© 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. 10-GEMMASCOPE-RES-131K
    4. 75637
    Prev
    Next
    INDEX
    Explanations

    when certain conditions apply

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

    The neuron fires on auxiliary-verb constructions (has/have/was/were, etc.) that introduce perfect- or passive-voice phrases.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    tgärder
    -1.36
     will
    -1.34
    ”.
    -1.21
     również
    -1.20
     in
    -1.20
     einige
    -1.18
     labai
    -1.13
    .”
    -1.10
    ñales
    -1.09
    人也
    -1.08
    POSITIVE LOGITS
    when
    1.31
     Sosial
    1.29
    ︓
    1.22
     ">=
    1.18
    とき
    1.18
    %
    
    1.18
    `;
    
    1.16
    ))
    
    1.16
    tiem
    1.14
     '))
    1.14
    Activations Density 0.045%

    No Known Activations