INDEX

Explanations

underestimate or discount importance

The neuron strongly activates on phrasal‐verb constructions made of a modifying particle plus a verb (e.g. “under-estimate,” “count-out,” “down-play”).

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

ʻ

-1.52

 Such

-1.34

¨

-1.31

 eſſ

-1.30

 πιο

-1.29

例えば

-1.28

 Already

-1.24

 Comparing

-1.21

 Previously

-1.21

 ſch

-1.21

POSITIVE LOGITS

 chances

1.98

 possibility

1.57

struktion

1.52

 septiembre

1.50

 Erscheinung

1.47

 espectacular

1.46

咴

1.41

場に

1.40

|<\

1.40

 coming

1.40

Activations Density 0.011%