© 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. 6016
    Prev
    Next
    INDEX
    Explanations

    structured data and references to coding constructs

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

    programming code, specifically code related to method execution and parameters.

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

    code patterns showing member/property access and chaining in programming (e.g., dot/arrow accesses and indexed lookups).

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

    object property or method access patterns in code, particularly when chaining multiple accesses with dot notation.

    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
     disambiguazione
    -0.69
    PhysRev
    -0.66
    cotch
    -0.49
    >`;
    -0.48
    CPtr
    -0.48
     deflate
    -0.47
    scriptId
    -0.46
     dbName
    -0.44
     الد
    -0.44
     gyhoeddwyd
    -0.44
    POSITIVE LOGITS
    GOTREF
    0.44
    MessageTagHelper
    0.40
    richTextPanel
    0.39
    findpost
    0.37
    aarrggbb
    0.36
     tinggal
    0.36
    GenerationType
    0.35
     rempliss
    0.34
    addPreferredGap
    0.33
     remplir
    0.31
    Activations Density 0.201%

    No Known Activations