© 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. 21368
    Prev
    Next
    INDEX
    Explanations

    works well with

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

    The neuron detects evaluative collocations where something “works/looks/etc.” + a preposition (“on,” “with,” “in,” etc.) + a target, i.e. phrases expressing that one thing fits or is compatible with another.

    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
     for
    -2.17
    calvin
    -1.86
    crocs
    -1.84
    timberland
    -1.84
    peanut
    -1.77
     as
    -1.77
    ralph
    -1.76
    garmin
    -1.70
    lacoste
    -1.67
    mushroom
    -1.66
    POSITIVE LOGITS
    family
    1.66
    i
    1.63
    Context
    1.61
    cache
    1.57
    flex
    1.52
    它
    1.48
     itinéraire
    1.47
    model
    1.47
    road
    1.46
     kablo
    1.45
    Activations Density 0.030%

    No Known Activations