INDEX

Explanations

recommendations and advice

The neuron detects occurrences of the word “recommend,” i.e. recommendation language.

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

mathrm

-2.58

-2.48

The

-2.44

 With

-2.33

has

-2.17

 That

-2.17

年来

-2.17

主要是

-2.16

 sometimes

-2.16

}$.

-2.14

POSITIVE LOGITS

蚡

2.48

鹀

2.44

 pérd

2.33

 WOMAN

2.33

阏

2.25

 musée

2.25

敺

2.25

 incrí

2.22

 レスト

2.20

 parlement

2.19

Activations Density 0.004%