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 - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 28111
    Prev
    Next
    INDEX
    Explanations

    specific materials and their concentrations in scientific contexts

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

    statistical analysis and models

    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-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
     reordered
    -0.34
     StringTokenizer
    -0.33
    ьаж
    -0.30
     urgently
    -0.29
    acamata
    -0.28
     uğ
    -0.28
    更好的
    -0.27
     odbior
    -0.27
    )))));
    -0.27
     endregion
    -0.27
    POSITIVE LOGITS
    windowFixed
    0.59
    RegistryLite
    0.54
    tonode
    0.52
    stateParams
    0.52
    PostExecute
    0.50
     لينك
    0.50
    ậc
    0.49
    MarshalTo
    0.49
    Soorten
    0.49
    DIPSETTING
    0.49
    Activations Density 0.201%

    No Known Activations