© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 102259
    Prev
    Next
    INDEX
    Explanations

    node metastasis

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

    hyphenated technical terms in scientific and medical contexts.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    The neuron consistently fires on the second element of hyphenated compounds (e.g. “node” in lymph‐node, “steel” in stainless‐steel, “stay” in short‐stay, “vivo” in in‐vivo, “stamp” in time‐stamp, “left” in top‐left).

    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
     was
    -2.53
     började
    -1.74
     his
    -1.73
     Saltar
    -1.64
    %">
    -1.52
    ']=
    -1.47
    )
    
    -1.43
     försö
    -1.42
    ⁈
    -1.41
     latter
    -1.38
    POSITIVE LOGITS
     misschien
    1.73
     schuldig
    1.55
    “
    1.47
    なもの
    1.46
     potrebbero
    1.45
     verhind
    1.43
     each
    1.43
    igens
    1.39
    になったら
    1.37
    なのです
    1.34
    Activations Density 0.259%

    No Known Activations