INDEX

Explanations

question imply

This neuron detects logical‐implication markers (e.g. “implies,” “Longrightarrow,” etc.) in mathematical expressions.

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

蹰

-1.57

 PUSTAKA

-1.47

Ivoire

-1.44

 tetap

-1.38

eeq

-1.35

lawesi

-1.31

<bos>

-1.28

gnose

-1.26

ürgen

-1.25

Revenir

-1.25

POSITIVE LOGITS

1.80

’

1.70

 there

1.67

1.50

1.41

 medica

1.39

1.38

1.37

綃

1.36

Дорогие

1.32

Activations Density 0.001%