INDEX

Explanations

process.env and Process.Start

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

the word "process" followed by a period in code contexts.

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

This neuron responds to the “.” member‐access operator in code (e.g. identifier followed by a dot).

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

'-

-2.03

-'

-1.80

 pouvons

-1.79

"-

-1.75

 dejó

-1.74

-'

-1.74

'-

-1.73

'[

-1.72

皤

-1.71

 escase

-1.70

POSITIVE LOGITS

層

1.79

 أيض

1.68

two

1.66

Ольга

1.62

 make

1.61

 what

1.60

 five

1.60

霂

1.59

Елена

1.59

1.56

Activations Density 0.001%

process.env and Process.Start

the word "process" followed by a period in code contexts.

This neuron responds to the “.” member‐access operator in code (e.g. identifier followed by a dot).

No Comments

No Known Activations

process.env and Process.Start

the word "process" followed by a period in code contexts.

This neuron responds to the “.” member‐access operator in code (e.g. identifier followed by a dot).

No Comments

No Known Activations