Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 1978
    Prev
    Next
    INDEX
    Explanations

    punctuation marks, particularly quotation marks and sentences

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

    foreign words and citations

    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
     InputDecoration
    -0.65
    __);
    -0.64
    +#+#
    -0.61
    GHIJKLM
    -0.60
     الرياضيه
    -0.57
     NDL
    -0.56
     Italijani
    -0.55
    IntoConstraints
    -0.54
    UPA
    -0.54
    }}_{\
    -0.54
    POSITIVE LOGITS
     computadoras
    0.63
     anún
    0.62
     Bewußt
    0.60
    uxxxx
    0.60
     cœurs
    0.59
     regalías
    0.59
     dueños
    0.58
     feroit
    0.57
     argint
    0.57
     enfans
    0.57
    Activations Density 0.033%

    No Known Activations