© 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. 452
    Prev
    Next
    INDEX
    Explanations

    so

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

    The neuron fires strongly on words that begin new sentences—especially sentence-initial discourse markers or adverbs (e.g. “Consequently,” “So,” “However,” “If,” “When,” etc.).

    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
    畢
    0.29
    (),
    0.28
            
    0.27
     चूंकि
    0.26
    由於
    0.26
    0
    0.25
    その他
    0.24
    สาว
    0.24
     기타
    0.24
    銨
    0.24
    POSITIVE LOGITS
     therefor
    0.41
     wouldn
    0.41
     deshalb
    0.40
    oooo
    0.39
     derfor
    0.39
     therefore
    0.37
     shouldn
    0.36
     daher
    0.36
     Daher
    0.36
     deswegen
    0.36
    Activations Density 0.480%

    No Known Activations