INDEX

Explanations

header files and definitions

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

underscores in programming code, particularly after hash symbols in preprocessor directives and before variable or function names.

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

The neuron fires on underscore‐laden tokens—i.e. standalone “_” or identifiers/macros containing “_”.

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

 réjou

-1.64

 énorm

-1.56

 leopardo

-1.51

 lampu

-1.50

 エッチ

-1.46

huu

-1.45

を利用

-1.45

 vêtements

-1.41

وحة

-1.40

⢌

-1.40

POSITIVE LOGITS

of

3.47

for

2.05

are

1.66

 other

1.59

 although

1.58

 should

1.46

 which

1.45

 that

1.43

 have

1.38

no

1.38

Activations Density 0.016%

header files and definitions

underscores in programming code, particularly after hash symbols in preprocessor directives and before variable or function names.

The neuron fires on underscore‐laden tokens—i.e. standalone “_” or identifiers/macros containing “_”.

No Comments

No Known Activations

header files and definitions

underscores in programming code, particularly after hash symbols in preprocessor directives and before variable or function names.

The neuron fires on underscore‐laden tokens—i.e. standalone “_” or identifiers/macros containing “_”.

No Comments

No Known Activations