© 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. 10948
    Prev
    Next
    INDEX
    Explanations

    proper ethical conduct

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

    The neuron is tuned to detect directive or advisory language—words and verbs that convey instructions, suggestions, or calls to action.

    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
     keuze
    -1.22
     generó
    -1.16
    udahkan
    -1.14
     yet
    -1.13
    了出去
    -1.13
    却被
    -1.12
     vanske
    -1.09
    charest
    -1.08
    ПРЕ
    -1.08
    atoga
    -1.06
    POSITIVE LOGITS
     what
    1.38
     benar
    1.30
     good
    1.27
     сначала
    1.16
     better
    1.12
     behave
    1.10
     best
    1.09
     proper
    1.09
     considerare
    1.06
     healthy
    1.04
    Activations Density 0.025%

    No Known Activations