INDEX

Explanations

punctuation and symbols

The neuron is keyed to square-bracket delimiters, i.e. the literal “\[” and “\]” tokens in the text.

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

 эту

-2.13

微妙

-2.03

-1.98

這些

-1.94

 blijkt

-1.91

<bos>

-1.86

 entsprechenden

-1.86

 oplossing

-1.84

);

-1.83

 vollständ

-1.83

POSITIVE LOGITS

 signifikan

2.09

꜆

2.06

 kommer

2.05

 mutta

2.05

 solides

2.03

揼

1.96

ter

1.95

 конечно

1.91

錞

1.88

 raiſ

1.84

Activations Density 0.043%