© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 32853
    Prev
    Next
    INDEX
    Explanations

    Hawthorne album names

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

    The neuron strongly activates on multi‐letter tokens that begin with a capital letter (e.g. class names, proper nouns, macro or identifier names) and ignores single‐letter caps or lowercase words.

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Negative Logits
    Vanity
    -0.90
     женщин
    -0.90
    还有
    -0.89
     latter
    -0.87
    MANY
    -0.86
    我又
    -0.85
    了一遍
    -0.85
     tych
    -0.85
    垦
    -0.85
     Şi
    -0.85
    POSITIVE LOGITS
     می‌دهد
    0.98
    嬗
    0.93
    דה
    0.93
    ><
    0.91
     Burgen
    0.90
     one
    0.88
    着脸
    0.87
     ".$
    0.87
    了口气
    0.85
     Pfalz
    0.84
    Activations Density 0.073%

    No Known Activations