Dataset Viewer
images
images listlengths 1
1
| problem
stringclasses 1
value | answer
sequencelengths 2
2
|
|---|---|---|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
4,
6
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
2,
3
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
4,
6
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
1,
4
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
4,
6
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
1,
3
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
2,
4
] |
|
<image>You are controlling a robotic arm in a 3D space. The end-effector can move in six directions: 1 = +x (forward), 2 = -x (backward), 3 = +y (right), 4 = -y (left), 6 = -z (down). The image you see is split horizontally into two halves: the UPPER half is a top-down view taken from directly above the end-effector, while the LOWER half is the default oblique perspective view that shows the whole robot, table, and cube. You need to execute two consecutive actions to reach the green cube. First, choose one number from [1,2,3,4,6] as your first action. After executing the first action, you will see the updated state and choose your second action. The green cube always appears near the end-effector's reachable zone. Output only the final answer inside <answer>…</answer>, in the format <answer>n</answer> for each action. | [
1,
3
] |
README.md exists but content is empty.
- Downloads last month
- 41