INDEX

Explanations

prepositions introducing context

This neuron fires strongly on common English function words—especially prepositions (e.g. “From,” “Under,” “On,” “By,” “With,” “At”) and articles (“a,” “the,” “of”)—i.e. it detects the beginnings of prepositional phrases.

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

もちろん

-1.27

 бампер

-1.23

やろ

-1.22

ตัวเอง

-1.09

內容簡介

-1.08

 tentunya

-1.05

 unangemess

-1.04

Ofic

-1.03

礎

-1.02

だけど

-1.01

POSITIVE LOGITS

 this

1.20

了些

1.16

 tisíc

1.13

 which

1.12

uitton

1.11

\&

1.10

をします

1.06

antwo

1.02

ごとの

1.00

𝙞

1.00

Activations Density 0.017%