INDEX

Explanations

access via a service

The neuron strongly activates on the word “via,” i.e. occurrences of that preposition indicating method or channel.

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

や

-1.34

用来

-1.26

 porción

-1.22

椋

-1.20

浒

-1.19

についても

-1.17

🧌

-1.16

 based

-1.14

歘

-1.11

🫃

-1.11

POSITIVE LOGITS

the

2.08

 their

2.02

1.93

an

1.92

our

1.64

its

1.45

After

1.42

During

1.40

Through

1.37

ええ

1.31

Activations Density 0.052%