INDEX

Explanations

any or anything

The neuron is sensitive to the indefinite quantifier “any” (and its close variants like “anything”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

总是

-1.56

𓆏

-1.45

</h2>

-1.43

possibly

-1.43

hamed

-1.42

¨

-1.41

always

-1.39

´

-1.38

 realmente

-1.38

 素描

-1.38

POSITIVE LOGITS

 that

2.03

 from

1.86

1.63

 what

1.59

 with

1.49

 because

1.42

 There

1.41

 That

1.41

one

1.41

on

1.41

Activations Density 0.053%