© 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. 9742
    Prev
    Next
    INDEX
    Explanations

    was the

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

    This neuron detects factual news‐style reporting cues—especially past‐tense auxiliaries (“was,” “viewed,” “were”), possessives (“their”), ordinals (“first”), and place/time markers typical of third-person event summaries.

    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
     acestei
    -1.16
    ☆☆☆☆☆
    -1.09
     социально
    -1.07
    ≔
    -1.06
     acestui
    -1.05
     thofe
    -1.03
     esat
    -1.01
     què
    -0.98
    volle
    -0.97
    這個
    -0.96
    POSITIVE LOGITS
     was
    1.26
     one
    1.23
     a
    1.14
    他才
    1.10
     todėl
    1.06
     an
    1.05
     only
    1.05
     another
    1.05
     merely
    1.03
     einem
    1.00
    Activations Density 0.044%

    No Known Activations