INDEX

Explanations

for instance

The neuron strongly activates on linking words or short phrases that introduce examples or relationships in technical prose (e.g. “connection to,” “for instance”).

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

 both

-1.41

 what

-1.39

was

-1.38

 relatively

-1.35

 не

-1.32

 llamar

-1.32

 mostly

-1.30

 principio

-1.29

 could

-1.25

 diferente

-1.24

POSITIVE LOGITS

to

1.52

 denkt

1.32

indicates

1.27

盉

1.27

 BECOME

1.24

 PROPOSED

1.23

umblr

1.23

particularly

1.23

 dolci

1.22

 COME

1.20

Activations Density 0.161%