INDEX

Explanations

seal, seals, sealed

The main thing this neuron does is detect technical terms related to seals and sealing (e.g., “seal,” “sealing,” “sealed”) in mechanical or engineering contexts.

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

dáms

-1.38

廨

-1.34

籏

-1.31

 umožňuje

-1.30

ઁ

-1.30

 現貨

-1.29

 nabízí

-1.28

注意事

-1.27

 Hinblick

-1.25

炱

-1.24

POSITIVE LOGITS

 that

1.42

 which

1.33

」

1.27

oak

1.26

1.21

 experience

1.19

he

1.18

="">

1.16

”.

1.12

 contact

1.10

Activations Density 0.059%