INDEX

Explanations

positive sentiment phrases

The neuron fires on sentence‐ or segment‐initial words (i.e. the first capitalized token at the start of a sentence or line).

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

 atve

-1.16

嚆

-1.09

 abito

-1.07

vård

-1.07

Wert

-1.05

zni

-1.04

晔

-1.03

月中

-0.97

ₒ

-0.96

tambah

-0.96

POSITIVE LOGITS

 also

1.27

1.21

 inoltre

0.96

他也

0.95

dibly

0.93

 начинает

0.91

0.88

อย่างไร

0.86

bolite

0.84

Activations Density 0.012%