INDEX

Explanations

directions followed by street names

The neuron strongly activates on tokens used in postal or location references—street numbers, cardinal‐direction abbreviations (NE, SE), room numbers, etc.

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

 weitere

-1.41

鸲

-1.37

 també

-1.37

 altres

-1.33

цима

-1.31

 considerando

-1.30

({

-1.28

!”

-1.27

烀

-1.25

onsors

-1.23

POSITIVE LOGITS

 from

1.73

of

1.43

 utterly

1.42

 which

1.36

 with

1.31

at

1.27

 obuv

1.25

 wholly

1.24

for

1.23

 through

1.22

Activations Density 0.024%