© 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
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 131k
    9. 20-GEMMASCOPE-RES-131K
    11. 8196
    Prev
    Next
    INDEX
    Explanations

    terms specifically related to danger or harmfulness

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

    dangerous situations

    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
     Inscrivez
    -0.54
     increí
    -0.52
     sonre
    -0.52
     âmes
    -0.50
     Verständnis
    -0.49
     AssemblyTitle
    -0.48
    iconque
    -0.48
     miniaturka
    -0.48
     auroit
    -0.48
     Мексичка
    -0.48
    POSITIVE LOGITS
     Dangerous
    1.11
     dangerous
    1.10
    dangerous
    1.07
    Dangerous
    1.04
     DANG
    0.81
    EROUS
    0.77
    hazardous
    0.72
     hazardous
    0.70
     опас
    0.66
     Hazardous
    0.65
    Activations Density 0.008%

    No Known Activations