INDEX

Explanations

foreign language words

The neuron fires most strongly on multi‐word verb constructions—especially auxiliary + participle or past‐tense verb phrases (e.g. “synced it to…,” “have been taught,” “would have given”).

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

 from

-1.61

 both

-1.33

 until

-1.30

 will

-1.29

one

-1.27

 again

-1.27

do

-1.24

to

-1.24

 where

-1.22

 after

-1.20

POSITIVE LOGITS

 Strasse

1.28

 fordi

1.27

yorum

1.25

 nyingi

1.20

uly

1.20

 senaste

1.20

 schuhe

1.19

ſus

1.19

 netts

1.17

 credere

1.17

Activations Density 0.215%