Displayed are triplets annotated with human referential utterances. A neural listener, trained with chairs, is used to color-code attention-wise the utterances and score each object. Underlined tokens where out-of-the-vocabulary and were ignored by the listener.
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker:
a round table top supported by four legs possibly Correctly guessed: True Confidences: 0.05, 0.02, 0.93 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker: has four legs , circular Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: circular table Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker:
round table with four legs that curl upwards Correctly guessed: True Confidences: 0.02, 0.18, 0.79 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: round Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker: modern and round Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: legs bend Correctly guessed: False Confidences: 0.68, 0.17, 0.15 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker: the table top is round Correctly guessed: True Confidences: 0.00, 0.01, 0.98 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker:
a round table that looks like an ash tray Correctly guessed: True Confidences: 0.02, 0.08, 0.90 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker: round table with four legs Correctly guessed: True Confidences: 0.01, 0.01, 0.98 |
||