© 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. 104500
    Prev
    Next
    INDEX
    Explanations

    system components and states

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

    The neuron fires on domain-specific technical nouns and related object terms (e.g. “valve,” “sump,” “pipe,” “sprayed,” “mist,” “calculations,” “result”), effectively picking out core components and actions in procedural or patent-style descriptions.

    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
     especially
    -1.20
     ehemal
    -1.18
     Además
    -1.17
     cómoda
    -1.15
     además
    -1.11
     but
    -1.10
    собенно
    -1.06
     özellikle
    -1.06
     podczas
    -1.06
     both
    -1.06
    POSITIVE LOGITS
     to
    1.05
     hopefully
    1.01
     ultimately
    1.00
     whence
    1.00
     again
    1.00
     boken
    1.00
     enten
    1.00
     remainder
    0.96
    继续
    0.93
     thereafter
    0.92
    Activations Density 0.162%

    No Known Activations