© 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. 34162
    Prev
    Next
    INDEX
    Explanations

    I believe

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

    The neuron is detecting first-person subjective commentary—i.e. words that signal the writer’s personal opinions or reactions (“I,” “we,” intensifiers like “really,” “believed,” 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
    Sloven
    -1.03
    absor
    -1.01
    Imgs
    -0.96
    Prez
    -0.96
     Bajos
    -0.96
    赚
    -0.96
    Tiet
    -0.94
    Klas
    -0.94
    ἶ
    -0.94
    Отзывы
    -0.92
    POSITIVE LOGITS
     watching
    1.28
     predictions
    1.13
     I
    1.12
    ranno
    1.10
     watch
    1.09
     think
    1.09
     see
    1.08
    predictions
    1.05
     nová
    1.03
     readers
    1.02
    Activations Density 0.046%

    No Known Activations