© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-27B-IT
    5. 16-GEMMASCOPE-2-TRANSCODER-262K
    7. 216844
    Prev
    Next
    INDEX
    Explanations

    The user wants a concise explanation (3-20 words) of a neuron's behavior.I need to find patterns in:- MAX_ACTIVATING_TOKENS: ius, N, explore, ,, ity, ar, as, ity, of, audible- TOKENS_AFTER_MAX_ACTIVATING_TOKEN: ., uan, ChatGPT, vinyl, is, qa, a, is, smart, range- TOP_POSITIVE_LOGITS: PES, propulsion, ┃, ън, misconception, PBS, ગ्यो, commissary, promotional, BIOS- TOP_ACTIVATING_TEXTS: - "Nuan (暖)" - Chinese word for warmth, related to food description. - "explore ChatGPT" - Suggests exploring AI tools. - "Polyvinyl Chloride (PVC or V)" / "vinyl flooring" - Material science/products.Chinese characters or warmth, explore ChatGPT, vinyl, smart gadget, audible range

    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_16_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
     añadi
    0.49
     म्हणजे
    0.46
    ról
    0.45
     أو
    0.45
    so
    0.44
    addTarget
    0.44
    كينا
    0.43
    人了
    0.43
    kowe
    0.43
    œurs
    0.43
    POSITIVE LOGITS
     PES
    0.49
     propulsion
    0.47
    ┃
    0.46
    ън
    0.45
     misconception
    0.45
     PBS
    0.44
     ગયો
    0.44
     commissary
    0.44
     promotional
    0.43
     BIOS
    0.43
    Activations Density 0.000%

    No Known Activations