INDEX

Explanations

Hugo, Ernst, apk, priori

The neuron fires on the Latin phrase “a priori,” i.e. it detects occurrences of the token priori (as part of “a priori”).

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

-2.88

<bos>

-2.75

oader

-2.73

ма

-2.50

Its

-2.48

ですが

-2.48

骜

-2.39

icoot

-2.34

asiswa

-2.31

oncesto

-2.31

POSITIVE LOGITS

3.25

2.86

ate

2.80

2.78

2.53

 отличается

2.50

 poderosos

2.50

</strong>

2.50

家で

2.44

of

2.42

Activations Density 0.028%