INDEX

Explanations

programming generics and linux systems

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

generic type parameters and angle brackets in programming contexts.

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

The neuron selectively lights up on capitalized identifiers and labels—e.g. proper names, titles, and technical terms (like “Pacman,” “Arch,” “Linux,” “Source,” “Author,” “Photo,” or code‐keywords like “new”).

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

only

-1.51

↵↵

-1.45

мії

-1.35

ORPG

-1.31

 from

-1.27

 "/";

-1.25

왼

-1.24

 більш

-1.22

 тільки

-1.22

'';

-1.22

POSITIVE LOGITS

of

1.73

he

1.29

 lakini

1.27

 were

1.24

ᚻ

1.21

也不

1.20

 ersche

1.19

甚至

1.18

ズニー

1.16

惲

1.15

Activations Density 0.005%

programming generics and linux systems

generic type parameters and angle brackets in programming contexts.

The neuron selectively lights up on capitalized identifiers and labels—e.g. proper names, titles, and technical terms (like “Pacman,” “Arch,” “Linux,” “Source,” “Author,” “Photo,” or code‐keywords like “new”).

No Comments

No Known Activations

programming generics and linux systems

generic type parameters and angle brackets in programming contexts.

The neuron selectively lights up on capitalized identifiers and labels—e.g. proper names, titles, and technical terms (like “Pacman,” “Arch,” “Linux,” “Source,” “Author,” “Photo,” or code‐keywords like “new”).

No Comments

No Known Activations