© 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
    2. Gemma-3-27B-IT
    3. 15-GEMMASCOPE-2-TRANSCODER-262K
    4. 14575
    Prev
    Next
    INDEX
    Explanations

    The top positive logits are a diverse set of words in different languages, and punctuation. The MAX_ACTIVATING_TOKENS list is dominated by "er", with one instance of "ref". The TOKENS_AFTER_MAX_ACTIVATING_TOKEN shows "Extension", "/", "are", ")", "ix" following "er" and "ref".The "er" token is very frequent, and in the context of "Extension", it might related to software extensions or code-related terms. However, the subsequent tokens like "/", ":**", "are" and the diversity in TOP_POSITIVE_LOGITS make it hard to pinpoint a single coherent, specific pattern related to language or coding contexts that is strongly supported across all lists simultaneously.The `TOP_POSITIVE_LOGITS` are very varied, suggesting the neuron might be involved in detecting *something* that can manifest in many different forms

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_15_width_262k_l0_small_affine
    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
     op
    0.21
     glomer
    0.21
     hardly
    0.20
     lymphoid
    0.20
     Milner
    0.20
     fortr
    0.19
     drept
    0.19
     primacy
    0.19
     Emeritus
    0.19
    Op
    0.19
    POSITIVE LOGITS
     інші
    0.23
    !')
    0.23
    STRUCTIONS
    0.22
    !")
    0.22
    untuk
    0.21
    ді
    0.21
     cachorro
    0.21
     फू
    0.20
     עבור
    0.20
    ฟ้า
    0.20
    Activations Density 0.001%

    No Known Activations