domains / families / research / iclr-reproducibility

iclr-reproducibility

separate

Use when strengthening reproducibility for ICLR papers, including seeds, variance, compute, datasets, implementation details, ethics statements, and reviewer-verifiable evidence. Use when writing the ICLR reproducibility statement, when a reviewer says a result is not verifiable, or when mapping each representation-learning claim to a seed, split, and command so anyone reading the permanent OpenReview record can check it.

When to use

• strengthening reproducibility for ICLR papers
• writing the ICLR reproducibility statement
• addressing reviewer claims that results are not verifiable
• mapping representation-learning claims to seeds, splits, and commands
• auditing reproducibility for experiments, simulations, data processing, human subjects, or benchmark comparisons
• recording seeds, variance, confidence intervals, test splits, preprocessing, early stopping, hyperparameter search, model selection, and compute budget
• distinguishing training compute from inference compute and reporting hardware details
• adding negative results and failure cases to explain boundary conditions

When not to use

• tasks not involving ICLR papers
• tasks not involving reproducibility, seeds, variance, compute, datasets, implementation details, ethics statements, or reviewer-verifiable evidence
• tasks not involving writing an ICLR reproducibility statement
• tasks not involving addressing reviewer claims that results are not verifiable
• tasks not involving mapping representation-learning claims to seeds, splits, and commands

Skill page

Source