© 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. 71274
    Prev
    Next
    INDEX
    Explanations

    usually, sometimes, often

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

    The neuron fires strongly on content-rich words (i.e. non‐stopwords) throughout running prose, effectively picking out substantive tokens and ignoring common function words.

    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
     was
    -1.84
     this
    -1.34
     today
    -1.23
     your
    -1.18
     encontré
    -1.08
     llevo
    -1.07
     همچنین
    -1.07
    遊ぶ
    -1.06
    this
    -1.05
    方は
    -1.05
    POSITIVE LOGITS
     usually
    2.72
    Usually
    2.09
     or
    2.08
     sometimes
    2.06
     either
    2.02
     often
    1.98
    usually
    1.98
     invariably
    1.96
     Usually
    1.87
     либо
    1.79
    Activations Density 0.075%

    No Known Activations