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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 13866
    Prev
    Next
    INDEX
    Explanations

    blocked off

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

    The neuron fires on multiword constructions where a past‐participle form of a verb combines with a following particle (e.g. slicked-back, walled-off, sectioned-off, paved over).

    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
     to
    -1.89
    jedi
    -1.59
    сії
    -1.56
     TAMBIÉN
    -1.55
    binden
    -1.45
    IZACIÓN
    -1.34
    骘
    -1.34
     configurar
    -1.34
     will
    -1.33
    
    -1.33
    POSITIVE LOGITS
    e
    1.69
    ks
    1.66
    but
    1.63
    They
    1.63
    h
    1.62
    u
    1.61
    −/−
    1.60
    hline
    1.52
    。“
    1.51
    with
    1.48
    Activations Density 0.015%

    No Known Activations