© 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
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 12729
    Prev
    Next
    INDEX
    Explanations

    expo and Tel Aviv events

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly activates on capitalized whole-word tokens—i.e. proper names or identifiers (like “Env”, “Expo”, “Tel”, “Aviv”)—so it’s looking for uppercase initial tokens that signal named entities or constants.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    𖤣
    -3.39
    骈
    -2.80
    らしいです
    -2.72
    玙
    -2.67
     馆
    -2.59
    玬
    -2.59
     nahil
    -2.58
    鲇
    -2.52
    獃
    -2.52
    牺
    -2.50
    POSITIVE LOGITS
    4
    4.06
    ’
    3.89
     '
    3.81
    9
    3.78
    3
    3.55
    7
    3.50
    8
    3.50
    on
    3.38
    2
    3.17
    5
    3.06
    Activations Density 0.001%

    No Known Activations