© 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. 1327
    Prev
    Next
    INDEX
    Explanations

    back to

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

    The neuron overwhelmingly activates on the word “back,” especially when used in phrasal contexts like “back to,” “go back,” or “back in.”

    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
     оставить
    0.62
    끝
    0.60
     получа
    0.58
    outcome
    0.57
    여러
    0.57
    ссе
    0.55
     thay
    0.55
     besondere
    0.55
     отклю
    0.55
    Transitions
    0.55
    POSITIVE LOGITS
    回到
    1.24
     semula
    1.20
     indietro
    1.19
     tilbake
    1.13
    に戻
    1.10
    戻
    1.08
    回来
    1.07
     retomar
    1.07
     terug
    1.04
     whence
    1.03
    Activations Density 0.443%

    No Known Activations