INDEX

Explanations

professionals asking questions

The neuron fires strongly on common grammatical “glue” tokens—high-frequency function words (e.g. for, and, of, the) and punctuation like commas.

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

 teenager

-1.27

locals

-1.25

 attacker

-1.14

Adults

-1.08

 insurer

-1.07

 believer

-1.05

titutions

-1.04

Customers

-1.03

 Businesses

-1.02

 établissement

-1.02

POSITIVE LOGITS

 female

1.16

多くの

1.15

 practicing

1.13

 successful

1.07

 polskich

1.06

 male

1.02

 those

0.98

 experienced

0.96

 erfolgre

0.92

广大

0.92

Activations Density 0.243%