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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 38-GEMMASCOPE-RES-16K
    11. 15676
    Prev
    Next
    INDEX
    Explanations

    references to harm and destruction, particularly in contexts involving environmental, personal, or reputational damage

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.38.hook_resid_post
    Hook Layer
    38
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    SerializeField
    -0.39
    bedarf
    -0.35
     problém
    -0.33
     Problems
    -0.33
     problems
    -0.33
     challenges
    -0.32
     ComVisible
    -0.32
     improved
    -0.32
     problem
    -0.32
     probl
    -0.32
    POSITIVE LOGITS
     livelihoods
    0.99
     reputations
    0.99
    せっかく
    0.98
    innocent
    0.98
     integrity
    0.96
     innocent
    0.94
    好不容易
    0.93
     precious
    0.91
    貴重な
    0.90
    precious
    0.89
    Activations Density 0.958%

    No Known Activations