© 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. 22-GEMMASCOPE-RES-131K
    7. 64751
    Prev
    Next
    INDEX
    Explanations

    reporting speech and facts

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

    This neuron fires strongly on content-bearing tokens—nouns, verbs, numbers, and other semantically important words—while ignoring function words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     just
    -1.42
     kemudian
    -1.30
     really
    -1.28
    حول
    -1.24
     both
    -1.23
     Particularly
    -1.17
     Perhaps
    -1.16
     Though
    -1.13
     потім
    -1.13
     extraordinarily
    -1.11
    POSITIVE LOGITS
     мою
    1.24
    塹
    1.19
    ​​​
    1.16
    ఽ
    1.15
     unangemessene
    1.14
    etary
    1.13
    GIVEREF
    1.13
    toon
    1.13
     мои
    1.13
     captivating
    1.13
    Activations Density 0.007%

    No Known Activations