© 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
    2. Gemma-3-27B-IT
    3. 11-GEMMASCOPE-2-TRANSCODER-262K
    4. 58829
    Prev
    Next
    INDEX
    Explanations

    This neuron seems to be activated by the token "Arch" followed by a period, especially when "Arch" is part of a proper noun or named location (like "Marble Arch"). It also shows activation for single uppercase letters like 'S' at the end of phrases, which are often followed by punctuation. The presence of emojis and varied international words suggests it might be sensitive to varied text inputs or specific stylistic choices in the data. The combination points towards recognizing specific entities or sentence endings, possibly within a broader context of historical or descriptive texts.The most compelling pattern is the occurrence of "Arch" followed by a period (or similar punctuation) and the frequent appearance of the "S" token before punctuation in different contexts within the activating texts.Therefore, a concise explanation would be:**arch. and s. punctuation**

    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_11_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
    سٹم
    0.61
    キャッシュ
    0.56
    u
    0.55
    iin
    0.52
    commutative
    0.52
    jdbcType
    0.52
    ି
    0.52
    ハンド
    0.51
    uintes
    0.51
    부는
    0.51
    POSITIVE LOGITS
     לה
    0.58
     sonar
    0.57
     המ
    0.55
    ﻅ
    0.52
     problemlos
    0.51
     propio
    0.50
     tepl
    0.50
     debuted
    0.49
     tick
    0.49
    ↵↵
    0.48
    Activations Density 0.001%

    No Known Activations