General Assistant Tasks on GAIA L3
28.7AccuracyExComm
Evaluation Results
| Method | Links | |
|---|---|---|
| ExCommBackbone=Qwen2.5-7B, N > 1?=true, Available Tools=C / F / S2026.05 | 28.7 | |
| Independent Scaling + SRBackbone=Qwen2.5-7B, N > 1?=true, Available Tools=C / F / S2026.05 | 26.5 | |
| Sequential Revision (SR)Backbone=Qwen2.5-7B, N > 1?=false, Available Tools=C / F / S2026.05 | 21 | |
| Independent ScalingBackbone=Qwen2.5-7B, N > 1?=true, Available Tools=C / F / S2026.05 | 20.1 | |
| Tree SearchBackbone=Qwen2.5-7B, N > 1?=true, Available Tools=C / F / S2026.05 | 16.7 | |
| Base AgentBackbone=Qwen2.5-7B, N > 1?=false, Available Tools=C / F / S2026.05 | 16.4 | |
| ExCommBackbone=Gemini-2.5-Flash-Lite, N > 1?=true, Available Tools=C / F / S2026.05 | 14.8 | |
| Tree SearchBackbone=Gemini-2.5-Flash-Lite, N > 1?=true, Available Tools=C / F / S2026.05 | 10.2 | |
| Independent ScalingBackbone=Gemini-2.5-Flash-Lite, N > 1?=true, Available Tools=C / F / S2026.05 | 9 | |
| Base AgentBackbone=Gemini-2.5-Flash-Lite, N > 1?=false, Available Tools=C / F / S2026.05 | 5.9 | |
| Independent Scaling + SRBackbone=Gemini-2.5-Flash-Lite, N > 1?=true, Available Tools=C / F / S2026.05 | 5.6 | |
| Sequential Revision (SR)Backbone=Gemini-2.5-Flash-Lite, N > 1?=false, Available Tools=C / F / S2026.05 | 4.9 |