1
Your word
measured
“the telephone is an annoying interruption”
Definitions come from WordNet, a hand-curated dictionary.
some abrupt occurrence that interrupts an ongoing activity
Panels 3, 4 and 5 are computed per model. Switch to see them disagree.
“the telephone is an annoying interruption”
Definitions come from WordNet, a hand-curated dictionary.
This is the tokenizer splitting text, not the model understanding it.
make is the dictionary opposite of breaks, yet this model puts it closer than 37 of the 42 synonyms it can score. Opposites share the sentences a word lives in, so cosine alone cannot tell them apart - which is why the dictionary seeds this list rather than the geometry.
Shade shows how close the model puts each word to breaks, 2 of 8 also appear in the dictionary. 42 of 100 dictionary synonyms are in this build; the rest have no vector to compare yet.
keep is the dictionary opposite of breaks, yet this model puts it closer than 34 of the 42 synonyms it can score. Opposites share the sentences a word lives in, so cosine alone cannot tell them apart - which is why the dictionary seeds this list rather than the geometry.
Shade shows how close the model puts each word to breaks, 2 of 8 also appear in the dictionary. 42 of 100 dictionary synonyms are in this build; the rest have no vector to compare yet.
repair is the dictionary opposite of breaks, yet this model puts it closer than 33 of the 42 synonyms it can score. Opposites share the sentences a word lives in, so cosine alone cannot tell them apart - which is why the dictionary seeds this list rather than the geometry.
Shade shows how close the model puts each word to breaks, 3 of 8 also appear in the dictionary. 42 of 100 dictionary synonyms are in this build; the rest have no vector to compare yet.
Model neighbours are distributional, not dictionary synonyms. Two words can be close because they appear in similar sentences, which is why an antonym can outscore a synonym here.
Scores are projections onto axes we defined from anchor words, not labels the model assigns.
These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to breaks. That gap is the model’s own learned association.
These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to breaks. That gap is the model’s own learned association.
These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to breaks. That gap is the model’s own learned association.
These are statistical associations in the training data, not the model thinking.