INDEX

Explanations

mat sat

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

This neuron detects proper names of products, brands, or titles (multiword capitalized entities).

oai_token-act-pair · o4-mini Triggered by @jyhe0408

words that are likely misspelled or contain unusual character combinations resulting from text corruption or spam.

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

keyword-stuffed, SEO-style phrase clusters—especially commercial product terms, brand/tech listings, and spammy marketing/diet language.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

等多

0.70

 rapidement

0.69

ن

0.68

自身

0.65

搅拌

0.64

 mustang

0.63

ونات

0.63

姐妹

0.63

 кош

0.63

襲

0.61

POSITIVE LOGITS

 deutschen

0.65

 unwillingness

0.65

 Zeitschrift

0.64

 proviso

0.62

 Informatika

0.62

 しかし

0.61

 internationalen

0.61

 Trustees

0.60

 enjoyable

0.59

 welchen

0.59

Activations Density 0.035%

mat sat

This neuron detects proper names of products, brands, or titles (multiword capitalized entities).

words that are likely misspelled or contain unusual character combinations resulting from text corruption or spam.

keyword-stuffed, SEO-style phrase clusters—especially commercial product terms, brand/tech listings, and spammy marketing/diet language.

No Comments

No Known Activations