© 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. 4794
    Prev
    Next
    INDEX
    Explanations

    vision (`scopic` followed by `vision`)

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

    Explanation of neuron 4 behavior: this neuron strongly activates on unusual, low‐frequency tokens—e.g. rare acronyms, domain names, technical jargon, or otherwise out-of-the-ordinary words.

    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
     V
    0.76
     R
    0.74
    իկ
    0.74
            
    0.74
     Quiet
    0.73
     Slope
    0.73
     Earrings
    0.73
     Candidates
    0.72
    াপ্ত
    0.71
     P
    0.70
    POSITIVE LOGITS
    ي
    0.93
    б
    0.86
    मोहन
    0.85
    égalité
    0.84
    ༣
    0.83
    bbero
    0.82
    مو
    0.82
    Anast
    0.82
    ન
    0.80
    Daten
    0.80
    Activations Density 0.000%

    No Known Activations