INDEX

Explanations

would say that or argue

This neuron detects conversational reporting verbs and pronouns in opinion‐framing phrases like “would say that…” or “…tell you.”

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

in

-1.67

so

-1.49

↵↵↵

-1.48

 jeune

-1.43

ранее

-1.34

 would

-1.30

ellants

-1.28

tifkan

-1.28

ação

-1.26

紹

-1.26

POSITIVE LOGITS

and

1.41

ዡ

1.38

anu

1.38

 moteurs

1.37

 biet

1.34

lho

1.33

 Mungkin

1.31

そろそろ

1.30

 gewor

1.28

常常

1.27

Activations Density 0.015%