© 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
    2. Gemma-3-27B-IT
    3. 11-GEMMASCOPE-2-TRANSCODER-262K
    4. 90468
    Prev
    Next
    INDEX
    Explanations

    This neuron is primarily associated with states of *enabling* or *disabling* certain functions or features. The `MAX_ACTIVATING_TOKENS` list is dominated by "Enable", "enable", "on", and "disable", "disable". The `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` list shows contexts like "editing", "mouse tracking", "events", "button", and "Remote Desktop", all of which are actions or components that can be turned on or off.The `TOP_POSITIVE_LOGITS` are quite diverse and don't directly relate to enabling/disabling, suggesting they might be background noise or related to very specific, perhaps multilingual, contexts where these enable/disable states appear.The most consistent and strong pattern is the toggling of states.Therefore, a concise explanation would be:enabling or disabling states

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_11_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    icioso
    0.52
    ubro
    0.51
    ലെസ്
    0.50
    cita
    0.50
    ),"
    0.49
    ],"
    0.49
    izion
    0.48
    性
    0.48
    ensional
    0.47
     پہ
    0.47
    POSITIVE LOGITS
    тяну
    0.54
    ם
    0.53
    Phi
    0.52
     bütün
    0.52
    प्ला
    0.51
     बिना
    0.50
     Dupont
    0.50
     Æ
    0.50
     Helmet
    0.50
    शरी
    0.49
    Activations Density 0.001%

    No Known Activations