© 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. 15810
    Prev
    Next
    INDEX
    Explanations

    bar to drink

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

    This neuron primarily lights up on relatively rare or content-heavy tokens—that is, low-frequency words (including proper names, specific nouns, and other information-dense terms).

    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
     অপহরণকারীদের
    0.60
     OFDM
    0.57
     dynamical
    0.57
     કચ્છ
    0.57
     conject
    0.55
     USPS
    0.55
     Newtonian
    0.55
     Dynamical
    0.55
     Montessori
    0.54
     Kite
    0.54
    POSITIVE LOGITS
     alcohol
    3.28
     алкого
    3.13
     alkohol
    3.08
    Alcohol
    3.06
    alcohol
    3.03
     alcoholic
    3.02
     Alcohol
    3.00
     booze
    3.00
     alkoh
    2.91
     Alkohol
    2.89
    Activations Density 0.723%

    No Known Activations