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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 1133
    Prev
    Next
    INDEX
    Explanations

    programming syntax and formatting indicators

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

    numerical values, letters, or scores in programming or mathematical contexts, especially as they appear in return statements, assignments, conditionals, or structured output.

    oai_token-act-pair · gpt-4.1-2025-04-14Triggered by @ywhp2c8tjm

    code snippets containing programming keywords and operators.

    oai_token-act-pair · gemini-2.0-flashTriggered by @ywhp2c8tjm

    Code or mathematical expressions

    np_acts-logits-general · gemini-2.0-flash

    true/false evaluation

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    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
    parsedMessage
    -0.64
    transQ
    -0.63
    octanol
    -0.61
    AndEndTag
    -0.56
    uxxxx
    -0.56
    脚注の使い方
    -0.56
     betweenstory
    -0.55
    таратура
    -0.54
    ſelben
    -0.54
    TagNumber
    -0.54
    POSITIVE LOGITS
     because
    0.40
    hyrchwyd
    0.38
     unless
    0.35
    unless
    0.33
     rather
    0.32
    身
    0.32
     immediate
    0.30
     automáticamente
    0.30
     due
    0.29
     الرياضيه
    0.29
    Activations Density 1.326%

    No Known Activations