© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 15424
    Prev
    Next
    INDEX
    Explanations

    Combinatorial, Background, .edu, .com, ECT

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

    The neuron fires on tokens that are parts of network addresses—domain name labels, URL punctuation (dots, slashes, colons), and numeric IP‐address segments.

    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
    之前
    -1.31
     difícil
    -1.29
     skill
    -1.27
     must
    -1.25
    包括
    -1.23
     squadre
    -1.20
    「
    -1.20
     set
    -1.20
    温和
    -1.19
    mahami
    -1.17
    POSITIVE LOGITS
    んじゃ
    1.66
    っていますが
    1.63
     получают
    1.54
    嬉しいです
    1.38
    Джере
    1.31
     appartenant
    1.29
     verantwort
    1.28
    attempts
    1.27
    衬衫
    1.27
     maternidade
    1.27
    Activations Density 0.049%

    No Known Activations