INDEX

Explanations

technical / legal text identifiers

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

whitespace characters, particularly tabs and spaces used for indentation in code.

oai_token-act-pair · claude-4-5-sonnet

Triggered by @bfgcn7frf5

The neuron activates on numeric literals (both integers and decimal numbers).

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

Negative Logits

儿的

-1.60

 They

-1.43

يًا

-1.41

 手绘

-1.40

紲

-1.37

 понравился

-1.35

繊細

-1.35

</h3>

-1.35

ladung

-1.34

//!

-1.34

POSITIVE LOGITS

 then

1.75

 such

1.71

 those

1.59

ktır

1.35

他在

1.34

 people

1.32

its

1.31

who

1.31

 rów

1.30

你在

1.28

Activations Density 0.002%

technical / legal text identifiers

whitespace characters, particularly tabs and spaces used for indentation in code.

The neuron activates on numeric literals (both integers and decimal numbers).

No Known Activations

technical / legal text identifiers

whitespace characters, particularly tabs and spaces used for indentation in code.

The neuron activates on numeric literals (both integers and decimal numbers).

No Known Activations