© 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
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 967
    Prev
    Next
    INDEX
    Explanations

    references to the necessity and efficiency of managing resources or operations

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

    phrases related to avoiding repetitive tasks or manual work.

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi

    language indicating that effort or steps are avoided or minimized—statements that something is not required, only needed once, or can be done without having to perform extra work.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    phrases describing the elimination or avoidance of a requirement or effort.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    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
     חיצוניים
    -0.56
    Autoritní
    -0.50
    providedIn
    -0.47
     Савезне
    -0.47
     Biôgrafia
    -0.44
    CloseOperation
    -0.42
    ніципалі
    -0.42
     rece
    -0.42
     AssemblyVersion
    -0.41
    djangoproject
    -0.41
    POSITIVE LOGITS
    findpost
    0.42
     äta
    0.40
     amanhã
    0.39
     niitä
    0.39
    echo
    0.39
    还得
    0.38
     unnecessarily
    0.37
     preocupação
    0.37
     camiset
    0.36
     menunggu
    0.36
    Activations Density 0.299%

    No Known Activations