© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 34577
    Prev
    Next
    INDEX
    Explanations

    looks like / sounded like

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

    The neuron is selectively activating on contentful, typically less common words—especially nouns, adjectives, and proper names—rather than on 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
     consideradas
    -0.87
     nacidos
    -0.82
     tuer
    -0.82
     polega
    -0.81
    glers
    -0.81
    穎
    -0.81
     У
    -0.81
     his
    -0.80
    優れた
    -0.80
     Sardegna
    -0.80
    POSITIVE LOGITS
     when
    1.04
     before
    1.02
     havent
    0.98
     might
    0.98
     could
    0.96
     whenever
    0.96
     wouldnt
    0.94
     végét
    0.91
     περισσότε
    0.89
     everytime
    0.88
    Activations Density 0.029%

    No Known Activations