© 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. 34-GEMMASCOPE-RES-131K
    7. 33588
    Prev
    Next
    INDEX
    Explanations

    earned

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

    The neuron fires on auxiliary “helping” verbs—especially forms of HAVE and BE (has, have, are, was, were)—marking perfective or passive constructions.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     zaraz
    -0.77
    fass
    -0.70
     vida
    -0.69
    ppus
    -0.67
    JNI
    -0.66
    middot
    -0.66
    ͦ
    -0.66
     fascic
    -0.65
     ов
    -0.65
    Hymen
    -0.65
    POSITIVE LOGITS
     earned
    6.56
     earn
    5.66
     earning
    5.03
     earns
    4.78
    Earn
    4.78
    earned
    4.75
     Earned
    4.53
     Earn
    4.03
    earn
    3.92
    earning
    3.47
    Activations Density 0.067%

    No Known Activations