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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 1266
    Prev
    Next
    INDEX
    Explanations

    special formatting or structural markers in the text

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

    occurrences of numerical values and references to years

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

    technical terms and questions

    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-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    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
    GeneratedValue
    -0.54
    pret
    -0.53
    hoven
    -0.53
    واس
    -0.50
     Rosenberg
    -0.49
     Martinez
    -0.48
     RELATIONS
    -0.46
     late
    -0.46
    apiVersion
    -0.46
     Bad
    -0.46
    POSITIVE LOGITS
    存于互联网档案馆
    0.86
     Numerade
    0.78
    awtextra
    0.75
    expandindo
    0.73
    featureID
    0.73
    новништво
    0.72
     initComponents
    0.72
    SharedDtor
    0.71
     propOrder
    0.71
     jScrollPane
    0.68
    Activations Density 0.027%

    No Known Activations