© 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. 34-GEMMASCOPE-RES-131K
    4. 98789
    Prev
    Next
    INDEX
    Explanations

    on- followed by a word

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

    This neuron detects the token “On” (or “on”) when it appears as a standalone prefix—especially in hyphenated or compound terms (e.g. “On-site,” “On-demand,” “On-line,” etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    お客
    -0.86
    を
    -0.84
     Surprisingly
    -0.84
     definitely
    -0.83
     highlighting
    -0.81
     brilliant
    -0.81
     possibly
    -0.81
     blatantly
    -0.80
    最高の
    -0.80
    基準
    -0.80
    POSITIVE LOGITS
    Demand
    1.14
     demand
    1.05
    wers
    1.03
    wner
    0.99
    ions
    0.97
     Demand
    0.96
     behalf
    0.94
    GOING
    0.94
     AVERAGE
    0.94
     average
    0.93
    Activations Density 0.043%

    No Known Activations