© 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. 54052
    Prev
    Next
    INDEX
    Explanations

    conjunctions, then sentences

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

    This neuron fires most strongly on very short tokens (especially 1–4 characters long, like common words, punctuation, or dashes).

    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
     granito
    -1.13
     aldrig
    -1.10
     rivi
    -1.09
     patate
    -1.05
    Actu
    -1.05
    ウール
    -1.05
    افته
    -1.02
     marte
    -1.02
     faciliter
    -1.02
     harde
    -1.02
    POSITIVE LOGITS
     so
    1.84
     поэтому
    1.64
     even
    1.50
     nor
    1.41
    只能
    1.39
     therefore
    1.33
     will
    1.30
     Even
    1.24
     only
    1.21
     Therefore
    1.20
    Activations Density 0.134%

    No Known Activations