© 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. 36509
    Prev
    Next
    INDEX
    Explanations

    but, tended, showed, suffer, affected

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

    The neuron is tuned to pick out quantitative and contrastive result statements in research‐style text—highlighting numbers, percentages, and connectors like “but” or “however.”

    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
     seharusnya
    -1.21
    vų
    -1.06
     运动
    -1.05
    Détails
    -1.05
     mulia
    -1.04
    úd
    -1.03
     النبات
    -1.03
     hôtel
    -1.02
    Praxis
    -1.02
     veau
    -1.02
    POSITIVE LOGITS
     they
    1.24
     "
    1.24
     модели
    1.20
     “
    1.19
     Informationen
    1.16
    řejmě
    1.16
     their
    1.09
     Mereka
    1.08
     trends
    1.07
     když
    1.06
    Activations Density 0.101%

    No Known Activations