© 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
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 8274
    Prev
    Next
    INDEX
    Explanations

    terms related to statistical mechanics and percolation theory

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

    numerical values related to physical properties like temperature, energy levels, and concentration in scientific texts.

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

    technical, equation-heavy physics text referencing quantitative parameters, thresholds, rates, and regimes

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

    mathematical and scientific expressions describing physical parameters, thresholds, and comparative relationships in technical 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
    IsMutable
    -0.60
    ➟
    -0.58
    Personendaten
    -0.57
    évaluateur
    -0.54
    ſelf
    -0.52
     للمعارف
    -0.52
     nonUne
    -0.52
    رشف
    -0.52
     protoimpl
    -0.51
    contentLoaded
    -0.50
    POSITIVE LOGITS
     disambiguazione
    0.45
     situations
    0.39
     beyond
    0.39
     ab
    0.38
    ousel
    0.35
     Smyth
    0.35
     areas
    0.35
    行
    0.35
    tschaft
    0.34
     Wiktionnaire
    0.34
    Activations Density 0.928%

    No Known Activations