INDEX

Explanations

wait followed by question or doubt

This neuron detects moments where the writer interjects with a surprised “Wait” (often followed by a comma and a question or clarification).

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

 néanmoins

-1.45

-1.29

 Verpflich

-1.27

-1.25

 étudié

-1.24

 certainement

-1.23

 conservé

-1.23

EVERY

-1.23

Лу

-1.22

 toutefois

-1.21

POSITIVE LOGITS

 this

1.71

 because

1.47

 their

1.41

 zafira

1.38

 Venedig

1.33

 whose

1.31

 before

1.30

thenburg

1.27

 without

1.26

 numerous

1.26

Activations Density 0.008%