© 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. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 13129
    Prev
    Next
    INDEX
    Explanations

    instances of the word "has."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    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
     themselves
    -0.51
     själva
    -0.47
    returnValue
    -0.47
    Preference
    -0.46
     they
    -0.46
     mereka
    -0.46
     forKey
    -0.43
    LikeLike
    -0.43
     Vie
    -0.42
    Expire
    -0.42
    POSITIVE LOGITS
    s
    0.84
    has
    0.81
     Has
    0.77
    Has
    0.75
    IntoConstraints
    0.75
    is
    0.74
    does
    0.73
     has
    0.73
    HAS
    0.71
     setIs
    0.71
    Activations Density 0.113%

    No Known Activations