Sequential Instructions

This is the sequential instructions dataset from Understanding the Effects of RLHF on LLM Generalisation and Diversity. The dataset is in the alpaca_eval format. For information about how the dataset was generated, see https://github.com/RobertKirk/stanford_alpaca. The instructions in the dataset generally have a sequence of steps we expect the model to complete all at once. In our work, we found that RLHF models generalise much better to this dataset than… See the full description on the dataset page: https://huggingface.co/datasets/UCL-DARK/sequential-instructions.

consisting of open-source video frames accompanied by story-like annotations.

This work discusses drinking water sampling efforts for lead in Flint, MI. This dataset is associated with the following publication: Lytle, D., M. Schock, K. Wait, K. Cahalan, V. Bosscher, A. Porter, and M. Deltoral. SEQUENTIAL DRINKING WATER SAMPLING AS A TOOL FOR EVALUATING LEAD IN FLINT, MICHIGAN. WATER RESEARCH. Elsevier Science Ltd, New York, NY, USA, 157: 40-54, (2019).

This data set comprises uncalibrated and calibrated data from the Cassini Composite Infrared Spectrometer (CIRS) instrument. The basic data is comprised of uncalibrated raw spectra, along with along with pointing and geometry information, and housekeeping information. Also included are calibrated power spectra, and documentation.

Toys

Sequential designs and competing risks methodology are both well established. Their combined use has recently received some attention from a theoretical perspective, but their joint application in practice has been discussed less. The aim of this paper is to provide the applied statistician with a basic understanding of both sequential design theory and competing risks methodology and how to combine them in practice. Relevant references to more detailed theoretical discussions are provided and all discussions are illustrated using a real case study. Extensive R and SAS code is provided in the online supplementary material.

Sequential Datasets.

Visual storytelling refers to the manner of describing a set of images rather than a single image, also known as multi-image captioning. Visual Storytelling Task (VST) takes a set of images as input and aims to generate a coherent story relevant to the input images. In this dataset, we bridge the gap and present a new dataset for expressive and coherent story creation. We present the Sequential Storytelling Image Dataset (SSID), consisting of open-source video frames accompanied by story-like annotations. In addition, we provide four annotations (i.e., stories) for each set of five images. The image sets are collected manually from publicly available videos in three domains: documentaries, lifestyle, and movies, and then annotated manually using Amazon Mechanical Turk. In summary, SSID dataset is comprised of 17,365 images, which resulted in a total of 3,473 unique sets of five images. Each set of images is associated with four ground truths, resulting in a total of 13,892 unique ground truths (i.e., written stories). And each ground truth is composed of five connected sentences written in the form of a story.

CompAct (Compositional Activities) presents a comprehensive benchmark for assessing the compositional generalization abilities of Sequential Multimodal Models. CompAct is a carefully constructed, perceptually grounded dataset set within a rich backdrop of egocentric kitchen activity videos. Each instance in our dataset is represented with a combination of raw video footage, naturally occurring sound, and crowd-sourced step-by-step descriptions. More importantly, our setup ensures that the individual concepts are consistently distributed across training and evaluation sets, while their compositions are novel in the evaluation set. We conduct a comprehensive assessment of several unimodal and multimodal models.

We develop a sequential Monte Carlo (SMC) algorithm for estimating Bayesian dynamic stochastic general equilibrium (DSGE) models; wherein a particle approximation to the posterior is built iteratively through tempering the likelihood. Using two empirical illustrations consisting of the Smets and Wouters model and a larger news shock model we show that the SMC algorithm is better suited for multimodal and irregular posterior distributions than the widely used random walk Metropolis-Hastings algorithm. We find that a more diffuse prior for the Smets and Wouters model improves its marginal data density and that a slight modification of the prior for the news shock model leads to drastic changes in the posterior inference about the importance of news shocks for fluctuations in hours worked. Unlike standard Markov chain Monte Carlo (MCMC) techniques; the SMC algorithm is well suited for parallel computing.

This dataset contains data and code to replicate main and supplemental figures for the related article published in Matter:

Title: A fluidic relaxation oscillator for reprogrammable sequential actuation in soft robots

DOI: 10.1016/j.matt.2022.06.002

In the article we introduce a simple and compact soft valve with intentional hysteresis, analogous to an electronic relaxation oscillator. By integrating the valve with a soft actuator, we transform a continuous inflow to cyclic activation. Importantly, we show that our circuits can activate up to five actuators in various sequences, and that we can physically reprogram the activation order by varying the (initial) conditions in the fluidic circuit. Moreover, we show the feasibility of our approach under more realistic conditions by building a four-legged robot.

This dataset contains measurement data and simulation files.

The data are recorded (in human-readable format) from experiments on our fluidic circuits (e.g., pressure, flow data), and are accompanied by MATLAB scripts for data processing as well as generating figures.

The simulation files are MATLAB and LTspice files for simulating our fluidic circuits making use of the analogy with electronic circuits. For more involved parameter sweeps we generate, run, and post-process LTspice input and result files using MATLAB. More details and instruction for use are provided in the included readme.txt files.

During the study period

This paper develops estimators for dynamic microeconomic models with serially correlated unobserved state variables using sequential Monte Carlo methods to estimate the parameters and the distribution of the unobservables. If persistent unobservables are ignored, the estimates can be subject to a dynamic form of sample selection bias. We focus on single-agent dynamic discrete-choice models and dynamic games of incomplete information. We propose a full-solution maximum likelihood procedure and a two-step method and use them to estimate an extended version of the capital replacement model of Rust with the original data and in a Monte Carlo study.

This dataset is about book series. It has 1 row and is filtered where the books is Sequential methods in statistics. It features 10 columns including number of authors, number of books, earliest publication date, and latest publication date.

deu05232/promptriever-ours-v8-sequential dataset hosted on Hugging Face and contributed by the HF Datasets community

This is a repository for the datasets handled by the algorithm, MSDT (Multi-valued and Sequential-labeled Decision Tree). All the datasets grouped into five database archive files used in the corresponding experiments can be downloaded from here.

Power for standardized effect sizes of (top) and (bottom) for each combination of randomization procedure and group sequential design for the z-test.

Global Home Sequential Compression Devices market size is expected to reach $1.58 billion by 2029 at 7.6%, segmented as by standard, non-portable models, tabletop models, clinical-grade devices

This dataset contains images and driving protocols used in the paper: "Counting and Sequential Information Processing in Mechanical Metamaterials", published in Physical Review Letters.

In this paper we demonstrate "beam counters"; metamaterials that count driving cycles. We demonstrate the counters sensitivity to various driving amplitudes and show how this might be used to infer more information from the applied driving and how to construct a "lock and key" metamaterial with an internal state that can only be reached with one unique input sequence.

The data in this replication package is hierarchically organized by making use of a directory per figure in the paper. Contained in this replication package are a number of files used in the figure as described below.

Experimental data from the measurement setup is stored in matched subdirectorys with a name, for example "001/" and a file "times_001.csv". Here the subdirectory contains images taken of the sample and the csv file contains the times at which the images were taken, the change in pixel intensity from one image to the next (sampled at a shorter interval than the saved images) and the position of the driving stage and a measured inductive position.

fig1

001/ - directory containing the original unedited figures comparing above and below D*

001_cropped_selection/ - directory containing a cropped selection of the images used in the figure

times_001.csv

fig2

002/ - directory containing images of the ten counter being compressed with marked m-beams

times_002.csv

Kymograph.tif - A kymograph image calculated from 002 by filtering the colored m-beams and taking stacking a single horizontal slice from all of the images.

kymograph.svg - The plotted horizontal position of the beam traces visible in Kymograph.tif

fig3

000/ - directory containing images of the compression of counter with uncut a-beams

000_cropped/

times_000.csv

004/ - directory containing images of the same counter with the a-beams slit cut

004_cropped/

times_004.csv

comparison/ - directory containing a selection of cropped images at comparable driving used in the figure

fig4

Alternative starting condition/

001/ - directory containing images of the ten-counter compressed in an alternative starting state

001_cropped/ - directory with cropped versions of the images in 001/ and further cropped versions in further subdirectories

001_selection/

times_001.csv

BBAC machine

006/ - directory containing images of the four counters making up the BBAC machine under a driving of BAC

006_state/ - selection of images used in the paper with a subdirectory contining cropped versions

times_006.csv

We introduce QBLink, a new dataset of about 18,000 question sequences, each sequence consists of three naturally occurring human-authored questions (totaling around 56,000 unique questions). The sequences themselves are also naturally occurring (i.e., we do not artificially combine individually-authored questions to form sequences), which allows us to focus more on the important connections between questions that should be incorporated to improve the end-to-end question answering accuracy. QBLink is based on the bonus questions of Quiz Bowl tournaments. Unlike previous work that only uses the starter (or tossup) questions, bonus questions are not interruptable (players always hear the complete question) and have greater variability in difficulty. Bonus questions start with a lead-in text, which sets the stage for the rest of the question, followed by a sequence of related questions.