INDEX

Explanations

human related suffixes

The main thing this neuron does is detect occurrences of the substring “human” (e.g. in “human,” “humanist,” “humanized”).

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

羕

-2.39

‪.‬‬

-2.38

 Anſ

-2.30

sprzedam

-2.23

་་

-2.23

bluza

-2.22

CurtirCurtir

-2.20

伋

-2.14

").

-2.14



-2.09

POSITIVE LOGITS

</b>

2.59

<b>

2.42

2.22

↵↵

2.14

with

2.05

2.03

1.98

<u>

1.96

or

1.92

1.81

Activations Density 0.006%