© 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. 107077
    Prev
    Next
    INDEX
    Explanations

    Russian 'т' followed by punctuation

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

    The neuron is highly sensitive to the Cyrillic letter “т,” activating strongly on tokens that are exactly “т” or begin/end with that letter.

    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
    <0xA3>
    -0.98
     atmosfer
    -0.98
     apartament
    -0.95
    ですが
    -0.93
     Regist
    -0.84
     trist
    -0.82
     pittores
    -0.81
    regist
    -0.81
     imponer
    -0.81
     assust
    -0.80
    POSITIVE LOGITS
     its
    1.03
    ができます
    0.95
     moeilijk
    0.90
    芊
    0.90
     betrekking
    0.90
     mascotas
    0.86
     jakby
    0.86
    ttar
    0.86
     geboten
    0.85
    をかける
    0.85
    Activations Density 0.003%

    No Known Activations