© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 20663
    Prev
    Next
    INDEX
    Explanations

    aware of / read all / played down

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

    This neuron detects reporting‐style verbs and related formal statement verbs (e.g. “said,” “asked,” “read,” “assure,” “watered down,” etc.) commonly used in news‐report prose.

    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
     chré
    -2.03
     religieuses
    -1.84
     of
    -1.69
     растений
    -1.66
     گیاه
    -1.65
     catholique
    -1.65
     roślin
    -1.63
     ALSO
    -1.59
     eccl
    -1.57
    ׇ
    -1.54
    POSITIVE LOGITS
    וּ
    2.72
    ּוֹ
    2.55
    ᾿
    2.00
    ַּ
    1.97
    ִּ
    1.93
    וֹ
    1.85
     бампер
    1.84
    ֶּ
    1.80
    sữa
    1.75
    ᾽
    1.70
    Activations Density 0.142%

    No Known Activations