© 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. 129176
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on tokens referring to things or people being “affected” (i.e. impacted) by something.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    affected individuals or parts

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

    words related to problems or negative impacts, particularly "affected" and "contaminated" items.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi
    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
     různ
    -1.36
    蠓
    -1.31
     Dinas
    -1.31
    蛏
    -1.28
     erila
    -1.25
     desafios
    -1.23
     continúas
    -1.20
     berücksich
    -1.19
     diversos
    -1.16
     critères
    -1.16
    POSITIVE LOGITS
     areas
    1.66
     from
    1.48
     items
    1.47
     with
    1.47
     individuals
    1.39
     area
    1.38
     fifteen
    1.21
     Lingkungan
    1.21
     situations
    1.19
     parts
    1.18
    Activations Density 0.039%

    No Known Activations