INDEX

Explanations

code syntax

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

closing braces or block-ending keywords in code that terminate functions or control structures.

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

This neuron detects closing delimiters—i.e. lines or tokens that end a block or expression (closing `}` or `)`).

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

-1.91

也许

-1.59

 “[

-1.55

ane

-1.51

ens

-1.49

ah

-1.47

！”

-1.45

("

-1.43

!”

-1.43

POSITIVE LOGITS

 verksam

1.69

 that

1.66

することも

1.49

瓛

1.48

 火锅

1.48

 geene

1.48

 arbejds

1.48

utom

1.47

𝗩

1.45

 그것

1.44

Activations Density 0.014%

code syntax

closing braces or block-ending keywords in code that terminate functions or control structures.

This neuron detects closing delimiters—i.e. lines or tokens that end a block or expression (closing `}` or `)`).

No Comments

No Known Activations

code syntax

closing braces or block-ending keywords in code that terminate functions or control structures.

This neuron detects closing delimiters—i.e. lines or tokens that end a block or expression (closing `}` or `)`).

No Comments

No Known Activations