AI alignment in the shape of Reinforcement Learning from Human Feedback (RLHF) is increasingly treated as a crucial ingredient for high performance large language models. Proximal Policy Optimization (PPO) has been positioned by recent literature as the canonical method for the RL part of RLHF. However, it involves both high computational cost and sensitive hyperparameter tuning. We posit that most of the motivational principles that led to the development of PPO are less of a practical concern in RLHF and advocate for a less computationally expensive method that preserves and even increases performance. We revisit the formulation of alignment from human preferences in the context of RL. Keeping simplicity as a guiding principle, we show that many components of PPO are unnecessary in an RLHF context and that far simpler REINFORCE-style optimization variants outperform both PPO and newly proposed "RL-free" methods such as DPO and RAFT. Our work suggests that careful adaptation to LLMs alignment characteristics enables benefiting from online RL optimization at low cost.

Online task scheduling serves an integral role for task-intensive applications in cloud computing and crowdsourcing. Optimal scheduling can enhance system performance, typically measured by the reward-to-cost ratio, under some task arrival distribution. On one hand, both reward and cost are dependent on task context (e.g., evaluation metric) and remain black-box in practice. These render reward and cost hard to model thus unknown before decision making. On the other hand, task arrival behaviors remain sensitive to factors like unpredictable system fluctuation whereby a prior estimation or the conventional assumption of arrival distribution (e.g., Poisson) may fail. This implies another practical yet often neglected challenge, i.e., uncertain task arrival distribution. Towards effective scheduling under a stationary environment with various uncertainties, we propose a double-optimistic learning based Robbins-Monro (DOL-RM) algorithm. Specifically, DOL-RM integrates a learning module that incorporates optimistic estimation for reward-to-cost ratio and a decision module that utilizes the Robbins-Monro method to implicitly learn task arrival distribution while making scheduling decisions. Theoretically, DOL-RM achieves convergence gap and no regret learning with a sub-linear regret of $O(T^{3/4})$, which is the first result for online task scheduling under uncertain task arrival distribution and unknown reward and cost. Our numerical results in a synthetic experiment and a real-world application demonstrate the effectiveness of DOL-RM in achieving the best cumulative reward-to-cost ratio compared with other state-of-the-art baselines.

Understanding citizens' values in participatory systems is crucial for citizen-centric policy-making. We envision a hybrid participatory system where participants make choices and provide motivations for those choices, and AI agents estimate their value preferences by interacting with them. We focus on situations where a conflict is detected between participants' choices and motivations, and propose methods for estimating value preferences while addressing detected inconsistencies by interacting with the participants. We operationalize the philosophical stance that "valuing is deliberatively consequential." That is, if a participant's choice is based on a deliberation of value preferences, the value preferences can be observed in the motivation the participant provides for the choice. Thus, we propose and compare value estimation methods that prioritize the values estimated from motivations over the values estimated from choices alone. Then, we introduce a disambiguation strategy that addresses the detected inconsistencies between choices and motivations by directly interacting with the participants. We evaluate the proposed methods on a dataset of a large-scale survey on energy transition. The results show that explicitly addressing inconsistencies between choices and motivations improves the estimation of an individual's value preferences. The disambiguation strategy does not show substantial improvements when compared to similar baselines--however, we discuss how the novelty of the approach can open new research avenues and propose improvements to address the current limitations.

Deep Learning (DL) models have rapidly advanced, focusing on achieving high performance through testing model accuracy and robustness. However, it is unclear whether DL projects, as software systems, are tested thoroughly or functionally correct when there is a need to treat and test them like other software systems. Therefore, we empirically study the unit tests in open-source DL projects, analyzing 9,129 projects from GitHub. We find that: 1) unit tested DL projects have positive correlation with the open-source project metrics and have a higher acceptance rate of pull requests, 2) 68% of the sampled DL projects are not unit tested at all, 3) the layer and utilities (utils) of DL models have the most unit tests. Based on these findings and previous research outcomes, we built a mapping taxonomy between unit tests and faults in DL projects. We discuss the implications of our findings for developers and researchers and highlight the need for unit testing in open-source DL projects to ensure their reliability and stability. The study contributes to this community by raising awareness of the importance of unit testing in DL projects and encouraging further research in this area.

Both text and video data are abundant on the internet and support large-scale self-supervised learning through next token or frame prediction. However, they have not been equally leveraged: language models have had significant real-world impact, whereas video generation has remained largely limited to media entertainment. Yet video data captures important information about the physical world that is difficult to express in language. To address this gap, we discuss an under-appreciated opportunity to extend video generation to solve tasks in the real world. We observe how, akin to language, video can serve as a unified interface that can absorb internet knowledge and represent diverse tasks. Moreover, we demonstrate how, like language models, video generation can serve as planners, agents, compute engines, and environment simulators through techniques such as in-context learning, planning and reinforcement learning. We identify major impact opportunities in domains such as robotics, self-driving, and science, supported by recent work that demonstrates how such advanced capabilities in video generation are plausibly within reach. Lastly, we identify key challenges in video generation that mitigate progress. Addressing these challenges will enable video generation models to demonstrate unique value alongside language models in a wider array of AI applications.

Effective information retrieval (IR) in settings with limited training data, particularly for complex queries, remains a challenging task. This paper introduces IR2, Information Regularization for Information Retrieval, a technique for reducing overfitting during synthetic data generation. This approach, representing a novel application of regularization techniques in synthetic data creation for IR, is tested on three recent IR tasks characterized by complex queries: DORIS-MAE, ArguAna, and WhatsThatBook. Experimental results indicate that our regularization techniques not only outperform previous synthetic query generation methods on the tasks considered but also reduce cost by up to 50%. Furthermore, this paper categorizes and explores three regularization methods at different stages of the query synthesis pipeline-input, prompt, and output-each offering varying degrees of performance improvement compared to models where no regularization is applied. This provides a systematic approach for optimizing synthetic data generation in data-limited, complex-query IR scenarios. All code, prompts and synthetic data are available at https://github.com/Info-Regularization/Information-Regularization.

Generative search engines have the potential to transform how people seek information online, but generated responses from existing large language models (LLMs)-backed generative search engines may not always be accurate. Nonetheless, retrieval-augmented generation exacerbates safety concerns, since adversaries may successfully evade the entire system by subtly manipulating the most vulnerable part of a claim. To this end, we propose evaluating the robustness of generative search engines in the realistic and high-risk setting, where adversaries have only black-box system access and seek to deceive the model into returning incorrect responses. Through a comprehensive human evaluation of various generative search engines, such as Bing Chat, PerplexityAI, and YouChat across diverse queries, we demonstrate the effectiveness of adversarial factual questions in inducing incorrect responses. Moreover, retrieval-augmented generation exhibits a higher susceptibility to factual errors compared to LLMs without retrieval. These findings highlight the potential security risks of these systems and emphasize the need for rigorous evaluation before deployment.

There are many settings where it is useful to predict and explain the success or failure of a dialogue. Circumplex theory from psychology models the social orientations (e.g., Warm-Agreeable, Arrogant-Calculating) of conversation participants and can be used to predict and explain the outcome of social interactions. Our work is novel in its systematic application of social orientation tags to modeling conversation outcomes. In this paper, we introduce a new data set of dialogue utterances machine-labeled with social orientation tags. We show that social orientation tags improve task performance, especially in low-resource settings, on both English and Chinese language benchmarks. We also demonstrate how social orientation tags help explain the outcomes of social interactions when used in neural models. Based on these results showing the utility of social orientation tags for dialogue outcome prediction tasks, we release our data sets, code, and models that are fine-tuned to predict social orientation tags on dialogue utterances.

Live performances of music are always charming, with the unpredictability of improvisation due to the dynamic between musicians and interactions with the audience. Jazz improvisation is a particularly noteworthy example for further investigation from a theoretical perspective. Here, we introduce a novel mathematical game theory model for jazz improvisation, providing a framework for studying music theory and improvisational methodologies. We use computational modeling, mainly reinforcement learning, to explore diverse stochastic improvisational strategies and their paired performance on improvisation. We find that the most effective strategy pair is a strategy that reacts to the most recent payoff (Stepwise Changes) with a reinforcement learning strategy limited to notes in the given chord (Chord-Following Reinforcement Learning). Conversely, a strategy that reacts to the partner's last note and attempts to harmonize with it (Harmony Prediction) strategy pair yields the lowest non-control payoff and highest standard deviation, indicating that picking notes based on immediate reactions to the partner player can yield inconsistent outcomes. On average, the Chord-Following Reinforcement Learning strategy demonstrates the highest mean payoff, while Harmony Prediction exhibits the lowest. Our work lays the foundation for promising applications beyond jazz: including the use of artificial intelligence (AI) models to extract data from audio clips to refine musical reward systems, and training machine learning (ML) models on existing jazz solos to further refine strategies within the game.

Hallucinations pose a significant challenge to the reliability and alignment of Large Language Models (LLMs), limiting their widespread acceptance beyond chatbot applications. Despite ongoing efforts, hallucinations remain a prevalent challenge in LLMs. The detection of hallucinations itself is also a formidable task, frequently requiring manual labeling or constrained evaluations. This paper introduces an automated scalable framework that combines benchmarking LLMs' hallucination tendencies with efficient hallucination detection. We leverage LLMs to generate challenging tasks related to hypothetical phenomena, subsequently employing them as agents for efficient hallucination detection. The framework is domain-agnostic, allowing the use of any language model for benchmark creation or evaluation in any domain. We introduce the publicly available HypoTermQA Benchmarking Dataset, on which state-of-the-art models' performance ranged between 3% and 11%, and evaluator agents demonstrated a 6% error rate in hallucination prediction. The proposed framework provides opportunities to test and improve LLMs. Additionally, it has the potential to generate benchmarking datasets tailored to specific domains, such as law, health, and finance.