© 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. 28306
    Prev
    Next
    INDEX
    Explanations

    a silly

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

    The neuron fires most strongly on the run of short function words that open a main predicate or noun phrase—e.g. “when I’m in a,” “what does a,” “One week in …”—i.e. it detects those four-word introductory function-word 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
     the
    -1.18
    ఠ
    -1.08
    atique
    -1.06
     huy
    -1.05
     quí
    -1.04
    oher
    -1.02
     his
    -0.98
    ově
    -0.98
    ཥ
    -0.97
     chua
    -0.96
    POSITIVE LOGITS
     it
    1.56
    ❊
    1.01
    こんばん
    1.01
    czegóły
    1.00
     ragu
    0.97
     djur
    0.96
     дуже
    0.92
    Inhalts
    0.92
    GRAFÍA
    0.91
     Realität
    0.91
    Activations Density 0.041%

    No Known Activations