INDEX

Explanations

foreign words

The neuron fires on the character bigram “av,” i.e. it detects occurrences of the substring “av” within words.

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

 カットソー

-2.44



-2.38

<bos>

-2.22

oramic

-2.14

↵↵

-2.13

 unor

-2.09

 arduous

-2.09

 hoher

-2.09

mė

-2.08

噌

-2.05

POSITIVE LOGITS

 کرده

2.67

紲

2.63

 peran

2.53

緜

2.48

 penting

2.38



2.36

 ſont

2.34

 camiset

2.27

 Scritto

2.23

 kacang

2.22

Activations Density 0.001%