INDEX

Explanations

run or execute

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

words related to code execution or running programs, particularly in non-English programming contexts.

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

The neuron activates on words meaning “execute” or “run” (e.g., forms of “execute,” “run,” “ejecuta,” “実行,” “выполня,” etc.) in various languages.

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

要不然

-2.28

出了一

-2.22

 been

-2.13

ны

-2.09

-1.99

那你

-1.97

-1.96

</h4>

-1.94

 revel

-1.92

 discomfort

-1.89

POSITIVE LOGITS

喼

2.91

趼

2.61

彣

2.50

’，

2.50

Ꮉ

2.45

蕯

2.41



2.41

оргаш

2.41

 kebaya

2.38

 malu

2.38

Activations Density 0.001%

run or execute

words related to code execution or running programs, particularly in non-English programming contexts.

The neuron activates on words meaning “execute” or “run” (e.g., forms of “execute,” “run,” “ejecuta,” “実行,” “выполня,” etc.) in various languages.

No Comments

No Known Activations

run or execute

words related to code execution or running programs, particularly in non-English programming contexts.

The neuron activates on words meaning “execute” or “run” (e.g., forms of “execute,” “run,” “ejecuta,” “実行,” “выполня,” etc.) in various languages.

No Comments

No Known Activations