INDEX

Explanations

adjective and adverb suffixes

The neuron activates primarily on short tokens that are common English word-ending suffixes (e.g. “ous,” “ed,” “ing,” “ically”), flagging morphological endings.

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

 novos

-1.25

将她

-1.24

 historischen

-1.22

‌

-1.21

低头

-1.19

􀂃

-1.17

 arī

-1.15



-1.14

เติ

-1.13

UALLY

-1.12

POSITIVE LOGITS

したのが

1.39

鱻

1.34

 Arzneimit

1.33

criminator

1.32

عية

1.29

Wondering

1.27

Until

1.24

’

1.23

盘

1.21

鏑

1.20

Activations Density 0.103%