© 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. 10-GEMMASCOPE-RES-131K
    7. 68408
    Prev
    Next
    INDEX
    Explanations

    why can't

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

    The neuron detects the contraction “can’t” (i.e. the “can + ’t” token pair) and its immediate function‐word context expressing inability or questioning why something “can’t” be.

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

    complaint-style questions expressing inability, especially “Why can’t/couldn’t …?” constructions about something not working or being possible.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    the phrase "can't" (especially in questions or complaints expressing frustration or inability).

    oai_token-act-pair · claude-4-5-sonnetTriggered 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
    .”
    -1.38
     gerais
    -1.34
     caseros
    -1.32
     expectations
    -1.31
     any
    -1.28
    ContainsKey
    -1.27
     aislados
    -1.26
     eigene
    -1.25
     comerciais
    -1.25
     nacidos
    -1.24
    POSITIVE LOGITS
    ↵
    1.53
    INE
    1.26
    ے
    1.16
     الآن
    1.15
    cadena
    1.15
    förmig
    1.13
     عندما
    1.13
    ческая
    1.13
    mær
    1.12
    बल
    1.11
    Activations Density 0.018%

    No Known Activations