© 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. 15-GEMMASCOPE-2-TRANSCODER-262K
    4. 14399
    Prev
    Next
    INDEX
    Explanations

    Let's analyze the provided lists to understand the neuron's behavior:* **MAX_ACTIVATING_TOKENS**: Victoria, supports, Court, the, pil, pile, Victoria, pile, Lake, the* **TOKENS_AFTER_MAX_ACTIVATING_TOKEN**: ., R, of, Law, adores, (, (, Turk, SecondLooking at these, several patterns emerge:1. **"Victoria" followed by "."**: This is a common occurrence for proper nouns ending sentences.2. **"Court" followed by "of"**: This strongly suggests phrases like "Court of Law" or similar legal contexts.3. **"pile"**: This appears multiple times. One instance is followed by '(', another by 'Turk'. It might be part of a word or a specific construct.4. **"Lake" followed by "the"**: Suggests "Lake the..." (though "Lake Victoria" is more common, here it's "Lake" then "the").Let's look at `TOP_POSITIVE_LOGITS`:startswith, za, niv, wide, ുണ്ട്, Ne, 🦐, utz, waar, 两大These are very diverse and don't immediately form a clear linguistic pattern. They might indicate specific, perhaps less common, contexts or languages.Let's look at `TOP_ACTIVATING_TEXTS`:* "into Lake Victoria. The White Nile then *exits* Lake Victoria and flows north into Sudan." -> **Victoria, Lake*** "Permanent Court of Arbitration (PCA) in The Hague, which in 2016 ruled that China's claims to

    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_15_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
     -"
    0.57
     "'
    0.55
     bushy
    0.49
     condos
    0.48
     deciduous
    0.48
     contour
    0.47
     ?",
    0.47
     बॉबी
    0.47
    ꒷
    0.46
     peaceful
    0.46
    POSITIVE LOGITS
    startswith
    0.63
    za
    0.59
    niv
    0.54
    wide
    0.53
    ുണ്ട്
    0.52
    Ne
    0.51
    🦐
    0.51
    utz
    0.50
    waar
    0.50
    两大
    0.50
    Activations Density 0.001%

    No Known Activations