© 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. 34-GEMMASCOPE-RES-131K
    4. 47856
    Prev
    Next
    INDEX
    Explanations

    numerical death tolls

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

    numerical counts of casualties (deaths, injuries, or missing people) in news articles.

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

    The neuron detects quantitative expressions—that is, standalone numbers or spelled‐out number words indicating counts.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    ⌜
    -0.87
    zoon
    -0.82
    neighbors
    -0.81
     alpes
    -0.79
     Эти
    -0.79
    Datas
    -0.78
    ventajas
    -0.75
    骄傲
    -0.75
     something
    -0.74
     zoon
    -0.74
    POSITIVE LOGITS
     people
    2.36
     persons
    1.94
     individuals
    1.40
    people
    1.40
     osób
    1.39
     died
    1.35
     person
    1.34
     were
    1.30
     personnes
    1.26
     personer
    1.23
    Activations Density 0.045%

    No Known Activations