© 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. 34-GEMMASCOPE-RES-131K
    7. 45308
    Prev
    Next
    INDEX
    Explanations

    quantifiers after "for"

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

    The neuron strongly activates on very high‐frequency tokens—especially numerals and common function words (e.g. “the,” “for,” number strings). It is detecting highly frequent/basic tokens.

    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
     one
    -0.98
     одного
    -0.89
     одной
    -0.86
    orkin
    -0.84
     одному
    -0.81
    広場
    -0.80
    そうに
    -0.80
    赊
    -0.79
     notare
    -0.79
    Shortest
    -0.79
    POSITIVE LOGITS
     most
    2.94
    大部分
    2.63
     much
    2.38
     majority
    2.19
     معظم
    2.17
    ほとんど
    2.08
     sebagian
    2.00
     vět
    1.93
     біль
    1.77
     большин
    1.77
    Activations Density 0.070%

    No Known Activations