© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 8195
    Prev
    Next
    INDEX
    Explanations

    node or * followed by punctuation/special characters

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

    The neuron doesn’t pick out any coherent linguistic category—it simply lights up on those odd, out-of-place numeric-noise spikes in the activations rather than on any real word pattern.

    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
     –
    0.92
    ل
    0.81
     To
    0.76
     Relation
    0.76
    ي
    0.76
     AD
    0.75
    पी
    0.74
    يب
    0.73
     Prepared
    0.73
    ari
    0.73
    POSITIVE LOGITS
    catcher
    0.94
    shells
    0.92
     pessoais
    0.92
     cheques
    0.89
     rashes
    0.86
    вался
    0.84
     surfers
    0.84
    )}$,
    0.84
    скія
    0.82
     borrar
    0.82
    Activations Density 0.000%

    No Known Activations