© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 26943
    Prev
    Next
    INDEX
    Explanations

    social media handles and names

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

    The neuron strongly fires on standalone capitalized tokens and name‐style chunks (e.g. parts of Twitter handles, artist names, titles), effectively spotting proper nouns and all‐caps 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
    -2.06
     or
    -2.03
     with
    -1.88
     this
    -1.83
     at
    -1.77
     for
    -1.76
     provide
    -1.74
     in
    -1.73
     helps
    -1.66
     reduce
    -1.66
    POSITIVE LOGITS
     ativar
    1.91
     素描
    1.77
     spéciaux
    1.74
    Forrás
    1.70
    鉺
    1.65
     aislamiento
    1.64
     atingir
    1.63
    Válasz
    1.63
    ܆
    1.59
     diriger
    1.58
    Activations Density 0.011%

    No Known Activations