© 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. 2550
    Prev
    Next
    INDEX
    Explanations

    The neuron activates strongly on tokens that head off enumerated lists of methods or action items—i.e. it picks out the key words at the start of bulleted or numbered instruction/method lists.

    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
     ஒவ்வ
    0.77
     EACH
    0.75
     THEN
    0.70
     INCLUDING
    0.70
     Rồi
    0.68
     각각
    0.68
    INCLUDING
    0.66
    +](
    0.65
     (
    0.63
     включая
    0.60
    POSITIVE LOGITS
    なども
    1.23
    などは
    1.16
     etc
    1.08
    などで
    1.06
    などの
    1.03
    などに
    1.03
    などが
    1.02
     போன்றவை
    1.02
    、
    1.02
     często
    1.02
    Activations Density 0.379%

    No Known Activations