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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 18-GEMMASCOPE-RES-16K
    6. 8650
    Prev
    Next
    INDEX
    Explanations

    discussions surrounding legal and financial matters, specifically related to obligations and disclosures

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

    "never"

    np_acts-logits-general · gemini-2.0-flash

    never completed or consummated

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.hook_resid_post
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ertale
    -0.63
    ########.
    -0.61
    AutoresizingMask
    -0.59
    stdc
    -0.58
     AssemblyCulture
    -0.58
     StatefulWidget
    -0.58
     AspNetCore
    -0.56
    TintMode
    -0.56
    Controllo
    -0.56
     automaticamente
    -0.55
    POSITIVE LOGITS
     never
    2.45
    never
    2.13
     Never
    2.02
    Never
    1.95
     nunca
    1.82
     nigdy
    1.73
    nunca
    1.66
     NEVER
    1.65
    NEVER
    1.62
     nooit
    1.58
    Activations Density 0.529%

    No Known Activations