INDEX

Explanations

forum discussions and features

The neuron primarily detects mentions of an online forum (i.e. occurrences of the token “forum”).

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

⚣

-2.48

からお

-2.34

潆

-2.31

訁

-2.30

･･

-2.25

してます

-2.22

genodigd

-2.22

 lecciones

-2.19

 trifft

-2.14

谞

-2.14

POSITIVE LOGITS

</strong>

3.00

ar

2.91

2.80

<i>

2.77

餮

2.33

the

2.30

 HAPPENED

2.27

2.25

to

2.23

 …”

2.23

Activations Density 0.007%