Sign in to view source links and access this dataset
Description
Samarth0710 compiled full peer-review data for 5,286 papers accepted to the NeurIPS 2025 conference, scraped from OpenReview. Each row contains the paper PDF, reviewer scores, written feedback, author rebuttals, and the final acceptance decision. The dataset was uploaded to Hugging Face on April 15, 2026.
Use Cases
Benchmarking LLM-generated reviews against human-written feedback based on the provided reviewer comments.
Analyzing correlations between reviewer scores and final acceptance decisions.
Training models to predict paper acceptance based on review scores and rebuttals.
Studying the structure and content of author rebuttals in response to reviewer critiques.
Strengths
Contains data for 5,286 accepted papers, providing a substantial sample.
Includes full paper PDFs, quantitative scores, and qualitative written feedback per review.
Captures the complete review cycle with author rebuttals and final decisions.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count for the specific dataset file is unknown, which may limit suitability assessment.
Data may reflect selection bias inherent to the NeurIPS 2025 conference submission pool.
Provenance
Source
OpenReview, the platform hosting NeurIPS 2025 peer reviews.
Collection Method
Scraped from OpenReview.
Time Range
NeurIPS 2025 conference cycle.
Freshness
Last updated 2026-04-15 09:52:20; freshness should be verified.
License is unknown; terms of use must be verified before download and application.