INDEX

Explanations

listing disparate items

The neuron is detecting numeric tokens—especially multi‐digit numbers like years or other numeric values.

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

-2.25

他有

-1.43

 вашего

-1.40

いました

-1.39

 имел

-1.33

此人

-1.27

 this

-1.26

 wohin

-1.22

いましたが

-1.21

 хорошего

-1.20

POSITIVE LOGITS

the

6.00

The

1.51

(≧

1.40

そして

1.39



1.31

chè

1.27

the

1.26

(￣

1.25

المراجع

1.23

(*´

1.21

Activations Density 0.189%