© 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. 103596
    Prev
    Next
    INDEX
    Explanations

    trail propagation, names, statistical

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

    The neuron strongly activates on dense, capitalized technical jargon—especially cryptanalysis or cipher‐algorithm terms and section headings (e.g. “Differential and Linear trail propagation in Noekeon”).

    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
     dusty
    -0.76
    entier
    -0.75
    cellation
    -0.75
    ophytes
    -0.74
    𝟮
    -0.70
    窯
    -0.69
    paced
    -0.67
     olla
    -0.67
    deville
    -0.67
     stew
    -0.66
    POSITIVE LOGITS
     statistical
    0.88
    Statistical
    0.85
     distinguish
    0.83
    dopodob
    0.76
    statistical
    0.76
    ésor
    0.74
     CRY
    0.74
    DDT
    0.74
     Alli
    0.74
    }}+\
    0.74
    Activations Density 0.034%

    No Known Activations