INDEX

Explanations

discussion or talk

The neuron fires on verbs and phrases indicating discussion or conversation (e.g. “discussed,” “talked about,” “we discussed”).

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

to

-1.73

 altri

-1.45

of

-1.41

 according

-1.40

 mentre

-1.39

-1.38

 jetzt

-1.34

 from

-1.33

you

-1.30

 самому

-1.28

POSITIVE LOGITS

 kteří

1.75

。」

1.66

鹮

1.61

 ktorí

1.53

 personlig

1.45

酈

1.41

sandalia

1.38

卌

1.36

 něko

1.34

赑

1.33

Activations Density 0.025%