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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 16-GEMMASCOPE-RES-65K
    6. 53509
    Prev
    Next
    INDEX
    Explanations

    themes of evasion and fleeing from danger or legal consequences

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

    avoid prosecution, arrest, or attacks

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 16-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_16/width_65k/average_l0_128
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.16.hook_resid_post
    Hook Layer
    16
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    icoot
    -0.68
    +:+
    -0.63
     themſelves
    -0.61
     chofe
    -0.57
     caufe
    -0.57
    createComponent
    -0.56
    igshid
    -0.56
     cauſe
    -0.56
    Zeneca
    -0.56
     poffe
    -0.55
    POSITIVE LOGITS
    Autoritní
    0.56
     evade
    0.55
     interviewers
    0.54
     creditors
    0.52
     impending
    0.50
     interviews
    0.50
     interview
    0.49
    invokeLater
    0.49
    SuccessListener
    0.49
    <table>
    0.48
    Activations Density 0.047%

    No Known Activations