Acceptance Prediction on ICLR 2024/2025 and NeurIPS 2023/2024 (test)
70.45AccuracyTF-1l-8h (AC+abs)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TF-1l-8h (AC+abs)Model Architecture=three-way transformer architecture, Configuration=1 layer, 8 heads, Input Features=Author-Capability + abstract2026.01 | 70.45 | 70.6 | 50.52 | 58.87 | |
| bertInput Features=title+abs2026.01 | 68.33 | 67.69 | 48.54 | 56.25 | |
| bert-largeInput Features=title+abs2026.01 | 67.95 | 67.23 | 51.57 | 56.49 | |
| SA-1Model Component=SA-12026.01 | 65.29 | 64.26 | 39.18 | 48.4 | |
| SA-2Model Component=SA-22026.01 | 65.21 | 63.4 | 40.63 | 49.25 | |
| TF-1l-8hModel Architecture=three-way transformer architecture, Configuration=1 layer, 8 heads2026.01 | 64.74 | 60.19 | 46.67 | 52.57 | |
| R1Model Component=R12026.01 | 64.64 | 62.42 | 40.72 | 48.83 | |
| TF-1l-1hModel Architecture=three-way transformer architecture, Configuration=1 layer, 1 head2026.01 | 64.64 | 60.34 | 45.89 | 52 | |
| bert-largeInput Features=ACI (Author-Capability-Idea)2026.01 | 64.55 | 62.89 | 40.2 | 48.46 | |
| Gated Ave.2026.01 | 64.48 | 61.6 | 42.85 | 49.99 | |
| bertInput Features=ACI (Author-Capability-Idea)2026.01 | 64.32 | 61.72 | 41.96 | 49.5 | |
| Gemini-3-preview2026.01 | 58.13 | 59.88 | 67.86 | 63.62 | |
| ChatGLM-4.72026.01 | 57.54 | 60.02 | 61.1 | 60.56 | |
| Claude-sonnet-4.52026.01 | 52.67 | 56.02 | 56.14 | 56.08 | |
| GPT-4o2026.01 | 52.03 | 54.59 | 64.45 | 59.11 | |
| Deepseek-V3.22026.01 | 51.52 | 55.81 | 64.86 | 60 | |
| Random Guess2026.01 | 49.91 | 55.34 | 49.65 | 52.34 |