INDEX

Explanations

actions leading to outcomes

The neuron fires on content words that describe actions, movements, or event transitions (e.g. verbs like “splits,” “head,” “led,” “doing,” and related particles like “off,” “to,” “for”).

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

γραμμα

-1.30

детский

-1.30

 corações

-1.27

Etimología

-1.27

показа

-1.27

abaikan

-1.24

venant

-1.23

嘤

-1.23

ޗ

-1.21

⢫

-1.20

POSITIVE LOGITS

 himself

1.98

its

1.55

his

1.49

 within

1.37

 herself

1.35

1.30

 сам

1.30

其

1.27

 every

1.26

in

1.24

Activations Density 0.250%