© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 73428
    Prev
    Next
    INDEX
    Explanations

    temporal context

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

    The neuron specializes in picking out discourse‐connecting words and adverbs (e.g. “but,” “so,” “when,” “because,” “quickly”) that signal shifts or relations between clauses.

    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
    ativeness
    -0.87
    itation
    -0.87
    
    -0.82
     אוי
    -0.81
    ạch
    -0.81
     lanka
    -0.81
     laboratorium
    -0.79
    styp
    -0.78
    contentType
    -0.77
    TestBed
    -0.77
    POSITIVE LOGITS
    当时
    1.88
    當時
    1.63
     during
    1.56
     podczas
    1.41
     durante
    1.38
     tijdens
    1.24
     sempat
    1.21
     originally
    1.13
     lors
    1.13
     last
    1.11
    Activations Density 0.092%

    No Known Activations