Agentic Workflow Performance Prediction on FLORA-Bench Math-GD 1.0 (test)
64.4AccuracyGLOW
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GLOWbase_LLM=Qwen3-1.7B, GNN=two-layer GAT, SBERT=all-MiniLM-L6-v22025.12 | 64.4 | 63.5 | |
| Agentic Predictor2025.12 | 62.9 | 61.8 | |
| Qwen3base_model=Qwen3-1.7B, fine-tuned=true2025.12 | 62 | 61.4 | |
| Graph Transformer2025.12 | 61.3 | 60.9 | |
| GCNII2025.12 | 61 | 59.1 | |
| One-For-All2025.12 | 60 | 59.9 | |
| GCN2025.12 | 59.8 | 60.1 | |
| GAT2025.12 | 59.4 | 58.7 |