© 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. 22-GEMMASCOPE-RES-131K
    7. 104398
    Prev
    Next
    INDEX
    Explanations

    like a

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

    The neuron is highly responsive to content‐bearing nouns (including nouny “-ing” forms), essentially picking out single‐word objects and concepts in the text.

    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
     revamped
    -1.13
     gruesome
    -1.10
    細い
    -1.04
    也知道
    -1.03
     dando
    -1.03
     insanely
    -1.03
    がる
    -1.00
     soledad
    -0.99
    色々な
    -0.98
     innovations
    -0.98
    POSITIVE LOGITS
     or
    2.11
    علي
    1.11
     out
    1.08
    üksek
    1.04
     where
    1.03
    หรือ
    1.02
    tinh
    1.02
    ５
    1.01
    wheres
    0.98
     whose
    0.96
    Activations Density 0.025%

    No Known Activations