© 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. 555
    Prev
    Next
    INDEX
    Explanations

    had difficulty or success

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

    The neuron is specifically detecting the token “had,” i.e. occurrences of the word “had.”

    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
     inaugura
    -1.39
     now
    -1.28
     currently
    -1.24
    UTOR
    -1.21
     quinta
    -1.15
     still
    -1.09
     Capítulo
    -1.09
     Cinéma
    -1.08
    Чтоб
    -1.08
     \
    -1.07
    POSITIVE LOGITS
     difficulty
    1.21
     problémy
    1.16
     céré
    1.14
     extrêmement
    1.10
     sådan
    1.09
     людини
    1.09
    racked
    1.09
     это
    1.07
    Luxem
    1.05
    狲
    1.05
    Activations Density 0.060%

    No Known Activations