INDEX

Explanations

names of places and institutions

The neuron fires strongly on multi-token named entities (proper names of institutions, companies, places, events, etc.).

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

的話

-1.23

它

-1.21

 экономи

-1.20

 delicado

-1.16

 espirituales

-1.15

Ỷ

-1.13

摂取

-1.11

 offenbar

-1.10

ぷ

-1.10

美味しいです

-1.09

POSITIVE LOGITS

 which

1.75

 where

1.31

 mostrarán

1.27

 TÉCNICA

1.27

 również

1.21

 célè

1.20

 beliau

1.18

 rése

1.17

idable

1.16

ū

1.11

Activations Density 0.068%