© 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. 13897
    Prev
    Next
    INDEX
    Explanations

    specific subjects and their related terms

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

    The neuron is a detector for compound‐term constructions—especially words joined by hyphens or “X by Y” patterns (e.g. audio-books, Unix-like, addition by a constant, multiplication by a constant).

    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
    UDES
    -0.89
    搅
    -0.89
    íqu
    -0.88
    殡
    -0.88
     объясни
    -0.85
    嚐
    -0.82
    HXLINE
    -0.82
    SourceChecksum
    -0.81
    那时候
    -0.81
     tato
    -0.81
    POSITIVE LOGITS
     dáms
    1.04
     samostat
    1.00
     második
    0.97
     umě
    0.94
     empieza
    0.93
     incluye
    0.92
     môže
    0.90
     poprzez
    0.90
    ついに
    0.90
     menggambarkan
    0.90
    Activations Density 0.045%

    No Known Activations