Dataset Viewer
Auto-converted to Parquet Duplicate
image
imagewidth (px)
300
720
question
stringlengths
27
164
choices
listlengths
2
6
answer
stringclasses
46 values
answer_type
stringclasses
5 values
auxiliary_images
images listlengths
0
4
capability_name
stringclasses
8 values
question_type_name
stringclasses
5 values
difficulty
stringclasses
3 values
id
stringlengths
26
26
sample_id
stringclasses
68 values
split
stringclasses
1 value
capability_id
stringclasses
8 values
question_type
stringclasses
6 values
template_id
stringclasses
22 values
template_variant_id
stringclasses
220 values
option_labels
listlengths
2
6
option_texts
listlengths
2
6
options_json
stringclasses
844 values
answer_format
stringclasses
2 values
metric_name
stringclasses
4 values
metric_family
stringclasses
3 values
scorer
stringclasses
4 values
scoring_rule
stringclasses
4 values
vlmeval_compatible
bool
1 class
difficulty_reason
stringclasses
21 values
gt_fields
listlengths
1
4
metadata_json
stringlengths
1.27k
6.95k
Which of these categories are actually present in the current image?
[ "A. plunger", "B. shower glass", "C. towel rack", "D. plate" ]
["B","C"]
multi_choice
visible object and category recognition
Multiple Choice Question
medium
ALFRED_GT_OBJECTIVE_002876
spatial_0024
test
C01_OBJECT_VISIBILITY
multi_choice
ALF_C01_VISIBLE_MULTISELECT
ALF_C01_VISIBLE_MULTISELECT_v07
[ "A", "B", "C", "D" ]
[ "plunger", "shower glass", "towel rack", "plate" ]
{"A":"plunger","B":"shower glass","C":"towel rack","D":"plate"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
set membership over multiple candidate categories
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"selected_categories":["TowelHolder","ShowerGlass"],"template_variant_id":"ALF_C01_VISIBLE_MULTISELECT_v07","visible_categories":["Bathtub","Candle","Cloth","CounterTop","Drawer","Faucet","GarbageCan","HandTowelHolder","ShowerDoor","ShowerGlass","Sink","SinkBasin","Towel","TowelHolder"]},"gt_lineage...
Which candidate category definitely appears in this image?
[ "A. vase", "B. mirror", "C. potato", "D. alarm clock" ]
D
single_choice
visible object and category recognition
Multiple Choice Question
simple
ALFRED_GT_OBJECTIVE_000002
base_0025
test
C01_OBJECT_VISIBILITY
single_choice
ALF_C01_VISIBLE_SINGLE_CHOICE
ALF_C01_VISIBLE_SINGLE_CHOICE_v02
[ "A", "B", "C", "D" ]
[ "vase", "mirror", "potato", "alarm clock" ]
{"A":"vase","B":"mirror","C":"potato","D":"alarm clock"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
single visible category selected among absent distractors
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"answer_category":"AlarmClock","distractors":["Mirror","Vase","Potato"],"template_variant_id":"ALF_C01_VISIBLE_SINGLE_CHOICE_v02","visible_categories":["AlarmClock","Book","DiningTable","Lamp","Mug","Painting","Pen","Pencil","RemoteControl"]},"gt_lineage":{"answer_margin":{"distractor_count":3,"visi...
Which range does the number of visible lettuce instances in the scene fall into?
[ "A. 0", "B. 1", "C. 2", "D. 3-4", "E. 5-8", "F. 9 or more" ]
C
interval_choice
Instance count and set size
Interval Selection Question
medium
ALFRED_GT_OBJECTIVE_004018
spatial_0018
test
C02_COUNTING_SET_SIZE
interval_choice
ALF_C02_CATEGORY_COUNT_INTERVAL
ALF_C02_CATEGORY_COUNT_INTERVAL_v03
[ "A", "B", "C", "D", "E", "F" ]
[ "0", "1", "2", "3-4", "5-8", "9 or more" ]
{"A":"0","B":"1","C":"2","D":"3-4","E":"5-8","F":"9 or more"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
category count mapped to interval; no raw numeric answer
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"category":"Lettuce","count":2,"interval":"2","template_variant_id":"ALF_C02_CATEGORY_COUNT_INTERVAL_v03"},"gt_lineage":{"answer_margin":{"count":2,"interval":"2"},"difficulty":"medium","difficulty_reason":"category count mapped to interval; no raw numeric answer","distractor_strategy":"fixed mutual...
Please arrange these categories in order of their appearance from most to least, paying particular attention to the exact numeric values and capital-letter markers.
[ "A. laptop", "B. drawer", "C. window" ]
["B","A","C"]
ordered_list
Instance count and set size
Sorting Question
medium
ALFRED_GT_OBJECTIVE_002768
base_0014
test
C02_COUNTING_SET_SIZE
ordered_list
ALF_C02_COUNT_ORDER
ALF_C02_COUNT_ORDER_v02
[ "A", "B", "C" ]
[ "laptop", "drawer", "window" ]
{"A":"laptop","B":"drawer","C":"window"}
list
Exact Order Accuracy
ordering
alfred_gt_objective_metrics.ordered_scores
predicted ordered option labels exactly equal gold order
true
multi-category count comparison with tie rejection
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"counts_by_category":{"Drawer":4,"Laptop":3,"Window":2},"ordered_categories":["Drawer","Laptop","Window"],"template_variant_id":"ALF_C02_COUNT_ORDER_v02"},"gt_lineage":{"answer_margin":{"counts_by_category":{"Drawer":4,"Laptop":3,"Window":2},"minimum_count_gap":1},"difficulty":"medium","difficulty_r...
Observe the positions of A, B, C, and D, which one is the farthest to the right in the image?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
D
single_choice
image plane spatial relationships
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_003761
long_horizon_0004
test
C03_IMAGE_PLANE_SPATIAL
marked_object_choice
ALF_C03_RIGHTMOST_MARKED
ALF_C03_RIGHTMOST_MARKED_v06
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
requires projected image-plane comparison among marked objects
[ "agent.position", "agent.rotation", "objects.position", "observations.visibleObjects" ]
{"answer_evidence":{"projected_x_by_option":{"A":0.07840155458722659,"B":0.2949772585750624,"C":0.36234218439369303,"D":0.9232288978927654},"template_variant_id":"ALF_C03_RIGHTMOST_MARKED_v06"},"gt_lineage":{"answer_margin":{"margin_meaning":"minimum projected-x gap from rightmost object","right_projected_x_gap":0.5609...
Which one is closest according to the current perspective?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
C
single_choice
Three-dimensional distance and proximity
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_003748
long_horizon_0000
test
C04_DISTANCE_DEPTH
marked_object_choice
ALF_C04_NEAREST_MARKED
ALF_C04_NEAREST_MARKED_v09
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
distance comparison among multiple marked candidates with near-tie rejection
[ "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":0.756179,"B":0.7298729,"C":0.426688939,"D":1.05674922},"template_variant_id":"ALF_C04_NEAREST_MARKED_v09"},"gt_lineage":{"answer_margin":{"nearest_gap_m":0.3032},"difficulty":"medium","difficulty_reason":"distance comparison among multiple marked candidates with near-tie ...
Observe the annotated image and give the order of these candidate objects from nearest to farthest.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["B","D","C","A"]
ordered_list
Three-dimensional distance and proximity
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006022
long_horizon_0000
test
C04_DISTANCE_DEPTH
ordered_list
ALF_C04_DISTANCE_ORDER
ALF_C04_DISTANCE_ORDER_v07
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
full ordering over four marked objects by tracker distance
[ "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":1.34494054,"B":0.331879735,"C":0.9266402,"D":0.601336956},"template_variant_id":"ALF_C04_DISTANCE_ORDER_v07"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":0.2695},"difficulty":"hard","difficulty_reason":"full ordering over four marked objects by tracke...
Choose an appropriate range for the number of pickable objects in R or on R.
[ "A. 0", "B. 1", "C. 2", "D. 3-4", "E. 5-8", "F. 9 or more" ]
C
interval_choice
container and containment relationship
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_002227
long_horizon_0011
test
C05_RECEPTACLE_CONTAINMENT
marked_object_choice
ALF_C05_RECEPTACLE_COUNT_INTERVAL
ALF_C05_RECEPTACLE_COUNT_INTERVAL_v10
[ "A", "B", "C", "D", "E", "F" ]
[ "0", "1", "2", "3-4", "5-8", "9 or more" ]
{"A":"0","B":"1","C":"2","D":"3-4","E":"5-8","F":"9 or more"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
counts pickupable members of one marked receptacle and maps to interval
[ "objects.history.receptacleObjectIds", "objects.history.pickupable" ]
{"answer_evidence":{"count":2,"interval":"2","pickupable_member_ids":["Cup|-01.77|+00.93|+00.52","Spatula|-01.35|+00.94|+00.65"],"receptacle_id":"CounterTop|-01.37|+00.95|+00.35","template_variant_id":"ALF_C05_RECEPTACLE_COUNT_INTERVAL_v10"},"gt_lineage":{"answer_margin":{"interval":"2","pickupable_member_count":2},"di...
Please select the labeled objects currently on R or inside R.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["A","C"]
multi_choice
container and containment relationship
Multiple Choice Question
hard
ALFRED_GT_OBJECTIVE_006098
long_horizon_0005
test
C05_RECEPTACLE_CONTAINMENT
multi_choice
ALF_C05_RECEPTACLE_MULTISELECT
ALF_C05_RECEPTACLE_MULTISELECT_v04
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects all marked objects linked to one receptacle membership set
[ "objects.history.receptacleObjectIds", "objects.history.pickupable" ]
{"answer_evidence":{"candidate_ids_by_option":{"A":"Lettuce|-01.76|+00.97|+00.09|LettuceSliced_1","B":"Knife|-01.79|+00.92|+00.36","C":"Tomato|-01.52|+00.96|+00.56","D":"Window|-01.43|+01.34|-00.35"},"member_ids":["Potato|-01.66|+00.92|+00.56","DishSponge|-01.52|+00.89|-00.05","Tomato|-01.52|+00.96|+00.56","Spatula|-01...
With the robot's current orientation as the reference, which direction is object A located?
[ "A. front", "B. left side", "C. right side" ]
A
single_choice
proximity reachability and egocentric orientation
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_002196
long_horizon_0001
test
C06_REACHABILITY_EGOCENTRIC
marked_object_choice
ALF_C06_EGO_DIRECTION
ALF_C06_EGO_DIRECTION_v06
[ "A", "B", "C" ]
[ "front", "left side", "right side" ]
{"A":"front","B":"left side","C":"right side"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
transforms marked object world position into agent-centered bearing sector
[ "agent.position", "agent.rotation", "objects.history.position", "observations.visibleObjects" ]
{"answer_evidence":{"bearing_deg":8.080652393249744e-06,"category":"AppleSliced","direction":"front","object_id":"Apple|-00.04|+00.97|-01.49|AppleSliced_0","template_variant_id":"ALF_C06_EGO_DIRECTION_v06"},"gt_lineage":{"answer_margin":{"bearing_deg":0.0,"sector":"front"},"difficulty":"medium","difficulty_reason":"tra...
Compare Figure A and Figure B, is the spoon visible in only one of the two images?
[ "A. is", "B. No" ]
A
yes_no
multi-frame visibility and trajectory order
Multiple-frame/Multiple-view Selection Question
medium
ALFRED_GT_OBJECTIVE_002001
spatial_0011
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
multi_view_choice
ALF_C07_VISIBILITY_CHANGE_YN
ALF_C07_VISIBILITY_CHANGE_YN_v06
[ "A", "B" ]
[ "is", "No" ]
{"A":"is","B":"No"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
compares category visibility across two ordered frames
[ "execution_steps.image_path", "observations.visibleObjects" ]
{"answer_evidence":{"category":"Spoon","changed":true,"template_variant_id":"ALF_C07_VISIBILITY_CHANGE_YN_v06","visible_in_A":true,"visible_in_B":false},"gt_lineage":{"answer_margin":{"visible_in_A":true,"visible_in_B":false},"difficulty":"medium","difficulty_reason":"compares category visibility across two ordered fra...
Which frame shows the TV as closer? Please give the image order from nearest to farthest.
[ "A. Figure A", "B. Figure B", "C. Figure C" ]
["B","C","A"]
ordered_list
multi-frame visibility and trajectory order
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006604
base_0006
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
ordered_list
ALF_C07_SAME_OBJECT_DISTANCE_ORDER
ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v03
[ "A", "B", "C" ]
[ "Figure A", "Figure B", "Figure C" ]
{"A":"Figure A","B":"Figure B","C":"Figure C"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
tracks same object identity across frames and orders distance values
[ "execution_steps.image_path", "objects.history.distance" ]
{"answer_evidence":{"category":"Television","distance_by_panel_m":{"A":3.84543324,"B":1.21628821,"C":2.50706148},"object_id":"Television|+00.83|+01.47|+03.73","template_variant_id":"ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v03"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":1.2907732699999999},"difficulty"...
Based on the current scene state, can A be judged as "cleaned"?
[ "A. is", "B. No" ]
B
yes_no
object state and operable attributes
question about selecting identified objects in the image
simple
ALFRED_GT_OBJECTIVE_000017
long_horizon_0003
test
C08_OBJECT_STATE_AFFORDANCE
marked_object_choice
ALF_C08_STATE_YN
ALF_C08_STATE_YN_v10
[ "A", "B" ]
[ "is", "No" ]
{"A":"is","B":"No"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
direct boolean state lookup for one marked visible object
[ "objects.history.isOpen", "objects.history.isToggledOn", "objects.history.isSliced", "objects.history.isDirty" ]
{"answer_evidence":{"asked_state_text":"Cleaned","category":"Mug","object_id":"Mug|-00.34|+00.91|-01.37","state_name_hidden":"isCleaned","state_value":false,"template_variant_id":"ALF_C08_STATE_YN_v10"},"gt_lineage":{"answer_margin":{"asked_value":true,"state_field_hidden":"isCleaned","state_value":false},"difficulty":...
A, B, C, D, please select all objects that are currently supported to be cut.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["B"]
multi_choice
object state and operable attributes
Multiple Choice Question
medium
ALFRED_GT_OBJECTIVE_002365
long_horizon_0002
test
C08_OBJECT_STATE_AFFORDANCE
multi_choice
ALF_C08_AFFORDANCE_MULTISELECT
ALF_C08_AFFORDANCE_MULTISELECT_v07
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects a GT affordance-defined set among marked object candidates
[ "objects.history.pickupable", "objects.history.openable", "objects.history.toggleable", "objects.history.sliceable" ]
{"answer_evidence":{"affordance_by_option":{"A":false,"B":true,"C":false,"D":false},"affordance_hidden":"sliceable","candidate_ids_by_option":{"A":"Sink|+01.54|+00.87|+01.64","B":"Potato|+01.55|+00.98|+01.67","C":"Tomato|+01.36|+01.00|+02.02|TomatoSliced_4","D":"Spatula|+01.41|+00.96|+01.59"},"template_variant_id":"ALF...
Select the categories that can be seen in the image from the following options.
[ "A. bowl", "B. remote control", "C. trash can", "D. spray bottle" ]
A
single_choice
visible object and category recognition
Multiple Choice Question
simple
ALFRED_GT_OBJECTIVE_000012
long_horizon_0009
test
C01_OBJECT_VISIBILITY
single_choice
ALF_C01_VISIBLE_SINGLE_CHOICE
ALF_C01_VISIBLE_SINGLE_CHOICE_v03
[ "A", "B", "C", "D" ]
[ "bowl", "remote control", "trash can", "spray bottle" ]
{"A":"bowl","B":"remote control","C":"trash can","D":"spray bottle"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
single visible category selected among absent distractors
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"answer_category":"Bowl","distractors":["SprayBottle","GarbageCan","RemoteControl"],"template_variant_id":"ALF_C01_VISIBLE_SINGLE_CHOICE_v03","visible_categories":["AppleSliced","Bowl","CellPhone","DiningTable","Fork","Fridge","Glassbottle","Pan","SaltShaker","Spatula"]},"gt_lineage":{"answer_margin...
Which range does the number of pillow instances in the scene fall into?
[ "A. 0", "B. 1", "C. 2", "D. 3-4", "E. 5-8", "F. 9 or more" ]
C
interval_choice
Instance count and set size
Interval Selection Question
medium
ALFRED_GT_OBJECTIVE_004030
base_0023
test
C02_COUNTING_SET_SIZE
interval_choice
ALF_C02_CATEGORY_COUNT_INTERVAL
ALF_C02_CATEGORY_COUNT_INTERVAL_v03
[ "A", "B", "C", "D", "E", "F" ]
[ "0", "1", "2", "3-4", "5-8", "9 or more" ]
{"A":"0","B":"1","C":"2","D":"3-4","E":"5-8","F":"9 or more"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
category count mapped to interval; no raw numeric answer
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"category":"Pillow","count":2,"interval":"2","template_variant_id":"ALF_C02_CATEGORY_COUNT_INTERVAL_v03"},"gt_lineage":{"answer_margin":{"count":2,"interval":"2"},"difficulty":"medium","difficulty_reason":"category count mapped to interval; no raw numeric answer","distractor_strategy":"fixed mutuall...
How many times do these categories appear in the figure? Please rank them from most to least.
[ "A. stove", "B. stove knob", "C. surface", "D. cabinet" ]
["D","B","A","C"]
ordered_list
Instance count and set size
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006846
spatial_0000
test
C02_COUNTING_SET_SIZE
ordered_list
ALF_C02_COUNT_ORDER
ALF_C02_COUNT_ORDER_v09
[ "A", "B", "C", "D" ]
[ "stove", "stove knob", "surface", "cabinet" ]
{"A":"stove","B":"stove knob","C":"surface","D":"cabinet"}
list
Exact Order Accuracy
ordering
alfred_gt_objective_metrics.ordered_scores
predicted ordered option labels exactly equal gold order
true
multi-category count comparison with tie rejection
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"counts_by_category":{"Cabinet":7,"CounterTop":2,"StoveBurner":3,"StoveKnob":4},"ordered_categories":["Cabinet","StoveKnob","StoveBurner","CounterTop"],"template_variant_id":"ALF_C02_COUNT_ORDER_v09"},"gt_lineage":{"answer_margin":{"counts_by_category":{"Cabinet":7,"CounterTop":2,"StoveBurner":3,"St...
Observe the positions of A, B, C, and D, which one is the closest to the left edge of the screen?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
D
single_choice
image plane spatial relationships
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_003785
long_horizon_0005
test
C03_IMAGE_PLANE_SPATIAL
marked_object_choice
ALF_C03_LEFTMOST_MARKED
ALF_C03_LEFTMOST_MARKED_v06
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
requires projected image-plane comparison among marked objects
[ "agent.position", "agent.rotation", "objects.position", "observations.visibleObjects" ]
{"answer_evidence":{"projected_x_by_option":{"A":0.7404387451364904,"B":0.30896600886618364,"C":0.3836951935166293,"D":0.1849870755250277},"template_variant_id":"ALF_C03_LEFTMOST_MARKED_v06"},"gt_lineage":{"answer_margin":{"left_projected_x_gap":0.124,"margin_meaning":"minimum projected-x gap from leftmost object"},"di...
Sort the candidate labeled objects from near to far.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["A","C","D","B"]
ordered_list
Three-dimensional distance and proximity
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006026
long_horizon_0004
test
C04_DISTANCE_DEPTH
ordered_list
ALF_C04_DISTANCE_ORDER
ALF_C04_DISTANCE_ORDER_v01
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
full ordering over four marked objects by tracker distance
[ "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":1.99366379,"B":3.18770862,"C":2.23671842,"D":2.48772573},"template_variant_id":"ALF_C04_DISTANCE_ORDER_v01"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":0.2431},"difficulty":"hard","difficulty_reason":"full ordering over four marked objects by tracker...
Please determine which candidate object has a placement/carrying relationship with R.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
C
single_choice
container and containment relationship
question about selecting identified objects in the image
hard
ALFRED_GT_OBJECTIVE_008672
base_0002
test
C05_RECEPTACLE_CONTAINMENT
marked_object_choice
ALF_C05_RECEPTACLE_MEMBER
ALF_C05_RECEPTACLE_MEMBER_v07
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
requires following receptacle membership and selecting among marked object candidates
[ "objects.history.receptacleObjectIds", "objects.history.pickupable" ]
{"answer_evidence":{"correct_object_id":"Towel|-00.49|+01.37|+00.90","member_ids":["Towel|-00.49|+01.37|+00.90"],"receptacle_id":"TowelHolder|-00.49|+01.35|+01.01","template_variant_id":"ALF_C05_RECEPTACLE_MEMBER_v07"},"gt_lineage":{"answer_margin":{"distractor_count":11,"member_count":1},"difficulty":"hard","difficult...
Please select multiple objects annotated above or inside R.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["B","D"]
multi_choice
container and containment relationship
Multiple Choice Question
hard
ALFRED_GT_OBJECTIVE_006118
long_horizon_0005
test
C05_RECEPTACLE_CONTAINMENT
multi_choice
ALF_C05_RECEPTACLE_MULTISELECT
ALF_C05_RECEPTACLE_MULTISELECT_v09
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects all marked objects linked to one receptacle membership set
[ "objects.history.receptacleObjectIds", "objects.history.pickupable" ]
{"answer_evidence":{"candidate_ids_by_option":{"A":"Window|-01.43|+01.34|-00.35","B":"Lettuce|-01.76|+00.97|+00.09|LettuceSliced_5","C":"Lettuce|-01.76|+00.97|+00.09|LettuceSliced_0","D":"Bread|-01.13|+00.97|+00.26"},"member_ids":["Potato|-01.66|+00.92|+00.56","DishSponge|-01.52|+00.89|-00.05","Tomato|-01.52|+00.96|+00...
Which of these labeled objects is within the current interactive proximity range?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
D
single_choice
proximity reachability and egocentric orientation
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_002199
long_horizon_0001
test
C06_REACHABILITY_EGOCENTRIC
marked_object_choice
ALF_C06_NEARBY_OBJECT
ALF_C06_NEARBY_OBJECT_v04
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
uses nearby GT and distance tie-breaking among marked candidates
[ "observations.nearbyObjects", "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":1.72227657,"B":1.01585686,"C":1.0014503,"D":0.379985183},"nearby_ids":["Knife|+00.66|+00.94|+00.30"],"template_variant_id":"ALF_C06_NEARBY_OBJECT_v04"},"gt_lineage":{"answer_margin":{"nearby_count":1,"nearest_distance_m":0.379985183},"difficulty":"medium","difficulty_reas...
Which image shows the showerhead first in the order from Figure A to Figure D?
[ "A. Figure A", "B. Figure B", "C. Figure C", "D. Figure D" ]
D
single_choice
multi-frame visibility and trajectory order
Multiple-frame/Multiple-view Selection Question
medium
ALFRED_GT_OBJECTIVE_002002
base_0012
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
multi_view_choice
ALF_C07_FIRST_VISIBLE_FRAME
ALF_C07_FIRST_VISIBLE_FRAME_v01
[ "A", "B", "C", "D" ]
[ "Figure A", "Figure B", "Figure C", "Figure D" ]
{"A":"Figure A","B":"Figure B","C":"Figure C","D":"Figure D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
requires scanning ordered panels for first visible occurrence
[ "execution_steps.image_path", "observations.visibleObjects" ]
{"answer_evidence":{"category":"ShowerHead","template_variant_id":"ALF_C07_FIRST_VISIBLE_FRAME_v01","visible_series":{"A":0,"B":0,"C":0,"D":1}},"gt_lineage":{"answer_margin":{"first_visible_panel":"D"},"difficulty":"medium","difficulty_reason":"requires scanning ordered panels for first visible occurrence","distractor_...
Track the same dressing table and arrange the images A, B, C from closest to farthest.
[ "A. Figure A", "B. Figure B", "C. Figure C" ]
["C","B","A"]
ordered_list
multi-frame visibility and trajectory order
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006608
spatial_0012
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
ordered_list
ALF_C07_SAME_OBJECT_DISTANCE_ORDER
ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v02
[ "A", "B", "C" ]
[ "Figure A", "Figure B", "Figure C" ]
{"A":"Figure A","B":"Figure B","C":"Figure C"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
tracks same object identity across frames and orders distance values
[ "execution_steps.image_path", "objects.history.distance" ]
{"answer_evidence":{"category":"Dresser","distance_by_panel_m":{"A":4.610656,"B":2.220506,"C":1.13231933},"object_id":"Dresser|-01.47|+00.03|-02.05","template_variant_id":"ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v02"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":1.0881866699999998},"difficulty":"hard","d...
Observe the annotated image and determine if object A is currently in the "open" state.
[ "A. is", "B. No" ]
B
yes_no
object state and operable attributes
question about selecting identified objects in the image
simple
ALFRED_GT_OBJECTIVE_000034
long_horizon_0001
test
C08_OBJECT_STATE_AFFORDANCE
marked_object_choice
ALF_C08_STATE_YN
ALF_C08_STATE_YN_v01
[ "A", "B" ]
[ "is", "No" ]
{"A":"is","B":"No"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
direct boolean state lookup for one marked visible object
[ "objects.history.isOpen", "objects.history.isToggledOn", "objects.history.isSliced", "objects.history.isDirty" ]
{"answer_evidence":{"asked_state_text":"Open","category":"Cabinet","object_id":"Cabinet|-00.71|+01.96|-00.82","state_name_hidden":"isOpen","state_value":false,"template_variant_id":"ALF_C08_STATE_YN_v01"},"gt_lineage":{"answer_margin":{"asked_value":true,"state_field_hidden":"isOpen","state_value":false},"difficulty":"...
Which of these candidate objects can execute "toggle"?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["D"]
multi_choice
object state and operable attributes
Multiple Choice Question
medium
ALFRED_GT_OBJECTIVE_002373
long_horizon_0011
test
C08_OBJECT_STATE_AFFORDANCE
multi_choice
ALF_C08_AFFORDANCE_MULTISELECT
ALF_C08_AFFORDANCE_MULTISELECT_v04
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects a GT affordance-defined set among marked object candidates
[ "objects.history.pickupable", "objects.history.openable", "objects.history.toggleable", "objects.history.sliceable" ]
{"answer_evidence":{"affordance_by_option":{"A":false,"B":false,"C":false,"D":true},"affordance_hidden":"toggleable","candidate_ids_by_option":{"A":"Lettuce|-02.79|+00.38|+04.58","B":"Apple|-01.27|+00.85|+00.23|AppleSliced_0","C":"Bread|-02.81|+01.24|+04.35","D":"CoffeeMachine|-02.96|+00.93|+03.61"},"template_variant_i...
Is the blind visible in this frame?
[ "A. is", "B. No" ]
A
yes_no
visible object and category recognition
Multiple Choice Question
simple
ALFRED_GT_OBJECTIVE_000014
base_0018
test
C01_OBJECT_VISIBILITY
single_choice
ALF_C01_VISIBLE_CATEGORY_YN
ALF_C01_VISIBLE_CATEGORY_YN_v08
[ "A", "B" ]
[ "is", "No" ]
{"A":"is","B":"No"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
single visible category membership lookup
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"category":"Blinds","present":true,"template_variant_id":"ALF_C01_VISIBLE_CATEGORY_YN_v08"},"gt_lineage":{"answer_margin":{},"difficulty":"simple","difficulty_reason":"single visible category membership lookup","distractor_strategy":"present category","gt_lineage_depth":1,"involved_gt_edges":["objec...
Based on the current viewpoint, which range does the total count of visible instances fall into?
[ "A. 0-10", "B. 11-25", "C. 26-50", "D. 51 or more" ]
B
interval_choice
Instance count and set size
Interval Selection Question
medium
ALFRED_GT_OBJECTIVE_004067
base_0016
test
C02_COUNTING_SET_SIZE
interval_choice
ALF_C02_TOTAL_COUNT_INTERVAL
ALF_C02_TOTAL_COUNT_INTERVAL_v04
[ "A", "B", "C", "D" ]
[ "0-10", "11-25", "26-50", "51 or more" ]
{"A":"0-10","B":"11-25","C":"26-50","D":"51 or more"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
all visible instances counted from GT and mapped to interval
[ "observations.visibleObjects" ]
{"answer_evidence":{"interval":"11-25","template_variant_id":"ALF_C02_TOTAL_COUNT_INTERVAL_v04","total_visible_instances":20},"gt_lineage":{"answer_margin":{"interval":"11-25","total_visible_instances":20},"difficulty":"medium","difficulty_reason":"all visible instances counted from GT and mapped to interval","distract...
Which of the objects labeled A, B, C, or D is located at the far right of the image?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
D
single_choice
image plane spatial relationships
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_003869
long_horizon_0011
test
C03_IMAGE_PLANE_SPATIAL
marked_object_choice
ALF_C03_RIGHTMOST_MARKED
ALF_C03_RIGHTMOST_MARKED_v10
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
requires projected image-plane comparison among marked objects
[ "agent.position", "agent.rotation", "objects.position", "observations.visibleObjects" ]
{"answer_evidence":{"projected_x_by_option":{"A":0.6669098048193336,"B":0.6752934248164326,"C":0.5526435104793105,"D":0.8902794214826664},"template_variant_id":"ALF_C03_RIGHTMOST_MARKED_v10"},"gt_lineage":{"answer_margin":{"margin_meaning":"minimum projected-x gap from rightmost object","right_projected_x_gap":0.215},"...
Which objects are closer and which are farther? Give the order from closest to farthest.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["D","C","A","B"]
ordered_list
Three-dimensional distance and proximity
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006046
long_horizon_0000
test
C04_DISTANCE_DEPTH
ordered_list
ALF_C04_DISTANCE_ORDER
ALF_C04_DISTANCE_ORDER_v03
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
full ordering over four marked objects by tracker distance
[ "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":3.336725,"B":3.75086045,"C":2.40442944,"D":1.78653193},"template_variant_id":"ALF_C04_DISTANCE_ORDER_v03"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":0.4141},"difficulty":"hard","difficulty_reason":"full ordering over four marked objects by tracker d...
R is the highlighted carrier. Please select all candidate objects that are currently placed on or inside R.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["D"]
multi_choice
container and containment relationship
Multiple Choice Question
hard
ALFRED_GT_OBJECTIVE_006188
long_horizon_0004
test
C05_RECEPTACLE_CONTAINMENT
multi_choice
ALF_C05_RECEPTACLE_MULTISELECT
ALF_C05_RECEPTACLE_MULTISELECT_v06
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects all marked objects linked to one receptacle membership set
[ "objects.history.receptacleObjectIds", "objects.history.pickupable" ]
{"answer_evidence":{"candidate_ids_by_option":{"A":"CounterTop|+00.09|+00.89|-01.52","B":"Kettle|+00.84|+00.91|+00.04","C":"Cabinet|-00.92|+01.67|-00.62","D":"Kettle|-00.42|+00.90|-00.92"},"member_ids":["Kettle|-00.42|+00.90|-00.92"],"receptacle_id":"StoveBurner|-00.42|+00.90|-00.92","template_variant_id":"ALF_C05_RECE...
From the robot's current position, is A more towards the front, left, or right?
[ "A. front", "B. left side", "C. right side" ]
A
single_choice
proximity reachability and egocentric orientation
question about selecting identified objects in the image
medium
ALFRED_GT_OBJECTIVE_002212
long_horizon_0003
test
C06_REACHABILITY_EGOCENTRIC
marked_object_choice
ALF_C06_EGO_DIRECTION
ALF_C06_EGO_DIRECTION_v02
[ "A", "B", "C" ]
[ "front", "left side", "right side" ]
{"A":"front","B":"left side","C":"right side"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
transforms marked object world position into agent-centered bearing sector
[ "agent.position", "agent.rotation", "objects.history.position", "observations.visibleObjects" ]
{"answer_evidence":{"bearing_deg":8.227975163747558,"category":"ButterKnife","direction":"front","object_id":"ButterKnife|-02.66|+00.72|-00.29","template_variant_id":"ALF_C06_EGO_DIRECTION_v02"},"gt_lineage":{"answer_margin":{"bearing_deg":8.228,"sector":"front"},"difficulty":"medium","difficulty_reason":"transforms ma...
Has the appearance of the pepper shaker changed from Figure A to Figure B?
[ "A. is", "B. No" ]
A
yes_no
multi-frame visibility and trajectory order
Multiple-frame/Multiple-view Selection Question
medium
ALFRED_GT_OBJECTIVE_002003
base_0026
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
multi_view_choice
ALF_C07_VISIBILITY_CHANGE_YN
ALF_C07_VISIBILITY_CHANGE_YN_v07
[ "A", "B" ]
[ "is", "No" ]
{"A":"is","B":"No"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
compares category visibility across two ordered frames
[ "execution_steps.image_path", "observations.visibleObjects" ]
{"answer_evidence":{"category":"PepperShaker","changed":true,"template_variant_id":"ALF_C07_VISIBILITY_CHANGE_YN_v07","visible_in_A":true,"visible_in_B":false},"gt_lineage":{"answer_margin":{"visible_in_A":true,"visible_in_B":false},"difficulty":"medium","difficulty_reason":"compares category visibility across two orde...
Please compare the distance of the same pillow from the camera in three frames and sort them from nearest to farthest.
[ "A. Figure A", "B. Figure B", "C. Figure C" ]
["B","A","C"]
ordered_list
multi-frame visibility and trajectory order
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006612
base_0020
test
C07_TEMPORAL_VISIBILITY_TRAJECTORY
ordered_list
ALF_C07_SAME_OBJECT_DISTANCE_ORDER
ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v09
[ "A", "B", "C" ]
[ "Figure A", "Figure B", "Figure C" ]
{"A":"Figure A","B":"Figure B","C":"Figure C"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
tracks same object identity across frames and orders distance values
[ "execution_steps.image_path", "objects.history.distance" ]
{"answer_evidence":{"category":"Pillow","distance_by_panel_m":{"A":1.30171025,"B":0.8482037,"C":1.65965343},"object_id":"Pillow|-01.24|+00.46|-00.03","template_variant_id":"ALF_C07_SAME_OBJECT_DISTANCE_ORDER_v09"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":0.35794318000000014},"difficulty":"hard",...
Which of these candidate objects can be cut?
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["A","C"]
multi_choice
object state and operable attributes
Multiple Choice Question
hard
ALFRED_GT_OBJECTIVE_006466
long_horizon_0001
test
C08_OBJECT_STATE_AFFORDANCE
multi_choice
ALF_C08_AFFORDANCE_MULTISELECT
ALF_C08_AFFORDANCE_MULTISELECT_v09
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Set Exact Match + F1
set_match
alfred_gt_objective_metrics.set_scores
set equality is primary; precision/recall/F1 are auxiliary
true
selects a GT affordance-defined set among marked object candidates
[ "objects.history.pickupable", "objects.history.openable", "objects.history.toggleable", "objects.history.sliceable" ]
{"answer_evidence":{"affordance_by_option":{"A":true,"B":false,"C":true,"D":false},"affordance_hidden":"sliceable","candidate_ids_by_option":{"A":"Potato|+00.89|+00.95|+00.33","B":"StoveKnob|-00.37|+00.79|-00.72","C":"Apple|-00.80|+00.97|-00.83","D":"Cabinet|+01.40|+01.87|-01.26"},"template_variant_id":"ALF_C08_AFFORDA...
Which type of object is currently in the field of view?
[ "A. cut lettuce", "B. table", "C. spatula", "D. lettuce" ]
A
single_choice
visible object and category recognition
Multiple Choice Question
simple
ALFRED_GT_OBJECTIVE_000024
long_horizon_0005
test
C01_OBJECT_VISIBILITY
single_choice
ALF_C01_VISIBLE_SINGLE_CHOICE
ALF_C01_VISIBLE_SINGLE_CHOICE_v04
[ "A", "B", "C", "D" ]
[ "cut lettuce", "table", "spatula", "lettuce" ]
{"A":"cut lettuce","B":"table","C":"spatula","D":"lettuce"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
single visible category selected among absent distractors
[ "observations.visibleObjects", "objects.objectType" ]
{"answer_evidence":{"answer_category":"LettuceSliced","distractors":["Lettuce","DiningTable","Spatula"],"template_variant_id":"ALF_C01_VISIBLE_SINGLE_CHOICE_v04","visible_categories":["Fridge","LettuceSliced"]},"gt_lineage":{"answer_margin":{"distractor_count":3,"visible_category_count":2},"difficulty":"simple","diffic...
How many visible objects are there, in what range should the overall count fall?
[ "A. 0-10", "B. 11-25", "C. 26-50", "D. 51 or more" ]
B
interval_choice
Instance count and set size
Interval Selection Question
medium
ALFRED_GT_OBJECTIVE_004195
spatial_0019
test
C02_COUNTING_SET_SIZE
interval_choice
ALF_C02_TOTAL_COUNT_INTERVAL
ALF_C02_TOTAL_COUNT_INTERVAL_v08
[ "A", "B", "C", "D" ]
[ "0-10", "11-25", "26-50", "51 or more" ]
{"A":"0-10","B":"11-25","C":"26-50","D":"51 or more"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
all visible instances counted from GT and mapped to interval
[ "observations.visibleObjects" ]
{"answer_evidence":{"interval":"11-25","template_variant_id":"ALF_C02_TOTAL_COUNT_INTERVAL_v08","total_visible_instances":14},"gt_lineage":{"answer_margin":{"interval":"11-25","total_visible_instances":14},"difficulty":"medium","difficulty_reason":"all visible instances counted from GT and mapped to interval","distract...
Label the object A which horizontal region of the image is it in?
[ "A. left side", "B. middle", "C. right side" ]
A
single_choice
image plane spatial relationships
question about selecting identified objects in the image
simple
ALFRED_GT_OBJECTIVE_000038
long_horizon_0011
test
C03_IMAGE_PLANE_SPATIAL
marked_object_choice
ALF_C03_REGION_CHOICE
ALF_C03_REGION_CHOICE_v01
[ "A", "B", "C" ]
[ "left side", "middle", "right side" ]
{"A":"left side","B":"middle","C":"right side"}
scalar
Accuracy
choice_accuracy
alfred_gt_objective_metrics.exact_match; VLMEvalKit ImageMCQ exact_matching-compatible
normalized predicted option letter equals gold option letter
true
single marked object projected to one of three coarse horizontal regions
[ "agent.position", "agent.rotation", "objects.position", "observations.visibleObjects" ]
{"answer_evidence":{"object_id":"Fridge|-02.80|+00.00|+04.43","projected_x":0.14868419342059785,"region":"left side","template_variant_id":"ALF_C03_REGION_CHOICE_v01"},"gt_lineage":{"answer_margin":{"projected_x":0.1487,"region":"left side"},"difficulty":"simple","difficulty_reason":"single marked object projected to o...
Arrange these labeled objects from closest to the camera to farthest.
[ "A. label object A", "B. label object B", "C. label object C", "D. Label object D" ]
["C","D","A","B"]
ordered_list
Three-dimensional distance and proximity
Sorting Question
hard
ALFRED_GT_OBJECTIVE_006077
long_horizon_0005
test
C04_DISTANCE_DEPTH
ordered_list
ALF_C04_DISTANCE_ORDER
ALF_C04_DISTANCE_ORDER_v05
[ "A", "B", "C", "D" ]
[ "label object A", "label object B", "label object C", "Label object D" ]
{"A":"label object A","B":"label object B","C":"label object C","D":"Label object D"}
list
Pairwise Accuracy
ordering
alfred_gt_objective_metrics.pairwise_accuracy
mean consistency over all ordered option-label pairs
true
full ordering over four marked objects by tracker distance
[ "objects.history.distance", "observations.visibleObjects" ]
{"answer_evidence":{"distance_by_option_m":{"A":0.9857383,"B":1.31967366,"C":0.488004416,"D":0.7289746},"template_variant_id":"ALF_C04_DISTANCE_ORDER_v05"},"gt_lineage":{"answer_margin":{"minimum_adjacent_distance_gap_m":0.241},"difficulty":"hard","difficulty_reason":"full ordering over four marked objects by tracker d...
End of preview. Expand in Data Studio

BenchForge ALFRED

Household activities and spatial understanding in ALFRED scenes.

The test split contains 10,000 English questions. The English edition preserves the original question IDs, option labels, answers, and scoring definitions. The main image and all referenced auxiliary images are embedded in the Parquet files as their original image bytes.

  • image: Main image supplied to the question.
  • auxiliary_images: Ordered list of additional images supplied with the question. An empty list means the task has no additional images.
  • question, choices, and answer: English question and answer choices, with the original option labels in the scoring answer.
  • data/test-00000-of-00001.jsonl: English text records with their original relative image paths for users of the existing benchmark layout.

Load the English test split using Hugging Face Datasets:

from datasets import load_dataset

dataset = load_dataset("jby20180901/BenchForge-alfred", split="test")

Use the Parquet edition to access images directly. For the JSONL edition, use the repository’s assets/ directory or extract assets.tar at the repository root if the dataset ships with an image archive.

Downloads last month
68