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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-2B
    7. Residual Stream - 65k
    9. 0-GEMMASCOPE-RES-65K
    11. 12884
    Prev
    Next
    INDEX
    Explanations

    the word "where" and its variations, indicating a focus on locations or places

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 0-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_0/width_65k/average_l0_73
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.0.hook_resid_post
    Hook Layer
    0
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     propOrder
    -0.90
    httphttps
    -0.81
     InputDecoration
    -0.78
    setVerticalGroup
    -0.75
    씩
    -0.71
     Administrativna
    -0.70
     <<<<<<<<<<<<<<
    -0.70
    FieldDescriptor
    -0.70
     Monfieur
    -0.69
     betreft
    -0.69
    POSITIVE LOGITS
     where
    1.23
    where
    1.02
     donde
    0.81
     what
    0.70
     où
    0.69
     how
    0.69
     onde
    0.66
     hvor
    0.63
     Where
    0.60
     WHERE
    0.59
    Activations Density 0.099%

    No Known Activations