© 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. 5733
    Prev
    Next
    INDEX
    Explanations

    references to theoretical models and their predictions in a scientific or technical context

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

    technical or computational language related to modeling, measurement, and interpolation parameters.

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

    theoretical models and comparisons between predicted and measured values.

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

    technical, formal descriptions of models, algorithms, measurements, and control systems, including code-like identifiers and stepwise procedural language.

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

    technical descriptions of computational models, simulations, and comparisons between predicted and measured values in scientific or engineering contexts.

    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
     ExecuteAsync
    -0.74
     surla
    -0.67
    OGND
    -0.57
    BASEPATH
    -0.56
    astify
    -0.56
    PreExecute
    -0.54
    Tikang
    -0.53
    CloseOperation
    -0.53
    uxxxx
    -0.53
     allAfrica
    -0.52
    POSITIVE LOGITS
     comparison
    0.34
    match
    0.33
     compare
    0.33
     Comparison
    0.33
    Match
    0.32
     Ternyata
    0.31
     اشت
    0.30
     compares
    0.30
     مت
    0.30
     matching
    0.29
    Activations Density 0.438%

    No Known Activations