© 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
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-2B
    7. Residual Stream - 16k
    9. 12-GEMMASCOPE-RES-16K
    11. 12342
    Prev
    Next
    INDEX
    Explanations

    symbols and mathematical notation related to expressions or equations in mathematical contexts

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

    The neuron fires on long technical identifiers and figure/image-like labels — strings with repeated words, many digits and dashes (e.g., image filenames or compound reference IDs).

    oai_token-act-pair · gpt-5-miniTriggered by @phamvanluong2003

    technical markup and code-like tokens, especially LaTeX/math commands, Markdown figure/link references, and programming-style identifiers or paths with underscores, hyphens, or dots.

    oai_token-act-pair · gpt-5Triggered by @areeb0102

    structured alphanumeric identifiers or codes in technical documents.

    oai_token-act-pair · deepseek-r1Triggered by @areeb0102

    fully qualified identifiers and structured namespace patterns (such as package paths, class hierarchies, and mathematical variable sequences).

    oai_token-act-pair · claude-4-5-haikuTriggered by @areeb0102
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 12-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_12/width_16k/average_l0_82
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.12.hook_resid_post
    Hook Layer
    12
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     betweenstory
    -1.30
     Италијани
    -1.19
    oredCriteria
    -1.12
    ArrowToggle
    -1.11
    NameInMap
    -1.09
    sizeCache
    -1.08
    ьаж
    -1.08
    OGND
    -1.06
    aarrggbb
    -1.04
    AsUp
    -1.02
    POSITIVE LOGITS
    \{\\
    0.68
     
    0.62
     the
    0.51
    enumi
    0.49
    ISTRATION
    0.45
    uests
    0.44
    UMNS
    0.44
    ITIONS
    0.44
     are
    0.44
    ercises
    0.43
    Activations Density 26.527%

    No Known Activations