© 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. 1927
    Prev
    Next
    INDEX
    Explanations

    target audience questions

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

    This neuron strongly activates on programming‐style code fragments and shell-script or diff-style lines (e.g. build commands, patch hunks, and code identifiers), and stays off on ordinary prose.

    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
     Rober
    0.65
     Proficiency
    0.61
     Hopper
    0.59
     Oriented
    0.59
     مؤس
    0.58
    მასრულ
    0.58
    ‬
    0.58
    
    0.58
    chinen
    0.57
    xcuser
    0.57
    POSITIVE LOGITS
    underbrace
    0.51
    場で
    0.49
     wonder
    0.45
    லக
    0.45
    ᄒ
    0.45
    提取
    0.44
    先
    0.44
     wrongs
    0.43
     rbind
    0.43
    架
    0.43
    Activations Density 0.004%

    No Known Activations