© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 180
    Prev
    Next
    INDEX
    Explanations

    The neuron is detecting the start‐of‐sequence (the “<bos>” token) in the input.

    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
    ,
    0.46
     set
    0.45
     up
    0.45
     over
    0.44
     into
    0.43
     back
    0.43
     sits
    0.43
    -
    0.43
     on
    0.42
     argument
    0.42
    POSITIVE LOGITS
     Elektrokh
    0.54
     Když
    0.51
     Trên
    0.50
     Ча
    0.48
    Công
    0.48
    Benzyl
    0.48
    Hydrochloride
    0.47
     Ли
    0.47
     Само
    0.47
     Políticas
    0.47
    Activations Density 2.725%

    No Known Activations