© 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. 130492
    Prev
    Next
    INDEX
    Explanations

    This neuron detects words and phrases that signal an exception or obstacle holding something up (e.g. “except,” “hang-up,” “holding up”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    identifying specific issues or confusion

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     arī
    -1.27
     other
    -1.09
     another
    -1.04
    impôt
    -1.04
    這樣
    -1.00
     also
    -1.00
    lüğü
    -0.99
     দে
    -0.98
     bună
    -0.98
    positorio
    -0.98
    POSITIVE LOGITS
     the
    1.22
     lack
    1.13
     how
    1.12
     zakresie
    1.08
     aspect
    1.07
     handling
    1.06
     poziomie
    1.06
    在于
    1.05
    蓁
    1.04
    handling
    0.99
    Activations Density 0.088%

    No Known Activations