INDEX

Explanations

message in code contexts

The neuron activates on the literal token “Message,” i.e. occurrences of the word “Message.”

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

 possono

-2.45

 largely

-2.44

偏偏

-2.39

 overly

-2.30

and

-2.28

 singularly

-2.28

 dando

-2.27

ية

-2.25

 predominantly

-2.25

на

-2.23

POSITIVE LOGITS

<bos>

2.91

 intentar

2.63

wiście

2.58

したいと思います

2.48

_{

2.47

鞴

2.42

 parachoque

2.39

韪

2.39

啖

2.38

鉝

2.36

Activations Density 0.004%