© 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. 72336
    Prev
    Next
    INDEX
    Explanations

    good and bad

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

    The neuron fires on strongly valenced words—especially positive/negative sentiment terms (e.g. “good,” “bad,” “ups,” “downs,” “positive,” “negative,” “triumphs,” “failures,” etc.).

    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
     invitamos
    -1.03
     adjacent
    -1.01
     a
    -1.00
    انجليز
    -0.94
     next
    -0.94
    纷纷
    -0.93
     geprü
    -0.92
     anyone
    -0.92
    Opts
    -0.91
    erapeutics
    -0.91
    POSITIVE LOGITS
     عندما
    1.39
     těch
    1.38
     also
    1.36
     cuando
    1.35
     sometimes
    1.32
     when
    1.30
     también
    1.23
     những
    1.21
    Cuando
    1.20
     occasionally
    1.17
    Activations Density 0.025%

    No Known Activations