INDEX

Explanations

names and technical abbreviations

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

capitalized technical abbreviations or proper nouns, particularly short acronyms and surnames.

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

The neuron fires on multi‐letter uppercase or title‐case tokens—i.e. acronyms, class/module names, and proper 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

 situato

-2.70

-2.55



-2.53

’

-2.50

’)

-2.45

-2.39

 nouveautés

-2.38

镲

-2.38

↵↵

-2.34

POSITIVE LOGITS

 embro

2.61

 большого

2.61

比べ

2.58

༛

2.50

膿

2.50

吶

2.45

比べると

2.44

比べて

2.33

";

2.31



2.31

Activations Density 0.040%

names and technical abbreviations

capitalized technical abbreviations or proper nouns, particularly short acronyms and surnames.

The neuron fires on multi‐letter uppercase or title‐case tokens—i.e. acronyms, class/module names, and proper names.

No Comments

No Known Activations

names and technical abbreviations

capitalized technical abbreviations or proper nouns, particularly short acronyms and surnames.

The neuron fires on multi‐letter uppercase or title‐case tokens—i.e. acronyms, class/module names, and proper names.

No Comments

No Known Activations