Agentic Workflow Performance Prediction on FLORA-Bench Coding-GD 1.0 (test)
85.1AccuracyGLOW
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GLOWbase_LLM=Qwen3-1.7B, GNN=two-layer GAT, SBERT=all-MiniLM-L6-v22025.12 | 85.1 | 77.3 | |
| Qwen3base_model=Qwen3-1.7B, fine-tuned=true2025.12 | 84.2 | 76.1 | |
| Agentic Predictor2025.12 | 83.4 | 75.9 | |
| GAT2025.12 | 83.3 | 75.1 | |
| Graph Transformer2025.12 | 83.2 | 75.2 | |
| GCNII2025.12 | 82.4 | 75.4 | |
| One-For-All2025.12 | 82.3 | 74.1 | |
| GCN2025.12 | 82.1 | 74.3 |