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 2026-09-22
#ModelRewardCostIn tokOut tokTimeStepsSubmits
1GPT-6 Astra0.2143$155.5978.1M233k19h 27m2996
2GPT-6 Astra0.1905$229.70113.1M288k19h 53m3646
3GPT-6 Astra0.1905$125.1064.1M212k19h 30m2616
4GPT-6 Astra0.0714$291.48141.3M320k19h 38m4026
5GPT-6 Astra0.0000$142.9574.8M253k19h 46m3117