© 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. 11681
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on words that signal user-focused product capabilities or instructions—e.g. second-person pronouns and modals (“you,” “can,” “allows,” “may,” “also,” etc.)—marking up technical or promotional descriptions of features.

    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
    例如
    0.93
    它
    0.89
    वादियों
    0.88
    ਗ
    0.86
    ਟ
    0.84
    稻
    0.82
    美國
    0.82
    ceding
    0.81
    insuku
    0.81
    达
    0.80
    POSITIVE LOGITS
     Bucure
    1.55
     aproape
    1.54
     pentru
    1.52
     Pentru
    1.48
     numai
    1.45
     Cluj
    1.44
     foarte
    1.42
     România
    1.41
     astfel
    1.39
    Pentru
    1.34
    Activations Density 0.062%

    No Known Activations