© 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-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 68386
    Prev
    Next
    INDEX
    Explanations

    differences and comparisons

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

    The neuron detects the markers and tokens of numbered or bullet‐style rule lists (e.g. “1.”, “2.”, “All”, “MUST”) that structure a set of formal instructions.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    izmus
    -1.07
    cuadro
    -0.98
     senaste
    -0.96
    伫
    -0.95
     contiennent
    -0.95
    Flere
    -0.93
    każ
    -0.92
     $_"
    -0.92
    esség
    -0.91
     multicolore
    -0.91
    POSITIVE LOGITS
     advantages
    1.25
     differences
    1.14
     difference
    1.06
     benefits
    1.05
     still
    1.05
     their
    1.03
     sharing
    1.01
     all
    1.01
     access
    1.00
     تمام
    0.98
    Activations Density 0.064%

    No Known Activations