Hello, everyone,
After conduct some data exploration, we found that it is only same instruction for each GameID.
Is it correct???
I think it should there are many different relative instructions for each GameID and it is more challenge than single instruction one.
Additionally, the columns “InitializedWorldPath” should tell us initial block (position, color).
But I don’t know how to build the block by these below information.
data_path : initial_world_states/builder-data/23-c135/step-2
data['tape'].split('\n') =>
['1 action start_recover_world_state',
'2 action select_and_place_block 56 -1 63 1 6.93311074709203e-24 64.12000274658203 4 -1.733277824122399e-24 0 -1',
'3 action select_and_place_block 56 -1 64 1 6.93311074709203e-24 64.12000274658203 4 -1.733277824122399e-24 0 -1',
'4 action select_and_place_block 56 -1 65 1 6.93311074709203e-24 64.12000274658203 4 -1.733277824122399e-24 0 -1',
'5 action select_and_place_block 56 -1 66 1 6.93311074709203e-24 64.12000274658203 4 -1.733277824122399e-24 0 -1',
'6 action select_and_place_block 56 -1 67 1 6.93311074709203e-24 64.12000274658203 4 -1.733277824122399e-24 0 -1',
'7 block_change (-1, 63, 1, 0, 56) (-1, 64, 1, 0, 56) (-1, 65, 1, 0, 56) (-1, 66, 1, 0, 56) (-1, 67, 1, 0, 56)',
'7 pos_change (-0.5948635840248595, 69.39999999999988, 3.1557175631906675)',
'7 set_look (-1.1160000000000003, 0.0020000000000000625)',
'8 action finish_recover_world_state',
'2741 action step_forward',
'']
Thank you a lot.