© 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-3-27B-IT
    5. 44-GEMMASCOPE-2-TRANSCODER-262K
    7. 5767
    Prev
    Next
    INDEX
    Explanations

    say "impossible"

    np_max-act-logits · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_44_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    மாண
    0.39
    andered
    0.38
     chắn
    0.36
     headwinds
    0.36
    <unused347>
    0.35
    aminophen
    0.34
    డానికి
    0.34
     голову
    0.34
    筮
    0.34
    คร
    0.34
    POSITIVE LOGITS
     impossible
    2.08
     Impossible
    1.97
    Impossible
    1.97
    impossible
    1.88
     impossibile
    1.86
     imposible
    1.83
     imposs
    1.70
     impos
    1.63
     Im
    1.61
     impossibility
    1.59
    Activations Density 0.211%

    No Known Activations