© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 14850
    Prev
    Next
    INDEX
    Explanations

    programming code

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

    Explanation of neuron 4 behavior: the main thing this neuron does is find unusually rare or technical words—tokens that occur infrequently or are domain-specific.

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

    periods or full stops that mark the end of sentences.

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

    salient, domain-specific keywords and markers—often proper nouns, technical terms, or connective signal words—that anchor the topic or structure of a passage.

    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
    ্যাট
    0.84
     obeys
    0.77
     differentiating
    0.71
    懑
    0.70
     infection
    0.68
     डायब
    0.68
    TextViewStyle
    0.66
     inoculum
    0.66
    م
    0.66
     Microbi
    0.64
    POSITIVE LOGITS
    ectors
    0.81
    obec
    0.80
     tais
    0.79
    l
    0.77
    r
    0.74
    वालों
    0.73
     marki
    0.72
     tên
    0.70
    cite
    0.70
    ianie
    0.70
    Activations Density 0.000%

    No Known Activations