INDEX

Explanations

river references

This neuron fires on geographic place‐name tokens—especially water features like “River,” “Reservoir,” 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

.”

-2.11

場所は

-1.64

这一刻

-1.63

一方で

-1.57

也沒

-1.56

According

-1.56

 changed

-1.50

墻

-1.46

 also

-1.45

賬

-1.45

POSITIVE LOGITS

unuz

1.59

 لیے

1.58

to

1.55

螣

1.54

ཥ

1.46

 allá

1.45

 naveg

1.45

 stille

1.42

↵↵

1.41

葙

1.40

Activations Density 0.084%