© 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-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 86778
    Prev
    Next
    INDEX
    Explanations

    reddish, urlopen, damage

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

    The neuron is looking for multi-syllable subword pieces – i.e. WordPiece/subword tokens that correspond to stems or affixes (like “fecha” + “dura” in “fechadura,” “urlopen,” “reddish,” “damage,” “cursor,” etc.).

    oai_token-act-pair · o4-miniTriggered 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
    癀
    -2.86
    труд
    -2.70
     また
    -2.69
     It
    -2.66
    ."
    -2.47
     he
    -2.42
    ↵↵
    -2.39
    喾
    -2.39
     この
    -2.31
     οπο
    -2.31
    POSITIVE LOGITS
    3
    3.36
    2
    3.30
    1
    3.28
    0
    3.20
    6
    3.16
    7
    2.83
    5
    2.80
     itſelf
    2.64
     Ukraina
    2.50
     zul
    2.47
    Activations Density 0.001%

    No Known Activations