INDEX

Explanations

in/for/to [word]

The neuron fires on the main “head” word of a clause (the key predicate or topic noun) – typically the content word immediately following auxiliaries like be/have or marking the sentence’s central concept.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 Algunos

-1.34

 Algunas

-1.32

rightarrow

-1.26

-1.23

 یکی

-1.23

 Allí

-1.21

 Tampoco

-1.20

 Dijo

-1.19

 právě

-1.15

 الأكثر

-1.15

POSITIVE LOGITS

it

1.63

</i>

1.37

缺点

1.35

 küm

1.30

şam

1.28

cipit

1.26

 time

1.25

龱

1.24

Huhu

1.23

产生了

1.21

Activations Density 0.211%