© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 8901
    Prev
    Next
    INDEX
    Explanations

    return to original state

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

    The neuron detects phrases that describe restoring or returning something to a prior or original state (e.g. “return to pre-X,” “restore to original,” “revert to its original…”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     then
    -1.31
     to
    -1.23
    ProgressHUD
    -1.16
     apapun
    -1.15
    </tfoot>
    -1.10
    being
    -1.09
    htë
    -1.09
     upbeat
    -1.09
    須賀
    -1.07
     their
    -1.06
    POSITIVE LOGITS
     instead
    1.75
     faster
    1.32
    ↵
    1.29
     quicker
    1.23
    шает
    1.20
    Instead
    1.20
     gamla
    1.16
     nadar
    1.16
     balkon
    1.16
     cheaper
    1.14
    Activations Density 0.036%

    No Known Activations