INDEX

Explanations

general article

The neuron spotlights attributive adjectives (e.g. “broader,” “open,” “central,” “potential”) that modify and qualify key noun phrases.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 these

-1.51

 while

-1.47

そのため

-1.23

は

-1.23

 during

-1.21

of

-1.21

 previous

-1.21

子は

-1.19

子が

-1.16

 recent

-1.13

POSITIVE LOGITS

様な

1.45

ところに

1.25

millimeters

1.18

)}$,

1.17

 regeringen

1.16

 років

1.10

uxxxx

1.09

 userDetails

1.09

いくつか

1.08

ɓ

1.08

Activations Density 0.096%