INDEX

Explanations

parenthetical speakers or 'I' statements

This neuron detects dialogue speaker labels—specifically, names (or labels) followed by a colon as used in transcripts or script‐style text.

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

看了一眼

-1.26

看了看

-1.25

$-$

-1.15

 seamlessly

-1.13

<td>

-1.13

-1.12

 today

-1.12

ramiento

-1.10

 carretera

-1.10

 this

-1.10

POSITIVE LOGITS

 eerste

1.29

Эти

1.24

詬

1.23

 first

1.16

 maart

1.14

いただきたい

1.14

otro

1.13

mita

1.13

 niečo

1.13

まず

1.12

Activations Density 0.051%