© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 5555
    Prev
    Next
    INDEX
    Explanations

    mathematical or technical symbols and notations in the text

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

    mathematical and scientific terminology, particularly related to physics equations and technical language.

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi

    The neuron identifies the conjunction "or" between items in lists or alternatives, especially in technical and scientific contexts.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    phrases that present alternatives or equivalences between terms, especially markers introducing aliases or clarifications.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    the word "or" when used to present alternatives or equivalences in technical and formal contexts.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    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
    2
    -0.36
      
    -0.32
     before
    -0.31
    1
    -0.31
     against
    -0.31
     during
    -0.31
    ↵
    -0.30
    3
    -0.28
     while
    -0.28
     had
    -0.28
    POSITIVE LOGITS
     autorytatywna
    1.08
    Tikang
    1.05
     ویکی‌پدی
    0.97
     queſta
    0.97
    <unused3>
    0.91
    <unused28>
    0.91
    <unused52>
    0.90
    [@BOS@]
    0.90
    <unused68>
    0.90
    <pad>
    0.90
    Activations Density 0.358%

    No Known Activations