© 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. 129774
    Prev
    Next
    INDEX
    Explanations

    The neuron preferentially lights up on longer, information‐rich content words (e.g. multi‐syllable nouns and adjectives) rather than on short or function words.

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

    non-english words

    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
    Leigh
    -0.90
     only
    -0.84
     but
    -0.80
    SysUser
    -0.78
     barely
    -0.76
    ？？？
    -0.76
    స్తు
    -0.76
     none
    -0.74
    Aunque
    -0.74
    小学生
    -0.73
    POSITIVE LOGITS
     frequentemente
    0.85
    踉
    0.81
    vements
    0.81
     melhorar
    0.79
    scn
    0.79
     商务
    0.77
     menambah
    0.76
     gillar
    0.75
     mostrarán
    0.73
     älskar
    0.73
    Activations Density 0.198%

    No Known Activations