INDEX

Explanations

quality of

The main thing this neuron does is detect occurrences of the word “quality,” especially in contexts like “high quality.”

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

をします

-1.54

睒

-1.41

閦

-1.34

nitř

-1.33

ジョー

-1.32

襌

-1.30

 spesielt

-1.29

Mengapa

-1.28

をしました

-1.28

 consejos

-1.28

POSITIVE LOGITS

and

2.59

to

1.52

1.49

 perdana

1.29

\&

1.29

<strong>

1.28

1.26

of

1.25

🌟

1.24

 yakni

1.23

Activations Density 0.026%