INDEX

Explanations

data handling

This neuron responds to occurrences of the token “data.”

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

or

-1.14

 first

-1.13

if

-1.12

but

-1.10

แรก

-1.09

AN

-1.09

一刻

-1.08

-1.05

Răsp

-1.05

 pasaba

-1.04

POSITIVE LOGITS

data

1.45

 theſe

1.31

 défauts

1.29

 data

1.27

数据

1.27

álise

1.27

 semaines

1.23

lidos

1.23

erobic

1.20

ㅛ

1.19

Activations Density 0.026%