INDEX

Explanations

introducing examples

This neuron detects when the text uses placeholder/example signals (especially the word “something” and “like”) to introduce an example or illustrative phrase.

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

↵↵↵↵↵↵↵↵↵↵

-1.47

↵↵↵↵↵↵↵↵↵↵↵

-1.46

-1.45

</h4>

-1.41

↵↵↵↵↵↵↵↵↵↵↵↵

-1.39

 dessas

-1.37

 ftate

-1.37

↵↵↵↵↵↵↵

-1.36

melons

-1.36

↵↵↵

-1.34

POSITIVE LOGITS

“

1.59

“

1.45

1.37

,:);

1.34

 after

1.33

 gefe

1.33

 einwand

1.30

 newly

1.28

out

1.23

 erwor

1.23

Activations Density 0.102%