INDEX

Explanations

observing rules and limits

This neuron detects occurrences of the word “observe” (in its noun or gerund forms like “observance” or “observing”) indicating adherence to rules, rituals, or protocols.

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

ть

-1.54

 hinder

-1.53

lidae

-1.52

 Assessing

-1.50

rógeno

-1.49

蘼

-1.48

ources

-1.46

咿

-1.45

 considérons

-1.45

פות

-1.45

POSITIVE LOGITS

on

1.88

</td>

1.78

you

1.58

and

1.55

</h5>

1.51

</i>

1.48

if

1.47

 between

1.47

ஂ

1.45

1.43

Activations Density 0.037%