© 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. 71463
    Prev
    Next
    INDEX
    Explanations

    person's birth or past role

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

    The neuron strongly activates on numeric date and year tokens (and the surrounding “was” linking verb) as they appear in biographical or birth‐date statements.

    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
     time
    -1.05
     will
    -1.04
     parco
    -0.98
     przeds
    -0.97
     werde
    -0.96
    過ごし
    -0.96
    内容は
    -0.94
     فيه
    -0.93
    Lengths
    -0.93
     volverá
    -0.92
    POSITIVE LOGITS
    étit
    1.16
    ierter
    1.15
     who
    1.15
    sprzedam
    1.09
    URLException
    1.09
     thoſe
    1.07
     whoſe
    1.07
    }):
    1.06
     theſe
    1.06
    Ꮴ
    1.05
    Activations Density 0.167%

    No Known Activations