INDEX

Explanations

code testing

The neuron strongly activates on the literal string “test” (e.g. in function names, script keys, JSON fields, 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

ふん

-1.47

 简介

-1.37

 客户

-1.34

Questão

-1.33

 介绍

-1.27

 gmin

-1.23

 sánh

-1.22

 diſ

-1.22

って

-1.21

epy

-1.20

POSITIVE LOGITS

 with

1.45

for

1.34

 使っ

1.33

test

1.30

 like

1.29

 through

1.29

код

1.29

mètres

1.29

飾

1.28



1.28

Activations Density 0.062%