© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 26712
    Prev
    Next
    INDEX
    Explanations

    describe capabilities or requirements

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

    The neuron strongly activates on very high-frequency English words—especially common verbs (e.g. “have,” “provide,” “be”) and frequent function words (e.g. “in,” “with”).

    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
     has
    -1.71
     or
    -1.67
     is
    -1.59
     was
    -1.40
     its
    -1.21
     does
    -1.21
     with
    -1.18
     before
    -1.16
     this
    -1.13
     totally
    -1.09
    POSITIVE LOGITS
     themselves
    3.66
     their
    2.66
    themselves
    2.53
     are
    2.11
     jejich
    1.92
     they
    1.88
    their
    1.80
    Their
    1.75
     έχουν
    1.62
     Their
    1.53
    Activations Density 0.142%

    No Known Activations