INDEX

Explanations

still continuing actions or states

The neuron fires strongly on the adverb “still” (and similar continuative/state-keeping words) wherever it appears.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 also

-2.23

 également

-1.91

 również

-1.88

 también

-1.66

 уже

-1.57

 already

-1.55

 גם

-1.54

 just

-1.53

 again

-1.50

 أيضا

-1.50

POSITIVE LOGITS

 ยัง

1.84

 remains

1.56

ing

1.45

STILL

1.41

remains

1.38

 trotz

1.37

稣

1.34

 retains

1.28

who

1.27

}>;

1.26

Activations Density 0.173%