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-2-2B
    4. Residual Stream - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 868
    Prev
    Next
    INDEX
    Explanations

    instructions or actions to improve effectiveness in organizational settings

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

    directives for improving organizational culture and engagement

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

    use feedback to promote accountability

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    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
    sizeCache
    -0.99
     protoimpl
    -0.92
    GEBURTSDATUM
    -0.90
    tagHelperRunner
    -0.87
    Hentet
    -0.86
     незавершена
    -0.84
     صوتيه
    -0.84
    Datuak
    -0.81
    OGND
    -0.81
    HideFlags
    -0.80
    POSITIVE LOGITS
     Keep
    0.88
     keep
    0.85
    Keep
    0.79
     Always
    0.77
     Know
    0.77
    Know
    0.74
    keep
    0.72
     always
    0.72
     don
    0.71
     know
    0.68
    Activations Density 0.024%

    No Known Activations