© 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. 13375
    Prev
    Next
    INDEX
    Explanations

    computed tomography

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

    stack traces in error logs and debugging outputs.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    The neuron primarily detects standalone numeric tokens—especially line numbers and other numbers in code error traces.

    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
     (
    -1.18
    поги
    -1.04
    </tfoot>
    -1.03
    lichte
    -1.03
    糠
    -1.01
    titch
    -1.01
     nive
    -1.00
     ngunit
    -1.00
     Livre
    -1.00
     guma
    -1.00
    POSITIVE LOGITS
    Meld
    1.15
     czemu
    1.15
    donde
    1.10
     которое
    1.06
    espero
    1.03
     dinosaurio
    1.02
    não
    1.02
     posibilidades
    1.01
    Mitar
    1.01
    這裡
    0.99
    Activations Density 0.006%

    No Known Activations