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. 1407
    Prev
    Next
    INDEX
    Explanations

    references to negligence and its consequences in legal contexts

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

    Possessive pronoun followed by "actions/conduct"

    np_acts-logits-general · gemini-2.0-flash

    possessive followed by action or conduct

    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
    ClientSize
    -0.64
    HtmlAttribute
    -0.59
    GenerationType
    -0.59
    FXMLLoader
    -0.59
     CascadeType
    -0.58
    TagHelper
    -0.58
     advantage
    -0.57
     Vorteil
    -0.57
     protoimpl
    -0.56
    UnusedPrivate
    -0.55
    POSITIVE LOGITS
     actions
    1.92
     Actions
    1.55
    actions
    1.52
     ACTIONS
    1.40
    Actions
    1.31
     behavior
    1.25
     behaviors
    1.24
     behaviour
    1.20
    ACTIONS
    1.17
     action
    1.17
    Activations Density 0.475%

    No Known Activations