© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 75027
    Prev
    Next
    INDEX
    Explanations

    help to serve

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

    The neuron fires on verbs that describe an author’s or user’s actions—especially in technical or problem‐solving contexts (e.g. “inspect,” “see,” “set up,” “research,” “written”).

    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
    或者是
    -1.40
    ᄑ
    -1.19
     they
    -1.16
     these
    -1.16
    ∷
    -1.15
    Kamu
    -1.12
    𝛽
    -1.08
    金
    -1.07
    ńcu
    -1.07
     possuem
    -1.07
    POSITIVE LOGITS
    struktion
    1.32
     länder
    1.27
     EXPERIMENTS
    1.25
     its
    1.23
    钡
    1.23
     semoga
    1.23
     млн
    1.23
     menyiapkan
    1.21
     memperoleh
    1.20
    RTLD
    1.20
    Activations Density 0.100%

    No Known Activations