INDEX

Explanations

prepositions indicating origin or affiliation

The neuron primarily picks out numeric tokens (numbers or identifiers containing digits) in the code.

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

for

-1.60

and

-1.48

！！！！！

-1.39

in

-1.38

 with

-1.37

与此同时

-1.30

を

-1.28

！！！！

-1.25

on

-1.23

～！

-1.17

POSITIVE LOGITS

 edades

1.56

んですか

1.55

感じが

1.55

感じです

1.53

 prévu

1.50

 opak

1.48

 entschieden

1.44

んだよね

1.44

 étrangère

1.43

 odkry

1.39

Activations Density 0.021%