© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 23915
    Prev
    Next
    INDEX
    Explanations

    cannot let go

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

    The neuron is highly sensitive to auxiliary and control verbs expressing ability or permission—particularly forms of “can’t,” “let,” and “allow (allowed to)” in common multi‐word phrases.

    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
     zoll
    -1.37
    Accès
    -1.31
     frauen
    -1.31
    kumpulan
    -1.30
    rigated
    -1.27
    filé
    -1.27
    真正的
    -1.26
     sánchez
    -1.26
     farbe
    -1.24
    璈
    -1.24
    POSITIVE LOGITS
     solchen
    1.41
     такую
    1.38
     такое
    1.32
    </h4>
    1.32
     другое
    1.31
    t
    1.31
     mente
    1.23
    呟
    1.23
     do
    1.22
     profesional
    1.22
    Activations Density 0.016%

    No Known Activations