INDEX

Explanations

counting, quantifiers, totals

This neuron detects proper‐name entities—especially organization, publication, and institution names (often in all–caps or title case) and their associated acronyms.

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

or

0.69

 μια

0.64

或

0.62

({

0.61

 rumors

0.61

 или

0.60

(!

0.59

 או

0.58

หรือ

0.58

POSITIVE LOGITS

 valamint

0.85

 وذلك

0.82

 conformément

0.74

 artistes

0.71

 selaku

0.69

 എന്നീ

0.69

 physico

0.68

 समस्त

0.68

 ஆகியவற்ற

0.68

সকল

0.67

Activations Density 0.027%