How AI reads conk | LLMsaurus

1 Your word measured

noun 5 senses 21 synonyms 1 antonyms

Definitions come from WordNet, a hand-curated dictionary.

2 To an AI, it's pieces measured

DeepSeek V3 con k 2 tokens

Mistral 7B con k 2 tokens

Qwen3 8B con k 2 tokens

This is the tokenizer splitting text, not the model understanding it.

3 Synonyms, antonyms, and the AI measured

Qwen3 8B

DeepSeek V3

Mistral 7B

This model splits conk 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.

The model’s neighbours

con0.74 congo0.58 convection0.58 concha0.58 confluence0.57 conformance0.57 conduction0.57 conrad0.56

Dictionary synonyms measured

die0.07 choke0.04 pass0.03 exit0.02 stall0.02 faint0.01 expire-0.03 decease perish go pass away kick the bucket cash in one's chips buy the farm give-up the ghost drop dead pop off croak snuff it swoon pass out

Dictionary antonyms measured

be born

Shade shows how close the model puts each word to conk. 7 of 21 dictionary synonyms are in this build; the rest have no vector to compare yet.

This model splits conk 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.

The model’s neighbours

con0.82 congo0.74 enkephalin0.71 enl0.70 atp0.66 anon0.66 ats0.66 comforter0.64

Dictionary synonyms measured

faint0.03 choke0.01 stall-0.01 expire-0.02 exit-0.02 pass-0.03 die-0.23 decease perish go pass away kick the bucket cash in one's chips buy the farm give-up the ghost drop dead pop off croak snuff it swoon pass out

Dictionary antonyms measured

be born

Shade shows how close the model puts each word to conk. 7 of 21 dictionary synonyms are in this build; the rest have no vector to compare yet.

This model splits conk 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.

The model’s neighbours

con0.81 cones0.73 conto0.67 congo0.64 concha0.61 conrad0.59 conquer0.59 conception0.58

Dictionary synonyms measured

choke0.15 expire0.11 stall0.10 pass0.07 die0.04 faint0.03 exit0.02 decease perish go pass away kick the bucket cash in one's chips buy the farm give-up the ghost drop dead pop off croak snuff it swoon pass out

Dictionary antonyms measured

be born

Shade shows how close the model puts each word to conk. 7 of 21 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.

4 Its personality measured

Qwen3 8B

DeepSeek V3

Mistral 7B

This model splits conk 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.

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

This model splits conk 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.

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

This model splits conk 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.

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

Scores are projections onto axes we defined from anchor words, not labels the model assigns.

5 Surprising neighbours measured

Qwen3 8B

DeepSeek V3

Mistral 7B

This model splits conk 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.

conodont0.561 lek0.546 murk0.533 polk0.529 dirk0.528

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to conk. That gap is the model’s own learned association.

This model splits conk 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.

classy0.633 helpdesk0.628 dui0.627 mym0.626 atar0.619

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to conk. That gap is the model’s own learned association.

This model splits conk 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.

concurrent0.580 conforms0.578 conferences0.577 conquest0.557 converse0.557

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to conk. That gap is the model’s own learned association.

These are statistical associations in the training data, not the model thinking.