© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 10890
    Prev
    Next
    INDEX
    Explanations

    time travel or scientific terms

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

    The neuron is not responding to any semantic content but to token position—it systematically activates more strongly on words appearing later in the sequence.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    tion
    0.86
     Dens
    0.75
    ت
    0.74
    tic
    0.71
    tions
    0.71
     disputed
    0.70
    یت
    0.70
     Dense
    0.70
     surge
    0.70
    শীল
    0.68
    POSITIVE LOGITS
    Ầ
    0.80
    ନା
    0.79
    ುದ
    0.79
     detailing
    0.77
     Éireann
    0.77
    âu
    0.75
    aży
    0.74
     raccol
    0.73
    waża
    0.73
    Nú
    0.73
    Activations Density 0.000%

    No Known Activations