INDEX

Explanations

used when describing

The neuron fires strongly on semantically contentful tokens (key nouns, verbs, and modifiers) rather than on common function words.

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

什么呢

-1.13

According

-1.08

 teneb

-1.06

 felici

-1.05

Ihe

-1.01

 любы

-1.00

 sali

-0.98

 verru

-0.98

 mits

-0.97

 FACULTY

-0.96

POSITIVE LOGITS

it

1.45

 werdet

1.22

 which

1.20

 reagieren

1.18

 ciertos

1.17

 głównie

1.11

 violencia

1.10

 kabát

1.09

 kré

1.08

imarães

1.07

Activations Density 0.059%