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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-2B
    5. 21-GEMMASCOPE-TRANSCODER-16K
    7. 14581
    Prev
    Next
    INDEX
    Explanations

    words and phrases about corruption and political scandals, especially those involving the Brazilian government and the Mensalão scandal.

    oai_token-act-pair · gemini-2.0-flash

    words associated with actions or statements

    np_token-act-pair-logits · gpt-4o-mini

    corruption scandals

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_21/width_16k/average_l0_13
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.21.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    +#+#
    -0.87
    ंदीखरीदारी
    -0.61
    rungsseite
    -0.61
    formik
    -0.60
     <<<<<<<<<<<<<<
    -0.60
     referenties
    -0.51
    Datuak
    -0.50
     jsPsych
    -0.49
    plusplus
    -0.49
    المكان
    -0.49
    POSITIVE LOGITS
     kick
    0.66
    DIPSETTING
    0.60
     Kick
    0.57
    kick
    0.56
    __':
    0.56
     consultato
    0.53
    Kick
    0.53
     middle
    0.52
    Influence
    0.52
     sweetheart
    0.52
    Activations Density 4.860%

    No Known Activations