© 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
    2. Gemma-3-27B-IT
    3. 27-GEMMASCOPE-2-TRANSCODER-262K
    4. 198835
    Prev
    Next
    INDEX
    Explanations

    ... An estimated 20 billion disposable diapers...", "Your extrapolation to AI is very logical. If we accept that the impetus for "animal rights" stems from human empathy...", "Also try to keep in mind that money is a tool..."The most frequent MAX_ACTIVATING_TOKEN is 'that'.The tokens immediately following 'that' in TOKENS_AFTER_MAX_ACTIVATING_TOKEN are: 'frustrating', 'doing', 'gender', 'the', 'the', 'technology', 'governments', 'disposable', 'the', 'money'.Let's look at the TOP_ACTIVATING_TEXTS to see how 'that' is used.Example 1: "We all know

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_27_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
    コスメ
    0.37
     насыщен
    0.36
     combinations
    0.35
     Voltaire
    0.35
     cosm
    0.35
     Fuchs
    0.34
     Intel
    0.34
     ferment
    0.34
     ఉన్నాయి
    0.34
     combinar
    0.34
    POSITIVE LOGITS
    ?:
    0.39
    iml
    0.38
    Dre
    0.38
    kolah
    0.37
     διαφορε
    0.37
    minipage
    0.37
    hale
    0.36
    Stere
    0.36
    摩
    0.36
    needed
    0.35
    Activations Density 0.001%

    No Known Activations