INDEX

Explanations

calculate, count, determine

The neuron detects words and phrases that denote actively examining, inspecting, or evaluating something (e.g. look at, determine, analyze, evaluate, visually).

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

。

-1.40

-1.37

by

-1.27

 belle

-1.25

-1.17

city

-1.16

↵

-1.14

 لیے

-1.14

 mesta

-1.13

POSITIVE LOGITS

ሖ

1.38

 obicei

1.35

haviors

1.35

 thấm

1.35

 tendenza

1.33

 nový

1.32

hejiang

1.30

ついでに

1.29

するための

1.27

也會

1.23

Activations Density 0.061%