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-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 1058
    Prev
    Next
    INDEX
    Explanations

    mentions of physical restraints or bindings

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

    Objects attached to ropes

    np_acts-logits-general · gemini-2.0-flash

    rope, hanging, lanyard, collar

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     kasarigan
    -0.53
     StatelessWidget
    -0.52
    makeText
    -0.47
    extAlignment
    -0.41
    MarshalTo
    -0.40
     utafitiHapana
    -0.38
     насељу
    -0.38
    astify
    -0.37
    GeneratedMessage
    -0.37
     decisión
    -0.36
    POSITIVE LOGITS
     rope
    0.77
     ropes
    0.71
     Rope
    0.68
     hanging
    0.66
     loops
    0.64
    Rope
    0.63
     cable
    0.62
     dangling
    0.62
     cables
    0.62
     loop
    0.62
    Activations Density 0.500%

    No Known Activations