© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 34982
    Prev
    Next
    INDEX
    Explanations

    annual publication

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

    This neuron detects multiword action phrases that describe performing or skipping steps—specifically verb + object (+ instrument/PP) constructions (e.g. “skipping some of the filler ones,” “hit them with the harder stuff”).

    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
     currently
    -1.29
     immediately
    -1.06
     presently
    -0.97
    Immediately
    -0.96
    navi
    -0.95
    目前
    -0.95
     they
    -0.92
     Immediately
    -0.91
    aktifkan
    -0.90
     initially
    -0.90
    POSITIVE LOGITS
     increasingly
    1.78
     more
    1.55
    不再
    1.45
     artık
    1.38
     стала
    1.27
     стали
    1.24
    越来越
    1.24
     coraz
    1.23
     יותר
    1.20
     стал
    1.16
    Activations Density 0.067%

    No Known Activations