INDEX

Explanations

string replace and trim

The neuron responds to occurrences of “replace” operations in code (e.g. calls to a replace function or method).

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

 such

-1.59

such

-1.55

but

-1.39

dê

-1.27

 城市

-1.26

驶

-1.20

 same

-1.20

<bos>

-1.20

 союз

-1.19

nikon

-1.18

POSITIVE LOGITS

keteers

1.43

 вместо

1.38

 instead

1.22

has

1.21

Теперь

1.20

animity

1.20

otry

1.19

anía

1.18

ématiques

1.18

 happened

1.17

Activations Density 0.007%