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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 18-GEMMASCOPE-RES-16K
    6. 3761
    Prev
    Next
    INDEX
    Explanations

    conversational elements and expressions in informal dialogue

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

    French "je" or "tu" followed by a verb

    np_acts-logits-general · gemini-2.0-flash

    je / tu statements

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.hook_resid_post
    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
     Herrschaft
    -0.69
     near
    -0.67
     peman
    -0.67
     common
    -0.66
     different
    -0.66
     higher
    -0.64
     great
    -0.62
     cambian
    -0.62
     very
    -0.62
     crean
    -0.61
    POSITIVE LOGITS
     дописавши
    1.10
     ControllerBase
    0.91
     فريبيس
    0.86
    featureID
    0.86
     Мексичка
    0.84
    msgTypes
    0.83
    RenderAtEndOf
    0.76
    parsedMessage
    0.73
     Roskov
    0.71
     мәкал
    0.71
    Activations Density 0.056%

    No Known Activations