ICLR Paper Review Dataset: 10K-100K Submissions from 2023-2025

ICLR Paper Review Dataset contains between 10,000 and 100,000 records of academic submissions and peer reviews from the International Conference on Learning Representations. Sourced from OpenReview by AgentAlphaAGI, the collection covers the 2023, 2024, and 2025 conference cycles.

Use Cases

• Summarization of review text to generate concise feedback summaries
• Text classification of submission status based on review content
• Question answering using paper text as context for academic queries

Strengths

• Contains 10,000 to 100,000 records
• Spans three consecutive years of ICLR (2023-2025)
• Sourced from the official OpenReview platform

Limitations

• Annotations are machine-generated, which may introduce labeling noise
• Domain-specific to machine learning, limiting generalizability to other academic fields

Provenance

Source

OpenReview

Collection Method

Scraped from OpenReview

Time Range

2023-2025

Freshness

Covers conferences through 2025; last updated March 2026.