INDEX

Explanations

Kalman filter configurations

The neuron fires on mentions of advanced state‐estimation and Bayesian‐filtering methods (e.g. “Kalman filter,” “Extended Kalman Filter,” “SLAM,” “Bayesian filtering”).

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

 cojín

-1.27

 geográfica

-1.26

觸

-1.25

 appels

-1.23

トナム

-1.23

 chambres

-1.23

 auriculares

-1.21

 brazo

-1.19

 hospod

-1.19

 店舗

-1.18

POSITIVE LOGITS

৯

1.45

 Possibly

1.43

via

1.43

園

1.37

 With

1.35

Now

1.35

也

1.28

 tappa

1.27

 until

1.24

 Either

1.22

Activations Density 0.044%