© 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. 12470
    Prev
    Next
    INDEX
    Explanations

    it involves

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

    The neuron consistently lights up on sentence-initial tokens (especially pronouns or demonstratives like “This,” “It,” “What,” etc.), effectively marking the beginnings of sentences.

    oai_token-act-pair · o4-miniTriggered 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
    ००
    0.98
     conceivably
    0.90
    ifiably
    0.88
     온
    0.85
    Č
    0.84
     tabloid
    0.84
     audiences
    0.83
     Importantly
    0.83
    k
    0.83
    Ở
    0.81
    POSITIVE LOGITS
     consiste
    1.02
    ния
    1.00
     ஏற்படுகிறது
    1.00
     путем
    0.97
    例えば
    0.91
    により
    0.90
    например
    0.90
     разнови
    0.90
     sayesinde
    0.90
     depends
    0.89
    Activations Density 6.030%

    No Known Activations