© 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-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 104304
    Prev
    Next
    INDEX
    Explanations

    several different

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

    The neuron is especially sensitive to semicolons and the start of list-style or segmented descriptive clauses, i.e. it marks the boundaries of multi-part listings or parallel descriptions.

    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
    »).
    -1.08
    ఠ
    -1.06
    ₣
    -1.05
     silikon
    -1.03
    ießen
    -1.02
     kram
    -1.02
    ENCIÓN
    -1.02
     bambu
    -1.02
    likely
    -1.01
    balar
    -1.00
    POSITIVE LOGITS
     cut
    0.95
     demselben
    0.91
     and
    0.89
     tätig
    0.87
    扼
    0.83
    몇
    0.82
    di
    0.82
    四人
    0.81
     salient
    0.80
     six
    0.80
    Activations Density 0.016%

    No Known Activations