© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 105998
    Prev
    Next
    INDEX
    Explanations

    verbs followed by prepositions or infinitives

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

    The neuron is tuned to spotting “how‐to” question phrasing—i.e. instructional queries that begin with or prominently feature words like “how,” “do,” “to,” and “you.”

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ּוֹ
    -2.44
    ְּ
    -2.00
    ִּ
    -1.91
    ַּ
    -1.80
    ֶּ
    -1.77
     for
    -1.73
     
    -1.70
    ,
    -1.69
     more
    -1.58
     there
    -1.54
    POSITIVE LOGITS
    ׇ
    2.39
     یا
    1.71
    ἴ
    1.70
    ἵ
    1.68
    '];
    1.67
    ֺ
    1.56
    Ṭ
    1.52
     botanique
    1.52
     supermercados
    1.51
     ܐ
    1.49
    Activations Density 0.256%

    No Known Activations