Despite numerous successes, the field of reinforcement learning (RL) remains far from matching the impressive generalisation power of human behaviour learning. One possible way to help bridge this gap be to provide RL agents with richer, more human-like feedback expressed in natural language. To investigate this idea, we first extend BabyAI to automatically generate language feedback from the environment dynamics and goal condition success. Then, we modify the Decision Transformer architecture to take advantage of this additional signal. We find that training with language feedback either in place of or in addition to the return-to-go or goal descriptions improves agents' generalisation performance, and that agents can benefit from feedback even when this is only available during training, but not at inference.

Artificial Intelligence (AI) is expected to play an instrumental role in the next generation of wireless systems, such as sixth-generation (6G) mobile network. However, massive data, energy consumption, training complexity, and sensitive data protection in wireless systems are all crucial challenges that must be addressed for training AI models and gathering intelligence and knowledge from distributed devices. Federated Learning (FL) is a recent framework that has emerged as a promising approach for multiple learning agents to build an accurate and robust machine learning models without sharing raw data. By allowing mobile handsets and devices to collaboratively learn a global model without explicit sharing of training data, FL exhibits high privacy and efficient spectrum utilization. While there are a lot of survey papers exploring FL paradigms and usability in 6G privacy, none of them has clearly addressed how FL can be used to improve the protocol stack and wireless operations. The main goal of this survey is to provide a comprehensive overview on FL usability to enhance mobile services and enable smart ecosystems to support novel use-cases. This paper examines the added-value of implementing FL throughout all levels of the protocol stack. Furthermore, it presents important FL applications, addresses hot topics, provides valuable insights and explicits guidance for future research and developments. Our concluding remarks aim to leverage the synergy between FL and future 6G, while highlighting FL's potential to revolutionize wireless industry and sustain the development of cutting-edge mobile services.

In this work, we introduce a self-supervised behavior cloning transformer for text games, which are challenging benchmarks for multi-step reasoning in virtual environments. Traditionally, Behavior Cloning Transformers excel in such tasks but rely on supervised training data. Our approach auto-generates training data by exploring trajectories (defined by common macro-action sequences) that lead to reward within the games, while determining the generality and utility of these trajectories by rapidly training small models then evaluating their performance on unseen development games. Through empirical analysis, we show our method consistently uncovers generalizable training data, achieving about 90\% performance of supervised systems across three benchmark text games.

In this paper, we investigate a problem of actively learning threshold in latent space, where the unknown reward $g(\gamma, v)$ depends on the proposed threshold $\gamma$ and latent value $v$ and it can be $only$ achieved if the threshold is lower than or equal to the unknown latent value. This problem has broad applications in practical scenarios, e.g., reserve price optimization in online auctions, online task assignments in crowdsourcing, setting recruiting bars in hiring, etc. We first characterize the query complexity of learning a threshold with the expected reward at most $\epsilon$ smaller than the optimum and prove that the number of queries needed can be infinitely large even when $g(\gamma, v)$ is monotone with respect to both $\gamma$ and $v$. On the positive side, we provide a tight query complexity $\tilde{\Theta}(1/\epsilon^3)$ when $g$ is monotone and the CDF of value distribution is Lipschitz. Moreover, we show a tight $\tilde{\Theta}(1/\epsilon^3)$ query complexity can be achieved as long as $g$ satisfies one-sided Lipschitzness, which provides a complete characterization for this problem. Finally, we extend this model to an online learning setting and demonstrate a tight $\Theta(T^{2/3})$ regret bound using continuous-arm bandit techniques and the aforementioned query complexity results.

We present PyThaiNLP, a free and open-source natural language processing (NLP) library for Thai language implemented in Python. It provides a wide range of software, models, and datasets for Thai language. We first provide a brief historical context of tools for Thai language prior to the development of PyThaiNLP. We then outline the functionalities it provided as well as datasets and pre-trained language models. We later summarize its development milestones and discuss our experience during its development. We conclude by demonstrating how industrial and research communities utilize PyThaiNLP in their work. The library is freely available at https://github.com/pythainlp/pythainlp.

With the recent developments in large language models (LLMs) and their widespread availability through open source models and/or low-cost APIs, several exciting products and applications are emerging, many of which are in the field of STEM educational technology for K-12 and university students. There is a need to evaluate these powerful language models on several benchmarks, in order to understand their risks and limitations. In this short paper, we summarize and analyze the performance of Bard, a popular LLM-based conversational service made available by Google, on the standardized Physics GRE examination.

We introduce the new task of generating Illustrated Instructions, i.e., visual instructions customized to a user's needs. We identify desiderata unique to this task, and formalize it through a suite of automatic and human evaluation metrics, designed to measure the validity, consistency, and efficacy of the generations. We combine the power of large language models (LLMs) together with strong text-to-image generation diffusion models to propose a simple approach called StackedDiffusion, which generates such illustrated instructions given text as input. The resulting model strongly outperforms baseline approaches and state-of-the-art multimodal LLMs; and in 30% of cases, users even prefer it to human-generated articles. Most notably, it enables various new and exciting applications far beyond what static articles on the web can provide, such as personalized instructions complete with intermediate steps and pictures in response to a user's individual situation.

Model merging aims to cheaply combine individual task-specific models into a single multitask model. In this work, we view past merging methods as leveraging different notions of a ''task subspace'' in which models are matched before being merged. We connect the task subspace of a given model to its loss landscape and formalize how this approach to model merging can be seen as solving a linear system of equations. While past work has generally been limited to linear systems that have a closed-form solution, we consider using the conjugate gradient method to find a solution. We show that using the conjugate gradient method can outperform closed-form solutions, enables merging via linear systems that are otherwise intractable to solve, and flexibly allows choosing from a wide variety of initializations and estimates for the ''task subspace''. We ultimately demonstrate that our merging framework called ''Matching Models in their Task Subspace'' (MaTS) achieves state-of-the-art results in multitask and intermediate-task model merging. We release all of the code and checkpoints used in our work at https://github.com/r-three/mats.

Artificial intelligence (AI) refers to the ability of machines or software to mimic or even surpass human intelligence in a given cognitive task. While humans learn by both induction and deduction, the success of current AI is rooted in induction, relying on its ability to detect statistical regularities in task input -- an ability learnt from a vast amount of training data using enormous computation resources. We examine the performance of such a statistical AI in a human task through the lens of four factors, including task learnability, statistical resource, computation resource, and learning techniques, and then propose a three-phase visual framework to understand the evolving relation between AI and jobs. Based on this conceptual framework, we develop a simple economic model of competition to show the existence of an inflection point for each occupation. Before AI performance crosses the inflection point, human workers always benefit from an improvement in AI performance, but after the inflection point, human workers become worse off whenever such an improvement occurs. To offer empirical evidence, we first argue that AI performance has passed the inflection point for the occupation of translation but not for the occupation of web development. We then study how the launch of ChatGPT, which led to significant improvement of AI performance on many tasks, has affected workers in these two occupations on a large online labor platform. Consistent with the inflection point conjecture, we find that translators are negatively affected by the shock both in terms of the number of accepted jobs and the earnings from those jobs, while web developers are positively affected by the very same shock. Given the potentially large disruption of AI on employment, more studies on more occupations using data from different platforms are urgently needed.

In recent years, knowledge distillation methods based on contrastive learning have achieved promising results on image classification and object detection tasks. However, in this line of research, we note that less attention is paid to semantic segmentation. Existing methods heavily rely on data augmentation and memory buffer, which entail high computational resource demands when applying them to handle semantic segmentation that requires to preserve high-resolution feature maps for making dense pixel-wise predictions. In order to address this problem, we present Augmentation-free Dense Contrastive Knowledge Distillation (Af-DCD), a new contrastive distillation learning paradigm to train compact and accurate deep neural networks for semantic segmentation applications. Af-DCD leverages a masked feature mimicking strategy, and formulates a novel contrastive learning loss via taking advantage of tactful feature partitions across both channel and spatial dimensions, allowing to effectively transfer dense and structured local knowledge learnt by the teacher model to a target student model while maintaining training efficiency. Extensive experiments on five mainstream benchmarks with various teacher-student network pairs demonstrate the effectiveness of our approach. For instance, the DeepLabV3-Res18|DeepLabV3-MBV2 model trained by Af-DCD reaches 77.03%|76.38% mIOU on Cityscapes dataset when choosing DeepLabV3-Res101 as the teacher, setting new performance records. Besides that, Af-DCD achieves an absolute mIOU improvement of 3.26%|3.04%|2.75%|2.30%|1.42% compared with individually trained counterpart on Cityscapes|Pascal VOC|Camvid|ADE20K|COCO-Stuff-164K. Code is available at https://github.com/OSVAI/Af-DCD