INDEX

Explanations

eventually took / sorting facilities / regulate redox

The neuron fires on semantically salient content words—especially verbs and other key terms that carry the main “information” in a sentence.

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

 epitom

-1.29

 Often

-1.27

 some

-1.21

 enhances

-1.20

 embracing

-1.20

 using

-1.16

 these

-1.16

 with

-1.15

 longstanding

-1.14

 comidas

-1.12

POSITIVE LOGITS

...............

1.46

 .........

1.38

 verschillende

1.37

蚨

1.33

婓

1.32

 ſie

1.30

 unbelievably

1.28

椟

1.27

 colossal

1.27

 différentes

1.27

Activations Density 0.049%