INDEX

Explanations

United States, numbers, and letters

The neuron is mainly triggered by precise numerical tokens, especially floating-point numbers.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 lemmas

0.63

 things

0.61

 resolvers

0.60

 arousal

0.59

 tinct

0.57

 subjekt

0.57

 lexical

0.56

 procurar

0.56

 NaCl

0.56

 generalizations

0.56

POSITIVE LOGITS

美国

0.57

 statunitense

0.56

美國

0.54

neapolis

0.52

éroport

0.52

 britannique

0.52

 미국

0.52

 británico

0.52

 Londres

0.50

 américaine

0.50

Activations Density 0.001%