INDEX

Explanations

and then

This neuron fires on words that are acting as sentence‐subject references to people—especially personal pronouns (“I,” “you,” “we,” “they”) and similar group nouns (e.g. “teenagers”).

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

was

-3.00

 đã

-1.84

 kellett

-1.80

 telah

-1.49

 sarà

-1.45

 было

-1.42

 heeft

-1.38

 jsme

-1.37

 proporcionan

-1.36

 admiten

-1.35

POSITIVE LOGITS

get

1.93

 becomes

1.90

 gets

1.85

 make

1.83

 goes

1.82

 κάν

1.80

try

1.76

 take

1.72

 होती

1.70

 decide

1.69

Activations Density 0.349%