© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 34409
    Prev
    Next
    INDEX
    Explanations

    all or nothing conditions

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

    This neuron activates on words and phrases that express strong obligations or instructions—modal verbs and verbs of requirement (e.g. “forced,” “should,” “have to,” “must”) and related command structures.

    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
     usually
    -1.11
     обычно
    -1.03
     Usually
    -1.00
    通常
    -0.93
     Authority
    -0.88
    ]:
    -0.85
     ότι
    -0.85
     meist
    -0.85
     kolejne
    -0.84
     authority
    -0.83
    POSITIVE LOGITS
     uniquement
    1.44
     исключительно
    1.19
     모든
    1.09
     exclusivement
    1.09
     everywhere
    1.09
     everything
    1.09
     exclusively
    1.06
     every
    1.03
     solely
    1.03
     всичко
    1.02
    Activations Density 0.053%

    No Known Activations