© 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. 31-GEMMASCOPE-2-RES-16K
    7. 10647
    Prev
    Next
    INDEX
    Explanations

    The neuron is tuned to spot technical‐instruction or configuration language—things like command-line flags, script-style tokens, and other formal “how-to”/settings jargon.

    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_31_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
        
    1.16
    することができる
    1.13
    证券投资基金
    1.08
    1.08
                                   
    1.05
       
    1.04
    Abbreviations
    0.95
    0.94
    ________________
    0.93
    することにより
    0.93
    POSITIVE LOGITS
     acknow
    1.98
     administra
    1.70
     construc
    1.70
     accom
    1.64
     manuf
    1.56
     combina
    1.55
     practition
    1.54
     ermög
    1.52
     classifica
    1.52
    ナトリ
    1.49
    Activations Density 0.124%

    No Known Activations