© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 4367
    Prev
    Next
    INDEX
    Explanations

    organizing contacts or components

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

    The neuron spotlights past‐tense verb forms (especially auxiliaries like “was,” “were,” and past participles such as “considered”).

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

    legal or formal documents, particularly terms of service, agreements, and contracts.

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

    common function words that act as grammatical glue—auxiliaries, prepositions, adverbs, and similar connectors—in formal or instructional prose.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    بال
    0.82
    Giant
    0.70
    ز
    0.68
     Alright
    0.68
     trillions
    0.68
     up
    0.66
    H
    0.65
    一来
    0.64
     rows
    0.63
    So
    0.63
    POSITIVE LOGITS
     бло
    0.98
     деп
    0.93
     следующим
    0.90
     INCIDENTAL
    0.89
     материалы
    0.89
    찻
    0.88
     photore
    0.88
     ब्ला
    0.86
     Дмитрий
    0.86
     фотографи
    0.84
    Activations Density 0.000%

    No Known Activations