© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 942
    Prev
    Next
    INDEX
    Explanations

    mentions of artists, designers, and elements of creative work

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

    references to artists, designers, or creative professionals.

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi

    attribution/credit lines that name creators or collaborators of works, often listing roles and proper names (e.g., artists, designers, photographers, mixers) and introduced by attribution cues.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    attribution phrases indicating the creator or author of creative works, particularly following "by".

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    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
    withIdentifier
    -0.43
     egentligen
    -0.42
    ContentAsync
    -0.42
     snippetHide
    -0.41
    ChildScrollView
    -0.41
     MaterialApp
    -0.40
    orologio
    -0.40
    MemoryWarning
    -0.37
     désolés
    -0.37
     tarifs
    -0.37
    POSITIVE LOGITS
     Administrativna
    0.59
    :✨
    0.52
    )++;
    0.50
    EndInit
    0.49
     grooms
    0.47
     JDK
    0.46
     KZ
    0.46
     QLabel
    0.45
     LayoutInflater
    0.45
    rizona
    0.44
    Activations Density 0.126%

    No Known Activations