INDEX

Explanations

good choice or way

This neuron detects words that signal requests for recommendations or advice—e.g. “best,” “good,” “way,” “place,” “choice,” or “starting point.”

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

 beſt

-1.55

 différents

-1.49

埛

-1.48

尨

-1.48

 acclaimed

-1.48

BACA

-1.47

纐

-1.44

፨

-1.43

後記

-1.42

</strong>

-1.41

POSITIVE LOGITS

to

2.28

for

1.88

has

1.76

1.58

 before

1.58

 While

1.57

 because

1.48

 Having

1.40

 Using

1.39

 should

1.38

Activations Density 0.118%