INDEX

Explanations

URLs and online content

The neuron fires on fragments of URLs and file paths—especially on dots and common file‐extension tokens (e.g. “.pdf,” “.jpg,” “/blob,” “/plain,” “HEAD”)—i.e. it detects URL/file‐path syntax.

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

 multiuso

-1.34

 Ibidem

-1.20

tableData

-1.20



-1.16

 produs

-1.15

 obfer

-1.15

 vectorstock

-1.13

 occafion

-1.12

 ingenieros

-1.12

 photolibrary

-1.10

POSITIVE LOGITS

 его

1.17

Proposition

1.07

nö

1.05

 jeho

1.04

},

1.04

now

1.03

Errorf

1.02

what

1.02

its

1.01

<0xEF>

1.00

Activations Density 0.050%