INDEX

Explanations

each time, each user, each instance

This neuron activates on inflected Russian words meaning “each” or “every.”

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

</strong>

-3.42

ar

-2.61

↵

-2.55

–

-2.45

他の

-2.42

</h3>

-2.39

↵↵

-2.33

<strong>

-2.30

 freundlichen

-2.20

POSITIVE LOGITS

縂

2.52

甡

2.44

vaders

2.41

鋮

2.39

لون

2.27

梣

2.20

forKey

2.20

titial

2.16



2.14

 psychi

2.13

Activations Density 0.002%