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. 915
    Prev
    Next
    INDEX
    Explanations

    patterns of code structure and formatting

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

    "code" followed by hexadecimal values

    np_acts-logits-general · gemini-2.0-flash

    code: followed by hex codes

    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
     viewWillAppear
    -0.45
     (!(
    -0.42
     pleinement
    -0.42
    (!(
    -0.39
     férias
    -0.38
    ster
    -0.37
    upsert
    -0.37
    blij
    -0.36
     maioria
    -0.35
    (++
    -0.34
    POSITIVE LOGITS
     code
    1.48
     Code
    1.42
    Code
    1.39
     setCode
    1.35
    code
    1.35
     CODE
    1.32
     getCode
    1.29
    CODE
    1.27
     Codes
    1.26
     codes
    1.23
    Activations Density 0.154%

    No Known Activations