INDEX

Explanations

punctuation followed by examples

The neuron responds to numerical and citation‐style tokens (e.g. section numbers, reference indices, numerical values, and related punctuation) typical of structured technical or academic text.

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

 this

-1.04

>?

-1.04

 samtidig

-0.96

odyear

-0.93

izowane

-0.91

 samtidigt

-0.90

ciss

-0.90

 only

-0.88

hythm

-0.88

argue

-0.88

POSITIVE LOGITS

这些

1.42

這些

1.27

 इन

1.22

都可以

1.21

These

1.17

 tất

1.11

 estos

1.10

Whatever

1.09

 هر

1.02

 biasanya

1.02

Activations Density 0.063%