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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    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. 2371
    Prev
    Next
    INDEX
    Explanations

    phrases related to dividing and cutting items or materials

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

    Dividing or splitting into parts

    np_acts-logits-general · gemini-2.0-flash

    splitting into halves

    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
     AssemblyTitle
    -0.59
     يتيمه
    -0.50
    RenderAtEndOf
    -0.48
    AlterField
    -0.48
    fromnode
    -0.47
    neſs
    -0.46
    OwnProperty
    -0.45
     Fondation
    -0.45
     désolés
    -0.44
     defStyle
    -0.43
    POSITIVE LOGITS
     halves
    1.18
     half
    1.13
     halve
    1.12
     halved
    1.01
    half
    1.00
     Half
    0.94
    Half
    0.92
    半分
    0.90
     HALF
    0.88
     mitad
    0.84
    Activations Density 0.106%

    No Known Activations