INDEX

Explanations

sounds like or something from

This neuron detects hedging or simile lead-in phrases of the form “like something you might hear from …,” i.e. a comparative introspection (“like,” “something you might hear from,” etc.).

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

 télécharge

-1.39

 также

-1.20

 perpé

-1.17

 décider

-1.16

继续

-1.16

 podoba

-1.16

 constater

-1.15

 DIFFERENT

-1.13

 siè

-1.13

 fidé

-1.13

POSITIVE LOGITS

 only

1.56

all

1.37

if

1.34

 many

1.29

has

1.25

had

1.25

 some

1.21

at

1.19

 should

1.13

 even

1.10

Activations Density 0.079%