INDEX

Explanations

available, access, opportunities, options

The neuron fires on content-bearing, low-frequency (“high information”) tokens—typically technical terms, unusual proper nouns, or other topic-specific vocabulary.

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

ה

1.57

{@

1.27

€™

1.24

с

1.23

()=>{

1.22

佟

1.21

אם

1.17

으로부터

1.17

专注

1.16

{$

1.13

POSITIVE LOGITS

AVAILABLE

1.68

र

1.33

ר

1.33

 AVAILABLE

1.31

 disponível

1.30

 Hanya

1.23

az

1.22

 Пі

1.21

 alterna

1.21

prima

1.20

Activations Density 5.620%