© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-2B
    3. 22-GEMMASCOPE-TRANSCODER-16K
    4. 10520
    Prev
    Next
    INDEX
    Explanations

    numerical data and the word 'and'

    oai_token-act-pair · gemini-2.0-flash

    math/code context

    np_max-act-logits · gemini-2.0-flash

    say "numbers" Method used: 2 — tokens are numeric, so say "numbers"

    np_max-act-logits · gpt-5-miniTriggered by @chenshw0109
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_22/width_16k/average_l0_15
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.22.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    7
    -0.79
    8
    -0.76
    6
    -0.67
    七
    -0.63
     seven
    -0.60
     EIGHT
    -0.59
     seventh
    -0.58
    八
    -0.58
     eight
    -0.57
    9
    -0.56
    POSITIVE LOGITS
    UrlResolution
    0.82
     مشين
    0.69
    ScopeManager
    0.68
     سكانية
    0.63
    ViewFeatures
    0.63
     estekak
    0.63
    Pratique
    0.62
    PyExc
    0.61
     agrí
    0.61
    MLLoader
    0.59
    Activations Density 1.206%

    No Known Activations