© 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-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 1333
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate causal relationships or dependencies, particularly in scientific or analytical contexts

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

    Follows "result in" or "resulting in"

    np_acts-logits-general · gemini-2.0-flash

    resulting in larger damage or susceptibility

    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-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    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
    abancı
    -0.35
     debido
    -0.35
     poiché
    -0.33
    Somewhere
    -0.33
     grâce
    -0.33
    Brazilian
    -0.32
     vez
    -0.31
    Étymologie
    -0.31
    American
    -0.30
    because
    -0.30
    POSITIVE LOGITS
    enumii
    0.59
     appreciably
    0.55
     createSprite
    0.54
    MethodManager
    0.53
    HtmlAttribute
    0.51
     appreciable
    0.49
     dAtA
    0.48
    fromUtf
    0.46
    parsedMessage
    0.46
     dramatically
    0.45
    Activations Density 0.933%

    No Known Activations