© 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. 10-GEMMASCOPE-RES-131K
    7. 130009
    Prev
    Next
    INDEX
    Explanations

    Wakanda, Quentin, bifida, golang, Florianopolis, CMV, Casablanca

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

    The neuron fires strongly on the opening sub‐token of uncommon multi‐syllable words—especially proper names and technical terms (e.g. “Wak,” “Xamarin,” “Quentin,” “Bang”).

    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
     gänzlich
    -2.86
     cũng
    -2.70
     einzigen
    -2.64
     Schlafzimmer
    -2.59
     jeweiligen
    -2.45
     zuneh
    -2.39
    薅
    -2.34
    笳
    -2.34
     Ähn
    -2.33
     vorhand
    -2.31
    POSITIVE LOGITS
    on
    3.36
    ные
    2.44
    ”，
    2.38
     était
    2.27
    was
    2.27
    ፨
    2.25
     citoy
    2.22
    ?”
    2.20
    鲀
    2.17
    .
    2.09
    Activations Density 0.028%

    No Known Activations