Automatically assessing classroom discussion quality is becoming increasingly feasible with the help of new NLP advancements such as large language models (LLMs). In this work, we examine how the assessment performance of 2 LLMs interacts with 3 factors that may affect performance: task formulation, context length, and few-shot examples. We also explore the computational efficiency and predictive consistency of the 2 LLMs. Our results suggest that the 3 aforementioned factors do affect the performance of the tested LLMs and there is a relation between consistency and performance. We recommend a LLM-based assessment approach that has a good balance in terms of predictive performance, computational efficiency, and consistency.

Can humans and artificial intelligences share concepts and communicate? 'Making AI Intelligible' shows that philosophical work on the metaphysics of meaning can help answer these questions. Herman Cappelen and Josh Dever use the externalist tradition in philosophy to create models of how AIs and humans can understand each other. In doing so, they illustrate ways in which that philosophical tradition can be improved. The questions addressed in the book are not only theoretically interesting, but the answers have pressing practical implications. Many important decisions about human life are now influenced by AI. In giving that power to AI, we presuppose that AIs can track features of the world that we care about (for example, creditworthiness, recidivism, cancer, and combatants). If AIs can share our concepts, that will go some way towards justifying this reliance on AI. This ground-breaking study offers insight into how to take some first steps towards achieving Interpretable AI.

Large Language Models (LLMs) have transformed NLP with their remarkable In-context Learning (ICL) capabilities. Automated assistants based on LLMs are gaining popularity; however, adapting them to novel tasks is still challenging. While colossal models excel in zero-shot performance, their computational demands limit widespread use, and smaller language models struggle without context. This paper investigates whether LLMs can generalize from labeled examples of predefined tasks to novel tasks. Drawing inspiration from biological neurons and the mechanistic interpretation of the Transformer architecture, we explore the potential for information sharing across tasks. We design a cross-task prompting setup with three LLMs and show that LLMs achieve significant performance improvements despite no examples from the target task in the context. Cross-task prompting leads to a remarkable performance boost of 107% for LLaMA-2 7B, 18.6% for LLaMA-2 13B, and 3.2% for GPT 3.5 on average over zero-shot prompting, and performs comparable to standard in-context learning. The effectiveness of generating pseudo-labels for in-task examples is demonstrated, and our analyses reveal a strong correlation between the effect of cross-task examples and model activation similarities in source and target input tokens. This paper offers a first-of-its-kind exploration of LLMs' ability to solve novel tasks based on contextual signals from different task examples.

Offline reinforcement learning (RL) allows learning sequential behavior from fixed datasets. Since offline datasets do not cover all possible situations, many methods collect additional data during online fine-tuning to improve performance. In general, these methods assume that the transition dynamics remain the same during both the offline and online phases of training. However, in many real-world applications, such as outdoor construction and navigation over rough terrain, it is common for the transition dynamics to vary between the offline and online phases. Moreover, the dynamics may vary during the online fine-tuning. To address this problem of changing dynamics from offline to online RL we propose a residual learning approach that infers dynamics changes to correct the outputs of the offline solution. At the online fine-tuning phase, we train a context encoder to learn a representation that is consistent inside the current online learning environment while being able to predict dynamic transitions. Experiments in D4RL MuJoCo environments, modified to support dynamics' changes upon environment resets, show that our approach can adapt to these dynamic changes and generalize to unseen perturbations in a sample-efficient way, whilst comparison methods cannot.

To combat the misuse of Large Language Models (LLMs), many recent studies have presented LLM-generated-text detectors with promising performance. When users instruct LLMs to generate texts, the instruction can include different constraints depending on the user's need. However, most recent studies do not cover such diverse instruction patterns when creating datasets for LLM detection. In this paper, we reveal that even task-oriented constraints -- constraints that would naturally be included in an instruction and are not related to detection-evasion -- cause existing powerful detectors to have a large variance in detection performance. We focus on student essay writing as a realistic domain and manually create task-oriented constraints based on several factors for essay quality. Our experiments show that the standard deviation (SD) of current detector performance on texts generated by an instruction with such a constraint is significantly larger (up to an SD of 14.4 F1-score) than that by generating texts multiple times or paraphrasing the instruction. We also observe an overall trend where the constraints can make LLM detection more challenging than without them. Finally, our analysis indicates that the high instruction-following ability of LLMs fosters the large impact of such constraints on detection performance.

High-quality preference datasets are essential for training reward models that can effectively guide large language models (LLMs) in generating high-quality responses aligned with human preferences. As LLMs become stronger and better aligned, permissively licensed preference datasets, such as Open Assistant, HH-RLHF, and HelpSteer need to be updated to remain effective for reward modeling. Methods that distil preference data from proprietary LLMs such as GPT-4 have restrictions on commercial usage imposed by model providers. To improve upon both generated responses and attribute labeling quality, we release HelpSteer2, a permissively licensed preference dataset (CC-BY-4.0). Using a powerful internal base model trained on HelpSteer2, we are able to achieve the SOTA score (92.0%) on Reward-Bench's primary dataset, outperforming currently listed open and proprietary models, as of June 12th, 2024. Notably, HelpSteer2 consists of only ten thousand response pairs, an order of magnitude fewer than existing preference datasets (e.g., HH-RLHF), which makes it highly efficient for training reward models. Our extensive experiments demonstrate that reward models trained with HelpSteer2 are effective in aligning LLMs. In particular, we propose SteerLM 2.0, a model alignment approach that can effectively make use of the rich multi-attribute score predicted by our reward models. HelpSteer2 is available at https://huggingface.co/datasets/nvidia/HelpSteer2 and code is available at https://github.com/NVIDIA/NeMo-Aligner

When using large language models (LLMs) in high-stakes applications, we need to know when we can trust their predictions. Some works argue that prompting high-performance LLMs is sufficient to produce calibrated uncertainties, while others introduce sampling methods that can be prohibitively expensive. In this work, we first argue that prompting on its own is insufficient to achieve good calibration and then show that fine-tuning on a small dataset of correct and incorrect answers can create an uncertainty estimate with good generalization and small computational overhead. We show that a thousand graded examples are sufficient to outperform baseline methods and that training through the features of a model is necessary for good performance and tractable for large open-source models when using LoRA. We also investigate the mechanisms that enable reliable LLM uncertainty estimation, finding that many models can be used as general-purpose uncertainty estimators, applicable not just to their own uncertainties but also the uncertainty of other models. Lastly, we show that uncertainty estimates inform human use of LLMs in human-AI collaborative settings through a user study.

Transformer encoder with connectionist temporal classification (CTC) framework is widely used for automatic speech recognition (ASR). However, knowledge distillation (KD) for ASR displays a problem of disagreement between teacher-student models in frame-level alignment which ultimately hinders it from improving the student model's performance. In order to resolve this problem, this paper introduces a self-knowledge distillation (SKD) method that guides the frame-level alignment during the training time. In contrast to the conventional method using separate teacher and student models, this study introduces a simple and effective method sharing encoder layers and applying the sub-model as the student model. Overall, our approach is effective in improving both the resource efficiency as well as performance. We also conducted an experimental analysis of the spike timings to illustrate that the proposed method improves performance by reducing the alignment disagreement.

Self-supervised learning of image representations by predicting future frames is a promising direction but still remains a challenge. This is because of the under-determined nature of frame prediction; multiple potential futures can arise from a single current frame. To tackle this challenge, in this paper, we revisit the idea of stochastic video generation that learns to capture uncertainty in frame prediction and explore its effectiveness for representation learning. Specifically, we design a framework that trains a stochastic frame prediction model to learn temporal information between frames. Moreover, to learn dense information within each frame, we introduce an auxiliary masked image modeling objective along with a shared decoder architecture. We find this architecture allows for combining both objectives in a synergistic and compute-efficient manner. We demonstrate the effectiveness of our framework on a variety of tasks from video label propagation and vision-based robot learning domains, such as video segmentation, pose tracking, vision-based robotic locomotion, and manipulation tasks. Code is available on the project webpage: https://sites.google.com/view/2024rsp.

Question generation for education assessments is a growing field within artificial intelligence applied to education. These question-generation tools have significant importance in the educational technology domain, such as intelligent tutoring systems and dialogue-based platforms. The automatic generation of assessment questions, which entail clear-cut answers, usually relies on syntactical and semantic indications within declarative sentences, which are then transformed into questions. Recent research has explored the generation of assessment educational questions in Arabic. The reported performance has been adversely affected by inherent errors, including sentence parsing inaccuracies, name entity recognition issues, and errors stemming from rule-based question transformation. Furthermore, the complexity of lengthy Arabic sentences has contributed to these challenges. This research presents an innovative Arabic question-generation system built upon a three-stage process: keywords and key phrases extraction, question generation, and subsequent ranking. The aim is to tackle the difficulties associated with automatically generating assessment questions in the Arabic language. The proposed approach and results show a precision of 83.50%, a recall of 78.68%, and an Fl score of 80.95%, indicating the framework high efficiency. Human evaluation further confirmed the model efficiency, receiving an average rating of 84%.