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. 686
    Prev
    Next
    INDEX
    Explanations

    mathematical expressions and calculations

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

    Mathematical division or fractions

    np_acts-logits-general · gemini-2.0-flash

    division by two

    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
     cenar
    -0.56
     zatím
    -0.54
    ⓘ
    -0.48
     omnibus
    -0.48
    ष्य
    -0.48
    inOrder
    -0.47
    }));
    
    -0.47
     الحره
    -0.46
    istible
    -0.44
    })));
    -0.44
    POSITIVE LOGITS
     Divide
    0.73
     divide
    0.72
    divide
    0.69
    quelize
    0.67
     halves
    0.65
     divided
    0.65
     divides
    0.62
    Divide
    0.61
     dividing
    0.60
     half
    0.60
    Activations Density 0.533%

    No Known Activations