INDEX

Explanations

self-descriptor words

The neuron activates on compound adjectives beginning with the prefix “self-.”

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

 fermés

-1.01

 normaux

-0.93

hã

-0.93

 okam

-0.91

Senegal

-0.91

CHREIB

-0.90

 méret

-0.90

års

-0.90

 држа

-0.90

 stratég

-0.90

POSITIVE LOGITS

 self

1.59

 chosen

1.22

 imposed

1.13

 identification

1.05

 confidence

1.03

 criticism

1.01

ActionCreators

1.00

 motivated

1.00

 claimed

0.98

 perceived

0.98

Activations Density 0.010%