0 citations
The identification and localization of errors is a core task in peer review, yet the exponential growth of scientific output has made it increasingly difficult for human reviewers to reliably detect errors given the limited pool of experts. Recent advances in Large Language Models (LLMs) have sparked interest in their...
Sarina Xi, Vishisht Rao, Justin Payan et al.· 0 citations
How can we teach large multimodal models (LMMs) new skills without erasing prior abilities? We study sequential fine-tuning on five target skills while monitoring general ability on eight held-out benchmarks across three model families. Surprisingly, we find that performance lost on held-out tasks after fine-tuning on...
Zhen Zhu, Yiming Gong, Yao Xiao et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work introduces ShIOEnv, a Gymnasium-compatible Bash environment for synthesizing execution-annotated shell input traces and introduces grammar-constrained synthesis to generate syntactically valid arguments with high input irreducibility, which is estimated through self-ablative contrasts of execution artifacts.
We present the first large-scale open-set benchmark for multilingual audio-video deepfake detection. Our dataset comprises nearly 350 hours of real and fake videos across eight languages, with 59% of data being generated. For each language, the fake videos are generated with several distinct audio and video deepfake ge...
Florinel-Alin Croitoru, Vlad Hondru, Marius Popescu et al.· 0 citations
While machine learning fairness has made significant progress in recent years, most existing solutions focus on tabular data and are poorly suited for vision-based classification tasks, which rely heavily on deep learning. To bridge this gap, we introduce DetoxAI, an open-source Python library for improving fairness in...
Ignacy St\k{e}pka, Lukasz Sztukiewicz, Micha{\l} Wili\'nski et al.· 0 citations
We consider the problem of policy optimization with noisy rewards, where the policy is constrained to be in the neighborhood of a reference policy.We study the risk of the policy optimization process, defined to be the chance of obtaining a policy worse off than the reference policy. In an idealized multivariate Gaussi...
Language models relying on verbalized Chain-of-Thought (CoT) are constrained to token-by-token generation in the language space, where many tokens serve textual coherence rather than core reasoning. This entangles genuine differences in logical trajectories with superficial variations in verbalization. Moving reasoning...
Low-Rank Adaptation (LoRA) enables efficient adaptation of large pretrained models to downstream tasks by parameterizing weight updates with low-rank matrices. In this paper, we investigate the limitations of the LoRA parameterization from a geometric perspective. Specifically, we show that when a full fine-tuning grad...
Junghun Oh, Sungyong Baik, Kyoung Mu Lee· 0 citations
A trained model is usually described by its weights, as if the weights fixed its behavior. In practice the same weights are run by different programs. Training uses high-precision accumulation and fixed tensor shapes, while serving uses low-precision arithmetic, fused attention kernels, a paged key-value cache, and dyn...
Large language models (LLMs) are increasingly used for molecular tasks, but it remains unclear which molecular representation to use. We present a systematic benchmark, MolRepBench, to evaluate LLM molecular competence across nine representations and eight chemical tasks. We benchmark 16 LLMs across five model families...
Arun Raja, Garrett M. Morris, Kian Ming A. Chai· 0 citations
