© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 2443
    Prev
    Next
    INDEX
    Explanations

    but rather

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

    The neuron is primarily detecting negation and causal-connective tokens (e.g. "not," "because," "for").

    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
     both
    0.74
     +
    0.68
    Both
    0.66
    both
    0.66
     sowohl
    0.66
     :
    0.64
     $+
    0.63
     BOTH
    0.63
     både
    0.60
     +(
    0.60
    POSITIVE LOGITS
     necessariamente
    1.20
     necessarily
    1.16
    necessarily
    1.06
     necesariamente
    1.00
     nécessairement
    0.98
     forcément
    0.89
     sondern
    0.85
    而是
    0.85
     ただ
    0.84
     siquiera
    0.81
    Activations Density 0.425%

    No Known Activations