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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 15-GEMMASCOPE-RES-16K
    6. 8549
    Prev
    Next
    INDEX
    Explanations

    text that highlights the importance of specific topics or themes

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 15-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_15/width_16k/average_l0_78
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Hook Layer
    15
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     peor
    -0.67
    ImageField
    -0.65
    Unnamed
    -0.60
     secours
    -0.57
     vapeur
    -0.57
    いざ
    -0.57
    nommen
    -0.56
     generales
    -0.54
     coklat
    -0.54
    ardia
    -0.54
    POSITIVE LOGITS
     emphasise
    1.31
     emphasize
    1.27
     emphasis
    1.27
     Emphasis
    1.25
     emphasizes
    1.18
     emphas
    1.17
     emphasised
    1.15
     emphasizing
    1.12
     emphasized
    1.12
     highlights
    1.11
    Activations Density 0.132%

    No Known Activations