© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-2B
    7. Residual Stream - 16k
    9. 4-GEMMASCOPE-RES-16K
    11. 1841
    Prev
    Next
    INDEX
    Explanations

    references to humor, particularly dark or self-deprecating humor

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 4-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_4/width_16k/average_l0_124
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.4.hook_resid_post
    Hook Layer
    4
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    AllowUser
    -0.49
    神宮
    -0.46
     küpe
    -0.45
    وزی
    -0.42
    endpush
    -0.41
     насеље
    -0.41
     immobilier
    -0.40
     serce
    -0.40
     Tabor
    -0.40
     citta
    -0.40
    POSITIVE LOGITS
     humor
    1.86
     humour
    1.81
     joke
    1.79
     Humor
    1.77
    humor
    1.72
     humorous
    1.68
     comedy
    1.67
     jokes
    1.65
     comedic
    1.62
    Humor
    1.57
    Activations Density 0.430%

    No Known Activations