© 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-3-12B-IT
    5. 12-GEMMASCOPE-2-RES-16K
    7. 2664
    Prev
    Next
    INDEX
    Explanations

    The lists provided (`MAX_ACTIVATING_TOKENS` and `TOKENS_AFTER_MAX_ACTIVATING_TOKEN`) suggest a pattern where the primary token is often followed by a digit. However, the `TOP_POSITIVE_LOGITS` and `TOP_ACTIVATING_TEXTS` seem to point towards specific programming keywords, technical terms, or single-letter identifiers followed by a number or another identifier.Let's re-evaluate based on the common elements.- `MAX_ACTIVATING_TOKENS`: `/`, `Y`, `C`, `Kotlin`, `Laser`, `K`, `GPU`, `GR`, `No`, `Blue`- `TOKENS_AFTER_MAX_ACTIVATING_TOKEN`: `2`, `Y`, `C`, `Kotlin`, `Laser`, `K`, `GPU`, `GR`, `No`, `Blue`This is *very* peculiar. It looks like the `MAX_ACTIVATING_TOKENS` and `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` lists are identical or very similar, which defeats the purpose of looking at "tokens after".2-letter abbreviations and capitalized words

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-it/resid_post/layer_12_width_16k_l0_medium
    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
    blooded
    2.30
    ars
    2.03
    我
    1.98
    닭
    1.87
    Ook
    1.83
    zelfde
    1.82
    meye
    1.81
     pesky
    1.81
     maanden
    1.80
    ara
    1.80
    POSITIVE LOGITS
    ]//
    1.86
    ](\
    1.85
     И
    1.70
     ограни
    1.70
    )$}
    1.70
    )$
    1.66
     использу
    1.65
    ></
    1.63
    \}=\
    1.59
     I
    1.58
    Activations Density 0.233%

    No Known Activations