INDEX

Explanations

excluded, discriminated, leeway

This neuron responds strongly to isolated subword fragments produced by line breaks or hyphenation (i.e. word pieces split awkwardly across lines).

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

？』

-1.59

 They

-1.31

つまり

-1.27

 процес

-1.27

种类

-1.27

ワード

-1.22

There

-1.20

でしょうね

-1.17

 after

-1.16

 before

-1.16

POSITIVE LOGITS

了自己的

1.66

⮑

1.41

了自己

1.38

atized

1.30

了些

1.30

olving

1.28

topping

1.27

 piłkarz

1.26

семь

1.25

 semelh

1.24

Activations Density 0.007%