ICLR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ICLR 2026 (test)
0.55Actionability Score
38
ICLR 2025 (test)
48.08Kendall Tau Correlation
32
ICLR submissions 2025
89.8Accuracy
17
gold-labeled ICLR (test)
86Accuracy
16
ICLR 2025 (500 papers)
0.172Recall @ 9
16
ICLR 2025 (test)
71.92Accuracy
15
ICLR-5k
2.26Actionability Score
12
ICLR 2024
4.22Absolute Novelty
12
ICLR
96.2Success Rate (SR)
12
ICLR 2025
6.4Average Review Score
10
ICLR LLM-as-a-Judge 2025 (test)
3.38Actionability Score
9
ICLR Human Evaluation 2025 (test)
3.46Actionability
9
ICLR gold-labeled (test)
74.24Jaccard Similarity
8
ICLR 2025
3.35Originality
8
ICLR 20.2:1
1.82Cohen's d
6
ICLR Poster 2025
82.4Guided Paper Percentage
6
ICLR Spotlight 2025
82.1Percentage Guided
6
ICLR Oral 2025
82.2Guidance Rate
6
ICLR Overall 2025
82.4Percentage Guided
6
ICLR
96.7Success Rate (SR)
5
ICLR 100-paper corpus 2026
3,024Caught
4
ICLR
54.4AUC
4
ICLR problems 2026 N=20 (test)
56Feasibility Win%
4
ICLR 2025
45.73Avg Cites
3
ICLR 50 submissions 2026
88.6Str-Cov
3