© 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. 31-GEMMASCOPE-2-RES-16K
    4. 259
    Prev
    Next
    INDEX
    Explanations

    The neuron detects strong discourse-level connectives—especially causal or adversative markers like “because (of)” and “despite” that signal explanations or contrasts.

    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
     тобто
    0.38
    したがって
    0.36
     を
    0.34
     Indicate
    0.34
    С
    0.33
     یعنی
    0.33
    ただし
    0.33
    <0x0D>
    0.32
     형태로
    0.32
     따라서
    0.31
    POSITIVE LOGITS
     unavailability
    0.78
     its
    0.71
     dwindling
    0.71
     banyaknya
    0.71
     lack
    0.70
     unwillingness
    0.70
     paucity
    0.68
     skyrocketing
    0.67
     what
    0.67
     наличи
    0.66
    Activations Density 2.605%

    No Known Activations