INDEX

Explanations

causes and reasons

This neuron activates on occurrences of the word “reason” (and its plural form “reasons”) in the text.

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

⌯

-1.55

летия

-1.55

蟄

-1.50

所で

-1.41

ſelves

-1.41

 bermanfaat

-1.39

噚

-1.39

 essenciais

-1.38

ッキリ

-1.37

 ところ

-1.37

POSITIVE LOGITS

 reason

1.74

 می‌توان

1.66

<td>

1.62

1.52

时候

1.48

 shouldn

1.41

 ability

1.40

 chance

1.38

 мою

1.35

 only

1.33

Activations Density 0.018%