Navigation on FrozenLake 5x5
0.87Cumulative RewardASK (32B)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ASK (32B)Model Size=32B, trained_on=8x82026.04 | 0.87 | 10.12 | 100 | 47.55 | |
| ASK (72B)Model Size=72B, trained_on=8x82026.04 | 0.86 | 16.18 | 100 | 50.51 | |
| ASK (1.5B)Model Size=1.5B, trained_on=8x82026.04 | 0.23 | 11.53 | 100 | 28.47 | |
| PPOtrained_on=8x82026.04 | 0.01 | 42.41 | — | — | |
| ASK (14B)Model Size=14B, trained_on=8x82026.04 | 0.01 | 58.11 | 100 | 47.27 | |
| ASK (0.5B)Model Size=0.5B, trained_on=8x82026.04 | 0 | 31.61 | 100 | 99.87 | |
| ASK (3B)Model Size=3B, trained_on=8x82026.04 | 0 | 50.92 | 100 | 41.07 | |
| ASK (7B)Model Size=7B, trained_on=8x82026.04 | 0 | 61 | 100 | 51.03 |