© 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
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 9710
    Prev
    Next
    INDEX
    Explanations

    philosophy and reward hacking

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

    This neuron fires on longer, content-rich wordpieces (complete words and multi-character tokens), while ignoring short or purely functional bits like single letters, punctuation, or common small words.

    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
    慵
    0.71
     অজানা
    0.70
     پیغام
    0.70
     اساس
    0.69
    法规
    0.68
     ګرځنده
    0.68
    deload
    0.68
     basée
    0.68
    윈
    0.68
     overlays
    0.67
    POSITIVE LOGITS
     philosophers
    1.32
     philosopher
    1.31
     Philosophical
    1.28
     Wittgenstein
    1.21
     Frege
    1.18
     philosophical
    1.16
     Kant
    1.13
     Philos
    1.10
     dox
    1.09
     deont
    1.08
    Activations Density 0.194%

    No Known Activations