© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 425
    Prev
    Next
    INDEX
    Explanations

    URL patterns and specific strings

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

    The neuron strongly responds to isolated alphanumeric ID‐style tokens (short strings mixing letters and digits, like codes or identifiers).

    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
     when
    -1.18
     after
    -1.13
     universale
    -1.11
     before
    -1.11
     has
    -1.06
     khawatir
    -1.06
     Before
    -1.05
     diperhatikan
    -1.04
     while
    -1.02
     When
    -1.01
    POSITIVE LOGITS
     vía
    1.22
     cómod
    1.08
    ered
    1.06
    áře
    1.06
     récemment
    1.04
    ables
    1.04
     grabación
    1.04
    ogram
    1.03
    directly
    1.02
    aires
    1.02
    Activations Density 0.002%

    No Known Activations