INDEX

Explanations

easy to do or use

The neuron specifically fires on the word “to” when it follows “easy,” i.e. on instances of the phrase “easy to.”

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

↵↵↵

-2.00

-1.87

 ‘‘

-1.84

-1.77

知道的

-1.77

必要がある

-1.74

めっちゃ

-1.73

 ermöglicht

-1.68

蹩

-1.67

previously

-1.66

POSITIVE LOGITS

 with

1.92

and

1.80

鏇

1.73

 Teknologi

1.73

ﺅ

1.70

 Mudah

1.70

 will

1.67

 sorella

1.66

 ανθρώ

1.66

rouw

1.63

Activations Density 0.017%