INDEX

Explanations

lists, resolution, attachment

The neuron fires strongly on content-heavy nouns—especially proper names and technical terms.

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

and

-1.34

 naszej

-1.29

<bos>

-1.27

 geïsole

-1.26

🫥

-1.18

 sterke

-1.17

Herzlich

-1.15

 våra

-1.15

。”

-1.13

 sentiers

-1.10

POSITIVE LOGITS

<td>

1.54

 while

1.45

 pude

1.41

VolleyError

1.40

件事情

1.36

iqué

1.36

<strong>

1.34

),

1.32

to

1.27

 include

1.27

Activations Density 0.104%