INDEX

Explanations

as it were

The neuron fires strongly on first- and second-person pronouns (e.g. “I,” “you”) and related small function words in those pronoun-led phrases.

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

 even

-0.98

 couldn

-0.84

 every

-0.82

_<

-0.80

 inwest

-0.77

никюр

-0.74

 maybe

-0.72

 incluso

-0.72

ButtonClicked

-0.71

pezif

-0.71

POSITIVE LOGITS

 were

2.56

was

1.83

 كانت

1.52

 были

1.47

Were

1.43

were

1.38

 بودم

1.34

 كان

1.26

 Were

1.24

 थे

1.20

Activations Density 0.026%