1
Your word
measured
“The amount of homework decreased towards the end of the semester”
Definitions come from WordNet, a hand-curated dictionary.
decrease in size, extent, or range
下がる is Japanese for decreased, and every measurement below was made on that English word. Definitions come from English WordNet; Open Multilingual WordNet supplies the Japanese headword, not a Japanese definition.
Panels 3, 4 and 5 are computed per model. Switch to see them disagree.
“The amount of homework decreased towards the end of the semester”
Definitions come from WordNet, a hand-curated dictionary.
This panel splits 下がる itself, not decreased — a tokenizer does not care what language it is fed. It is the only panel here that does.
This is the tokenizer splitting text, not the model understanding it.
increased is the dictionary opposite of decreased, yet this model puts it closer than every one of the 4 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 decreased, 2 of 8 also appear in the dictionary. 4 of 6 dictionary synonyms are in this build; the rest have no vector to compare yet.
increased is the dictionary opposite of decreased, yet this model puts it closer than 3 of the 4 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 decreased, 2 of 8 also appear in the dictionary. 4 of 6 dictionary synonyms are in this build; the rest have no vector to compare yet.
This model splits decreased into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.
increased is the dictionary opposite of decreased, yet this model puts it closer than 3 of the 4 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 decreased. 4 of 6 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.
This model splits decreased into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.
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 decreased. 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 decreased. That gap is the model’s own learned association.
This model splits decreased into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.
These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to decreased. That gap is the model’s own learned association.
These are statistical associations in the training data, not the model thinking.