© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 27150
    Prev
    Next
    INDEX
    Explanations

    numbers between 100 and 200

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

    The neuron consistently lights up on specialized or technical content words and proper names (e.g. “Naruto,” “Excel,” “R,” “representing,” “multiplication,” “Nancy O’s”), so in effect it’s detecting domain-specific or named‐entity tokens.

    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
     thoſe
    -1.09
     recensement
    -1.09
    uguetes
    -1.09
    ကိုးကား
    -1.07
    ӝ
    -1.06
     MÉXICO
    -1.05
    FormTagHelper
    -1.05
     потім
    -1.02
     communiqué
    -1.02
    Chanel
    -1.01
    POSITIVE LOGITS
     would
    1.14
     in
    1.08
     as
    1.05
     example
    1.04
    4
    1.02
    1
    1.00
     may
    0.98
    man
    0.97
     but
    0.96
     given
    0.96
    Activations Density 0.008%

    No Known Activations