© 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. 24-GEMMASCOPE-2-TRANSCODER-262K
    4. 179715
    Prev
    Next
    INDEX
    Explanations

    The MAX_ACTIVATING_TOKENS list shows "dot" repeating several times, and also "T".The TOKENS_AFTER_MAX_ACTIVATING_TOKEN list shows "com", "]", and "c" following these.The TOP_ACTIVATING_TEXTS show instances like "dot com", "[dot]com", "[dot]c0m", "d0T c0m", "[dot] com".The neuron seems to be focused on detecting or flagging variations of ".com" and other domain-like structures. The presence of "dot" and explicit domain extensions like "com" in MAX_ACTIVATING_TOKENS, combined with suffixes like "com" and brackets in TOKENS_AFTER_MAX_ACTIVATING_TOKEN, strongly suggests this. The TOP_ACTIVATING_TEXTS confirm this by showing URLs and references to domains, often in a "dot something" or "[dot] something" format. The TOP_POSITIVE_LOGITS like "Youtube" and "유튜브" (Korean for YouTube) could indicate that the neuron flags often associated with online platforms or web addresses.dot com and domain-like structures

    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_24_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
     Drilling
    0.42
    atória
    0.41
    ರುವುದು
    0.40
     drilling
    0.40
     Abstracts
    0.39
    𝗔
    0.39
    auxite
    0.38
     Feria
    0.38
    ienz
    0.38
     porosity
    0.37
    POSITIVE LOGITS
    Youtube
    0.43
    全
    0.42
     godt
    0.41
    真人
    0.38
    V
    0.38
    ujacy
    0.38
     유튜브
    0.37
    喵
    0.37
     meteen
    0.37
    团队
    0.37
    Activations Density 0.000%

    No Known Activations