© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 16546
    Prev
    Next
    INDEX
    Explanations

    technical terms and identifiers

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

    The neuron is firing most strongly on unusually long or rare (low‐frequency) tokens—that is, it flags infrequent/rare subwords in the text.

    oai_token-act-pair · o4-mini
    Triggered 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
    Negative Logits
    次代
    -0.97
    laint
    -0.93
    ὸ
    -0.92
    一月
    -0.90
     ΤΟ
    -0.88
     nostru
    -0.86
    Oby
    -0.86
    nós
    -0.86
     bloss
    -0.85
     링크
    -0.82
    POSITIVE LOGITS
     unangemessene
    0.95
     kriminal
    0.89
     asisten
    0.85
     ilg
    0.85
     potrebbe
    0.85
    izat
    0.85
     guste
    0.84
    ục
    0.84
    щенный
    0.83
    cities
    0.83
    Activations Density 0.076%

    No Known Activations