INDEX

Explanations

Japanese particle で

The neuron fires on Japanese (non-Latin) text sequences, i.e. it detects Japanese characters (kanji/kana) in the input.

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

𝑧

-2.44

-2.42

す

-2.38

いい

-2.34

-2.16

'),

-2.09

すべての

-2.05

}$

-2.03

and

-2.02

م

-2.02

POSITIVE LOGITS

2.59

いて

2.44

いえば

2.28

 skupina

2.27

 einige

2.23

 Whether

2.19

 Erwachsene

2.17

睃

2.16

之后

2.16

 yelek

2.14

Activations Density 0.000%