© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26
    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-27B-IT
    3. 40-GEMMASCOPE-2-TRANSCODER-262K
    4. 241320
    Prev
    Next
    INDEX
    Explanations

    The pattern seems to be `chrome://` followed by an internal path or section of the Chrome browser. The `TOP_POSITIVE_LOGITS` are overwhelmingly `flags` and `flag`. This suggests the neuron might be related to Chrome's internal flags or settings pages.Let's combine these observations.- `MAX_ACTIVATING_TOKENS` has `chrome` and `:`.- `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` has `://`.- `TOP_POSITIVE_LOGITS` has `flags`.This strongly points to `chrome://flags`.I need to form a concise phrase (3-20 words)."chrome://flags" is 2 words, which is too short.The instructions say "The explanation could be a single word, or phrase, or pattern.""chrome://flags" is a specific pattern.However, the rules also say "Avoid simply listing all the tokens. Instead, try to find patterns." and "The explanation should be specific."Looking at the `TOP_ACTIVATING_TEXTS`:- `chrome://settings/system`- `chrome://extensions/`- `chrome://extensions`- `chrome://bookmarks`- `about:flags` (related concept, but not exactly `chrome://flags`)- `chrome://extensions/shortcuts`The occurrences of `chrome://` are very frequent.chrome://flags

    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_40_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Negative Logits
     responses
    0.40
     మూవీ
    0.40
     ramos
    0.39
     Truly
    0.38
     ruido
    0.38
    HTML
    0.38
    atsu
    0.38
     Responses
    0.37
     Builders
    0.37
    vamente
    0.36
    POSITIVE LOGITS
     flags
    1.13
    Flags
    1.11
     flag
    1.09
     Flags
    1.09
     Flag
    1.08
    flags
    1.06
    Flag
    1.05
    flag
    1.03
    FLAG
    0.94
     FLAG
    0.89
    Activations Density 0.001%

    No Known Activations