© 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. 68403
    Prev
    Next
    INDEX
    Explanations

    sequential steps

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

    The neuron is most strongly activated by past-participle verb forms (i.e. tokens ending in “-ed”).

    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
     our
    -1.23
     when
    -1.20
     helping
    -1.13
     this
    -1.09
     help
    -1.08
     using
    -1.07
     always
    -1.05
     simply
    -1.04
     love
    -1.03
     giving
    -1.03
    POSITIVE LOGITS
     then
    1.80
     Then
    1.49
     vervolgens
    1.38
    还要
    1.32
     sitten
    1.30
     затем
    1.27
     THEN
    1.24
    还需要
    1.20
    還要
    1.20
    才能
    1.18
    Activations Density 0.065%

    No Known Activations