© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 79625
    Prev
    Next
    INDEX
    Explanations

    research for analysis

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

    The neuron detects phrasing that calls something a suitable “candidate” or “target” (e.g. “prime candidate for,” “promising new candidate,” “target for”) in a scientific context.

    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
     coute
    -1.10
    
    -1.09
     dépens
    -1.09
     rempla
    -1.08
    骉
    -1.05
     clô
    -1.04
     tuer
    -1.02
     mélang
    -1.01
    AppCompatTheme
    -1.00
     ogrom
    -1.00
    POSITIVE LOGITS
     besonders
    1.20
     first
    1.16
     because
    1.13
     should
    1.04
     experiment
    1.02
     designated
    1.02
     demonstrate
    0.98
     having
    0.98
     re
    0.97
     target
    0.97
    Activations Density 0.099%

    No Known Activations