© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 80032
    Prev
    Next
    INDEX
    Explanations

    code syntax elements and types

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron flags programming‐language declaration tokens—things like type names (e.g. “bool,” “string,” “class”), import/preprocessor directives (e.g. “#import”), attributes (e.g. “InlineData”), and similar code‐structure keywords.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     through
    -1.38
     their
    -1.34
     over
    -1.12
     sexto
    -1.11
     epitom
    -1.02
    ﻷ
    -1.02
     part
    -0.99
     first
    -0.99
     focused
    -0.98
     allows
    -0.97
    POSITIVE LOGITS
     frequentemente
    1.09
     namentlich
    1.06
     ジェ
    1.06
    envoyer
    1.05
     erwähnt
    1.05
     praticamente
    1.00
    viating
    0.99
    abetes
    0.98
     kanskje
    0.97
    Germania
    0.97
    Activations Density 0.000%

    No Known Activations