INDEX

Explanations

articles, products, or items followed by descriptive words

The neuron strongly fires on the key subject‐defining nouns in a technical or how-to question, i.e. the main object or product name that the question is about.

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

 agarre

-1.02

 femininas

-0.94

 tejidos

-0.94

 Krank

-0.94

 feminina

-0.94

Dé

-0.93

kých

-0.92

varphi

-0.91

 گوشت

-0.89

 decirle

-0.89

POSITIVE LOGITS

 from

2.77

for

2.63

to

2.59

by

1.47

เพื่อ

1.29

 для

1.21

via

1.14

 чтобы

1.13

 that

1.08

 από

1.08

Activations Density 0.071%