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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 20-GEMMASCOPE-RES-16K
    11. 6431
    Prev
    Next
    INDEX
    Explanations

    explicit references to problem-solving and modification in technical contexts

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

    problem-solving and action-oriented language, particularly related to technical troubleshooting and fixing issues.

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

    instructions for solving or fixing problems, particularly in technical contexts.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    solution-oriented troubleshooting language that proposes or reports specific fixes or modifications to resolve a problem.

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

    instructions or suggestions for fixing technical problems or making corrections.

    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
     CanadaChoose
    -0.46
     nélkül
    -0.45
    Попис
    -0.44
     presupposes
    -0.44
    CppMethod
    -0.41
    hetamine
    -0.41
    авто
    -0.41
    Viited
    -0.41
    ⓧ
    -0.40
    TintMode
    -0.40
    POSITIVE LOGITS
     corrected
    0.48
    ftagPool
    0.48
     reverse
    0.45
     виправи
    0.43
     fix
    0.43
     umge
    0.42
     IMPRO
    0.42
     correction
    0.42
     improved
    0.41
     remedy
    0.41
    Activations Density 0.424%

    No Known Activations