Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. MLP Out - 65k
    5. 18-GEMMASCOPE-MLP-65K
    6. 52
    Prev
    Next
    INDEX
    Explanations

    syntactic structures and programming constructs

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-mlp-65k
    Configuration
    google/gemma-scope-2b-pt-mlp/layer_18/width_65k/average_l0_88
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.18.hook_mlp_out
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    openConnection
    -0.51
    __*/
    -0.50
    Datuak
    -0.47
    jandro
    -0.47
     flesta
    -0.47
     melu
    -0.46
     nối
    -0.45
    Skocz
    -0.45
     مُعرِّف
    -0.44
    /***/
    -0.43
    POSITIVE LOGITS
     Diſ
    0.71
     Reſ
    0.68
     Inſ
    0.66
     '\\;'
    0.65
    AntiForgeryToken
    0.65
     Anſ
    0.65
    ſelf
    0.65
     uſe
    0.64
     whoſe
    0.63
     Италијани
    0.63
    Activations Density 0.055%

    No Known Activations