© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 1177
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly responds wherever there’s a long, information‐dense span (e.g. a full, multi-clause sentence), essentially flagging unusually lengthy or complex sentences.

    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_31_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
    𝗹
    0.64
    の
    0.63
     ਖ
    0.63
    ㄟ
    0.61
    łej
    0.60
    ㄤ
    0.60
    ের
    0.59
    щий
    0.59
    ــــ
    0.59
     أ
    0.58
    POSITIVE LOGITS
     kulture
    0.64
     ಎಂಬುದ
    0.64
     bureaucrats
    0.64
     peacetime
    0.62
     esemplare
    0.59
    राजनी
    0.59
     ecco
    0.59
    ความเป็น
    0.59
     franchises
    0.58
     kass
    0.58
    Activations Density 0.955%

    No Known Activations