INDEX

Explanations

whose possession or descriptor

The neuron activates on the relative pronoun “whose,” spotting possessive relative clauses.

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

骘

-3.00

ቍ

-2.64

鈊

-2.36

脔

-2.28

咘

-2.22

");

-2.17

阄

-2.13

enchymal

-2.13

衎

-2.11

 سایر

-2.05

POSITIVE LOGITS

 endeavors

2.67

нутрен

2.55

(?)

2.52

‘

2.42

 endeavours

2.36

стреча

2.33

 また

2.27

is

2.27

ほうが

2.17

al

2.16

Activations Density 0.006%