© 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. 37-GEMMASCOPE-2-TRANSCODER-262K
    4. 126583
    Prev
    Next
    INDEX
    Explanations

    The neuron is detecting API request handlers, specifically focusing on the `req` object within server-side code. This is evident from the abundant `req` tokens appearing after potential activation points, and the presence of terms like 'Servlet', 'WebServer', and code examples using `req, res` in routes.My explanation will be: **api request handlers**This phrase fits the criteria:- Concise (3 words).- Explains what the neuron detects/predicts.- Based on patterns found in the lists (especially `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` and `TOP_ACTIVATING_TEXTS`).- Avoids forbidden starting phrases, capitalization, and mentions of "tokens".api request handlers

    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_37_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
    gift
    0.43
    鿁
    0.42
    Blo
    0.41
    }^{*}$
    0.40
    chini
    0.40
     ರೀ
    0.40
    ('\
    0.39
    ("\
    0.38
    inem
    0.38
    匤
    0.38
    POSITIVE LOGITS
     Servlet
    0.46
    ».
    0.39
    ောင်
    0.38
     बालो
    0.37
    秃
    0.37
    WebServer
    0.37
    陟
    0.36
    jetbrains
    0.36
    Servlet
    0.36
     converge
    0.36
    Activations Density 0.000%

    No Known Activations