© 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. 27458
    Prev
    Next
    INDEX
    Explanations

    bodyguard | secret service agent | firefighter | bounty hunter

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

    The neuron is effectively tracking how far into the document you are, with its activation rising the later a token appears.

    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
     its
    -1.30
     различных
    -1.13
    Américaine
    -1.11
     hivyo
    -1.07
     belki
    -1.06
     muhimu
    -1.02
     apartament
    -1.00
    וֹ
    -0.99
    Soorten
    -0.99
     depositphotos
    -0.98
    POSITIVE LOGITS
    Iné
    1.33
    Gön
    1.24
     isn
    1.23
    ׇ
    1.18
     сыром
    1.16
    upakan
    1.15
     didn
    1.15
     köt
    1.12
     combustível
    1.09
     wasn
    1.09
    Activations Density 0.047%

    No Known Activations