© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 67418
    Prev
    Next
    INDEX
    Explanations

    documentation and technical terms

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

    The neuron selectively activates on long, content-bearing tokens—especially technical or domain-specific terms (e.g. enzyme names, file formats, class/package identifiers), as opposed to short function words.

    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
    </em>
    -1.34
    inder
    -1.15
    걍
    -1.13
    )、
    -1.06
     */
    -1.05
    ють
    -1.05
    !)
    -1.03
     où
    -1.03
    +
    -1.03
    ,:),
    -1.01
    POSITIVE LOGITS
    踯
    1.41
     vrienden
    1.40
     mijne
    1.33
     cliquer
    1.28
     både
    1.25
     搞笑
    1.23
     célè
    1.22
     siè
    1.22
     سامسون
    1.21
    鷙
    1.21
    Activations Density 0.302%

    No Known Activations