INDEX

Explanations

removing specific data elements

This neuron fires on the key noun following “remove,” i.e. the thing being deleted (parameters, headers, numbers, rows, 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

 this

-1.67

 unrelenting

-1.32

Regardless

-1.30

 these

-1.30

With

-1.27

谘

-1.27

钲

-1.24

薗

-1.23

 Regardless

-1.23

When

-1.23

POSITIVE LOGITS

to

1.52

 from

1.43

for

1.38

by

1.27

 about

1.27

1.25

 gelungen

1.20

 Beleuchtung

1.20

 nowy

1.18

 Remove

1.17

Activations Density 0.122%