Dungeons & Dragons (D&D) is a tabletop roleplaying game with complex natural language interactions between players and hidden state information. Recent work has shown that large language models (LLMs) that have access to state information can generate higher quality game turns than LLMs that use dialog history alone. However, previous work used game state information that was heuristically created and was not a true gold standard game state. We present FIREBALL, a large dataset containing nearly 25,000 unique sessions from real D&D gameplay on Discord with true game state info. We recorded game play sessions of players who used the Avrae bot, which was developed to aid people in playing D&D online, capturing language, game commands and underlying game state information. We demonstrate that FIREBALL can improve natural language generation (NLG) by using Avrae state information, improving both automated metrics and human judgments of quality. Additionally, we show that LLMs can generate executable Avrae commands, particularly after finetuning.

Some of these meteorites fall in regions on Earth where fireball observatory networks are active, making it possible to record the trajectory of the fireball as it ablates material from the originating meteoroid. For some fireballs, this data can then be used to simulate both forward and backward in time to predict where the resulting meteorite landed on Earth and where the meteoroid originated in the solar system. Thus, recovering and analyzing these'orbital meteorites' with constrained, prior orbits provides an incredibly unique insight into the geology of the asteroid belt and the nature of mass transfer between the belt and the inner solar system. The Desert Fireball Network (DFN) (Bland et al. 2012; Howie et al. 2017) is one of many organizations (Oberst et al. 1998; Spurný et al. 2006; Trigo-Rodríguez et al. 2006; Olech et al. 2006; Colas et al. 2015; Devillepoix et al. 2020) that makes this possible.

A BSTRACT Evolutionary-based optimization approaches have recently shown promising results in domains such as Atari and robot locomotion but less so in solving 3D tasks directly from pixels. This paper presents a method called Deep Innovation Protection (DIP) that allows training complex world models end-to-end for such 3D environments. The main idea behind the approach is to employ multiobjective optimization to temporally reduce the selection pressure on specific components in a world model, allowing other components to adapt. We investigate the emergent representations of these evolved networks, which learn a model of the world without the need for a specific forward-prediction loss. 1 I NTRODUCTION The ability of the brain to model the world arose from the process of evolution. It evolved because it helped organisms to survive and strive in their particular environments and not because such forward prediction was explicitly optimized for. In contrast to the emergent neural representations in nature, current world model approaches are often directly rewarded for their ability to predict future states of the environment (Schmidhuber, 1990; Ha & Schmidhuber, 2018; Hafner et al., 2018; Wayne et al., 2018). While it is undoubtedly useful to be able to explicitly encourage a model to predict what will happen next, in this paper we are interested in what type of representations can emerge from the less directed process of artificial evolution and what ingredients might be necessary to encourage the emergence of such predictive abilities. In particular, we are building on the recently introduced world model architecture introduced by Ha & Schmidhuber (2018). This agent model contains three different components: (1) a visual module, mapping high-dimensional inputs to a lower-dimensional representative code, (2) an LSTM-based memory component, and (3) a controller component that takes input from the visual and memory module to determine the agent's next action.

Gaining a better understanding of how and what machine learning systems learn is important to increase confidence in their decisions and catalyze further research. In this paper, we analyze the predictions made by a specific type of recurrent neural network, mixture density RNNs (MD-RNNs). These networks learn to model predictions as a combination of multiple Gaussian distributions, making them particularly interesting for problems where a sequence of inputs may lead to several distinct future possibilities. An example is learning internal models of an environment, where different events may or may not occur, but where the average over different events is not meaningful. By analyzing the predictions made by trained MD-RNNs, we find that their different Gaussian components have two complementary roles: 1) Separately modeling different stochastic events and 2) Separately modeling scenarios governed by different rules. These findings increase our understanding of what is learned by predictive MD-RNNs, and open up new research directions for further understanding how we can benefit from their self-organizing model decomposition.

Can you launch fireworks from your drone? OK, before I answer this question I have my own question: Why? Guys, why would you want to put fireworks on your drone? Fireworks are cool and drones are cool. Therefore fireworks on drones are cool to the power of two, I guess. As a PSA, let's say that my official stance is that drones should just be drones and fireworks should just be fireworks.

The world is living under the threat of nuclear war and a terrifying simulation reveals what would happen if a nuclear bomb went off in a major city. As well as looking at the destruction, scientists used the computer model to work out how people would behave if the worst-case scenario struck. An entire city block was obliterated instantly and buildings blasted for a mile in almost every direction. Researchers found people who did nothing were most likely to die with nearly 280,000 people killed in just 48 hours. In the dystopian-like version of The Sims, researchers simulated a nuke exploding in Washington DC (pictured).

Wouldn't it be nice if you could simply take your videos, photos and music and tell an application to turn those into a nice video presentation? With Windows Story Remix, that's what Microsoft is trying to attempt. This new application, which will launch with the Windows 10 Fall Creators Update later this year, combines all of the new features in this upcoming version of Windows 10 into a simple to use, machine learning-based application that automatically creates professional-looking videos for you. A product like this always looks great in on-stage demos, of course, but if the final result is anything like what the company showed today, then it's definitely worth a closer look. After you give the application the videos and photos you want it to work with -- and maybe a soundtrack, too -- it sets off to build the video.