© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 636
    Prev
    Next
    INDEX
    Explanations

    cannot beat

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

    The neuron strongly activates on very common function words and pronouns (e.g. “it,” “is,” “the,” “we,” “everyone”), effectively detecting high-frequency filler tokens.

    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
     ಹಾಗೂ
    0.77
     solamente
    0.75
     natomiast
    0.72
     tzw
    0.70
     sieci
    0.68
     تقوم
    0.68
    ことにより
    0.68
    逕
    0.67
     oraz
    0.67
     demeure
    0.66
    POSITIVE LOGITS
     Atención
    0.84
     incompar
    0.80
     hắn
    0.80
     Coordin
    0.80
     cultivators
    0.79
     XNUMX
    0.78
     domine
    0.77
     Xuan
    0.76
     netizens
    0.75
     subordinates
    0.75
    Activations Density 0.171%

    No Known Activations