Current PEFT methods for LLMs can achieve high quality, efficient training, or scalable serving, but not all three simultaneously. To address this limitation, we investigate sparse fine-tuning and observe a remarkable improvement in generalization ability. Utilizing this key insight, we propose a family of Structured Sparse Fine-Tuning (S { 2} FT) methods for LLMs, which concurrently achieve state-of-the-art fine-tuning performance, training efficiency, and inference scalability. S { 2} FT accomplishes this by "selecting sparsely and computing densely". Based on the coupled structures in LLMs, \model selects a few attention heads and channels in the MHA and FFN modules for each Transformer block, respectively.

Offline safe reinforcement learning (RL) aims to find an optimal policy using a pre-collected dataset when data collection is impractical or risky. We propose a novel linear programming (LP) based primal-dual algorithm for convex MDPs that incorporates uncertainty'' parameters to improve data efficiency while requiring only partial data coverage assumption. Our theoretical results achieve a sample complexity of \mathcal{O}(1/(1-\gamma)\sqrt{n}) under general function approximation, improving the current state-of-the-art by a factor of 1/(1-\gamma), where n is the number of data samples in an offline dataset, and \gamma is the discount factor. The numerical experiments validate our theoretical findings, demonstrating the practical efficacy of our approach in achieving improved safety and learning efficiency in safe offline settings.

The Lion optimizer has been a promising competitor with the AdamW for training large AI models, with advantages in memory, computation, and sample efficiency. In this paper, we introduce Distributed Lion, an innovative adaptation of Lion for distributed training environments. Leveraging the sign operator in Lion, our Distributed Lion only requires to communicate binary or lower-precision vectorsbetween workers to the center server, significantly reducing the communication cost. Our theoretical analysis confirms Distributed Lion's convergence properties. Empirical results demonstrate its robustness across a range of tasks, worker counts, and batch sizes, on both vision and language problems.

Sample efficiency is critical for online Reinforcement Learning from Human Feedback (RLHF). While existing works investigate sample-efficient online exploration strategies, the potential of utilizing misspecified yet relevant reward models to accelerate learning remains underexplored. This paper studies how to transfer knowledge from those imperfect reward models in online RLHF. We start by identifying a novel property of the KL-regularized RLHF objective: \emph{a policy's ability to cover the optimal policy is captured by its sub-optimality}. Building on this insight, we propose a theoretical transfer learning algorithm with provable benefits compared to standard online learning. Our approach achieves low regret in the early stage by quickly adapting to the best available source reward models without prior knowledge of their quality, and over time, it attains an $\tilde{O}(\sqrt{T})$ regret bound \emph{independent} of structural complexity measures. Inspired by our theoretical findings, we develop an empirical algorithm with improved computational efficiency, and demonstrate its effectiveness empirically in summarization tasks.

This paper presents new estimators for Off Policy Evaluation (OPE) based on likelihoods and argues that the new estimators are better than Importance Sampling (IS). The paper provides strong theoretical guarantees of the estimators, and demonstrates their through simple experiments. The reviewers agree that the paper is well written overall and the proposed methods are technically sound and likely to be built upon by the community. One reviewer is unsure if the proposed methods will be practical in RL applications. The experiments are performed on very simple tasks.

AI can take on the task of lead scoring, providing you with a prioritised list of leads based on previous deals. Your sales teams can address the prospects who are most likely to convert into high-value sales first. A CRM-integrated AI can also tell you which leads are most likely to close. Sales Cloud, alongside Einstein Opportunity Insights, will automatically assess customer sentiment, competitor involvement, and prospect engagement to understand whether a prospect is likely to buy. For your existing clients, Einstein Account Insights can keep tabs on company updates and activities, and alert your sales teams when an opportunity presents itself.

Professor James Elder, who is a co-author of a study published by York University, says that deep convolutional neural networks (DCNNs) do not perceive objects as humans do, with configural shape perception, which could be risky in real-time AI applications. The study was reported in the iScience -- a Cell Press journal. "Deep Learning Models Are Unsuccessful in Capturing the Configural Manner of Human Shape Perception" is a joint study by Elder, a York Research Chair in Human and Computer Vision and a Co-Director of York's Centre for AI & Society, and Nicholas Baker, an Assistant Psychology Professor at Loyola College in Chicago, a former VISTA postdoctoral fellow at York. To discover how the human brain and DCNNs process complete, configural object properties, the scientists used novel visual stimuli known as "Frankensteins." Frankensteins are simply objects that have been taken apart and put back together the wrong way around.

Identifying the right talent remains challenging for HR professionals and recruiters. Companies rely on artificial intelligence (AI) technology to perform tasks like screening resumes and scheduling interviews. But are AI recruitment tools ethical and efficient? There are benefits and pitfalls of automated hiring practices. AI-driven technology and applicant tracking systems (ATS) have become indispensable tools for companies worldwide.

As a B2B marketing professional, you already understand how important customer relationships are, and how much time and effort goes into crafting a mutually beneficial relationship between your brand and a potential customer. But how do you go about building and nurturing these customer relationships efficiently and effectively? With the power of artificial intelligence integrated with your marketing automation platform, you can make smart decisions about your marketing more quickly than ever and connect with customers at scale, all while cutting down on wasted time and missed opportunities. Read on to learn how AI can make your B2B marketing more efficient so you can build and nurture customer relationships faster than ever before. High-value leads present great opportunities for your business. But it often takes several interactions to match a lead with high-value qualifiers and identify them as a VIP (very important prospect).