INDEX

Explanations

exploratory questions or considerations

The neuron detects cue words and verbs that signal recommendations or instructions (e.g. consider, must, propose, investigate).

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

帐户

-1.23

澳大利亚

-1.22

Föld

-1.20

 протяжении

-1.19

Morfologia

-1.18

 {}),

-1.16

㈦

-1.16

𝒙

-1.15

Mű

-1.15

pherals

-1.13

POSITIVE LOGITS

2.00

 where

1.56

 here

1.53

 that

1.48

 what

1.44

how

1.36

to

1.35

now

1.33

 other

1.24

why

1.24

Activations Density 0.027%