INDEX

Explanations

title followed by statement or action

The neuron flags author‐byline and attribution phrases—i.e. verbs and cues that introduce who is saying or noting something.

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

。

-1.18

 Erinnerungen

-1.16

--)

-1.13

૧

-1.13

 (!__

-1.11

𝐢

-1.11

。”

-1.09

either

-1.09

。《

-1.07

-1.06

POSITIVE LOGITS

 because

1.38

who

1.34

Cuándo

1.19

 когда

1.13

örü

1.12

 уже

1.10

 when

1.09

わかります

1.08

mě

1.07

руют

1.07

Activations Density 0.127%