INDEX

Explanations

The neuron strongly activates on the follow-up word “so” (and similarly “yes”) in conditional question clauses like “If so, …” or “If yes, …,” marking that pivot in a question.

if yes, how

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

مشخصات

-1.49

 mostrarán

-1.29

 penyakit

-1.22

معرفی

-1.21

توضیحات

-1.19

)}>

-1.14

 perasaan

-1.11

)?;

-1.11

;$

-1.10

échange

-1.10

POSITIVE LOGITS

 then

2.97

 Then

1.91

1.80

Then

1.77

why

1.67

 entonces

1.63

的话

1.60

 thì

1.59

 então

1.57

then

1.50

Activations Density 0.012%