© 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. 33887
    Prev
    Next
    INDEX
    Explanations

    Georgia, DISTRICT, Paper

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

    The neuron flags unusually rare or specific tokens—such as numbers, file‐path fragments, and low‐frequency proper nouns—while ignoring common words.

    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
     it
    -1.27
     this
    -1.21
     their
    -1.17
     from
    -1.14
    An
    -1.09
     It
    -1.09
    add
    -1.06
     when
    -1.05
    Anmerkungen
    -1.04
    énagement
    -1.03
    POSITIVE LOGITS
    がありますが
    1.70
    があり
    1.63
    があるので
    1.59
    もありました
    1.36
    もありますが
    1.35
    もあります
    1.25
    もあって
    1.23
     attente
    1.20
    があります
    1.20
    乁
    1.18
    Activations Density 0.002%

    No Known Activations