© 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. 22-GEMMASCOPE-RES-131K
    4. 98439
    Prev
    Next
    INDEX
    Explanations

    frequency or possibility

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

    The neuron strongly activates on high-frequency function words (e.g. “the,” “is,” “and,” “because”) and punctuation, effectively marking common “stop” words and separators.

    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
    Expl
    -1.02
    ECONOM
    -1.01
    GIRLS
    -0.97
    Arach
    -0.92
    PROFESSIONAL
    -0.90
    不需要
    -0.90
     ähn
    -0.90
     ähnlich
    -0.89
     돌
    -0.89
    Dile
    -0.89
    POSITIVE LOGITS
    Sometimes
    1.36
    Often
    1.34
    Many
    1.30
     sometimes
    1.26
     often
    1.22
    Several
    1.21
     oftentimes
    1.13
     many
    1.12
    Some
    1.05
     можуть
    1.05
    Activations Density 0.068%

    No Known Activations