© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 2564
    Prev
    Next
    INDEX
    Explanations

    positional index

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

    The neuron strongly fires on words that denote relative positions or parts in procedural instructions (e.g. “other,” “bottom,” “last,” “each,” “both,” “left”).

    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
    Path
    0.61
    Cutting
    0.60
    President
    0.59
    Ad
    0.58
    Foi
    0.58
    誼
    0.57
    Ed
    0.56
    Ин
    0.56
    Shi
    0.55
    Aspect
    0.55
    POSITIVE LOGITS
     two
    1.11
     three
    1.08
     drei
    0.96
     beiden
    0.95
     four
    0.94
     deux
    0.92
     halves
    0.91
     remaining
    0.88
     trois
    0.86
     zwei
    0.82
    Activations Density 0.271%

    No Known Activations