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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 131k
    9. 9-GEMMASCOPE-RES-131K
    11. 53345
    Prev
    Next
    INDEX
    Explanations

    Superscript characters (^) used to denote chemical or mathematical notation, often appearing in scientific or technical contexts.

    eleuther_acts_top20 · claude-3-5-sonnet-20240620Triggered by @nrbln

    unusual or nonsensical characters and symbols in the text

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

    The marked tokens predominantly indicate specialized notations or symbols frequently used in scientific, technical, or mathematical contexts, such as superscripts, subscripts, chemical notation, and special characters in formatting-related expressions.

    eleuther_acts_top20 · gpt-4oTriggered by @nrbln

    Superscripted tokens, often chemical formulas or mathematical expressions.

    eleuther_acts_top20 · gemini-1.5-proTriggered by @nrbln

    This neuron appears to activate on sequences of characters that are not standard alpha-numeric characters, suggesting it detects unusual or non-standard text formatting or possibly "noise". It's also possible that it is activating on uncommon diacritical marks that are included alongside alpha-numeric characters.

    oai_token-act-pair · gemini-1.5-proTriggered by @nrbln

    superscript characters and symbols, particularly those used in scientific notation.

    oai_token-act-pair · claude-3-5-sonnet-20240620Triggered by @nrbln

    special formatting characters and symbols

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.9.hook_resid_post
    Hook Layer
    9
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ManagedBean
    -0.48
    fromnode
    -0.47
     autorytatywna
    -0.47
    Derbyniad
    -0.46
     catalytic
    -0.46
    Diweddarwch
    -0.45
     Stephenson
    -0.45
     Servicios
    -0.44
    migrationBuilder
    -0.44
    הערות
    -0.44
    POSITIVE LOGITS
    ̪
    0.52
    ̈́
    0.50
    ̲
    0.47
    ͖
    0.47
    ̸
    0.45
    ̻
    0.45
    ̶
    0.45
    ͋
    0.43
    ͫ
    0.41
    ͢
    0.40
    Activations Density 0.163%

    No Known Activations