© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 1439
    Prev
    Next
    INDEX
    Explanations

    The neuron fires strongly on standalone label‐style tokens that introduce or separate sections (e.g. “Features:”, “Descriptions:”, “INSERT INTO …”), as well as brief salutations or sign-off phrases (“Thanks,” “Hi …”), so it is primarily detecting structural headings and headers in the text.

    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
    ’
    0.71
     ჩატი
    0.69
    რც
    0.69
    <unused2140>
    0.68
     tenía
    0.67
     miała
    0.67
    <unused411>
    0.66
     tenir
    0.63
    .​​
    0.63
     термин
    0.63
    POSITIVE LOGITS
    ↵
    2.47
    <0x0D>
    1.63
                                   
    1.44
                                  
    1.40
                                 
    1.34
                              
    1.31
                               
    1.29
                      
    1.26
                         
    1.26
                                
    1.26
    Activations Density 0.038%

    No Known Activations