© 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. 11062
    Prev
    Next
    INDEX
    Explanations

    The neuron identifies discourse‐level “pivot” words and transition markers (e.g. I decided, if not, traditional, let’s, that, don’t, always, permission) that signal shifts or emphasis in the text.

    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
     kilos
    0.77
    }`}
    0.75
     площа
    0.71
    }^{-
    0.70
     Industry
    0.69
     Technology
    0.69
     dots
    0.69
    டியில்
    0.69
    }}$-
    0.68
     ở
    0.68
    POSITIVE LOGITS
    ت
    0.87
    igt
    0.84
    ת
    0.82
    ان
    0.79
    на
    0.79
     executar
    0.77
     できる
    0.75
    ის
    0.74
     rede
    0.74
    ించ
    0.73
    Activations Density 0.000%

    No Known Activations