INDEX

Explanations

the same

The neuron is highly tuned to detect the word “work,” especially when it appears as part of technical or compound terms (e.g., “body work”).

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

 mới

-1.27

too

-1.23

หน้านี้

-1.13

реи

-1.11

理石

-1.11

-1.10

 прика

-1.09

 Both

-1.09

 вернуть

-1.09

方は

-1.08

POSITIVE LOGITS

由于

1.30

 andando

1.18

 stessi

1.16

 stuks

1.15

)}\

1.12

 createdAt

1.08

чів

1.06

 Keuangan

1.04

віта

1.04

 blauw

1.04

Activations Density 0.041%