© 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. 84236
    Prev
    Next
    INDEX
    Explanations

    trend

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

    The neuron fires on the word “trend” (and similar statistical‐movement cues), i.e. it spots when the text is talking about trends.

    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
    VIATIONS
    -1.62
     a
    -1.61
     obtained
    -1.52
     '</
    -1.49
     both
    -1.47
    نع
    -1.45
     alojamientos
    -1.43
     květ
    -1.41
     viewed
    -1.41
    довлет
    -1.38
    POSITIVE LOGITS
    ta
    2.03
     trends
    1.63
    ion
    1.53
    ner
    1.53
     debilit
    1.52
    en
    1.48
     intenta
    1.47
    幸好
    1.46
    he
    1.42
    ter
    1.41
    Activations Density 0.020%

    No Known Activations