The interest in linear complexity models for large language models is on the rise, although their scaling capacity remains uncertain. In this study, we present the scaling laws for linear complexity language models to establish a foundation for their scalability. Specifically, we examine the scaling behaviors of three efficient linear architectures. These include TNL, a linear attention model with data-independent decay; HGRN2, a linear RNN with data-dependent decay; and cosFormer2, a linear attention model without decay. We also include LLaMA as a baseline architecture for softmax attention for comparison. These models were trained with six variants, ranging from 70M to 7B parameters on a 300B-token corpus, and evaluated with a total of 1,376 intermediate checkpoints on various downstream tasks. These tasks include validation loss, commonsense reasoning, and information retrieval and generation. The study reveals that existing linear complexity language models exhibit similar scaling capabilities as conventional transformer-based models while also demonstrating superior linguistic proficiency and knowledge retention.

Deep reinforcement learning has the potential to address various scientific problems. In this paper, we implement an optics simulation environment for reinforcement learning based controllers. The environment captures the essence of nonconvexity, nonlinearity, and time-dependent noise inherent in optical systems, offering a more realistic setting. Subsequently, we provide the benchmark results of several reinforcement learning algorithms on the proposed simulation environment. The experimental findings demonstrate the superiority of off-policy reinforcement learning approaches over traditional control algorithms in navigating the intricacies of complex optical control environments. The code of the paper is available at https://github.com/Walleclipse/Reinforcement-Learning-Pulse-Stacking.

The ability to recognize analogies is fundamental to human cognition. Existing benchmarks to test word analogy do not reveal the underneath process of analogical reasoning of neural models. Holding the belief that models capable of reasoning should be right for the right reasons, we propose a first-of-its-kind Explainable Knowledge-intensive Analogical Reasoning benchmark (E-KAR). Our benchmark consists of 1,655 (in Chinese) and 1,251 (in English) problems sourced from the Civil Service Exams, which require intensive background knowledge to solve. More importantly, we design a free-text explanation scheme to explain whether an analogy should be drawn, and manually annotate them for each and every question and candidate answer. Empirical results suggest that this benchmark is very challenging for some state-of-the-art models for both explanation generation and analogical question answering tasks, which invites further research in this area.

Elden Ring is the front-runner for 2022's game of the year. Reviewers are fawning over it. It's the title the entire gaming community is talking about and that everyone wants to play. The hype sounds like The Legend of Zelda: Breath of the Wild all over again, and that one ended up being so beloved it became one of the best-selling video games of all time. Elden Ring, however, will never achieve that status--the gameplay is just too grueling to appeal to every player.

Some developers have opted to put control over accessibility or difficulty into the hands of players. Mega blockbuster "The Last of Us Part II," for example, provides granular but flexible difficulty options like reducing accuracy of enemy fire and removing the ability for foe's to flank you. A disabled gamer may not be able to complete a complicated combination of button mashing in a fighting game, or execute the carefully timed parries needed to defeat a boss in an action game, without the ability to tweak the difficulty to match their needs. Malleable settings allow players to tailor their experience, along with a suite of accessibility options for the hearing impaired, better visibility for players with impaired vision, and more.

I think a lot of new users hear'Demon's Souls' and they go'Oh that's game's really difficult.' It's challenging, but very fair,

Due to the realization that deep reinforcement learning algorithms trained on high-dimensional tasks can strongly overfit to their training environments, there have been several studies that investigated the generalization performance of these algorithms. However, there has been no similar study that evaluated the generalization performance of algorithms that were specifically designed for generalization, i.e. meta-reinforcement learning algorithms. In this paper, we assess the generalization performance of these algorithms by leveraging high-dimensional, procedurally generated environments. We find that these algorithms can display strong overfitting when they are evaluated on challenging tasks. We also observe that scalability to high-dimensional tasks with sparse rewards remains a significant problem among many of the current meta-reinforcement learning algorithms. With these results, we highlight the need for developing meta-reinforcement learning algorithms that can both generalize and scale.

For some time now, my colleague Dave Thier and I have argued back and forth about the merits of an'easy mode' for notoriously challenging games like Dark Souls. So the debate continues with Nioh, Team Ninja's excellent new Samurai action-RPG which draws a lot of inspiration from such infamously challenging games as Dark Souls and Ninja Gaiden. Dave's argument essentially boils down to this: Nioh's difficulty will turn off more casual players or players who don't have the kind of time to sink into such a challenging game. This is both foolish (as it limits sales) and snobby (as it leaves the game accessible only to the hardcore audience) and would be easily mitigated by the inclusion of an easy mode, making it more accessible to all while leaving the core experience unscathed. Every game we play sets out to do something different.