© 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
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 47939
    Prev
    Next
    INDEX
    Explanations

    arithmetic mean, polygons, segments, photns

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

    The neuron selectively activates on technical identifiers and compound tokens—especially those with hyphens or other punctuation in them (e.g. “PH-TH,” “non-,” “two-,” file‐path fragments, header fields), marking hyphenated or punctuated jargon and code‐style names.

    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
    entown
    -0.78
    responseCode
    -0.77
     EDWIN
    -0.76
    甶
    -0.74
    主力
    -0.74
     GEORG
    -0.73
    륜
    -0.71
    ^{+}$
    -0.71
    iyle
    -0.71
    tily
    -0.70
    POSITIVE LOGITS
     poté
    0.81
    WriteTagHelper
    0.79
    Поли
    0.79
     earthen
    0.76
    ticias
    0.76
    サー
    0.75
    関
    0.73
    배
    0.73
    öss
    0.72
     Magneto
    0.71
    Activations Density 0.002%

    No Known Activations