INDEX

Explanations

instantiating objects or classes

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

the dot operator used to access object properties or methods in code.

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

The neuron detects member‐access operators in code (like PHP’s “->” or Python’s “.”).

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

and

-1.51

-1.45

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.40

ions

-1.40

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.40

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.38

기도

-1.32

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.32

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.31

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.30

POSITIVE LOGITS

of

1.62

 which

1.55

 depois

1.45

 działalności

1.37

 indicate

1.30

 deseo

1.30

 used

1.29

 אשר

1.28

游客

1.27

Activations Density 0.008%

instantiating objects or classes

the dot operator used to access object properties or methods in code.

The neuron detects member‐access operators in code (like PHP’s “->” or Python’s “.”).

No Comments

No Known Activations

instantiating objects or classes

the dot operator used to access object properties or methods in code.

The neuron detects member‐access operators in code (like PHP’s “->” or Python’s “.”).

No Comments

No Known Activations