© 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. 97932
    Prev
    Next
    INDEX
    Explanations

    "the" followed by quotation mark

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

    This neuron isn’t picking out any linguistic pattern in the text—it simply fires on tokens that occur at specific fixed positions in the sequence (i.e. it’s a positional‐encoding neuron).

    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
     autora
    -0.88
    €™
    -0.85
     engan
    -0.83
    某种
    -0.82
    所谓
    -0.82
     apropiado
    -0.82
    ǻ
    -0.81
    失效
    -0.81
     arru
    -0.81
    白衣
    -0.81
    POSITIVE LOGITS
     "
    1.13
     “
    1.09
     “…
    0.90
     «
    0.89
    fasis
    0.88
    eara
    0.88
    ″
    0.85
     "\
    0.84
    ropractor
    0.79
     „
    0.79
    Activations Density 0.001%

    No Known Activations