INDEX

Explanations

version: or value:

The neuron strongly responds to markup delimiters and element names in XML‐style (and similar) configuration files—that is, it detects when you’re inside “<…>” tags.

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

tando

-1.39

瘢

-1.25

 全国

-1.25

wanne

-1.22

tür

-1.17

Odpo

-1.16

 Bakter

-1.16

美味しいです

-1.14

 Konkur

-1.11

 oči

-1.11

POSITIVE LOGITS

1.42

1.20

="

1.18

'.

1.15

 nawet

1.13

\_

1.11

1.10

 aktiviert

1.09

뭄

1.09

Activations Density 0.107%