© 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-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 33258
    Prev
    Next
    INDEX
    Explanations

    Code documentation and multilingual content

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

    The neuron is picking up on chunks of CJK text (e.g. Chinese or Japanese character sequences).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     from
    -2.59
     their
    -2.59
     as
    -2.48
     only
    -2.41
     by
    -2.39
     because
    -2.33
     like
    -2.31
     before
    -2.31
     an
    -2.23
     not
    -2.23
    POSITIVE LOGITS
     faciliter
    1.98
     Dinas
    1.98
     triomphe
    1.84
    Aantal
    1.80
     penser
    1.80
     ACTUAL
    1.77
     verhind
    1.77
     relâche
    1.77
    Nieuwe
    1.74
     goû
    1.73
    Activations Density 0.005%

    No Known Activations