© 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. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 63429
    Prev
    Next
    INDEX
    Explanations

    attributes related to characteristics and comparisons

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

    big, small, alive

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

    instances of a user asking for or referring to their own name (e.g., "What is my name?").

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0

    direct questions or prompts asking for the speaker’s name or personal identity (requests to guess who they are).

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     AssemblyCompany
    -0.55
    XmlAccessorType
    -0.52
    BagLayout
    -0.48
    ölkerung
    -0.48
     وتسجيلات
    -0.47
    RegressionTest
    -0.45
    httphttps
    -0.44
    faßt
    -0.43
     manageable
    -0.42
     říká
    -0.42
    POSITIVE LOGITS
     مشين
    0.56
     snippetHide
    0.43
     peuple
    0.41
     deleteById
    0.41
     wifi
    0.41
     odeur
    0.41
     scanf
    0.40
     éter
    0.40
    verwijspagina
    0.40
    destroyAll
    0.40
    Activations Density 0.275%

    No Known Activations