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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 16350
    Prev
    Next
    INDEX
    Explanations

    elements related to structured data or APIs, particularly focusing on contact information and response logs

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

    Data elements in programming code, particularly in format strings, method calls, and object properties. Common patterns include variable names, field accesses with methods like getName() or getThread(), string formatting with delimiters like parentheses, commas that separate parameters, and identifiers like author, year, city, or name that represent data fields in structured formats.

    eleuther_acts_top20 · claude-3-7-sonnet-20250219Triggered by @lu-christina
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
    IsMutable
    -0.48
    exitRule
    -0.47
    yntaxException
    -0.45
     StatefulWidget
    -0.45
    RegistryLite
    -0.44
    ➯
    -0.44
    تقاوى
    -0.42
    Numerology
    -0.41
     viewDidLoad
    -0.41
    fjspx
    -0.40
    POSITIVE LOGITS
    webElementXpaths
    0.45
    riwal
    0.41
     Complement
    0.37
    jamientos
    0.36
     concre
    0.36
    драт
    0.35
    buttonBar
    0.35
    itects
    0.34
     complement
    0.34
     cohort
    0.34
    Activations Density 0.042%

    No Known Activations