© 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. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 31-GEMMASCOPE-RES-16K
    11. 6779
    Prev
    Next
    INDEX
    Explanations

    text related to numerical data and references to scientific findings

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    This neuron activates for document formatting elements like line breaks, section separators, footnote markers, and numbered list items, particularly in academic or technical documents.

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

    begin-of-sequence tokens and structural markers that denote the start of new documents or sections.

    oai_token-act-pair · claude-4-5-haikuTriggered by @emiglarou
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 31-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_31/width_16k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.31.hook_resid_post
    Hook Layer
    31
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    //
    -0.53
     faſt
    -0.51
     neceff
    -0.51
     poffe
    -0.50
     fubject
    -0.48
     itſelf
    -0.47
     INTERESAR
    -0.45
     Houſe
    -0.45
     fhort
    -0.44
    inSlope
    -0.43
    POSITIVE LOGITS
    Captor
    0.47
    ۜ
    0.46
    setCellStyle
    0.45
    findOneAndUpdate
    0.45
    UnsafeEnabled
    0.44
     SANS
    0.43
     tartalmaz
    0.43
     decker
    0.42
    IActionResult
    0.42
    UNSIGNED
    0.42
    Activations Density 1.537%

    No Known Activations