© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-27B-IT
    5. 40-GEMMASCOPE-2-RES-262K
    7. 1579
    Prev
    Next
    INDEX
    Explanations

    what and how questions

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

    the neuron detects prominent instructional or question-oriented phrases — e.g., headings, emphasized action items, and direct questions.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0

    the neuron detects question words and question-like prompts (tokens that introduce or mark interrogative sentences).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/resid_post/layer_40_width_262k_l0_medium
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ቱም
    0.44
     السبب
    0.41
    інки
    0.40
    之所以
    0.38
    ાસ
    0.37
     dichos
    0.36
     needn
    0.36
    maked
    0.34
    ետ
    0.34
    }.$
    0.34
    POSITIVE LOGITS
     What
    2.23
    What
    2.17
     what
    2.09
    what
    1.92
     How
    1.91
     how
    1.79
    How
    1.79
     क्या
    1.73
    どのような
    1.64
     ماذا
    1.59
    Activations Density 0.103%

    No Known Activations