© 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. 95496
    Prev
    Next
    INDEX
    Explanations

    movie titles and specific names

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

    The neuron fires on tokens consisting mainly of uppercase letters (e.g. titles, acronyms, or all-caps 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
     szen
    -1.30
    )、
    -1.20
    eing
    -1.12
    espół
    -1.10
    𝐌
    -1.09
    」、
    -1.07
    ४
    -1.04
    usive
    -1.02
     vun
    -1.02
     tjen
    -1.02
    POSITIVE LOGITS
     thwarted
    1.21
     lessened
    1.15
     impeded
    1.14
     imparts
    1.11
     devast
    1.10
     nabízí
    1.10
     tumultuous
    1.09
     magnificent
    1.09
     strolled
    1.08
     significantly
    1.07
    Activations Density 0.103%

    No Known Activations