Navigation on FrozenLake 6x6
0.75RewardASK (72B)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ASK (72B)Model Size=72B, trained_on=8x82026.04 | 0.75 | 21.58 | 100 | 50.33 | |
| ASK (32B)Model Size=32B, trained_on=8x82026.04 | 0.69 | 22.92 | 100 | 52.35 | |
| ASK (1.5B)Model Size=1.5B, trained_on=8x82026.04 | 0.11 | 8.6 | 100 | 28.96 | |
| ASK (14B)Model Size=14B, trained_on=8x82026.04 | 0.01 | 60.31 | 100 | 51.5 | |
| PPOtrained_on=8x82026.04 | 0 | 36.8 | — | — | |
| ASK (0.5B)Model Size=0.5B, trained_on=8x82026.04 | 0 | 31.61 | 100 | 100 | |
| ASK (3B)Model Size=3B, trained_on=8x82026.04 | 0 | 50.38 | 100 | 41.43 | |
| ASK (7B)Model Size=7B, trained_on=8x82026.04 | 0 | 61 | 100 | 50.31 |