© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 1064
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on words that introduce subjective reactions or judgments—e.g. “Sounds,” “Feels,” “Looks,” “Sounds like,” etc.—marking evaluative or opinion-framing phrases.

    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
    י
    0.70
    তে
    0.60
    nder
    0.59
     compañ
    0.57
     withd
    0.57
    ্ল্ড
    0.57
    шча
    0.56
    ς
    0.56
    నో
    0.54
    γν
    0.54
    POSITIVE LOGITS
    urés
    0.56
     Looks
    0.55
     оригинала
    0.54
    trimenti
    0.54
    Looks
    0.54
    utations
    0.54
    像是
    0.54
    گاه
    0.53
     seperti
    0.52
    荒
    0.52
    Activations Density 0.055%

    No Known Activations