© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 7146
    Prev
    Next
    INDEX
    Explanations

    photo or man

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

    The neuron strongly activates on generic blog‐comment phrases expressing personal reaction or thanks (e.g. “every,” “owner,” “look into,” “thank,” “recognize,” “boys excited,” etc.), i.e. boilerplate comment‐spam expressions of appreciation or inquiry.

    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
     Μα
    0.55
     banished
    0.55
    ylvania
    0.55
    ‰
    0.54
     Omar
    0.53
     unhappy
    0.52
    vue
    0.51
     asi
    0.51
    糟糕
    0.51
     คะแนน
    0.50
    POSITIVE LOGITS
     setzte
    0.56
     அமைக்க
    0.54
     Herzen
    0.52
    odymium
    0.52
     বিদ্রোহীদের
    0.52
     acceler
    0.51
    بيض
    0.51
    เป็นการ
    0.51
     accel
    0.51
     anún
    0.50
    Activations Density 0.005%

    No Known Activations