INDEX

Explanations

ask parents that

The neuron strongly activates on the main topical or content words that typically appear in the leading sentence of a new section or topic.

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

我们

-1.36

 doulou

-1.27

іє

-1.24

 identifiant

-1.16

更に

-1.13

 デザイン

-1.13

ﭽ

-1.12

aconda

-1.11

 searched

-1.10

のほうが

-1.09

POSITIVE LOGITS

my

2.59

 this

2.09

 myself

1.82

and

1.52

1.31

 that

1.27

 aceea

1.23

 этого

1.19

 lavoratori

1.17

 például

1.15

Activations Density 0.451%