© 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
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 117600
    Prev
    Next
    INDEX
    Explanations

    giant, large, high, burning, setting

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

    The neuron is acting like a position counter, firing in proportion to how far into the document a token appears (i.e. encoding absolute token position).

    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
     despite
    -0.92
     irrespective
    -0.85
    präche
    -0.83
     without
    -0.83
    беле
    -0.82
    exports
    -0.82
     kiedy
    -0.80
     might
    -0.80
    踐
    -0.78
    美子
    -0.78
    POSITIVE LOGITS
     giant
    1.48
     replica
    1.30
    giant
    1.24
     огром
    1.13
     faux
    1.13
     mock
    1.13
     simulated
    1.12
     gigantic
    1.10
    Giant
    1.04
    faux
    1.03
    Activations Density 0.056%

    No Known Activations