© 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. 22-GEMMASCOPE-RES-131K
    7. 14822
    Prev
    Next
    INDEX
    Explanations

    explaining results, problems, or outcomes

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

    The neuron detects metadiscourse verbs and phrases where the author signals analysis or intent (e.g. “attempts to understand,” “explain,” “found,” “knows”).

    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
     aidant
    -1.30
     amélior
    -1.26
     régle
    -1.20
     étend
    -1.17
     deserto
    -1.14
     associés
    -1.12
     compét
    -1.12
     самы
    -1.12
     conséquence
    -1.11
     manuale
    -1.10
    POSITIVE LOGITS
     также
    1.38
    AuthContext
    1.26
    Reklama
    1.26
    BitConverter
    1.21
    šak
    1.20
     höl
    1.17
    InputBorder
    1.15
    warzys
    1.13
    となり
    1.11
     advertencia
    1.11
    Activations Density 0.013%

    No Known Activations