© 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. 10-GEMMASCOPE-RES-131K
    7. 58406
    Prev
    Next
    INDEX
    Explanations

    truncated, generated, action

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

    The neuron fires on technical‐documentation keywords—especially past-participle tags like “generated” and “truncated” (and related code annotations such as “action”)—marking boilerplate code comments and error‐message terms.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    ↵
    -1.95
     The
    -1.71
    猻
    -1.66
     или
    -1.58
    ꜄
    -1.57
    嘥
    -1.55
     we
    -1.47
     imitate
    -1.45
    </i>
    -1.44
     akin
    -1.43
    POSITIVE LOGITS
    0
    1.97
    ዣ
    1.76
    
    1.73
    靣
    1.71
    3
    1.71
    ILED
    1.68
    淠
    1.67
    姈
    1.66
    VERTISEMENT
    1.64
    ⽌
    1.64
    Activations Density 0.001%

    No Known Activations