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. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 13839
    Prev
    Next
    INDEX
    Explanations

    characters involved in criminal activities or undercover operations

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

    Third-person pronouns preceding verbs

    np_acts-logits-general · gemini-2.0-flash

    names followed by actions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71
    Prompts (Dashboard)
    36,864 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
     meines
    -0.71
     Wicidata
    -0.71
     mojej
    -0.66
     meiner
    -0.65
     kadang
    -0.60
     saya
    -0.57
    ScopeManager
    -0.57
    optString
    -0.57
    SPATH
    -0.57
    imetsu
    -0.56
    POSITIVE LOGITS
     threatens
    0.71
     convinces
    0.71
     flashback
    0.68
     drunken
    0.68
     Meanwhile
    0.67
     unknowingly
    0.66
     accidentally
    0.66
     overheard
    0.66
     unwittingly
    0.65
     blackmail
    0.65
    Activations Density 0.562%

    No Known Activations