INDEX

Explanations

advert and readFile actions

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

the word "advert" or programming-related terms like "readFile".

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

This neuron fires on tokens that contain the substring “advert,” i.e. references to advertisements (e.g. advert, adverts, advertorial).

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

ing

-2.89

ة

-2.88

ي

-2.28

начально

-2.25

⢌

-2.20

榉

-2.17

啬

-2.17

婞

-2.16

).

-2.08

 με

-2.08

POSITIVE LOGITS

3.06

2.78

you

2.52

2.36

2.28

2.25

之为

2.22

2.20

Activations Density 0.001%

advert and readFile actions

the word "advert" or programming-related terms like "readFile".

This neuron fires on tokens that contain the substring “advert,” i.e. references to advertisements (e.g. advert, adverts, advertorial).

No Comments

No Known Activations

advert and readFile actions

the word "advert" or programming-related terms like "readFile".

This neuron fires on tokens that contain the substring “advert,” i.e. references to advertisements (e.g. advert, adverts, advertorial).

No Comments

No Known Activations