INDEX

Explanations

sequence following events

The neuron fires on the final word just before a sentence- or clause-ending punctuation (the last token before a period or comma).

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

 also

-1.43

and

-1.30

 kalori

-1.23

-1.22

 tunik

-1.22

 these

-1.20

 babak

-1.17

aussi

-1.15

 και

-1.10

 周年

-1.07

POSITIVE LOGITS

 then

2.41

 سپس

1.67

 entonces

1.63

 Then

1.53

Then

1.52

 Entonces

1.42

 następnie

1.36

接下来

1.34

 násled

1.27

 eventually

1.27

Activations Density 0.046%