© 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. 67501
    Prev
    Next
    INDEX
    Explanations

    here followed by location

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

    The neuron activates most strongly on temporal and locational cues (e.g. “here,” “on Wednesday,” “said on Saturday,” “In the two months since”), i.e. phrases that situate a news report in time and place.

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

    location and time references in news articles, particularly place names followed by datelines or temporal markers.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    news-style datelines and reporting cues that indicate the place, time/date, and wire agency of a news report.

    oai_token-act-pair · gpt-5Triggered 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
    異なる
    -0.90
    ඞ
    -0.88
     interesantes
    -0.88
    惮
    -0.87
    他现在
    -0.86
     **/
    -0.85
    opheles
    -0.84
     aqueles
    -0.84
    poxy
    -0.82
     voulait
    -0.82
    POSITIVE LOGITS
     here
    2.98
     здесь
    1.85
    这里
    1.66
    here
    1.60
    Here
    1.57
     اینجا
    1.54
     aquí
    1.52
    這裡
    1.52
     disini
    1.52
    这里的
    1.48
    Activations Density 0.087%

    No Known Activations