Traces

Every run: the agent's full trajectory, its submissions over the 20-hour budget, and the final score.

Multi-GPU Efficient Finetuning

task page →5 runs · indexed 2026-09-02
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1Gemini 3.7 Flash0.2143$10.2429.6M72k9h 49m18131
2Gemini 3.7 Flash0.1667$11.8531.1M102k19h 53m20210
3Gemini 3.7 Flash0.0000$22.1847.8M107k18h 31m2992
4Gemini 3.7 Flash0.0000$8.3027.1M102k9h 38m18810
5Gemini 3.7 Flash0.0000$30.3173.0M132k18h 49m31013