© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 86835
    Prev
    Next
    INDEX
    Explanations

    offset of

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

    The neuron detects numerical adjustment terms—especially “offset” (and related forms like offX/Y/Z) as well as drift, gain, and similar calibration or shift parameters.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     fantástica
    -1.68
    ?”
    -1.62
    禤
    -1.59
    糴
    -1.53
     kommunizieren
    -1.52
     visualizar
    -1.51
    IA
    -1.48
    IES
    -1.45
     estimular
    -1.44
    standing
    -1.43
    POSITIVE LOGITS
    '
    1.77
     of
    1.75
     &
    1.63
    Oye
    1.43
    ledad
    1.41
    麻辣
    1.40
     independ
    1.39
    柠檬
    1.38
    gründung
    1.36
     én
    1.34
    Activations Density 0.030%

    No Known Activations