Agentic Workflow Performance Prediction on FLORA-Bench Coding-AF 1.0 (test)
0.846AccuracyGLOW
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GLOWbase_LLM=Qwen3-1.7B, GNN=two-layer GAT, SBERT=all-MiniLM-L6-v22025.12 | 0.846 | 0.754 | |
| Agentic Predictor2025.12 | 0.832 | 0.739 | |
| GAT2025.12 | 0.829 | 0.721 | |
| GCN2025.12 | 0.827 | 0.713 | |
| Graph Transformer2025.12 | 0.827 | 0.723 | |
| GCNII2025.12 | 0.822 | 0.716 | |
| One-For-All2025.12 | 0.822 | 0.728 | |
| Qwen3base_model=Qwen3-1.7B, fine-tuned=true2025.12 | 0.814 | 0.724 |