© 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. 28274
    Prev
    Next
    INDEX
    Explanations

    used for processing or storing

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

    The neuron is keyed to multiword verb + preposition/particle constructions (phrasal verbs and similar verb-phrase patterns like “stand up to,” “slice through,” “fit to,” “for you to,” etc.).

    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
     it
    -1.53
     by
    -1.20
     when
    -1.09
     if
    -1.09
     there
    -1.07
     or
    -1.05
     in
    -1.03
     If
    -1.01
    helf
    -0.96
     because
    -0.95
    POSITIVE LOGITS
     your
    1.41
    ทั้ง
    1.12
     cả
    1.10
     qualsiasi
    1.10
     zarówno
    1.06
     busy
    1.05
    いろんな
    1.05
    也将
    1.04
     nawet
    1.03
    思わず
    1.02
    Activations Density 0.043%

    No Known Activations