© 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. 10-GEMMASCOPE-RES-131K
    7. 22391
    Prev
    Next
    INDEX
    Explanations

    a/an followed by condition

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

    The neuron is tracking each token’s absolute position in the document (i.e. it “fires” in proportion to how far into the text the token occurs).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     militer
    -1.81
     these
    -1.70
     visualiser
    -1.59
     Istri
    -1.58
    ♪♪
    -1.53
    ότη
    -1.52
     relais
    -1.52
     strategis
    -1.50
     Fø
    -1.49
     vider
    -1.48
    POSITIVE LOGITS
    </b>
    1.98
     chance
    1.84
     beginning
    1.71
     time
    1.71
     nobre
    1.66
     disposiciones
    1.66
    舠
    1.65
     welcomed
    1.62
     I
    1.59
     also
    1.59
    Activations Density 0.121%

    No Known Activations