INDEX

Explanations

programming and code snippets

np_acts-logits-general · gemini-2.5-flash-lite

the substring "коммента" (in Cyrillic text).

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron strongly activates on the Russian root “коммента” (as in комментарий), so it’s detecting tokens related to “comment.”

oai_token-act-pair · o4-mini Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

).

-3.06

襞

-2.91

-2.39

,’”

-2.36

 adventures

-2.23

通勤

-2.19

籐

-2.19

休憩

-2.16

嚙

-2.16

放心

-2.14

POSITIVE LOGITS

的

2.58

鹇

2.56

2.53

2.44

er

2.44

 potrze

2.36

墊

2.34

 bonté

2.33

}^{

2.30

2.20

Activations Density 0.000%

programming and code snippets

the substring "коммента" (in Cyrillic text).

The neuron strongly activates on the Russian root “коммента” (as in комментарий), so it’s detecting tokens related to “comment.”

No Comments

No Known Activations

programming and code snippets

the substring "коммента" (in Cyrillic text).

The neuron strongly activates on the Russian root “коммента” (as in комментарий), so it’s detecting tokens related to “comment.”

No Comments

No Known Activations