© 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-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 3107
    Prev
    Next
    INDEX
    Explanations

    let's and similar phrases

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly activates on the substring “Let” (often at the start of a token), e.g. in words or identifiers like “Let’s,” “Letchworth,” or a code macro named “LET.”

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     būs
    -0.88
    ͋
    -0.88
     mentre
    -0.84
    âteaux
    -0.83
    ですが
    -0.83
    g
    -0.82
     izman
    -0.81
     Exhaus
    -0.78
     Synchronous
    -0.76
    丟
    -0.76
    POSITIVE LOGITS
    tering
    1.25
     ensamb
    0.97
    鲤
    0.93
    bede
    0.92
    dled
    0.91
    uce
    0.91
    atau
    0.90
     cartón
    0.90
    niem
    0.90
     karton
    0.90
    Activations Density 0.033%

    No Known Activations