INDEX

Explanations

possible to

This neuron detects words and phrases that express possibility or impossibility (e.g. “possible to,” “impossible to,” “не получается”).

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

руются

-1.45

\'

-1.34

</em>

-1.29

-1.26

</b>

-1.22

-1.20

 στα

-1.18

ائے

-1.18

``

-1.15

ärsk

-1.15

POSITIVE LOGITS

 prácticamente

1.42

 کیلومتر

1.26

 vorkommen

1.22

 малю

1.18

 dichas

1.16

 powiedzieć

1.13

breviations

1.12

 modded

1.10

 hlavně

1.09

従業

1.09

Activations Density 0.115%