INDEX

Explanations

programming languages and software

The neuron is sensitive to the first “headline” word or title token at the start of each block (e.g. section headers or question titles).

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

and

-1.55

-1.40

-1.34

льное

-1.33

𝘛

-1.30

}=\

-1.27

ㅂ

-1.22

 both

-1.22

«

-1.20

 giang

-1.18

POSITIVE LOGITS

ar

1.49

 erzählen

1.34

ΠΟ

1.25

heute

1.24

viembre

1.24

Ekster

1.23

 sogenannten

1.22

㐂

1.22

每一个

1.21

danke

1.20

Activations Density 0.411%