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-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 80191
    Prev
    Next
    INDEX
    Explanations

    sequences of whitespace or indentation patterns in code

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

    programming code snippets

    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
     Administrativna
    -0.73
     unknownFields
    -0.68
     estekak
    -0.62
    ConstraintMaker
    -0.61
    Datuak
    -0.60
    évaluateur
    -0.59
     '\\;'
    -0.59
     ErrIntOverflow
    -0.57
    extAlignment
    -0.57
     otomatig
    -0.56
    POSITIVE LOGITS
    ConverterFactory
    0.40
     uğ
    0.35
     propOrder
    0.34
     iub
    0.32
     sepen
    0.31
    ganza
    0.31
    InstanceState
    0.31
    帳に追加
    0.28
    Queue
    0.28
     hor
    0.27
    Activations Density 0.002%

    No Known Activations