INDEX

Explanations

vers or words starting with Vers

The neuron consistently lights up on any token containing the substring “vers” (e.g. parts of “versions,” “versatility,” “Verschoor,” “VERSUS,” 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

”—

-2.52



-2.41

 rústica

-2.39

邸

-2.36

 colorida

-2.25

 respectivas

-2.23

毐

-2.22

 certificado

-2.20

眥

-2.20

”)

-2.17

POSITIVE LOGITS

vers

4.22

."

3.05

2.73

</h3>

2.72

ޏ

2.64

2.56

,"

2.47

where

2.25

-"

2.25

We

2.19

Activations Density 0.004%