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. 10-GEMMASCOPE-RES-16K
    6. 10478
    Prev
    Next
    INDEX
    Explanations

    the word "whatever" and its variations, indicating a focus on expressions of indifference or lack of specificity

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 10-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_10/width_16k/average_l0_77
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    liferay
    -0.75
    en
    -0.68
    :^{
    -0.65
    instein
    -0.64
     \
    -0.62
     cen
    -0.62
    I
    -0.60
     Rat
    -0.59
    an
    -0.59
     Pack
    -0.59
    POSITIVE LOGITS
     Whatever
    1.40
    atever
    1.36
    Whatever
    1.31
     whatever
    1.27
    whatever
    1.20
     whichever
    1.15
     Whoever
    1.14
    osoever
    1.11
    Whoever
    1.10
    hichever
    1.08
    Activations Density 0.044%

    No Known Activations