© 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. 22-GEMMASCOPE-RES-131K
    7. 92996
    Prev
    Next
    INDEX
    Explanations

    possessive pronouns and states

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

    The neuron is responding to very common “stop‐words” and high‐frequency tokens (articles, auxiliaries, pronouns, etc.) rather than contentful terms.

    oai_token-act-pair · o4-miniTriggered 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
    Not in Any Lists

    No Comments

    Negative Logits
    illir
    -1.05
    ;',
    -1.01
     });
    -0.96
     oscuros
    -0.94
    ama
    -0.93
     relacionada
    -0.92
    ၎
    -0.91
    僉
    -0.90
    你可以
    -0.90
     organes
    -0.90
    POSITIVE LOGITS
     their
    1.73
     that
    1.16
     its
    1.10
     have
    1.09
     whose
    1.04
     only
    1.00
     Their
    0.96
    pastebin
    0.92
     Regarding
    0.88
    filetype
    0.88
    Activations Density 0.090%

    No Known Activations