Search-augmented Multi-turn Tool Calling on Bamboogle (Success Rate)
65.32Success RateENVRL-GiGPO
Evaluation results
| Method | Links | |
|---|---|---|
| ENVRL-GiGPOBackbone=Qwen2.5-3B-Instruct2026.06 | 65.32 | |
| GiGPOBackbone=Qwen2.5-3B-Instruct2026.06 | 64.52 |
| Method | Links | |
|---|---|---|
| ENVRL-GiGPOBackbone=Qwen2.5-3B-Instruct2026.06 | 65.32 | |
| GiGPOBackbone=Qwen2.5-3B-Instruct2026.06 | 64.52 |