INDEX

Explanations

and

This neuron fires strongly on technical or domain-specific terminology—especially data-engineering jargon and formal documentation phrases.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 hanging

1.36

 plays

1.32

 sebagainya

1.23

 είναι

1.20

 เงี้ย

1.14

 explored

1.12

?,?,

1.12

 heritage

1.05

 पूरे

1.03

 takže

0.97

POSITIVE LOGITS

and

5.09

 અને

4.88

và

4.84

 ਅਤੇ

4.81

และ

4.78

 और

4.75

 आणि

4.75

 এবং

4.72

 και

4.69

и

4.66

Activations Density 13.153%