List your company

Navigation on FrozenLake 7x7

0.68RewardASK (72B)
-0.02720.15640.340.5236Apr 2, 2026

Evaluation results

MethodLinks
ASK (72B)Model Size=72B, trained_on=8x82026.040.6827.110050.88
ASK (32B)Model Size=32B, trained_on=8x82026.040.5830.210053.36
ASK (1.5B)Model Size=1.5B, trained_on=8x82026.040.18.0810029.14
PPOtrained_on=8x82026.04030.22——
ASK (0.5B)Model Size=0.5B, trained_on=8x82026.04025.2810099.95
ASK (3B)Model Size=3B, trained_on=8x82026.04053.5110044.23
ASK (7B)Model Size=7B, trained_on=8x82026.0406110051.57
ASK (14B)Model Size=14B, trained_on=8x82026.0406110050.51