© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 49161
    Prev
    Next
    INDEX
    Explanations

    above copyright

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

    The neuron fires on the first “header” token at the top of a snippet—things like copyright or license headers, include/import lines, major declarations (e.g. function signatures)—i.e. the token that begins a new code or license block.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    syntactic separators and structural markers in code and license headers, especially punctuation that introduces lists or separates parameters.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    newline characters followed by whitespace or comments in source code, particularly in license headers and function signatures.

    oai_token-act-pair · claude-4-5-sonnetTriggered 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
    ImageIO
    -0.82
     moyenne
    -0.79
     Tomar
    -0.78
     THOMPSON
    -0.76
     SafeMath
    -0.75
    zessin
    -0.74
    SurfaceView
    -0.73
    ","\
    -0.72
     voetbal
    -0.71
    Ꮐ
    -0.71
    POSITIVE LOGITS
     response
    1.04
     HttpServlet
    0.88
     javax
    0.88
    response
    0.87
     кото
    0.85
    diğinde
    0.84
     responses
    0.83
    yet
    0.80
     Response
    0.79
    兑换
    0.78
    Activations Density 0.000%

    No Known Activations