INDEX

Explanations

reading and readers

The neuron activates on occurrences of the word “read” (in its various forms, e.g. “read,” “reading”).

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

’

-2.75

-2.20

),

-2.08

 лучше

-1.88

 neuen

-1.85

How

-1.83

-1.81

 Typically

-1.81

两个

-1.78

’,

-1.76

POSITIVE LOGITS

猡

2.44

鉺

2.17

 سامسون

2.08

驒

1.98

鷥

1.98

 zima

1.97

 speel

1.88

 giorgio

1.87

 読む

1.84

 haup

1.81

Activations Density 0.056%