INDEX

Explanations

Nick SabanQ:

This neuron fires on the very first token of a new paragraph or code block (i.e. the initial word or marker beginning a text chunk).

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

 that

-1.55

to

-1.45

or

-1.25

[{{

-1.10

 powiedzieć

-1.09

 penonton

-1.07

 dapatkan

-1.04

ceuticals

-1.04

kym

-1.03

🇿

-1.02

POSITIVE LOGITS

།

1.01

 opciones

0.98

Nasc

0.96

sizePolicy

0.96

では

0.96

でも

0.94

 extrem

0.93

人気

0.92

シャツ

0.91

 erbjud

0.90

Activations Density 0.000%