© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 128446
    Prev
    Next
    INDEX
    Explanations

    This neuron detects structured metadata tokens—especially numeric identifiers and punctuation in file paths, document headers, and XML/HTML-style tags.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    Summer Heat, Selective decrease, The following, package com

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     of
    -1.14
     your
    -1.13
    瀝
    -1.13
     remarqu
    -1.09
     neque
    -1.07
    為に
    -1.07
     erila
    -1.04
    çou
    -1.03
     ي
    -1.02
     fondament
    -1.02
    POSITIVE LOGITS
    〬
    1.45
    Ľ
    1.27
     داشتند
    1.22
    llac
    1.17
     other
    1.13
    न्दर्भ
    1.13
    encounter
    1.13
     dones
    1.12
     Berikut
    1.11
     piacevole
    1.11
    Activations Density 0.024%

    No Known Activations