© 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. 31-GEMMASCOPE-2-RES-16K
    7. 6335
    Prev
    Next
    INDEX
    Explanations

    approximately 30-45 minutes

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

    The neuron is selectively firing on numeric tokens and other out‐of‐vocabulary/oddly tokenized items (e.g. decimal numbers like “2.5,” UI-style status words like “loading,” or rare adjectives such as “incredible”).

    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
    acoli
    0.71
    顼
    0.67
    hapi
    0.65
    ᅱ
    0.64
    螌
    0.64
     Abubakar
    0.63
    カフェ
    0.63
    adilan
    0.63
    giore
    0.62
    edic
    0.62
    POSITIVE LOGITS
     neighbors
    0.56
     vect
    0.55
     video
    0.55
     storylines
    0.54
     ):
    0.53
     vector
    0.52
     reen
    0.52
    RC
    0.52
     spearheaded
    0.52
     RC
    0.51
    Activations Density 0.000%

    No Known Activations