Dataset Viewer
Auto-converted to Parquet Duplicate
maze_string
stringlengths
1.18k
1.22k
exp_string
stringlengths
131
7k
reached_goal
bool
2 classes
reward
float64
0
1
length
int64
10
501
maze_spec
dict
start_pos
dict
goal_pos
listlengths
6
6
goal_reached
dict
intended_goal
dict
maze_index
int64
0
9k
data_source
stringclasses
1 value
prompt
stringlengths
1.45k
1.49k
completion
stringlengths
154
7.02k
ability
stringclasses
1 value
reward_model
dict
extra_info
dict
start 1 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 1 0 state 2 1 state 3 2 state 4 3 state 5 4 state 6 5 state 7 6 state 8 7 state 9 8 state 8 9 state 9 10 state 9 11 state 10 12 state 11 13 state 12 14 state 13 14 state 12 13 state 13 14 state 14 15 state 14 16 state 15 15
true
1
21
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 1, "y": 0 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 1 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 1 0 state 2 1 state 3 2 state 4 3 state 5 4 state 6 5 state 7 6 state 8 7 state 9 8 state 8 9 state 9 10 state 9 11 state 10 12 state 11 13 state 12 14 state 13 14 state 12 13 state 13 14 state 14 15 state 14 16 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 1 0 \n state 2 1 \n state 3 2 \n state 4 3 \n state 5 4 \n state 6 5 \n state 7 6 \n state 8 7 \n state 9 8 \n state 8 9 \n state 9 10 \n state 9 11 \n state 10 12 \n state 11 13 \n state 12 14 \n state 13 14 \n state 12 13 \n state 13 14 \n state 14 15 \n state 14 16 \n state 15 15 \n\...
{ "handover_type": "actor", "index": 0, "length": 21, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 5 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 2 5 state 3 6 state 3 7 state 3 8 state 3 9 state 4 10 state 5 11 state 6 12 state 6 13 state 7 14 state 8 13 state 9 14 state 10 15 state 10 16 state 11 17 state 12 16 state 13 16 state 14 17 state 15 16 state 15 15
true
1
20
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 5 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 5 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 5 state 3 6 state 3 7 state 3 8 state 3 9 state 4 10 state 5 11 state 6 12 state 6 13 state 7 14 state 8 13 state 9 14 state 10 15 state 10 16 state 11 17 state 12 16 state 13 16 state 14 17 state 15 16 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 2 5 \n state 3 6 \n state 3 7 \n state 3 8 \n state 3 9 \n state 4 10 \n state 5 11 \n state 6 12 \n state 6 13 \n state 7 14 \n state 8 13 \n state 9 14 \n state 10 15 \n state 10 16 \n state 11 17 \n state 12 16 \n state 13 16 \n state 14 17 \n state 15 16 \n state 15 15 \n\n </tra...
{ "handover_type": "ref", "index": 1, "length": 20, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 3 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 3 6 state 3 7 state 4 6 state 5 7 state 6 8 state 7 7 state 8 7 state 9 7 state 10 7 state 11 6 state 12 7 state 13 7 state 14 7 state 15 7 state 16 7 state 17 7 state 18 6 state 19 7 state 20 7 state 21 7 state 22 7
true
1
21
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 3, "y": 6 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 22, "y": 7 }
{ "x": 22, "y": 7 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 3 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 3 6 state 3 7 state 4 6 state 5 7 state 6 8 state 7 7 state 8 7 state 9 7 state 10 7 state 11 6 state 12 7 state 13 7 state 14 7 state 15 7 state 16 7 state 17 7 state 18 6 state 19 7 state 20 7 state 21 7 state 22 7 </trajectory>
maze_navigation
{ "ground_truth": "\n state 3 6 \n state 3 7 \n state 4 6 \n state 5 7 \n state 6 8 \n state 7 7 \n state 8 7 \n state 9 7 \n state 10 7 \n state 11 6 \n state 12 7 \n state 13 7 \n state 14 7 \n state 15 7 \n state 16 7 \n state 17 7 \n state 18 6 \n state 19 7 \n state 20 7 \n state 21 7 \n state 22 7 \n\n </...
{ "handover_type": "actor", "index": 2, "length": 21, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 1 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 1 0 state 2 1 state 3 2 state 4 3 state 5 4 state 4 5 state 5 6 state 5 7 state 6 8 state 7 7 state 8 8 state 9 9 state 10 10 state 11 11 state 10 12 state 11 12 state 12 13 state 13 14 state 14 15 state 15 15
true
1
20
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 1, "y": 0 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 1 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 1 0 state 2 1 state 3 2 state 4 3 state 5 4 state 4 5 state 5 6 state 5 7 state 6 8 state 7 7 state 8 8 state 9 9 state 10 10 state 11 11 state 10 12 state 11 12 state 12 13 state 13 14 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 1 0 \n state 2 1 \n state 3 2 \n state 4 3 \n state 5 4 \n state 4 5 \n state 5 6 \n state 5 7 \n state 6 8 \n state 7 7 \n state 8 8 \n state 9 9 \n state 10 10 \n state 11 11 \n state 10 12 \n state 11 12 \n state 12 13 \n state 13 14 \n state 14 15 \n state 15 15 \n\n </trajectory...
{ "handover_type": "ref", "index": 3, "length": 20, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 2 1 state 3 2 state 4 3 state 5 4 state 6 5 state 6 6 state 7 7 state 8 8 state 9 9 state 10 10 state 11 11 state 12 12 state 12 13 state 13 14 state 14 15 state 15 15
true
1
16
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 1 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 1 state 3 2 state 4 3 state 5 4 state 6 5 state 6 6 state 7 7 state 8 8 state 9 9 state 10 10 state 11 11 state 12 12 state 12 13 state 13 14 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 2 1 \n state 3 2 \n state 4 3 \n state 5 4 \n state 6 5 \n state 6 6 \n state 7 7 \n state 8 8 \n state 9 9 \n state 10 10 \n state 11 11 \n state 12 12 \n state 12 13 \n state 13 14 \n state 14 15 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "ref", "index": 4, "length": 16, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 0 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 0 6 state 1 5 state 2 6 state 2 5 state 3 6 state 3 5 state 4 6 state 3 7 state 3 8 state 3 7 state 3 8 state 3 9 state 4 10 state 5 10 state 6 11 state 7 12 state 8 13 state 9 14 state 10 15 state 11 15 state 12 16 state 13 16 state 14 15 state 15 15
true
1
24
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 0, "y": 6 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 0 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 0 6 state 1 5 state 2 6 state 2 5 state 3 6 state 3 5 state 4 6 state 3 7 state 3 8 state 3 7 state 3 8 state 3 9 state 4 10 state 5 10 state 6 11 state 7 12 state 8 13 state 9 14 state 10 15 state 11 15 state 12 16 state 13 16 state 14 15 state 15 15 </tra...
maze_navigation
{ "ground_truth": "\n state 0 6 \n state 1 5 \n state 2 6 \n state 2 5 \n state 3 6 \n state 3 5 \n state 4 6 \n state 3 7 \n state 3 8 \n state 3 7 \n state 3 8 \n state 3 9 \n state 4 10 \n state 5 10 \n state 6 11 \n state 7 12 \n state 8 13 \n state 9 14 \n state 10 15 \n state 11 15 \n state 12 16 \n state 13...
{ "handover_type": "ref", "index": 5, "length": 24, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 1 4 state 2 5 state 3 6 state 3 7 state 3 8 state 2 8 state 3 8 state 3 7 state 2 8 state 3 9 state 4 10 state 4 11 state 5 12 state 6 13 state 7 14 state 8 15 state 7 15 state 7 16 state 7 17 state 7 16 state 7 17 state 8 18 state 7 19 state 7 20 state 7 21 state...
true
1
26
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 1, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 7, "y": 22 }
{ "x": 7, "y": 22 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 1 4 state 2 5 state 3 6 state 3 7 state 3 8 state 2 8 state 3 8 state 3 7 state 2 8 state 3 9 state 4 10 state 4 11 state 5 12 state 6 13 state 7 14 state 8 15 state 7 15 state 7 16 state 7 17 state 7 16 state 7 17 state 8 18 state 7 19 state 7 20 state 7 21 ...
maze_navigation
{ "ground_truth": "\n state 1 4 \n state 2 5 \n state 3 6 \n state 3 7 \n state 3 8 \n state 2 8 \n state 3 8 \n state 3 7 \n state 2 8 \n state 3 9 \n state 4 10 \n state 4 11 \n state 5 12 \n state 6 13 \n state 7 14 \n state 8 15 \n state 7 15 \n state 7 16 \n state 7 17 \n state 7 16 \n state 7 17 \n state 8 1...
{ "handover_type": "actor", "index": 6, "length": 26, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 3 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 3 4 state 4 5 state 5 6 state 5 7 state 6 8 state 7 9 state 7 10 state 7 11 state 6 11 state 7 12 state 6 12 state 6 13 state 7 14 state 7 15 state 7 16 state 7 17 state 7 18 state 7 19 state 7 20 state 7 21 state 7 22
true
1
21
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 3, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 7, "y": 22 }
{ "x": 7, "y": 22 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 3 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 3 4 state 4 5 state 5 6 state 5 7 state 6 8 state 7 9 state 7 10 state 7 11 state 6 11 state 7 12 state 6 12 state 6 13 state 7 14 state 7 15 state 7 16 state 7 17 state 7 18 state 7 19 state 7 20 state 7 21 state 7 22 </trajectory>
maze_navigation
{ "ground_truth": "\n state 3 4 \n state 4 5 \n state 5 6 \n state 5 7 \n state 6 8 \n state 7 9 \n state 7 10 \n state 7 11 \n state 6 11 \n state 7 12 \n state 6 12 \n state 6 13 \n state 7 14 \n state 7 15 \n state 7 16 \n state 7 17 \n state 7 18 \n state 7 19 \n state 7 20 \n state 7 21 \n state 7 22 \n\n ...
{ "handover_type": "actor", "index": 7, "length": 21, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 2 3 state 3 4 state 4 5 state 5 6 state 5 5 state 6 6 state 7 7 state 8 8 state 9 9 state 10 10 state 9 11 state 10 12 state 11 11 state 12 12 state 12 13 state 13 14 state 14 15 state 15 15
true
1
18
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 3 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 3 state 3 4 state 4 5 state 5 6 state 5 5 state 6 6 state 7 7 state 8 8 state 9 9 state 10 10 state 9 11 state 10 12 state 11 11 state 12 12 state 12 13 state 13 14 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 2 3 \n state 3 4 \n state 4 5 \n state 5 6 \n state 5 5 \n state 6 6 \n state 7 7 \n state 8 8 \n state 9 9 \n state 10 10 \n state 9 11 \n state 10 12 \n state 11 11 \n state 12 12 \n state 12 13 \n state 13 14 \n state 14 15 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "actor", "index": 8, "length": 18, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 24 19 wall 13 13 wall 26 6 wall 4 7 wall 6 7 wall 9 29 wall 14 13 wall 4 19 wall 19 21 wall 13 1 wall 23 9 wall 26 13 wall 18 27 wall 18 10 wall 29 26 wall 19 26 wall 6 29 wall 24 11 wall 20 28 ...
state 1 4 state 2 5 state 3 6 state 3 7 state 2 8 state 3 9 state 4 10 state 5 11 state 6 12 state 6 13 state 7 14 state 8 15 state 9 15 state 10 15 state 11 16 state 11 15 state 12 16 state 13 16 state 14 17 state 15 16 state 15 15
true
1
21
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 1, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
0
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 1 4 state 2 5 state 3 6 state 3 7 state 2 8 state 3 9 state 4 10 state 5 11 state 6 12 state 6 13 state 7 14 state 8 15 state 9 15 state 10 15 state 11 16 state 11 15 state 12 16 state 13 16 state 14 17 state 15 16 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 1 4 \n state 2 5 \n state 3 6 \n state 3 7 \n state 2 8 \n state 3 9 \n state 4 10 \n state 5 11 \n state 6 12 \n state 6 13 \n state 7 14 \n state 8 15 \n state 9 15 \n state 10 15 \n state 11 16 \n state 11 15 \n state 12 16 \n state 13 16 \n state 14 17 \n state 15 16 \n state 15 15 ...
{ "handover_type": "ref", "index": 9, "length": 21, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 1 4 state 2 5 state 3 6 state 4 7 state 5 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 13 state 11 14 state 12 15 state 12 16 state 13 15 state 14 14 state 15 15
true
1
16
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 1, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 1 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 1 4 state 2 5 state 3 6 state 4 7 state 5 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 13 state 11 14 state 12 15 state 12 16 state 13 15 state 14 14 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 1 4 \n state 2 5 \n state 3 6 \n state 4 7 \n state 5 8 \n state 6 9 \n state 7 10 \n state 8 11 \n state 9 12 \n state 10 13 \n state 11 14 \n state 12 15 \n state 12 16 \n state 13 15 \n state 14 14 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "ref", "index": 10, "length": 16, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 5 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 5 0 state 6 1 state 7 2 state 8 3 state 9 4 state 10 5 state 10 6 state 10 5 state 11 5 state 12 6 state 13 7 state 12 8 state 12 9 state 13 8 state 14 9 state 15 9 state 15 10 state 15 11 state 16 12 state 16 13 state 15 14 state 15 15
true
1
22
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 5, "y": 0 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 5 0 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 5 0 state 6 1 state 7 2 state 8 3 state 9 4 state 10 5 state 10 6 state 10 5 state 11 5 state 12 6 state 13 7 state 12 8 state 12 9 state 13 8 state 14 9 state 15 9 state 15 10 state 15 11 state 16 12 state 16 13 state 15 14 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 5 0 \n state 6 1 \n state 7 2 \n state 8 3 \n state 9 4 \n state 10 5 \n state 10 6 \n state 10 5 \n state 11 5 \n state 12 6 \n state 13 7 \n state 12 8 \n state 12 9 \n state 13 8 \n state 14 9 \n state 15 9 \n state 15 10 \n state 15 11 \n state 16 12 \n state 16 13 \n state 15 14 \n...
{ "handover_type": "actor", "index": 11, "length": 22, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 2 6 state 3 5 state 4 4 state 5 3 state 6 2 state 6 1 state 7 0 state 8 0 state 9 0 state 10 0 state 11 0 state 12 1 state 13 0 state 14 0 state 15 0 state 16 0 state 15 1 state 16 2 state 16 1 state 17 0 state 18 0 state 19 0 state 20 0 state 21 1 state 22 0 stat...
true
1
33
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 6 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 29, "y": 0 }
{ "x": 29, "y": 0 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 6 state 3 5 state 4 4 state 5 3 state 6 2 state 6 1 state 7 0 state 8 0 state 9 0 state 10 0 state 11 0 state 12 1 state 13 0 state 14 0 state 15 0 state 16 0 state 15 1 state 16 2 state 16 1 state 17 0 state 18 0 state 19 0 state 20 0 state 21 1 state 22 0 ...
maze_navigation
{ "ground_truth": "\n state 2 6 \n state 3 5 \n state 4 4 \n state 5 3 \n state 6 2 \n state 6 1 \n state 7 0 \n state 8 0 \n state 9 0 \n state 10 0 \n state 11 0 \n state 12 1 \n state 13 0 \n state 14 0 \n state 15 0 \n state 16 0 \n state 15 1 \n state 16 2 \n state 16 1 \n state 17 0 \n state 18 0 \n state 19...
{ "handover_type": "actor", "index": 12, "length": 33, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 2 4 state 3 5 state 4 6 state 5 7 state 6 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 13 state 11 13 state 12 14 state 13 15 state 14 15 state 15 15
true
1
15
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 4 state 3 5 state 4 6 state 5 7 state 6 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 13 state 11 13 state 12 14 state 13 15 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 2 4 \n state 3 5 \n state 4 6 \n state 5 7 \n state 6 8 \n state 6 9 \n state 7 10 \n state 8 11 \n state 9 12 \n state 10 13 \n state 11 13 \n state 12 14 \n state 13 15 \n state 14 15 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "actor", "index": 13, "length": 15, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 5 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 5 3 state 4 4 state 5 5 state 6 6 state 6 7 state 7 8 state 8 9 state 9 8 state 10 9 state 9 10 state 10 11 state 11 12 state 12 13 state 12 14 state 13 15 state 14 15 state 15 15
true
1
17
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 5, "y": 3 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 5 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 5 3 state 4 4 state 5 5 state 6 6 state 6 7 state 7 8 state 8 9 state 9 8 state 10 9 state 9 10 state 10 11 state 11 12 state 12 13 state 12 14 state 13 15 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 5 3 \n state 4 4 \n state 5 5 \n state 6 6 \n state 6 7 \n state 7 8 \n state 8 9 \n state 9 8 \n state 10 9 \n state 9 10 \n state 10 11 \n state 11 12 \n state 12 13 \n state 12 14 \n state 13 15 \n state 14 15 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "actor", "index": 14, "length": 17, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 5 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 5 1 state 4 2 state 3 3 state 3 2 state 2 1 state 1 2 state 0 3 state 0 4 state 0 5 state 0 6 state 1 7 state 0 6 state 0 7 state 1 7 state 0 8 state 0 9 state 0 10 state 0 11 state 0 12 state 1 13 state 1 14 state 1 15 state 1 16 state 1 15 state 0 16 state 1 17 ...
true
1
40
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 5, "y": 1 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 0, "y": 29 }
{ "x": 0, "y": 29 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 5 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 5 1 state 4 2 state 3 3 state 3 2 state 2 1 state 1 2 state 0 3 state 0 4 state 0 5 state 0 6 state 1 7 state 0 6 state 0 7 state 1 7 state 0 8 state 0 9 state 0 10 state 0 11 state 0 12 state 1 13 state 1 14 state 1 15 state 1 16 state 1 15 state 0 16 state ...
maze_navigation
{ "ground_truth": "\n state 5 1 \n state 4 2 \n state 3 3 \n state 3 2 \n state 2 1 \n state 1 2 \n state 0 3 \n state 0 4 \n state 0 5 \n state 0 6 \n state 1 7 \n state 0 6 \n state 0 7 \n state 1 7 \n state 0 8 \n state 0 9 \n state 0 10 \n state 0 11 \n state 0 12 \n state 1 13 \n state 1 14 \n state 1 15 \n s...
{ "handover_type": "actor", "index": 15, "length": 40, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 6 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 6 4 state 5 5 state 4 6 state 4 5 state 4 4 state 3 5 state 2 6 state 1 7 state 0 8 state 0 9 state 0 10 state 0 11 state 0 12 state 0 13 state 0 12 state 0 13 state 1 14 state 1 15 state 1 14 state 1 15 state 2 16 state 1 17 state 0 18 state 1 18 state 0 19 state...
true
1
37
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 6, "y": 4 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 0, "y": 29 }
{ "x": 0, "y": 29 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 6 4 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 6 4 state 5 5 state 4 6 state 4 5 state 4 4 state 3 5 state 2 6 state 1 7 state 0 8 state 0 9 state 0 10 state 0 11 state 0 12 state 0 13 state 0 12 state 0 13 state 1 14 state 1 15 state 1 14 state 1 15 state 2 16 state 1 17 state 0 18 state 1 18 state 0 19 ...
maze_navigation
{ "ground_truth": "\n state 6 4 \n state 5 5 \n state 4 6 \n state 4 5 \n state 4 4 \n state 3 5 \n state 2 6 \n state 1 7 \n state 0 8 \n state 0 9 \n state 0 10 \n state 0 11 \n state 0 12 \n state 0 13 \n state 0 12 \n state 0 13 \n state 1 14 \n state 1 15 \n state 1 14 \n state 1 15 \n state 2 16 \n state 1 1...
{ "handover_type": "actor", "index": 16, "length": 37, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 3 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 3 1 state 4 2 state 5 3 state 6 4 state 7 3 state 7 4 state 8 5 state 9 6 state 10 7 state 11 8 state 12 9 state 13 10 state 14 10 state 15 11 state 15 12 state 15 11 state 15 12 state 14 13 state 15 14 state 15 15
true
1
20
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 3, "y": 1 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 3 1 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 3 1 state 4 2 state 5 3 state 6 4 state 7 3 state 7 4 state 8 5 state 9 6 state 10 7 state 11 8 state 12 9 state 13 10 state 14 10 state 15 11 state 15 12 state 15 11 state 15 12 state 14 13 state 15 14 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 3 1 \n state 4 2 \n state 5 3 \n state 6 4 \n state 7 3 \n state 7 4 \n state 8 5 \n state 9 6 \n state 10 7 \n state 11 8 \n state 12 9 \n state 13 10 \n state 14 10 \n state 15 11 \n state 15 12 \n state 15 11 \n state 15 12 \n state 14 13 \n state 15 14 \n state 15 15 \n\n </traje...
{ "handover_type": "actor", "index": 17, "length": 20, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 0 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 0 3 state 0 2 state 1 3 state 2 4 state 3 5 state 4 6 state 5 7 state 5 8 state 6 7 state 7 7 state 8 8 state 9 7 state 10 7 state 11 6 state 12 7 state 13 8 state 14 7 state 15 7 state 15 8 state 16 7 state 17 8 state 18 7 state 19 7 state 20 7 state 21 8 state 2...
true
1
26
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 0, "y": 3 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 22, "y": 7 }
{ "x": 22, "y": 7 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 0 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 0 3 state 0 2 state 1 3 state 2 4 state 3 5 state 4 6 state 5 7 state 5 8 state 6 7 state 7 7 state 8 8 state 9 7 state 10 7 state 11 6 state 12 7 state 13 8 state 14 7 state 15 7 state 15 8 state 16 7 state 17 8 state 18 7 state 19 7 state 20 7 state 21 8 st...
maze_navigation
{ "ground_truth": "\n state 0 3 \n state 0 2 \n state 1 3 \n state 2 4 \n state 3 5 \n state 4 6 \n state 5 7 \n state 5 8 \n state 6 7 \n state 7 7 \n state 8 8 \n state 9 7 \n state 10 7 \n state 11 6 \n state 12 7 \n state 13 8 \n state 14 7 \n state 15 7 \n state 15 8 \n state 16 7 \n state 17 8 \n state 18 7 ...
{ "handover_type": "ref", "index": 18, "length": 26, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 6 2 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 5 12 wall 7 1 wall 6 11 wall 29 6 wall 18 6 wall 17 23 wall 2 20 wall 4 29 wall 21 7 wall 21 12 wall 7 20 wall 27 9 wall 21 17 wall 26 7 wall 7 28 wall 12 21 wall 13 9 wall 23 17 wall 17 7 wall...
state 6 2 state 7 3 state 8 3 state 9 4 state 10 5 state 9 6 state 10 6 state 10 7 state 11 8 state 12 7 state 12 6 state 13 7 state 14 7 state 15 7 state 16 8 state 16 7 state 17 8 state 18 7 state 18 8 state 19 7 state 20 7 state 21 8 state 22 7
true
1
23
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 6, "y": 2 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 22, "y": 7 }
{ "x": 22, "y": 7 }
1
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 6 2 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 6 2 state 7 3 state 8 3 state 9 4 state 10 5 state 9 6 state 10 6 state 10 7 state 11 8 state 12 7 state 12 6 state 13 7 state 14 7 state 15 7 state 16 8 state 16 7 state 17 8 state 18 7 state 18 8 state 19 7 state 20 7 state 21 8 state 22 7 </trajectory>
maze_navigation
{ "ground_truth": "\n state 6 2 \n state 7 3 \n state 8 3 \n state 9 4 \n state 10 5 \n state 9 6 \n state 10 6 \n state 10 7 \n state 11 8 \n state 12 7 \n state 12 6 \n state 13 7 \n state 14 7 \n state 15 7 \n state 16 8 \n state 16 7 \n state 17 8 \n state 18 7 \n state 18 8 \n state 19 7 \n state 20 7 \n stat...
{ "handover_type": "actor", "index": 19, "length": 23, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 2 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 17 1 wall 19 6 wall 5 20 wall 7 28 wall 25 1 wall 21 5 wall 19 10 wall 16 15 wall 1 25 wall 23 8 wall 23 7 wall 28 10 wall 9 18 wall 0 9 wall 25 8 wall 16 28 wall 21 1 wall 15 28 wall 9 16 wall...
state 2 3 state 3 4 state 3 3 state 4 2 state 5 1 state 6 2 state 7 3 state 6 4 state 7 5 state 8 6 state 9 7 state 10 8 state 9 9 state 10 10 state 11 11 state 12 12 state 13 13 state 14 14 state 15 14 state 14 15 state 15 15
true
1
21
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 2, "y": 3 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
2
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 2 3 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 2 3 state 3 4 state 3 3 state 4 2 state 5 1 state 6 2 state 7 3 state 6 4 state 7 5 state 8 6 state 9 7 state 10 8 state 9 9 state 10 10 state 11 11 state 12 12 state 13 13 state 14 14 state 15 14 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 2 3 \n state 3 4 \n state 3 3 \n state 4 2 \n state 5 1 \n state 6 2 \n state 7 3 \n state 6 4 \n state 7 5 \n state 8 6 \n state 9 7 \n state 10 8 \n state 9 9 \n state 10 10 \n state 11 11 \n state 12 12 \n state 13 13 \n state 14 14 \n state 15 14 \n state 14 15 \n state 15 15 \n\n ...
{ "handover_type": "actor", "index": 20, "length": 21, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
start 3 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 goal 15 15 wall 17 1 wall 19 6 wall 5 20 wall 7 28 wall 25 1 wall 21 5 wall 19 10 wall 16 15 wall 1 25 wall 23 8 wall 23 7 wall 28 10 wall 9 18 wall 0 9 wall 25 8 wall 16 28 wall 21 1 wall 15 28 wall 9 16 wall...
state 3 6 state 4 7 state 5 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 11 state 11 12 state 12 13 state 13 14 state 14 15 state 15 15
true
1
13
{ "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ], "height": 30, "starts": [ { "x": 0, ...
{ "x": 3, "y": 6 }
[ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, "y": 7 }, { "x": 15, "y": 15 } ]
{ "x": 15, "y": 15 }
{ "x": 15, "y": 15 }
2
maze_navigation
Navigate through the maze from start to goal. You will receive a reward based on whether you reach the goal. The reward is purely based on the trajectory you follow. No amount of formatting or other text will be rewarded. <maze> start 3 6 goal 29 29 goal 29 0 goal 0 29 goal 7 22 goal 22 7 g...
state 3 6 state 4 7 state 5 8 state 6 9 state 7 10 state 8 11 state 9 12 state 10 11 state 11 12 state 12 13 state 13 14 state 14 15 state 15 15 </trajectory>
maze_navigation
{ "ground_truth": "\n state 3 6 \n state 4 7 \n state 5 8 \n state 6 9 \n state 7 10 \n state 8 11 \n state 9 12 \n state 10 11 \n state 11 12 \n state 12 13 \n state 13 14 \n state 14 15 \n state 15 15 \n\n </trajectory>", "style": "rule" }
{ "handover_type": "actor", "index": 21, "length": 13, "maze_spec": { "goals": [ { "x": 29, "y": 29 }, { "x": 29, "y": 0 }, { "x": 0, "y": 29 }, { "x": 7, "y": 22 }, { "x": 22, ...
End of preview. Expand in Data Studio
README.md exists but content is empty.
Downloads last month
9