© 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. 10-GEMMASCOPE-RES-131K
    7. 47844
    Prev
    Next
    INDEX
    Explanations

    regex patterns and code

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

    The neuron is tuned to spotting numeric tokens—digits and related quantifier syntax (e.g. `{n,m}`, ranges like `1-9`, `\d`, etc.)—especially within regular‐expression patterns.

    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
     laboratorium
    -1.52
    ar
    -1.52
     ⓒ
    -1.48
    mense
    -1.48
     mits
    -1.48
    😰
    -1.46
    饜
    -1.45
    𝑼
    -1.45
     inox
    -1.44
     mense
    -1.44
    POSITIVE LOGITS
     tries
    1.48
     âme
    1.42
    НИМА
    1.41
     katholischen
    1.38
    zbęd
    1.37
    ()){
    1.36
     after
    1.34
    ４
    1.33
     ahora
    1.33
     esos
    1.32
    Activations Density 0.022%

    No Known Activations