INDEX

Explanations

d- prefix sequences

The neuron strongly activates on standalone sentence-ending periods (full stops).

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

嵛

-2.52

 interesante

-2.25

឵

-2.09

</sub>

-2.02

⎆

-2.02

</h3>

-2.00

vantaged

-1.98

งิน

-1.98

 distraught

-1.96

Ꭾ

-1.95

POSITIVE LOGITS

”,

2.28

\%,

2.19

2.13

鲆

2.03

翾

2.03

،

1.98

(),

1.95

ING

1.94

1.89

🥹

1.88

Activations Density 0.029%