INDEX

Explanations

numbers and scores

This neuron detects mentions of team win‐loss records or “winless” streaks (i.e. parts of phrases giving a team’s sporting record).

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

 bereits

-1.25

 that

-1.24

to

-1.03

-1.02

 völlig

-1.01

ุป

-1.00

[{

-0.95

与此同时

-0.93

 plough

-0.92

 właśnie

-0.92

POSITIVE LOGITS

 lác

1.27

🪖

1.05

Ă

1.05

>>,

1.02

 nélk

1.01

 ventes

1.00

couch

1.00

 équipements

1.00

ATOM

0.99

 felb

0.98

Activations Density 0.012%