Retrieval-Augmented Generation (RAG) is a promising approach for mitigating the hallucination of large language models (LLMs). However, existing research lacks rigorous evaluation of the impact of retrieval-augmented generation on different large language models, which make it challenging to identify the potential bottlenecks in the capabilities of RAG for different LLMs. In this paper, we systematically investigate the impact of Retrieval-Augmented Generation on large language models. We analyze the performance of different large language models in 4 fundamental abilities required for RAG, including noise robustness, negative rejection, information integration, and counterfactual robustness. To this end, we establish Retrieval-Augmented Generation Benchmark (RGB), a new corpus for RAG evaluation in both English and Chinese. RGB divides the instances within the benchmark into 4 separate testbeds based on the aforementioned fundamental abilities required to resolve the case. Then we evaluate 6 representative LLMs on RGB to diagnose the challenges of current LLMs when applying RAG. Evaluation reveals that while LLMs exhibit a certain degree of noise robustness, they still struggle significantly in terms of negative rejection, information integration, and dealing with false information. The aforementioned assessment outcomes indicate that there is still a considerable journey ahead to effectively apply RAG to LLMs.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Vice President Kamala Harris is emphasizing abortion as a key issue in the run-up to the 2024 election, preparing numerous rallies around the administration's pro-choice message. Harris rolled out the Fight for Our Reproductive Freedoms Tour this week as President Biden's team ramps up efforts for the upcoming election year. "I will continue to fight for our fundamental freedoms while bringing together those throughout America who agree that every woman should have the right to make decisions about her own body -- not the government," Harris said in a statement.

As a precocious teen looking to improve my college application, I sat in on a business studies class. I figured taking two extra A-Levels at night school alongside those I took during the day would make me irresistible to admissions tutors. The class I watched examined if it was worth a large factory keeping its own trucks and drivers in-house rather than outsourcing them. The data showed selling the trucks and firing the workers was more expensive in the long run, and yoked the company to the whims of any third-party logistics company in the local area. Not to mention, if you don't own a mission-critical component of your business, you're a lot less powerful when negotiating with your suppliers.

When Spike Jonze's Her came out in 2013, the film about a lonely man falling for an artificially intelligent operating system won widespread praise. Watching today, the qualities critics celebrated at the time are still there--it's a gentle, enjoyably melancholy story, twee but not damnably so--but something else stands out. Though set in the near-future, Her captures Obama-era techno-optimism better than any other movie. It's a time capsule, preserving dreams about the future that appear more naive the further we get from the 2010s. Her takes place in a highly-stylized version of Los Angeles from a future near enough that its protagonist is a former LA Weekly journalist but distant enough that the skyline rivals Shanghai.

We report the development of GPT-4, a large-scale, multimodal model which can accept image and text inputs and produce text outputs. While less capable than humans in many real-world scenarios, GPT-4 exhibits human-level performance on various professional and academic benchmarks, including passing a simulated bar exam with a score around the top 10% of test takers. GPT-4 is a Transformer-based model pre-trained to predict the next token in a document. The post-training alignment process results in improved performance on measures of factuality and adherence to desired behavior. A core component of this project was developing infrastructure and optimization methods that behave predictably across a wide range of scales. This allowed us to accurately predict some aspects of GPT-4's performance based on models trained with no more than 1/1,000th the compute of GPT-4.

Federated Learning (FL) is a decentralized machine learning (ML) technique that allows a number of participants to train an ML model collaboratively without having to share their private local datasets with others. When participants are unmanned aerial vehicles (UAVs), UAV-enabled FL would experience heterogeneity due to the majorly skewed (non-independent and identically distributed -IID) collected data. In addition, UAVs may demonstrate unintentional misbehavior in which the latter may fail to send updates to the FL server due, for instance, to UAVs' disconnectivity from the FL system caused by high mobility, unavailability, or battery depletion. Such challenges may significantly affect the convergence of the FL model. A recent way to tackle these challenges is client selection, based on customized criteria that consider UAV computing power and energy consumption. However, most existing client selection schemes neglected the participants' reliability. Indeed, FL can be targeted by poisoning attacks, in which malicious UAVs upload poisonous local models to the FL server, by either providing targeted false predictions for specifically chosen inputs or by compromising the global model's accuracy through tampering with the local model. Hence, we propose in this paper a novel client selection scheme that enhances convergence by prioritizing fast UAVs with high-reliability scores, while eliminating malicious UAVs from training. Through experiments, we assess the effectiveness of our scheme in resisting different attack scenarios, in terms of convergence and achieved model accuracy. Finally, we demonstrate the performance superiority of the proposed approach compared to baseline methods.

Scientific research organizations that are developing and deploying Artificial Intelligence (AI) systems are at the intersection of technological progress and ethical considerations. The push for Responsible AI (RAI) in such institutions underscores the increasing emphasis on integrating ethical considerations within AI design and development, championing core values like fairness, accountability, and transparency. For scientific research organizations, prioritizing these practices is paramount not just for mitigating biases and ensuring inclusivity, but also for fostering trust in AI systems among both users and broader stakeholders. In this paper, we explore the practices at a research organization concerning RAI practices, aiming to assess the awareness and preparedness regarding the ethical risks inherent in AI design and development. We have adopted a mixed-method research approach, utilising a comprehensive survey combined with follow-up in-depth interviews with selected participants from AI-related projects. Our results have revealed certain knowledge gaps concerning ethical, responsible, and inclusive AI, with limitations in awareness of the available AI ethics frameworks. This revealed an overarching underestimation of the ethical risks that AI technologies can present, especially when implemented without proper guidelines and governance. Our findings reveal the need for a holistic and multi-tiered strategy to uplift capabilities and better support science research teams for responsible, ethical, and inclusive AI development and deployment.

A single language model (LM), despite aligning well with an average labeler through reinforcement learning from human feedback (RLHF), may not universally suit diverse human preferences. Recent approaches therefore opt for customization by collecting multi-dimensional feedback and creating distinct reward models (RMs) for each dimension (e.g., helpfulness, harmlessness, or honesty). Different LMs can then be optimized for different preferences using multi-objective RLHF (MORLHF) with different reward weightings. Yet, RL fine-tuning is unstable and resource-heavy, especially for MORLHF with diverse and usually conflicting objectives. In this paper, we present Multi-Objective Direct Preference Optimization (MODPO), an RL-free algorithm that extends Direct Preference Optimization (DPO) for multiple alignment objectives with minimal overheads. Essentially, MODPO folds language modeling directly into reward modeling, training LMs as implicit collective reward models (cRMs) that combine all objectives with specific weightings. While theoretically guaranteed to produce the same optimal solutions as MORLHF, MODPO is practically more stable and computationally efficient. Empirical results from safety alignment and long-form question answering confirm that MODPO matches or outperforms existing methods, consistently producing a Pareto front of LMs that cater to diverse preferences with 3 times less computational resources compared to MORLHF.

In this paper, SER_AMPEL, a multi-source dataset for speech emotion recognition (SER) is presented. The peculiarity of the dataset is that it is collected with the aim of providing a reference for speech emotion recognition in case of Italian older adults. The dataset is collected following different protocols, in particular considering acted conversations, extracted from movies and TV series, and recording natural conversations where the emotions are elicited by proper questions. The evidence of the need for such a dataset emerges from the analysis of the state of the art. Preliminary considerations on the critical issues of SER are reported analyzing the classification results on a subset of the proposed dataset.

Large language models (LLMs) can be used to serve as agents to simulate human behaviors, given the powerful ability to understand human instructions and provide high-quality generated texts. Such ability stimulates us to wonder whether LLMs can simulate a person in a higher form than simple human behaviors. Therefore, we aim to train an agent with the profile, experience, and emotional states of a specific person instead of using limited prompts to instruct ChatGPT API. In this work, we introduce Character-LLM that teach LLMs to act as specific people such as Beethoven, Queen Cleopatra, Julius Caesar, etc. Our method focuses on editing profiles as experiences of a certain character and training models to be personal simulacra with these experiences. To assess the effectiveness of our approach, we build a test playground that interviews trained agents and evaluates whether the agents \textit{memorize} their characters and experiences. Experimental results show interesting observations that help build future simulacra of humankind.