© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 9344
    Prev
    Next
    INDEX
    Explanations

    step then away

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

    The neuron fires on action words (verbs) that describe procedural steps or state changes (e.g. move, bring, back, convert, open, hide).

    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
     énfasis
    -1.23
     garantizar
    -1.21
     necesitan
    -1.20
     asegurarse
    -1.16
     necesitar
    -1.15
     utilice
    -1.14
     eficacia
    -1.13
     enfati
    -1.10
     létre
    -1.09
     dépens
    -1.09
    POSITIVE LOGITS
     again
    2.06
     back
    1.77
     to
    1.50
     another
    1.47
     other
    1.41
     kembali
    1.33
     now
    1.23
    Another
    1.23
     or
    1.22
     went
    1.21
    Activations Density 0.045%

    No Known Activations