INDEX

Explanations

The neuron activates strongly on Chinese (CJK) characters, i.e. it detects Chinese text segments.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nonché

1.35

 geralmente

1.22

 iako

1.21

 terwijl

1.20

 czyli

1.17

 società

1.16

 etiquette

1.16

 sedangkan

1.16

 meskipun

1.13

 valamint

1.10

POSITIVE LOGITS

把

1.13

用

1.12

做

1.04

得

1.03

有

1.01

有一

0.98

以

0.96

被

0.95

用到

0.95

为

0.94

Activations Density 0.010%