1167 research outputs found
Sort by
Chaos_Engineering_Requirements_ExperimentPlan
No full textThe dataset comprises one *.xlsx file that links historically observed incidents with resilience mechanisms and includes the derived experiment plan
Processed data for Lifetime of the first 4⁺ state of ¹³²Te
No full textThis dataset includes the figures of the manuscript "Lifetime of the first 4⁺ state of ¹³²Te". Additionally, the data displayed in the figures and the Python script to create the figures from the data are published here.
The raw data from the experiment cannot be published due to the size of around 2 Tb but can be made available on reasonable request.1.
Aletheia: What Makes RLVR For Code Verifiers Tick?
No full textMulti-domain thinking verifiers trained via Reinforcement Learning from Verifiable Rewards (RLVR) are a prominent fixture of the Large Language Model (LLM) post-training pipeline, owing to their ability to robustly rate and rerank model outputs. However, the adoption of such verifiers towards code generation has been comparatively sparse, with execution feedback constituting the dominant signal. Nonetheless, code verifiers remain valuable toward judging model outputs in scenarios where execution feedback is hard to obtain and are a potentially powerful addition to the code generation post-training toolbox. To this end, we create and open-source Aletheia, a controlled testbed that enables execution-grounded evaluation of code verifiers' robustness across disparate policy models and covariate shifts. We examine components of the RLVR-based verifier training recipe widely credited for its success: (1) intermediate thinking traces, (2) learning from negative samples, and (3) on-policy training. While experiments show the optimality of RLVR, we uncover important opportunities to simplify the recipe. Particularly, despite code verification being amenable to training- and inference-time scaling, on-policy learning stands out as the key component at smaller verifier sizes, and thinking-based training emerges as the most important component at larger scales
CORE-T: COherent REtrieval of Tables for Text-to-SQL
No full textWe present three preprocessed text-to-SQL benchmarks (BIRD, SPIDER and MMQA). We preprocessed these datasets to follow our open-book setting by merging tables from
multiple DBs (or question-specific schemas for MMQA) into a single retrieval corpus per benchmark. We provide the preprocessed data as well as their corresponding SQL databases.1.
Input files for simulations with SolidLBM
No full textThis data set contains the input files for the simulations conducted as part of the doctoral thesis.
The files are:
Python scripts for running simulations (lattice Boltzmann, finite elements) to produce raw data and the input files defining the specific simulation; and Python scripts for the post-processing of raw data and figure preparation.
This needs the SolidLBM- or SolidLBMeval-module, repsectively.
Input files are text-files with different file extensions (.geo, .par, .bc, .toml) that describe parameters of the simulation. Additionally, .msh and .bc_alt files are generated by the 'mesher'-module of SolidLBM and are needed as input for the 'solver'. The FE simulations require a ,toml file as input for the parameters.
The raw data is produced either as .vtu files (LB) or .xdmf files with HDF5 (FE). Examples of the produced raw output data is given for the TRT convergence study in trt-convergence-study.tar.
Otherwise, the scripts and input files are sufficient to run simulations and produce the output data.
The scripts for post-processing of data produce plots as visulisations of the simulation results. Intermediate results are saved as .npz (Numpy archives) for faster reproduction of plots without recalculation. Examples of this are also included in trt-convergence-study.tar.
The TRT parameter study (trt-param-study.tar) further builds upon the evaluation of data compiled in a spreadsheet (.ods).This data set contains the input files for the simulations conducted as part of the doctoral thesis
Supplementary information to eddy-resolving Reynolds-stress modeling of single and particle-laden turbulent flow in concentric annulus configurations with wall rotation
No full textSupplementary data associated with the dissertation "eddy-resolving Reynolds-stress modeling of single and particle-laden turbulent flow in concentric annulus configurations with wall rotation", including figure datasets. The data is organized according to the chapters of the dissertation and is provided in a compressed .zip archive that must be extracted prior to use
Author-in-the-Loop Response Generation and Evaluation: Integrating Author Expertise and Intent in Responses to Peer Review
No full textRe3Align, a new large-scale dataset for author-in-the-loop response generation, comprising 3.4k complete paper records (review, response, paper and revised paper) with 440k sentence-level edit annotations and 15k aligned review–response–edit triplets. This dataset is a supplement to the paper: Author-in-the-Loop Response Generation and Evaluation: Integrating Author Expertise and Intent in Responses to Peer Review. Please refer to the paper for more details (URL TBD)
SciCoQA
No full textWe present SciCoQA, a dataset for detecting discrepancies between scientific publications and their codebases to ensure faithful implementations. We construct SciCoQA from GitHub issues and reproducibility papers, and to scale our dataset, we propose a synthetic data generation method for constructing paper-code discrepancies. We analyze the paper-code discrepancies in detail and propose discrepancy types and categories to better understand the occurring mismatches. In total, our dataset consists of 611 paper-code discrepancies (81 real, 530 synthetic), spanning diverse computational science disciplines, including AI, Physics, Quantitative Biology, and others.v1.
On the relation between self diffusion and structural relaxation in the presence of strong mobility gradients across confinements: water in nanopores as a case study
No full textDatasets of the figures shown in the article with the same title as this submission. Original manuscript submitted to The Journal of Chemical Physics in December 2025. Revised version submitted to The Journal of Chemical Physics in January 2026
Transkripte Experteninterviews zu Robotern in der Pflege
Full text linkIm Sommer 2024 fanden vier Interviews leitfadengestützt, online statt, die Experten aus dem Bereich Forschung für Roboter in der Pflege befragt haben. Die Namen sind anonymisiert mit AR für Applied Research, R für Research, U für User. Zuerst wird der Leitfaden für die Interviews gezeigt und dann die vier Transkripte der Interviews.V