Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

The ensmallen library for flexible numerical optimization

Authors: Ryan R. Curtin, Marcus Edel, Rahul Ganesh Prabhu, Suryoday Basak, Zhihao Lou, Conrad Sanderson

JMLR 2021 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical comparisons show that ensmallen outperforms other frameworks while providing more functionality. The paper presents performance tables (Table 1 and Table 2) comparing runtimes for optimizing linear regression parameters and training logistic regression models on various datasets.
Researcher Affiliation Collaboration The author affiliations include a mix of industry (Relational AI, Epsilon) and academic/research institutions (Free University of Berlin, Birla Institute of Technology and Science Pilani, University of Texas at Arlington, Data61/CSIRO, and Griffith University).
Pseudocode Yes Figure 1 provides an "Example implementation of an objective function class for linear regression and usage of the L-BFGS optimizer", which includes structured C++ code demonstrating the methodology.
Open Source Code Yes The library is available at https://ensmallen.org and is distributed under the permissive BSD license. The source code and documentation are freely available at https://ensmallen.org.
Open Datasets Yes The paper uses "various real datasets from the UCI dataset repository (Lichman, 2013)", which is a well-known public dataset repository with proper attribution.
Dataset Splits No The paper mentions using "Highly noisy random data" and datasets from the "UCI dataset repository," but it does not specify any training/test/validation splits, percentages, or methodology for data partitioning needed to reproduce the data partitioning.
Hardware Specification Yes The experiments were performed on an AMD Ryzen 7 2700X with 64GB RAM, with g++ 10.2.0, Julia 1.5.2, Python 3.8.5, and Octave 6.1.0.
Software Dependencies Yes The paper specifies software versions: "g++ 10.2.0, Julia 1.5.2, Python 3.8.5, and Octave 6.1.0.".
Experiment Setup Yes The paper specifies a key experimental setting: "In each framework, the provided L-BFGS optimizer is limited to 10 iterations."