INDEX

Explanations

programming and web technologies

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

the word "Control" in technical contexts, particularly in HTTP headers and programming code.

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

The neuron fires on code-or doc-specific identifiers or keywords—especially capitalized or hyphenated terms like class/method names, headers, package names, and other technical “names.”

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

</h1>

-1.80

-1.75

-1.68

-1.66

During

-1.65

When

-1.65

-1.63

-1.60

Some

-1.58

POSITIVE LOGITS

ruguay

2.27

咝

1.81

𓂀

1.70

 accommodates

1.70

ᐠ

1.70

玗

1.66

 skru

1.65

riezmann

1.65

⢅

1.65

 違い

1.63

Activations Density 0.003%

programming and web technologies

the word "Control" in technical contexts, particularly in HTTP headers and programming code.

The neuron fires on code-or doc-specific identifiers or keywords—especially capitalized or hyphenated terms like class/method names, headers, package names, and other technical “names.”

No Comments

No Known Activations

programming and web technologies

the word "Control" in technical contexts, particularly in HTTP headers and programming code.

The neuron fires on code-or doc-specific identifiers or keywords—especially capitalized or hyphenated terms like class/method names, headers, package names, and other technical “names.”

No Comments

No Known Activations