© 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. 67365
    Prev
    Next
    INDEX
    Explanations

    transition to contrastive statements

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

    The neuron is sensitive to sentence boundaries, strongly activating on end‐of‐sentence punctuation (periods) and on the first word of a new sentence (often a capitalized token).

    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
    arkar
    -0.82
    嵋
    -0.80
    LoginComponent
    -0.79
    rors
    -0.79
    ifdef
    -0.78
     Dinge
    -0.78
     ET
    -0.76
    ленко
    -0.76
    пкой
    -0.75
    obox
    -0.74
    POSITIVE LOGITS
     nor
    1.45
    Instead
    1.27
    However
    1.26
    Nevertheless
    1.24
     Instead
    1.24
     Nevertheless
    1.16
     tampoco
    1.09
    むしろ
    1.09
     even
    1.05
    Nonetheless
    1.02
    Activations Density 0.065%

    No Known Activations